xref: /openbsd-src/gnu/usr.bin/perl/toke.c (revision 50b7afb2c2c0993b0894d4e34bf857cb13ed9c80)
1 /*    toke.c
2  *
3  *    Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4  *    2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others
5  *
6  *    You may distribute under the terms of either the GNU General Public
7  *    License or the Artistic License, as specified in the README file.
8  *
9  */
10 
11 /*
12  *  'It all comes from here, the stench and the peril.'    --Frodo
13  *
14  *     [p.719 of _The Lord of the Rings_, IV/ix: "Shelob's Lair"]
15  */
16 
17 /*
18  * This file is the lexer for Perl.  It's closely linked to the
19  * parser, perly.y.
20  *
21  * The main routine is yylex(), which returns the next token.
22  */
23 
24 /*
25 =head1 Lexer interface
26 
27 This is the lower layer of the Perl parser, managing characters and tokens.
28 
29 =for apidoc AmU|yy_parser *|PL_parser
30 
31 Pointer to a structure encapsulating the state of the parsing operation
32 currently in progress.  The pointer can be locally changed to perform
33 a nested parse without interfering with the state of an outer parse.
34 Individual members of C<PL_parser> have their own documentation.
35 
36 =cut
37 */
38 
39 #include "EXTERN.h"
40 #define PERL_IN_TOKE_C
41 #include "perl.h"
42 #include "dquote_static.c"
43 
44 #define new_constant(a,b,c,d,e,f,g)	\
45 	S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
46 
47 #define pl_yylval	(PL_parser->yylval)
48 
49 /* XXX temporary backwards compatibility */
50 #define PL_lex_brackets		(PL_parser->lex_brackets)
51 #define PL_lex_allbrackets	(PL_parser->lex_allbrackets)
52 #define PL_lex_fakeeof		(PL_parser->lex_fakeeof)
53 #define PL_lex_brackstack	(PL_parser->lex_brackstack)
54 #define PL_lex_casemods		(PL_parser->lex_casemods)
55 #define PL_lex_casestack        (PL_parser->lex_casestack)
56 #define PL_lex_defer		(PL_parser->lex_defer)
57 #define PL_lex_dojoin		(PL_parser->lex_dojoin)
58 #define PL_lex_expect		(PL_parser->lex_expect)
59 #define PL_lex_formbrack        (PL_parser->lex_formbrack)
60 #define PL_lex_inpat		(PL_parser->lex_inpat)
61 #define PL_lex_inwhat		(PL_parser->lex_inwhat)
62 #define PL_lex_op		(PL_parser->lex_op)
63 #define PL_lex_repl		(PL_parser->lex_repl)
64 #define PL_lex_starts		(PL_parser->lex_starts)
65 #define PL_lex_stuff		(PL_parser->lex_stuff)
66 #define PL_multi_start		(PL_parser->multi_start)
67 #define PL_multi_open		(PL_parser->multi_open)
68 #define PL_multi_close		(PL_parser->multi_close)
69 #define PL_preambled		(PL_parser->preambled)
70 #define PL_sublex_info		(PL_parser->sublex_info)
71 #define PL_linestr		(PL_parser->linestr)
72 #define PL_expect		(PL_parser->expect)
73 #define PL_copline		(PL_parser->copline)
74 #define PL_bufptr		(PL_parser->bufptr)
75 #define PL_oldbufptr		(PL_parser->oldbufptr)
76 #define PL_oldoldbufptr		(PL_parser->oldoldbufptr)
77 #define PL_linestart		(PL_parser->linestart)
78 #define PL_bufend		(PL_parser->bufend)
79 #define PL_last_uni		(PL_parser->last_uni)
80 #define PL_last_lop		(PL_parser->last_lop)
81 #define PL_last_lop_op		(PL_parser->last_lop_op)
82 #define PL_lex_state		(PL_parser->lex_state)
83 #define PL_rsfp			(PL_parser->rsfp)
84 #define PL_rsfp_filters		(PL_parser->rsfp_filters)
85 #define PL_in_my		(PL_parser->in_my)
86 #define PL_in_my_stash		(PL_parser->in_my_stash)
87 #define PL_tokenbuf		(PL_parser->tokenbuf)
88 #define PL_multi_end		(PL_parser->multi_end)
89 #define PL_error_count		(PL_parser->error_count)
90 
91 #ifdef PERL_MAD
92 #  define PL_endwhite		(PL_parser->endwhite)
93 #  define PL_faketokens		(PL_parser->faketokens)
94 #  define PL_lasttoke		(PL_parser->lasttoke)
95 #  define PL_nextwhite		(PL_parser->nextwhite)
96 #  define PL_realtokenstart	(PL_parser->realtokenstart)
97 #  define PL_skipwhite		(PL_parser->skipwhite)
98 #  define PL_thisclose		(PL_parser->thisclose)
99 #  define PL_thismad		(PL_parser->thismad)
100 #  define PL_thisopen		(PL_parser->thisopen)
101 #  define PL_thisstuff		(PL_parser->thisstuff)
102 #  define PL_thistoken		(PL_parser->thistoken)
103 #  define PL_thiswhite		(PL_parser->thiswhite)
104 #  define PL_thiswhite		(PL_parser->thiswhite)
105 #  define PL_nexttoke		(PL_parser->nexttoke)
106 #  define PL_curforce		(PL_parser->curforce)
107 #else
108 #  define PL_nexttoke		(PL_parser->nexttoke)
109 #  define PL_nexttype		(PL_parser->nexttype)
110 #  define PL_nextval		(PL_parser->nextval)
111 #endif
112 
113 static const char* const ident_too_long = "Identifier too long";
114 
115 #ifdef PERL_MAD
116 #  define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
117 #  define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
118 #else
119 #  define CURMAD(slot,sv)
120 #  define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
121 #endif
122 
123 #define XENUMMASK  0x3f
124 #define XFAKEEOF   0x40
125 #define XFAKEBRACK 0x80
126 
127 #ifdef USE_UTF8_SCRIPTS
128 #   define UTF (!IN_BYTES)
129 #else
130 #   define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || ( !(PL_parser->lex_flags & LEX_IGNORE_UTF8_HINTS) && (PL_hints & HINT_UTF8)))
131 #endif
132 
133 /* The maximum number of characters preceding the unrecognized one to display */
134 #define UNRECOGNIZED_PRECEDE_COUNT 10
135 
136 /* In variables named $^X, these are the legal values for X.
137  * 1999-02-27 mjd-perl-patch@plover.com */
138 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
139 
140 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
141 
142 /* LEX_* are values for PL_lex_state, the state of the lexer.
143  * They are arranged oddly so that the guard on the switch statement
144  * can get by with a single comparison (if the compiler is smart enough).
145  *
146  * These values refer to the various states within a sublex parse,
147  * i.e. within a double quotish string
148  */
149 
150 /* #define LEX_NOTPARSING		11 is done in perl.h. */
151 
152 #define LEX_NORMAL		10 /* normal code (ie not within "...")     */
153 #define LEX_INTERPNORMAL	 9 /* code within a string, eg "$foo[$x+1]" */
154 #define LEX_INTERPCASEMOD	 8 /* expecting a \U, \Q or \E etc          */
155 #define LEX_INTERPPUSH		 7 /* starting a new sublex parse level     */
156 #define LEX_INTERPSTART		 6 /* expecting the start of a $var         */
157 
158 				   /* at end of code, eg "$x" followed by:  */
159 #define LEX_INTERPEND		 5 /* ... eg not one of [, { or ->          */
160 #define LEX_INTERPENDMAYBE	 4 /* ... eg one of [, { or ->              */
161 
162 #define LEX_INTERPCONCAT	 3 /* expecting anything, eg at start of
163 				        string or after \E, $foo, etc       */
164 #define LEX_INTERPCONST		 2 /* NOT USED */
165 #define LEX_FORMLINE		 1 /* expecting a format line               */
166 #define LEX_KNOWNEXT		 0 /* next token known; just return it      */
167 
168 
169 #ifdef DEBUGGING
170 static const char* const lex_state_names[] = {
171     "KNOWNEXT",
172     "FORMLINE",
173     "INTERPCONST",
174     "INTERPCONCAT",
175     "INTERPENDMAYBE",
176     "INTERPEND",
177     "INTERPSTART",
178     "INTERPPUSH",
179     "INTERPCASEMOD",
180     "INTERPNORMAL",
181     "NORMAL"
182 };
183 #endif
184 
185 #ifdef ff_next
186 #undef ff_next
187 #endif
188 
189 #include "keywords.h"
190 
191 /* CLINE is a macro that ensures PL_copline has a sane value */
192 
193 #ifdef CLINE
194 #undef CLINE
195 #endif
196 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
197 
198 #ifdef PERL_MAD
199 #  define SKIPSPACE0(s) skipspace0(s)
200 #  define SKIPSPACE1(s) skipspace1(s)
201 #  define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
202 #  define PEEKSPACE(s) skipspace2(s,0)
203 #else
204 #  define SKIPSPACE0(s) skipspace(s)
205 #  define SKIPSPACE1(s) skipspace(s)
206 #  define SKIPSPACE2(s,tsv) skipspace(s)
207 #  define PEEKSPACE(s) skipspace(s)
208 #endif
209 
210 /*
211  * Convenience functions to return different tokens and prime the
212  * lexer for the next token.  They all take an argument.
213  *
214  * TOKEN        : generic token (used for '(', DOLSHARP, etc)
215  * OPERATOR     : generic operator
216  * AOPERATOR    : assignment operator
217  * PREBLOCK     : beginning the block after an if, while, foreach, ...
218  * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
219  * PREREF       : *EXPR where EXPR is not a simple identifier
220  * TERM         : expression term
221  * LOOPX        : loop exiting command (goto, last, dump, etc)
222  * FTST         : file test operator
223  * FUN0         : zero-argument function
224  * FUN0OP       : zero-argument function, with its op created in this file
225  * FUN1         : not used, except for not, which isn't a UNIOP
226  * BOop         : bitwise or or xor
227  * BAop         : bitwise and
228  * SHop         : shift operator
229  * PWop         : power operator
230  * PMop         : pattern-matching operator
231  * Aop          : addition-level operator
232  * Mop          : multiplication-level operator
233  * Eop          : equality-testing operator
234  * Rop          : relational operator <= != gt
235  *
236  * Also see LOP and lop() below.
237  */
238 
239 #ifdef DEBUGGING /* Serve -DT. */
240 #   define REPORT(retval) tokereport((I32)retval, &pl_yylval)
241 #else
242 #   define REPORT(retval) (retval)
243 #endif
244 
245 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
246 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
247 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
248 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
249 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
250 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
251 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
252 #define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
253 #define FTST(f)  return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
254 #define FUN0(f)  return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
255 #define FUN0OP(f)  return (pl_yylval.opval=f, CLINE, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0OP))
256 #define FUN1(f)  return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
257 #define BOop(f)  return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
258 #define BAop(f)  return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
259 #define SHop(f)  return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
260 #define PWop(f)  return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
261 #define PMop(f)  return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
262 #define Aop(f)   return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
263 #define Mop(f)   return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
264 #define Eop(f)   return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
265 #define Rop(f)   return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
266 
267 /* This bit of chicanery makes a unary function followed by
268  * a parenthesis into a function with one argument, highest precedence.
269  * The UNIDOR macro is for unary functions that can be followed by the //
270  * operator (such as C<shift // 0>).
271  */
272 #define UNI3(f,x,have_x) { \
273 	pl_yylval.ival = f; \
274 	if (have_x) PL_expect = x; \
275 	PL_bufptr = s; \
276 	PL_last_uni = PL_oldbufptr; \
277 	PL_last_lop_op = f; \
278 	if (*s == '(') \
279 	    return REPORT( (int)FUNC1 ); \
280 	s = PEEKSPACE(s); \
281 	return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
282 	}
283 #define UNI(f)    UNI3(f,XTERM,1)
284 #define UNIDOR(f) UNI3(f,XTERMORDORDOR,1)
285 #define UNIPROTO(f,optional) { \
286 	if (optional) PL_last_uni = PL_oldbufptr; \
287 	OPERATOR(f); \
288 	}
289 
290 #define UNIBRACK(f) UNI3(f,0,0)
291 
292 /* grandfather return to old style */
293 #define OLDLOP(f) \
294 	do { \
295 	    if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC) \
296 		PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC; \
297 	    pl_yylval.ival = (f); \
298 	    PL_expect = XTERM; \
299 	    PL_bufptr = s; \
300 	    return (int)LSTOP; \
301 	} while(0)
302 
303 #define COPLINE_INC_WITH_HERELINES		    \
304     STMT_START {				     \
305 	CopLINE_inc(PL_curcop);			      \
306 	if (PL_parser->lex_shared->herelines)	       \
307 	    CopLINE(PL_curcop) += PL_parser->lex_shared->herelines, \
308 	    PL_parser->lex_shared->herelines = 0;		     \
309     } STMT_END
310 
311 
312 #ifdef DEBUGGING
313 
314 /* how to interpret the pl_yylval associated with the token */
315 enum token_type {
316     TOKENTYPE_NONE,
317     TOKENTYPE_IVAL,
318     TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
319     TOKENTYPE_PVAL,
320     TOKENTYPE_OPVAL
321 };
322 
323 static struct debug_tokens {
324     const int token;
325     enum token_type type;
326     const char *name;
327 } const debug_tokens[] =
328 {
329     { ADDOP,		TOKENTYPE_OPNUM,	"ADDOP" },
330     { ANDAND,		TOKENTYPE_NONE,		"ANDAND" },
331     { ANDOP,		TOKENTYPE_NONE,		"ANDOP" },
332     { ANONSUB,		TOKENTYPE_IVAL,		"ANONSUB" },
333     { ARROW,		TOKENTYPE_NONE,		"ARROW" },
334     { ASSIGNOP,		TOKENTYPE_OPNUM,	"ASSIGNOP" },
335     { BITANDOP,		TOKENTYPE_OPNUM,	"BITANDOP" },
336     { BITOROP,		TOKENTYPE_OPNUM,	"BITOROP" },
337     { COLONATTR,	TOKENTYPE_NONE,		"COLONATTR" },
338     { CONTINUE,		TOKENTYPE_NONE,		"CONTINUE" },
339     { DEFAULT,		TOKENTYPE_NONE,		"DEFAULT" },
340     { DO,		TOKENTYPE_NONE,		"DO" },
341     { DOLSHARP,		TOKENTYPE_NONE,		"DOLSHARP" },
342     { DORDOR,		TOKENTYPE_NONE,		"DORDOR" },
343     { DOROP,		TOKENTYPE_OPNUM,	"DOROP" },
344     { DOTDOT,		TOKENTYPE_IVAL,		"DOTDOT" },
345     { ELSE,		TOKENTYPE_NONE,		"ELSE" },
346     { ELSIF,		TOKENTYPE_IVAL,		"ELSIF" },
347     { EQOP,		TOKENTYPE_OPNUM,	"EQOP" },
348     { FOR,		TOKENTYPE_IVAL,		"FOR" },
349     { FORMAT,		TOKENTYPE_NONE,		"FORMAT" },
350     { FORMLBRACK,	TOKENTYPE_NONE,		"FORMLBRACK" },
351     { FORMRBRACK,	TOKENTYPE_NONE,		"FORMRBRACK" },
352     { FUNC,		TOKENTYPE_OPNUM,	"FUNC" },
353     { FUNC0,		TOKENTYPE_OPNUM,	"FUNC0" },
354     { FUNC0OP,		TOKENTYPE_OPVAL,	"FUNC0OP" },
355     { FUNC0SUB,		TOKENTYPE_OPVAL,	"FUNC0SUB" },
356     { FUNC1,		TOKENTYPE_OPNUM,	"FUNC1" },
357     { FUNCMETH,		TOKENTYPE_OPVAL,	"FUNCMETH" },
358     { GIVEN,		TOKENTYPE_IVAL,		"GIVEN" },
359     { HASHBRACK,	TOKENTYPE_NONE,		"HASHBRACK" },
360     { IF,		TOKENTYPE_IVAL,		"IF" },
361     { LABEL,		TOKENTYPE_PVAL,		"LABEL" },
362     { LOCAL,		TOKENTYPE_IVAL,		"LOCAL" },
363     { LOOPEX,		TOKENTYPE_OPNUM,	"LOOPEX" },
364     { LSTOP,		TOKENTYPE_OPNUM,	"LSTOP" },
365     { LSTOPSUB,		TOKENTYPE_OPVAL,	"LSTOPSUB" },
366     { MATCHOP,		TOKENTYPE_OPNUM,	"MATCHOP" },
367     { METHOD,		TOKENTYPE_OPVAL,	"METHOD" },
368     { MULOP,		TOKENTYPE_OPNUM,	"MULOP" },
369     { MY,		TOKENTYPE_IVAL,		"MY" },
370     { NOAMP,		TOKENTYPE_NONE,		"NOAMP" },
371     { NOTOP,		TOKENTYPE_NONE,		"NOTOP" },
372     { OROP,		TOKENTYPE_IVAL,		"OROP" },
373     { OROR,		TOKENTYPE_NONE,		"OROR" },
374     { PACKAGE,		TOKENTYPE_NONE,		"PACKAGE" },
375     { PEG,		TOKENTYPE_NONE,		"PEG" },
376     { PLUGEXPR,		TOKENTYPE_OPVAL,	"PLUGEXPR" },
377     { PLUGSTMT,		TOKENTYPE_OPVAL,	"PLUGSTMT" },
378     { PMFUNC,		TOKENTYPE_OPVAL,	"PMFUNC" },
379     { POSTDEC,		TOKENTYPE_NONE,		"POSTDEC" },
380     { POSTINC,		TOKENTYPE_NONE,		"POSTINC" },
381     { POWOP,		TOKENTYPE_OPNUM,	"POWOP" },
382     { PREDEC,		TOKENTYPE_NONE,		"PREDEC" },
383     { PREINC,		TOKENTYPE_NONE,		"PREINC" },
384     { PRIVATEREF,	TOKENTYPE_OPVAL,	"PRIVATEREF" },
385     { QWLIST,		TOKENTYPE_OPVAL,	"QWLIST" },
386     { REFGEN,		TOKENTYPE_NONE,		"REFGEN" },
387     { RELOP,		TOKENTYPE_OPNUM,	"RELOP" },
388     { REQUIRE,		TOKENTYPE_NONE,		"REQUIRE" },
389     { SHIFTOP,		TOKENTYPE_OPNUM,	"SHIFTOP" },
390     { SUB,		TOKENTYPE_NONE,		"SUB" },
391     { THING,		TOKENTYPE_OPVAL,	"THING" },
392     { UMINUS,		TOKENTYPE_NONE,		"UMINUS" },
393     { UNIOP,		TOKENTYPE_OPNUM,	"UNIOP" },
394     { UNIOPSUB,		TOKENTYPE_OPVAL,	"UNIOPSUB" },
395     { UNLESS,		TOKENTYPE_IVAL,		"UNLESS" },
396     { UNTIL,		TOKENTYPE_IVAL,		"UNTIL" },
397     { USE,		TOKENTYPE_IVAL,		"USE" },
398     { WHEN,		TOKENTYPE_IVAL,		"WHEN" },
399     { WHILE,		TOKENTYPE_IVAL,		"WHILE" },
400     { WORD,		TOKENTYPE_OPVAL,	"WORD" },
401     { YADAYADA,		TOKENTYPE_IVAL,		"YADAYADA" },
402     { 0,		TOKENTYPE_NONE,		NULL }
403 };
404 
405 /* dump the returned token in rv, plus any optional arg in pl_yylval */
406 
407 STATIC int
408 S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
409 {
410     dVAR;
411 
412     PERL_ARGS_ASSERT_TOKEREPORT;
413 
414     if (DEBUG_T_TEST) {
415 	const char *name = NULL;
416 	enum token_type type = TOKENTYPE_NONE;
417 	const struct debug_tokens *p;
418 	SV* const report = newSVpvs("<== ");
419 
420 	for (p = debug_tokens; p->token; p++) {
421 	    if (p->token == (int)rv) {
422 		name = p->name;
423 		type = p->type;
424 		break;
425 	    }
426 	}
427 	if (name)
428 	    Perl_sv_catpv(aTHX_ report, name);
429 	else if ((char)rv > ' ' && (char)rv <= '~')
430 	    Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
431 	else if (!rv)
432 	    sv_catpvs(report, "EOF");
433 	else
434 	    Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
435 	switch (type) {
436 	case TOKENTYPE_NONE:
437 	    break;
438 	case TOKENTYPE_IVAL:
439 	    Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
440 	    break;
441 	case TOKENTYPE_OPNUM:
442 	    Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
443 				    PL_op_name[lvalp->ival]);
444 	    break;
445 	case TOKENTYPE_PVAL:
446 	    Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
447 	    break;
448 	case TOKENTYPE_OPVAL:
449 	    if (lvalp->opval) {
450 		Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
451 				    PL_op_name[lvalp->opval->op_type]);
452 		if (lvalp->opval->op_type == OP_CONST) {
453 		    Perl_sv_catpvf(aTHX_ report, " %s",
454 			SvPEEK(cSVOPx_sv(lvalp->opval)));
455 		}
456 
457 	    }
458 	    else
459 		sv_catpvs(report, "(opval=null)");
460 	    break;
461 	}
462         PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
463     };
464     return (int)rv;
465 }
466 
467 
468 /* print the buffer with suitable escapes */
469 
470 STATIC void
471 S_printbuf(pTHX_ const char *const fmt, const char *const s)
472 {
473     SV* const tmp = newSVpvs("");
474 
475     PERL_ARGS_ASSERT_PRINTBUF;
476 
477     PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
478     SvREFCNT_dec(tmp);
479 }
480 
481 #endif
482 
483 static int
484 S_deprecate_commaless_var_list(pTHX) {
485     PL_expect = XTERM;
486     deprecate("comma-less variable list");
487     return REPORT(','); /* grandfather non-comma-format format */
488 }
489 
490 /*
491  * S_ao
492  *
493  * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
494  * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
495  */
496 
497 STATIC int
498 S_ao(pTHX_ int toketype)
499 {
500     dVAR;
501     if (*PL_bufptr == '=') {
502 	PL_bufptr++;
503 	if (toketype == ANDAND)
504 	    pl_yylval.ival = OP_ANDASSIGN;
505 	else if (toketype == OROR)
506 	    pl_yylval.ival = OP_ORASSIGN;
507 	else if (toketype == DORDOR)
508 	    pl_yylval.ival = OP_DORASSIGN;
509 	toketype = ASSIGNOP;
510     }
511     return toketype;
512 }
513 
514 /*
515  * S_no_op
516  * When Perl expects an operator and finds something else, no_op
517  * prints the warning.  It always prints "<something> found where
518  * operator expected.  It prints "Missing semicolon on previous line?"
519  * if the surprise occurs at the start of the line.  "do you need to
520  * predeclare ..." is printed out for code like "sub bar; foo bar $x"
521  * where the compiler doesn't know if foo is a method call or a function.
522  * It prints "Missing operator before end of line" if there's nothing
523  * after the missing operator, or "... before <...>" if there is something
524  * after the missing operator.
525  */
526 
527 STATIC void
528 S_no_op(pTHX_ const char *const what, char *s)
529 {
530     dVAR;
531     char * const oldbp = PL_bufptr;
532     const bool is_first = (PL_oldbufptr == PL_linestart);
533 
534     PERL_ARGS_ASSERT_NO_OP;
535 
536     if (!s)
537 	s = oldbp;
538     else
539 	PL_bufptr = s;
540     yywarn(Perl_form(aTHX_ "%s found where operator expected", what), UTF ? SVf_UTF8 : 0);
541     if (ckWARN_d(WARN_SYNTAX)) {
542 	if (is_first)
543 	    Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
544 		    "\t(Missing semicolon on previous line?)\n");
545 	else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
546 	    const char *t;
547 	    for (t = PL_oldoldbufptr; (isWORDCHAR_lazy_if(t,UTF) || *t == ':');
548                                                             t += UTF ? UTF8SKIP(t) : 1)
549 		NOOP;
550 	    if (t < PL_bufptr && isSPACE(*t))
551 		Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
552 			"\t(Do you need to predeclare %"SVf"?)\n",
553 		    SVfARG(newSVpvn_flags(PL_oldoldbufptr, (STRLEN)(t - PL_oldoldbufptr),
554                                    SVs_TEMP | (UTF ? SVf_UTF8 : 0))));
555 	}
556 	else {
557 	    assert(s >= oldbp);
558 	    Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
559 		    "\t(Missing operator before %"SVf"?)\n",
560                     SVfARG(newSVpvn_flags(oldbp, (STRLEN)(s - oldbp),
561                                     SVs_TEMP | (UTF ? SVf_UTF8 : 0))));
562 	}
563     }
564     PL_bufptr = oldbp;
565 }
566 
567 /*
568  * S_missingterm
569  * Complain about missing quote/regexp/heredoc terminator.
570  * If it's called with NULL then it cauterizes the line buffer.
571  * If we're in a delimited string and the delimiter is a control
572  * character, it's reformatted into a two-char sequence like ^C.
573  * This is fatal.
574  */
575 
576 STATIC void
577 S_missingterm(pTHX_ char *s)
578 {
579     dVAR;
580     char tmpbuf[3];
581     char q;
582     if (s) {
583 	char * const nl = strrchr(s,'\n');
584 	if (nl)
585 	    *nl = '\0';
586     }
587     else if (isCNTRL(PL_multi_close)) {
588 	*tmpbuf = '^';
589 	tmpbuf[1] = (char)toCTRL(PL_multi_close);
590 	tmpbuf[2] = '\0';
591 	s = tmpbuf;
592     }
593     else {
594 	*tmpbuf = (char)PL_multi_close;
595 	tmpbuf[1] = '\0';
596 	s = tmpbuf;
597     }
598     q = strchr(s,'"') ? '\'' : '"';
599     Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
600 }
601 
602 #include "feature.h"
603 
604 /*
605  * Check whether the named feature is enabled.
606  */
607 bool
608 Perl_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
609 {
610     dVAR;
611     char he_name[8 + MAX_FEATURE_LEN] = "feature_";
612 
613     PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
614 
615     assert(CURRENT_FEATURE_BUNDLE == FEATURE_BUNDLE_CUSTOM);
616 
617     if (namelen > MAX_FEATURE_LEN)
618 	return FALSE;
619     memcpy(&he_name[8], name, namelen);
620 
621     return cBOOL(cop_hints_fetch_pvn(PL_curcop, he_name, 8 + namelen, 0,
622 				     REFCOUNTED_HE_EXISTS));
623 }
624 
625 /*
626  * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
627  * utf16-to-utf8-reversed.
628  */
629 
630 #ifdef PERL_CR_FILTER
631 static void
632 strip_return(SV *sv)
633 {
634     const char *s = SvPVX_const(sv);
635     const char * const e = s + SvCUR(sv);
636 
637     PERL_ARGS_ASSERT_STRIP_RETURN;
638 
639     /* outer loop optimized to do nothing if there are no CR-LFs */
640     while (s < e) {
641 	if (*s++ == '\r' && *s == '\n') {
642 	    /* hit a CR-LF, need to copy the rest */
643 	    char *d = s - 1;
644 	    *d++ = *s++;
645 	    while (s < e) {
646 		if (*s == '\r' && s[1] == '\n')
647 		    s++;
648 		*d++ = *s++;
649 	    }
650 	    SvCUR(sv) -= s - d;
651 	    return;
652 	}
653     }
654 }
655 
656 STATIC I32
657 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
658 {
659     const I32 count = FILTER_READ(idx+1, sv, maxlen);
660     if (count > 0 && !maxlen)
661 	strip_return(sv);
662     return count;
663 }
664 #endif
665 
666 /*
667 =for apidoc Amx|void|lex_start|SV *line|PerlIO *rsfp|U32 flags
668 
669 Creates and initialises a new lexer/parser state object, supplying
670 a context in which to lex and parse from a new source of Perl code.
671 A pointer to the new state object is placed in L</PL_parser>.  An entry
672 is made on the save stack so that upon unwinding the new state object
673 will be destroyed and the former value of L</PL_parser> will be restored.
674 Nothing else need be done to clean up the parsing context.
675 
676 The code to be parsed comes from I<line> and I<rsfp>.  I<line>, if
677 non-null, provides a string (in SV form) containing code to be parsed.
678 A copy of the string is made, so subsequent modification of I<line>
679 does not affect parsing.  I<rsfp>, if non-null, provides an input stream
680 from which code will be read to be parsed.  If both are non-null, the
681 code in I<line> comes first and must consist of complete lines of input,
682 and I<rsfp> supplies the remainder of the source.
683 
684 The I<flags> parameter is reserved for future use.  Currently it is only
685 used by perl internally, so extensions should always pass zero.
686 
687 =cut
688 */
689 
690 /* LEX_START_SAME_FILTER indicates that this is not a new file, so it
691    can share filters with the current parser.
692    LEX_START_DONT_CLOSE indicates that the file handle wasn't opened by the
693    caller, hence isn't owned by the parser, so shouldn't be closed on parser
694    destruction. This is used to handle the case of defaulting to reading the
695    script from the standard input because no filename was given on the command
696    line (without getting confused by situation where STDIN has been closed, so
697    the script handle is opened on fd 0)  */
698 
699 void
700 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, U32 flags)
701 {
702     dVAR;
703     const char *s = NULL;
704     yy_parser *parser, *oparser;
705     if (flags && flags & ~LEX_START_FLAGS)
706 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_start");
707 
708     /* create and initialise a parser */
709 
710     Newxz(parser, 1, yy_parser);
711     parser->old_parser = oparser = PL_parser;
712     PL_parser = parser;
713 
714     parser->stack = NULL;
715     parser->ps = NULL;
716     parser->stack_size = 0;
717 
718     /* on scope exit, free this parser and restore any outer one */
719     SAVEPARSER(parser);
720     parser->saved_curcop = PL_curcop;
721 
722     /* initialise lexer state */
723 
724 #ifdef PERL_MAD
725     parser->curforce = -1;
726 #else
727     parser->nexttoke = 0;
728 #endif
729     parser->error_count = oparser ? oparser->error_count : 0;
730     parser->copline = NOLINE;
731     parser->lex_state = LEX_NORMAL;
732     parser->expect = XSTATE;
733     parser->rsfp = rsfp;
734     parser->rsfp_filters =
735       !(flags & LEX_START_SAME_FILTER) || !oparser
736         ? NULL
737         : MUTABLE_AV(SvREFCNT_inc(
738             oparser->rsfp_filters
739              ? oparser->rsfp_filters
740              : (oparser->rsfp_filters = newAV())
741           ));
742 
743     Newx(parser->lex_brackstack, 120, char);
744     Newx(parser->lex_casestack, 12, char);
745     *parser->lex_casestack = '\0';
746     Newxz(parser->lex_shared, 1, LEXSHARED);
747 
748     if (line) {
749 	STRLEN len;
750 	s = SvPV_const(line, len);
751 	parser->linestr = flags & LEX_START_COPIED
752 			    ? SvREFCNT_inc_simple_NN(line)
753 			    : newSVpvn_flags(s, len, SvUTF8(line));
754 	sv_catpvs(parser->linestr, "\n;");
755     } else {
756 	parser->linestr = newSVpvs("\n;");
757     }
758     parser->oldoldbufptr =
759 	parser->oldbufptr =
760 	parser->bufptr =
761 	parser->linestart = SvPVX(parser->linestr);
762     parser->bufend = parser->bufptr + SvCUR(parser->linestr);
763     parser->last_lop = parser->last_uni = NULL;
764     parser->lex_flags = flags & (LEX_IGNORE_UTF8_HINTS|LEX_EVALBYTES
765 				 |LEX_DONT_CLOSE_RSFP);
766 
767     parser->in_pod = parser->filtered = 0;
768 }
769 
770 
771 /* delete a parser object */
772 
773 void
774 Perl_parser_free(pTHX_  const yy_parser *parser)
775 {
776     PERL_ARGS_ASSERT_PARSER_FREE;
777 
778     PL_curcop = parser->saved_curcop;
779     SvREFCNT_dec(parser->linestr);
780 
781     if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
782 	PerlIO_clearerr(parser->rsfp);
783     else if (parser->rsfp && (!parser->old_parser ||
784 		(parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
785 	PerlIO_close(parser->rsfp);
786     SvREFCNT_dec(parser->rsfp_filters);
787     SvREFCNT_dec(parser->lex_stuff);
788     SvREFCNT_dec(parser->sublex_info.repl);
789 
790     Safefree(parser->lex_brackstack);
791     Safefree(parser->lex_casestack);
792     Safefree(parser->lex_shared);
793     PL_parser = parser->old_parser;
794     Safefree(parser);
795 }
796 
797 void
798 Perl_parser_free_nexttoke_ops(pTHX_  yy_parser *parser, OPSLAB *slab)
799 {
800 #ifdef PERL_MAD
801     I32 nexttoke = parser->lasttoke;
802 #else
803     I32 nexttoke = parser->nexttoke;
804 #endif
805     PERL_ARGS_ASSERT_PARSER_FREE_NEXTTOKE_OPS;
806     while (nexttoke--) {
807 #ifdef PERL_MAD
808 	if (S_is_opval_token(parser->nexttoke[nexttoke].next_type
809 				& 0xffff)
810 	 && parser->nexttoke[nexttoke].next_val.opval
811 	 && parser->nexttoke[nexttoke].next_val.opval->op_slabbed
812 	 && OpSLAB(parser->nexttoke[nexttoke].next_val.opval) == slab) {
813 		op_free(parser->nexttoke[nexttoke].next_val.opval);
814 		parser->nexttoke[nexttoke].next_val.opval = NULL;
815 	}
816 #else
817 	if (S_is_opval_token(parser->nexttype[nexttoke] & 0xffff)
818 	 && parser->nextval[nexttoke].opval
819 	 && parser->nextval[nexttoke].opval->op_slabbed
820 	 && OpSLAB(parser->nextval[nexttoke].opval) == slab) {
821 	    op_free(parser->nextval[nexttoke].opval);
822 	    parser->nextval[nexttoke].opval = NULL;
823 	}
824 #endif
825     }
826 }
827 
828 
829 /*
830 =for apidoc AmxU|SV *|PL_parser-E<gt>linestr
831 
832 Buffer scalar containing the chunk currently under consideration of the
833 text currently being lexed.  This is always a plain string scalar (for
834 which C<SvPOK> is true).  It is not intended to be used as a scalar by
835 normal scalar means; instead refer to the buffer directly by the pointer
836 variables described below.
837 
838 The lexer maintains various C<char*> pointers to things in the
839 C<PL_parser-E<gt>linestr> buffer.  If C<PL_parser-E<gt>linestr> is ever
840 reallocated, all of these pointers must be updated.  Don't attempt to
841 do this manually, but rather use L</lex_grow_linestr> if you need to
842 reallocate the buffer.
843 
844 The content of the text chunk in the buffer is commonly exactly one
845 complete line of input, up to and including a newline terminator,
846 but there are situations where it is otherwise.  The octets of the
847 buffer may be intended to be interpreted as either UTF-8 or Latin-1.
848 The function L</lex_bufutf8> tells you which.  Do not use the C<SvUTF8>
849 flag on this scalar, which may disagree with it.
850 
851 For direct examination of the buffer, the variable
852 L</PL_parser-E<gt>bufend> points to the end of the buffer.  The current
853 lexing position is pointed to by L</PL_parser-E<gt>bufptr>.  Direct use
854 of these pointers is usually preferable to examination of the scalar
855 through normal scalar means.
856 
857 =for apidoc AmxU|char *|PL_parser-E<gt>bufend
858 
859 Direct pointer to the end of the chunk of text currently being lexed, the
860 end of the lexer buffer.  This is equal to C<SvPVX(PL_parser-E<gt>linestr)
861 + SvCUR(PL_parser-E<gt>linestr)>.  A NUL character (zero octet) is
862 always located at the end of the buffer, and does not count as part of
863 the buffer's contents.
864 
865 =for apidoc AmxU|char *|PL_parser-E<gt>bufptr
866 
867 Points to the current position of lexing inside the lexer buffer.
868 Characters around this point may be freely examined, within
869 the range delimited by C<SvPVX(L</PL_parser-E<gt>linestr>)> and
870 L</PL_parser-E<gt>bufend>.  The octets of the buffer may be intended to be
871 interpreted as either UTF-8 or Latin-1, as indicated by L</lex_bufutf8>.
872 
873 Lexing code (whether in the Perl core or not) moves this pointer past
874 the characters that it consumes.  It is also expected to perform some
875 bookkeeping whenever a newline character is consumed.  This movement
876 can be more conveniently performed by the function L</lex_read_to>,
877 which handles newlines appropriately.
878 
879 Interpretation of the buffer's octets can be abstracted out by
880 using the slightly higher-level functions L</lex_peek_unichar> and
881 L</lex_read_unichar>.
882 
883 =for apidoc AmxU|char *|PL_parser-E<gt>linestart
884 
885 Points to the start of the current line inside the lexer buffer.
886 This is useful for indicating at which column an error occurred, and
887 not much else.  This must be updated by any lexing code that consumes
888 a newline; the function L</lex_read_to> handles this detail.
889 
890 =cut
891 */
892 
893 /*
894 =for apidoc Amx|bool|lex_bufutf8
895 
896 Indicates whether the octets in the lexer buffer
897 (L</PL_parser-E<gt>linestr>) should be interpreted as the UTF-8 encoding
898 of Unicode characters.  If not, they should be interpreted as Latin-1
899 characters.  This is analogous to the C<SvUTF8> flag for scalars.
900 
901 In UTF-8 mode, it is not guaranteed that the lexer buffer actually
902 contains valid UTF-8.  Lexing code must be robust in the face of invalid
903 encoding.
904 
905 The actual C<SvUTF8> flag of the L</PL_parser-E<gt>linestr> scalar
906 is significant, but not the whole story regarding the input character
907 encoding.  Normally, when a file is being read, the scalar contains octets
908 and its C<SvUTF8> flag is off, but the octets should be interpreted as
909 UTF-8 if the C<use utf8> pragma is in effect.  During a string eval,
910 however, the scalar may have the C<SvUTF8> flag on, and in this case its
911 octets should be interpreted as UTF-8 unless the C<use bytes> pragma
912 is in effect.  This logic may change in the future; use this function
913 instead of implementing the logic yourself.
914 
915 =cut
916 */
917 
918 bool
919 Perl_lex_bufutf8(pTHX)
920 {
921     return UTF;
922 }
923 
924 /*
925 =for apidoc Amx|char *|lex_grow_linestr|STRLEN len
926 
927 Reallocates the lexer buffer (L</PL_parser-E<gt>linestr>) to accommodate
928 at least I<len> octets (including terminating NUL).  Returns a
929 pointer to the reallocated buffer.  This is necessary before making
930 any direct modification of the buffer that would increase its length.
931 L</lex_stuff_pvn> provides a more convenient way to insert text into
932 the buffer.
933 
934 Do not use C<SvGROW> or C<sv_grow> directly on C<PL_parser-E<gt>linestr>;
935 this function updates all of the lexer's variables that point directly
936 into the buffer.
937 
938 =cut
939 */
940 
941 char *
942 Perl_lex_grow_linestr(pTHX_ STRLEN len)
943 {
944     SV *linestr;
945     char *buf;
946     STRLEN bufend_pos, bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
947     STRLEN linestart_pos, last_uni_pos, last_lop_pos, re_eval_start_pos;
948     linestr = PL_parser->linestr;
949     buf = SvPVX(linestr);
950     if (len <= SvLEN(linestr))
951 	return buf;
952     bufend_pos = PL_parser->bufend - buf;
953     bufptr_pos = PL_parser->bufptr - buf;
954     oldbufptr_pos = PL_parser->oldbufptr - buf;
955     oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
956     linestart_pos = PL_parser->linestart - buf;
957     last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
958     last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
959     re_eval_start_pos = PL_parser->lex_shared->re_eval_start ?
960                             PL_parser->lex_shared->re_eval_start - buf : 0;
961 
962     buf = sv_grow(linestr, len);
963 
964     PL_parser->bufend = buf + bufend_pos;
965     PL_parser->bufptr = buf + bufptr_pos;
966     PL_parser->oldbufptr = buf + oldbufptr_pos;
967     PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
968     PL_parser->linestart = buf + linestart_pos;
969     if (PL_parser->last_uni)
970 	PL_parser->last_uni = buf + last_uni_pos;
971     if (PL_parser->last_lop)
972 	PL_parser->last_lop = buf + last_lop_pos;
973     if (PL_parser->lex_shared->re_eval_start)
974         PL_parser->lex_shared->re_eval_start  = buf + re_eval_start_pos;
975     return buf;
976 }
977 
978 /*
979 =for apidoc Amx|void|lex_stuff_pvn|const char *pv|STRLEN len|U32 flags
980 
981 Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
982 immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
983 reallocating the buffer if necessary.  This means that lexing code that
984 runs later will see the characters as if they had appeared in the input.
985 It is not recommended to do this as part of normal parsing, and most
986 uses of this facility run the risk of the inserted characters being
987 interpreted in an unintended manner.
988 
989 The string to be inserted is represented by I<len> octets starting
990 at I<pv>.  These octets are interpreted as either UTF-8 or Latin-1,
991 according to whether the C<LEX_STUFF_UTF8> flag is set in I<flags>.
992 The characters are recoded for the lexer buffer, according to how the
993 buffer is currently being interpreted (L</lex_bufutf8>).  If a string
994 to be inserted is available as a Perl scalar, the L</lex_stuff_sv>
995 function is more convenient.
996 
997 =cut
998 */
999 
1000 void
1001 Perl_lex_stuff_pvn(pTHX_ const char *pv, STRLEN len, U32 flags)
1002 {
1003     dVAR;
1004     char *bufptr;
1005     PERL_ARGS_ASSERT_LEX_STUFF_PVN;
1006     if (flags & ~(LEX_STUFF_UTF8))
1007 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_pvn");
1008     if (UTF) {
1009 	if (flags & LEX_STUFF_UTF8) {
1010 	    goto plain_copy;
1011 	} else {
1012 	    STRLEN highhalf = 0;    /* Count of variants */
1013 	    const char *p, *e = pv+len;
1014 	    for (p = pv; p != e; p++) {
1015 		if (! UTF8_IS_INVARIANT(*p)) {
1016                     highhalf++;
1017                 }
1018             }
1019 	    if (!highhalf)
1020 		goto plain_copy;
1021 	    lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len+highhalf);
1022 	    bufptr = PL_parser->bufptr;
1023 	    Move(bufptr, bufptr+len+highhalf, PL_parser->bufend+1-bufptr, char);
1024 	    SvCUR_set(PL_parser->linestr,
1025 	    	SvCUR(PL_parser->linestr) + len+highhalf);
1026 	    PL_parser->bufend += len+highhalf;
1027 	    for (p = pv; p != e; p++) {
1028 		U8 c = (U8)*p;
1029 		if (! UTF8_IS_INVARIANT(c)) {
1030 		    *bufptr++ = UTF8_TWO_BYTE_HI(c);
1031 		    *bufptr++ = UTF8_TWO_BYTE_LO(c);
1032 		} else {
1033 		    *bufptr++ = (char)c;
1034 		}
1035 	    }
1036 	}
1037     } else {
1038 	if (flags & LEX_STUFF_UTF8) {
1039 	    STRLEN highhalf = 0;
1040 	    const char *p, *e = pv+len;
1041 	    for (p = pv; p != e; p++) {
1042 		U8 c = (U8)*p;
1043 		if (UTF8_IS_ABOVE_LATIN1(c)) {
1044 		    Perl_croak(aTHX_ "Lexing code attempted to stuff "
1045 				"non-Latin-1 character into Latin-1 input");
1046 		} else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(p, e)) {
1047 		    p++;
1048 		    highhalf++;
1049 		} else if (! UTF8_IS_INVARIANT(c)) {
1050 		    /* malformed UTF-8 */
1051 		    ENTER;
1052 		    SAVESPTR(PL_warnhook);
1053 		    PL_warnhook = PERL_WARNHOOK_FATAL;
1054 		    utf8n_to_uvuni((U8*)p, e-p, NULL, 0);
1055 		    LEAVE;
1056 		}
1057 	    }
1058 	    if (!highhalf)
1059 		goto plain_copy;
1060 	    lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len-highhalf);
1061 	    bufptr = PL_parser->bufptr;
1062 	    Move(bufptr, bufptr+len-highhalf, PL_parser->bufend+1-bufptr, char);
1063 	    SvCUR_set(PL_parser->linestr,
1064 	    	SvCUR(PL_parser->linestr) + len-highhalf);
1065 	    PL_parser->bufend += len-highhalf;
1066 	    p = pv;
1067 	    while (p < e) {
1068 		if (UTF8_IS_INVARIANT(*p)) {
1069 		    *bufptr++ = *p;
1070                     p++;
1071 		}
1072 		else {
1073                     assert(p < e -1 );
1074 		    *bufptr++ = TWO_BYTE_UTF8_TO_UNI(*p, *(p+1));
1075 		    p += 2;
1076                 }
1077 	    }
1078 	} else {
1079 	  plain_copy:
1080 	    lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len);
1081 	    bufptr = PL_parser->bufptr;
1082 	    Move(bufptr, bufptr+len, PL_parser->bufend+1-bufptr, char);
1083 	    SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) + len);
1084 	    PL_parser->bufend += len;
1085 	    Copy(pv, bufptr, len, char);
1086 	}
1087     }
1088 }
1089 
1090 /*
1091 =for apidoc Amx|void|lex_stuff_pv|const char *pv|U32 flags
1092 
1093 Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1094 immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1095 reallocating the buffer if necessary.  This means that lexing code that
1096 runs later will see the characters as if they had appeared in the input.
1097 It is not recommended to do this as part of normal parsing, and most
1098 uses of this facility run the risk of the inserted characters being
1099 interpreted in an unintended manner.
1100 
1101 The string to be inserted is represented by octets starting at I<pv>
1102 and continuing to the first nul.  These octets are interpreted as either
1103 UTF-8 or Latin-1, according to whether the C<LEX_STUFF_UTF8> flag is set
1104 in I<flags>.  The characters are recoded for the lexer buffer, according
1105 to how the buffer is currently being interpreted (L</lex_bufutf8>).
1106 If it is not convenient to nul-terminate a string to be inserted, the
1107 L</lex_stuff_pvn> function is more appropriate.
1108 
1109 =cut
1110 */
1111 
1112 void
1113 Perl_lex_stuff_pv(pTHX_ const char *pv, U32 flags)
1114 {
1115     PERL_ARGS_ASSERT_LEX_STUFF_PV;
1116     lex_stuff_pvn(pv, strlen(pv), flags);
1117 }
1118 
1119 /*
1120 =for apidoc Amx|void|lex_stuff_sv|SV *sv|U32 flags
1121 
1122 Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1123 immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1124 reallocating the buffer if necessary.  This means that lexing code that
1125 runs later will see the characters as if they had appeared in the input.
1126 It is not recommended to do this as part of normal parsing, and most
1127 uses of this facility run the risk of the inserted characters being
1128 interpreted in an unintended manner.
1129 
1130 The string to be inserted is the string value of I<sv>.  The characters
1131 are recoded for the lexer buffer, according to how the buffer is currently
1132 being interpreted (L</lex_bufutf8>).  If a string to be inserted is
1133 not already a Perl scalar, the L</lex_stuff_pvn> function avoids the
1134 need to construct a scalar.
1135 
1136 =cut
1137 */
1138 
1139 void
1140 Perl_lex_stuff_sv(pTHX_ SV *sv, U32 flags)
1141 {
1142     char *pv;
1143     STRLEN len;
1144     PERL_ARGS_ASSERT_LEX_STUFF_SV;
1145     if (flags)
1146 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_sv");
1147     pv = SvPV(sv, len);
1148     lex_stuff_pvn(pv, len, flags | (SvUTF8(sv) ? LEX_STUFF_UTF8 : 0));
1149 }
1150 
1151 /*
1152 =for apidoc Amx|void|lex_unstuff|char *ptr
1153 
1154 Discards text about to be lexed, from L</PL_parser-E<gt>bufptr> up to
1155 I<ptr>.  Text following I<ptr> will be moved, and the buffer shortened.
1156 This hides the discarded text from any lexing code that runs later,
1157 as if the text had never appeared.
1158 
1159 This is not the normal way to consume lexed text.  For that, use
1160 L</lex_read_to>.
1161 
1162 =cut
1163 */
1164 
1165 void
1166 Perl_lex_unstuff(pTHX_ char *ptr)
1167 {
1168     char *buf, *bufend;
1169     STRLEN unstuff_len;
1170     PERL_ARGS_ASSERT_LEX_UNSTUFF;
1171     buf = PL_parser->bufptr;
1172     if (ptr < buf)
1173 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1174     if (ptr == buf)
1175 	return;
1176     bufend = PL_parser->bufend;
1177     if (ptr > bufend)
1178 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1179     unstuff_len = ptr - buf;
1180     Move(ptr, buf, bufend+1-ptr, char);
1181     SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - unstuff_len);
1182     PL_parser->bufend = bufend - unstuff_len;
1183 }
1184 
1185 /*
1186 =for apidoc Amx|void|lex_read_to|char *ptr
1187 
1188 Consume text in the lexer buffer, from L</PL_parser-E<gt>bufptr> up
1189 to I<ptr>.  This advances L</PL_parser-E<gt>bufptr> to match I<ptr>,
1190 performing the correct bookkeeping whenever a newline character is passed.
1191 This is the normal way to consume lexed text.
1192 
1193 Interpretation of the buffer's octets can be abstracted out by
1194 using the slightly higher-level functions L</lex_peek_unichar> and
1195 L</lex_read_unichar>.
1196 
1197 =cut
1198 */
1199 
1200 void
1201 Perl_lex_read_to(pTHX_ char *ptr)
1202 {
1203     char *s;
1204     PERL_ARGS_ASSERT_LEX_READ_TO;
1205     s = PL_parser->bufptr;
1206     if (ptr < s || ptr > PL_parser->bufend)
1207 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_to");
1208     for (; s != ptr; s++)
1209 	if (*s == '\n') {
1210 	    COPLINE_INC_WITH_HERELINES;
1211 	    PL_parser->linestart = s+1;
1212 	}
1213     PL_parser->bufptr = ptr;
1214 }
1215 
1216 /*
1217 =for apidoc Amx|void|lex_discard_to|char *ptr
1218 
1219 Discards the first part of the L</PL_parser-E<gt>linestr> buffer,
1220 up to I<ptr>.  The remaining content of the buffer will be moved, and
1221 all pointers into the buffer updated appropriately.  I<ptr> must not
1222 be later in the buffer than the position of L</PL_parser-E<gt>bufptr>:
1223 it is not permitted to discard text that has yet to be lexed.
1224 
1225 Normally it is not necessarily to do this directly, because it suffices to
1226 use the implicit discarding behaviour of L</lex_next_chunk> and things
1227 based on it.  However, if a token stretches across multiple lines,
1228 and the lexing code has kept multiple lines of text in the buffer for
1229 that purpose, then after completion of the token it would be wise to
1230 explicitly discard the now-unneeded earlier lines, to avoid future
1231 multi-line tokens growing the buffer without bound.
1232 
1233 =cut
1234 */
1235 
1236 void
1237 Perl_lex_discard_to(pTHX_ char *ptr)
1238 {
1239     char *buf;
1240     STRLEN discard_len;
1241     PERL_ARGS_ASSERT_LEX_DISCARD_TO;
1242     buf = SvPVX(PL_parser->linestr);
1243     if (ptr < buf)
1244 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1245     if (ptr == buf)
1246 	return;
1247     if (ptr > PL_parser->bufptr)
1248 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1249     discard_len = ptr - buf;
1250     if (PL_parser->oldbufptr < ptr)
1251 	PL_parser->oldbufptr = ptr;
1252     if (PL_parser->oldoldbufptr < ptr)
1253 	PL_parser->oldoldbufptr = ptr;
1254     if (PL_parser->last_uni && PL_parser->last_uni < ptr)
1255 	PL_parser->last_uni = NULL;
1256     if (PL_parser->last_lop && PL_parser->last_lop < ptr)
1257 	PL_parser->last_lop = NULL;
1258     Move(ptr, buf, PL_parser->bufend+1-ptr, char);
1259     SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - discard_len);
1260     PL_parser->bufend -= discard_len;
1261     PL_parser->bufptr -= discard_len;
1262     PL_parser->oldbufptr -= discard_len;
1263     PL_parser->oldoldbufptr -= discard_len;
1264     if (PL_parser->last_uni)
1265 	PL_parser->last_uni -= discard_len;
1266     if (PL_parser->last_lop)
1267 	PL_parser->last_lop -= discard_len;
1268 }
1269 
1270 /*
1271 =for apidoc Amx|bool|lex_next_chunk|U32 flags
1272 
1273 Reads in the next chunk of text to be lexed, appending it to
1274 L</PL_parser-E<gt>linestr>.  This should be called when lexing code has
1275 looked to the end of the current chunk and wants to know more.  It is
1276 usual, but not necessary, for lexing to have consumed the entirety of
1277 the current chunk at this time.
1278 
1279 If L</PL_parser-E<gt>bufptr> is pointing to the very end of the current
1280 chunk (i.e., the current chunk has been entirely consumed), normally the
1281 current chunk will be discarded at the same time that the new chunk is
1282 read in.  If I<flags> includes C<LEX_KEEP_PREVIOUS>, the current chunk
1283 will not be discarded.  If the current chunk has not been entirely
1284 consumed, then it will not be discarded regardless of the flag.
1285 
1286 Returns true if some new text was added to the buffer, or false if the
1287 buffer has reached the end of the input text.
1288 
1289 =cut
1290 */
1291 
1292 #define LEX_FAKE_EOF 0x80000000
1293 #define LEX_NO_TERM  0x40000000
1294 
1295 bool
1296 Perl_lex_next_chunk(pTHX_ U32 flags)
1297 {
1298     SV *linestr;
1299     char *buf;
1300     STRLEN old_bufend_pos, new_bufend_pos;
1301     STRLEN bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
1302     STRLEN linestart_pos, last_uni_pos, last_lop_pos;
1303     bool got_some_for_debugger = 0;
1304     bool got_some;
1305     if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF|LEX_NO_TERM))
1306 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk");
1307     linestr = PL_parser->linestr;
1308     buf = SvPVX(linestr);
1309     if (!(flags & LEX_KEEP_PREVIOUS) &&
1310 	    PL_parser->bufptr == PL_parser->bufend) {
1311 	old_bufend_pos = bufptr_pos = oldbufptr_pos = oldoldbufptr_pos = 0;
1312 	linestart_pos = 0;
1313 	if (PL_parser->last_uni != PL_parser->bufend)
1314 	    PL_parser->last_uni = NULL;
1315 	if (PL_parser->last_lop != PL_parser->bufend)
1316 	    PL_parser->last_lop = NULL;
1317 	last_uni_pos = last_lop_pos = 0;
1318 	*buf = 0;
1319 	SvCUR(linestr) = 0;
1320     } else {
1321 	old_bufend_pos = PL_parser->bufend - buf;
1322 	bufptr_pos = PL_parser->bufptr - buf;
1323 	oldbufptr_pos = PL_parser->oldbufptr - buf;
1324 	oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
1325 	linestart_pos = PL_parser->linestart - buf;
1326 	last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
1327 	last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
1328     }
1329     if (flags & LEX_FAKE_EOF) {
1330 	goto eof;
1331     } else if (!PL_parser->rsfp && !PL_parser->filtered) {
1332 	got_some = 0;
1333     } else if (filter_gets(linestr, old_bufend_pos)) {
1334 	got_some = 1;
1335 	got_some_for_debugger = 1;
1336     } else if (flags & LEX_NO_TERM) {
1337 	got_some = 0;
1338     } else {
1339 	if (!SvPOK(linestr))   /* can get undefined by filter_gets */
1340 	    sv_setpvs(linestr, "");
1341 	eof:
1342 	/* End of real input.  Close filehandle (unless it was STDIN),
1343 	 * then add implicit termination.
1344 	 */
1345 	if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
1346 	    PerlIO_clearerr(PL_parser->rsfp);
1347 	else if (PL_parser->rsfp)
1348 	    (void)PerlIO_close(PL_parser->rsfp);
1349 	PL_parser->rsfp = NULL;
1350 	PL_parser->in_pod = PL_parser->filtered = 0;
1351 #ifdef PERL_MAD
1352 	if (PL_madskills && !PL_in_eval && (PL_minus_p || PL_minus_n))
1353 	    PL_faketokens = 1;
1354 #endif
1355 	if (!PL_in_eval && PL_minus_p) {
1356 	    sv_catpvs(linestr,
1357 		/*{*/";}continue{print or die qq(-p destination: $!\\n);}");
1358 	    PL_minus_n = PL_minus_p = 0;
1359 	} else if (!PL_in_eval && PL_minus_n) {
1360 	    sv_catpvs(linestr, /*{*/";}");
1361 	    PL_minus_n = 0;
1362 	} else
1363 	    sv_catpvs(linestr, ";");
1364 	got_some = 1;
1365     }
1366     buf = SvPVX(linestr);
1367     new_bufend_pos = SvCUR(linestr);
1368     PL_parser->bufend = buf + new_bufend_pos;
1369     PL_parser->bufptr = buf + bufptr_pos;
1370     PL_parser->oldbufptr = buf + oldbufptr_pos;
1371     PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
1372     PL_parser->linestart = buf + linestart_pos;
1373     if (PL_parser->last_uni)
1374 	PL_parser->last_uni = buf + last_uni_pos;
1375     if (PL_parser->last_lop)
1376 	PL_parser->last_lop = buf + last_lop_pos;
1377     if (got_some_for_debugger && (PERLDB_LINE || PERLDB_SAVESRC) &&
1378 	    PL_curstash != PL_debstash) {
1379 	/* debugger active and we're not compiling the debugger code,
1380 	 * so store the line into the debugger's array of lines
1381 	 */
1382 	update_debugger_info(NULL, buf+old_bufend_pos,
1383 	    new_bufend_pos-old_bufend_pos);
1384     }
1385     return got_some;
1386 }
1387 
1388 /*
1389 =for apidoc Amx|I32|lex_peek_unichar|U32 flags
1390 
1391 Looks ahead one (Unicode) character in the text currently being lexed.
1392 Returns the codepoint (unsigned integer value) of the next character,
1393 or -1 if lexing has reached the end of the input text.  To consume the
1394 peeked character, use L</lex_read_unichar>.
1395 
1396 If the next character is in (or extends into) the next chunk of input
1397 text, the next chunk will be read in.  Normally the current chunk will be
1398 discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1399 then the current chunk will not be discarded.
1400 
1401 If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1402 is encountered, an exception is generated.
1403 
1404 =cut
1405 */
1406 
1407 I32
1408 Perl_lex_peek_unichar(pTHX_ U32 flags)
1409 {
1410     dVAR;
1411     char *s, *bufend;
1412     if (flags & ~(LEX_KEEP_PREVIOUS))
1413 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_peek_unichar");
1414     s = PL_parser->bufptr;
1415     bufend = PL_parser->bufend;
1416     if (UTF) {
1417 	U8 head;
1418 	I32 unichar;
1419 	STRLEN len, retlen;
1420 	if (s == bufend) {
1421 	    if (!lex_next_chunk(flags))
1422 		return -1;
1423 	    s = PL_parser->bufptr;
1424 	    bufend = PL_parser->bufend;
1425 	}
1426 	head = (U8)*s;
1427 	if (UTF8_IS_INVARIANT(head))
1428 	    return head;
1429 	if (UTF8_IS_START(head)) {
1430 	    len = UTF8SKIP(&head);
1431 	    while ((STRLEN)(bufend-s) < len) {
1432 		if (!lex_next_chunk(flags | LEX_KEEP_PREVIOUS))
1433 		    break;
1434 		s = PL_parser->bufptr;
1435 		bufend = PL_parser->bufend;
1436 	    }
1437 	}
1438 	unichar = utf8n_to_uvuni((U8*)s, bufend-s, &retlen, UTF8_CHECK_ONLY);
1439 	if (retlen == (STRLEN)-1) {
1440 	    /* malformed UTF-8 */
1441 	    ENTER;
1442 	    SAVESPTR(PL_warnhook);
1443 	    PL_warnhook = PERL_WARNHOOK_FATAL;
1444 	    utf8n_to_uvuni((U8*)s, bufend-s, NULL, 0);
1445 	    LEAVE;
1446 	}
1447 	return unichar;
1448     } else {
1449 	if (s == bufend) {
1450 	    if (!lex_next_chunk(flags))
1451 		return -1;
1452 	    s = PL_parser->bufptr;
1453 	}
1454 	return (U8)*s;
1455     }
1456 }
1457 
1458 /*
1459 =for apidoc Amx|I32|lex_read_unichar|U32 flags
1460 
1461 Reads the next (Unicode) character in the text currently being lexed.
1462 Returns the codepoint (unsigned integer value) of the character read,
1463 and moves L</PL_parser-E<gt>bufptr> past the character, or returns -1
1464 if lexing has reached the end of the input text.  To non-destructively
1465 examine the next character, use L</lex_peek_unichar> instead.
1466 
1467 If the next character is in (or extends into) the next chunk of input
1468 text, the next chunk will be read in.  Normally the current chunk will be
1469 discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1470 then the current chunk will not be discarded.
1471 
1472 If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1473 is encountered, an exception is generated.
1474 
1475 =cut
1476 */
1477 
1478 I32
1479 Perl_lex_read_unichar(pTHX_ U32 flags)
1480 {
1481     I32 c;
1482     if (flags & ~(LEX_KEEP_PREVIOUS))
1483 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_unichar");
1484     c = lex_peek_unichar(flags);
1485     if (c != -1) {
1486 	if (c == '\n')
1487 	    COPLINE_INC_WITH_HERELINES;
1488 	if (UTF)
1489 	    PL_parser->bufptr += UTF8SKIP(PL_parser->bufptr);
1490 	else
1491 	    ++(PL_parser->bufptr);
1492     }
1493     return c;
1494 }
1495 
1496 /*
1497 =for apidoc Amx|void|lex_read_space|U32 flags
1498 
1499 Reads optional spaces, in Perl style, in the text currently being
1500 lexed.  The spaces may include ordinary whitespace characters and
1501 Perl-style comments.  C<#line> directives are processed if encountered.
1502 L</PL_parser-E<gt>bufptr> is moved past the spaces, so that it points
1503 at a non-space character (or the end of the input text).
1504 
1505 If spaces extend into the next chunk of input text, the next chunk will
1506 be read in.  Normally the current chunk will be discarded at the same
1507 time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> then the current
1508 chunk will not be discarded.
1509 
1510 =cut
1511 */
1512 
1513 #define LEX_NO_NEXT_CHUNK 0x80000000
1514 
1515 void
1516 Perl_lex_read_space(pTHX_ U32 flags)
1517 {
1518     char *s, *bufend;
1519     bool need_incline = 0;
1520     if (flags & ~(LEX_KEEP_PREVIOUS|LEX_NO_NEXT_CHUNK))
1521 	Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_space");
1522 #ifdef PERL_MAD
1523     if (PL_skipwhite) {
1524 	sv_free(PL_skipwhite);
1525 	PL_skipwhite = NULL;
1526     }
1527     if (PL_madskills)
1528 	PL_skipwhite = newSVpvs("");
1529 #endif /* PERL_MAD */
1530     s = PL_parser->bufptr;
1531     bufend = PL_parser->bufend;
1532     while (1) {
1533 	char c = *s;
1534 	if (c == '#') {
1535 	    do {
1536 		c = *++s;
1537 	    } while (!(c == '\n' || (c == 0 && s == bufend)));
1538 	} else if (c == '\n') {
1539 	    s++;
1540 	    PL_parser->linestart = s;
1541 	    if (s == bufend)
1542 		need_incline = 1;
1543 	    else
1544 		incline(s);
1545 	} else if (isSPACE(c)) {
1546 	    s++;
1547 	} else if (c == 0 && s == bufend) {
1548 	    bool got_more;
1549 #ifdef PERL_MAD
1550 	    if (PL_madskills)
1551 		sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1552 #endif /* PERL_MAD */
1553 	    if (flags & LEX_NO_NEXT_CHUNK)
1554 		break;
1555 	    PL_parser->bufptr = s;
1556 	    COPLINE_INC_WITH_HERELINES;
1557 	    got_more = lex_next_chunk(flags);
1558 	    CopLINE_dec(PL_curcop);
1559 	    s = PL_parser->bufptr;
1560 	    bufend = PL_parser->bufend;
1561 	    if (!got_more)
1562 		break;
1563 	    if (need_incline && PL_parser->rsfp) {
1564 		incline(s);
1565 		need_incline = 0;
1566 	    }
1567 	} else {
1568 	    break;
1569 	}
1570     }
1571 #ifdef PERL_MAD
1572     if (PL_madskills)
1573 	sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1574 #endif /* PERL_MAD */
1575     PL_parser->bufptr = s;
1576 }
1577 
1578 /*
1579  * S_incline
1580  * This subroutine has nothing to do with tilting, whether at windmills
1581  * or pinball tables.  Its name is short for "increment line".  It
1582  * increments the current line number in CopLINE(PL_curcop) and checks
1583  * to see whether the line starts with a comment of the form
1584  *    # line 500 "foo.pm"
1585  * If so, it sets the current line number and file to the values in the comment.
1586  */
1587 
1588 STATIC void
1589 S_incline(pTHX_ const char *s)
1590 {
1591     dVAR;
1592     const char *t;
1593     const char *n;
1594     const char *e;
1595     line_t line_num;
1596 
1597     PERL_ARGS_ASSERT_INCLINE;
1598 
1599     COPLINE_INC_WITH_HERELINES;
1600     if (!PL_rsfp && !PL_parser->filtered && PL_lex_state == LEX_NORMAL
1601      && s+1 == PL_bufend && *s == ';') {
1602 	/* fake newline in string eval */
1603 	CopLINE_dec(PL_curcop);
1604 	return;
1605     }
1606     if (*s++ != '#')
1607 	return;
1608     while (SPACE_OR_TAB(*s))
1609 	s++;
1610     if (strnEQ(s, "line", 4))
1611 	s += 4;
1612     else
1613 	return;
1614     if (SPACE_OR_TAB(*s))
1615 	s++;
1616     else
1617 	return;
1618     while (SPACE_OR_TAB(*s))
1619 	s++;
1620     if (!isDIGIT(*s))
1621 	return;
1622 
1623     n = s;
1624     while (isDIGIT(*s))
1625 	s++;
1626     if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
1627 	return;
1628     while (SPACE_OR_TAB(*s))
1629 	s++;
1630     if (*s == '"' && (t = strchr(s+1, '"'))) {
1631 	s++;
1632 	e = t + 1;
1633     }
1634     else {
1635 	t = s;
1636 	while (!isSPACE(*t))
1637 	    t++;
1638 	e = t;
1639     }
1640     while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
1641 	e++;
1642     if (*e != '\n' && *e != '\0')
1643 	return;		/* false alarm */
1644 
1645     line_num = atoi(n)-1;
1646 
1647     if (t - s > 0) {
1648 	const STRLEN len = t - s;
1649 	SV *const temp_sv = CopFILESV(PL_curcop);
1650 	const char *cf;
1651 	STRLEN tmplen;
1652 
1653 	if (temp_sv) {
1654 	    cf = SvPVX(temp_sv);
1655 	    tmplen = SvCUR(temp_sv);
1656 	} else {
1657 	    cf = NULL;
1658 	    tmplen = 0;
1659 	}
1660 
1661 	if (!PL_rsfp && !PL_parser->filtered) {
1662 	    /* must copy *{"::_<(eval N)[oldfilename:L]"}
1663 	     * to *{"::_<newfilename"} */
1664 	    /* However, the long form of evals is only turned on by the
1665 	       debugger - usually they're "(eval %lu)" */
1666 	    char smallbuf[128];
1667 	    char *tmpbuf;
1668 	    GV **gvp;
1669 	    STRLEN tmplen2 = len;
1670 	    if (tmplen + 2 <= sizeof smallbuf)
1671 		tmpbuf = smallbuf;
1672 	    else
1673 		Newx(tmpbuf, tmplen + 2, char);
1674 	    tmpbuf[0] = '_';
1675 	    tmpbuf[1] = '<';
1676 	    memcpy(tmpbuf + 2, cf, tmplen);
1677 	    tmplen += 2;
1678 	    gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
1679 	    if (gvp) {
1680 		char *tmpbuf2;
1681 		GV *gv2;
1682 
1683 		if (tmplen2 + 2 <= sizeof smallbuf)
1684 		    tmpbuf2 = smallbuf;
1685 		else
1686 		    Newx(tmpbuf2, tmplen2 + 2, char);
1687 
1688 		if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
1689 		    /* Either they malloc'd it, or we malloc'd it,
1690 		       so no prefix is present in ours.  */
1691 		    tmpbuf2[0] = '_';
1692 		    tmpbuf2[1] = '<';
1693 		}
1694 
1695 		memcpy(tmpbuf2 + 2, s, tmplen2);
1696 		tmplen2 += 2;
1697 
1698 		gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
1699 		if (!isGV(gv2)) {
1700 		    gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
1701 		    /* adjust ${"::_<newfilename"} to store the new file name */
1702 		    GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
1703 		    /* The line number may differ. If that is the case,
1704 		       alias the saved lines that are in the array.
1705 		       Otherwise alias the whole array. */
1706 		    if (CopLINE(PL_curcop) == line_num) {
1707 			GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
1708 			GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
1709 		    }
1710 		    else if (GvAV(*gvp)) {
1711 			AV * const av = GvAV(*gvp);
1712 			const I32 start = CopLINE(PL_curcop)+1;
1713 			I32 items = AvFILLp(av) - start;
1714 			if (items > 0) {
1715 			    AV * const av2 = GvAVn(gv2);
1716 			    SV **svp = AvARRAY(av) + start;
1717 			    I32 l = (I32)line_num+1;
1718 			    while (items--)
1719 				av_store(av2, l++, SvREFCNT_inc(*svp++));
1720 			}
1721 		    }
1722 		}
1723 
1724 		if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
1725 	    }
1726 	    if (tmpbuf != smallbuf) Safefree(tmpbuf);
1727 	}
1728 	CopFILE_free(PL_curcop);
1729 	CopFILE_setn(PL_curcop, s, len);
1730     }
1731     CopLINE_set(PL_curcop, line_num);
1732 }
1733 
1734 #ifdef PERL_MAD
1735 /* skip space before PL_thistoken */
1736 
1737 STATIC char *
1738 S_skipspace0(pTHX_ char *s)
1739 {
1740     PERL_ARGS_ASSERT_SKIPSPACE0;
1741 
1742     s = skipspace(s);
1743     if (!PL_madskills)
1744 	return s;
1745     if (PL_skipwhite) {
1746 	if (!PL_thiswhite)
1747 	    PL_thiswhite = newSVpvs("");
1748 	sv_catsv(PL_thiswhite, PL_skipwhite);
1749 	sv_free(PL_skipwhite);
1750 	PL_skipwhite = 0;
1751     }
1752     PL_realtokenstart = s - SvPVX(PL_linestr);
1753     return s;
1754 }
1755 
1756 /* skip space after PL_thistoken */
1757 
1758 STATIC char *
1759 S_skipspace1(pTHX_ char *s)
1760 {
1761     const char *start = s;
1762     I32 startoff = start - SvPVX(PL_linestr);
1763 
1764     PERL_ARGS_ASSERT_SKIPSPACE1;
1765 
1766     s = skipspace(s);
1767     if (!PL_madskills)
1768 	return s;
1769     start = SvPVX(PL_linestr) + startoff;
1770     if (!PL_thistoken && PL_realtokenstart >= 0) {
1771 	const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
1772 	PL_thistoken = newSVpvn(tstart, start - tstart);
1773     }
1774     PL_realtokenstart = -1;
1775     if (PL_skipwhite) {
1776 	if (!PL_nextwhite)
1777 	    PL_nextwhite = newSVpvs("");
1778 	sv_catsv(PL_nextwhite, PL_skipwhite);
1779 	sv_free(PL_skipwhite);
1780 	PL_skipwhite = 0;
1781     }
1782     return s;
1783 }
1784 
1785 STATIC char *
1786 S_skipspace2(pTHX_ char *s, SV **svp)
1787 {
1788     char *start;
1789     const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
1790     const I32 startoff = s - SvPVX(PL_linestr);
1791 
1792     PERL_ARGS_ASSERT_SKIPSPACE2;
1793 
1794     s = skipspace(s);
1795     PL_bufptr = SvPVX(PL_linestr) + bufptroff;
1796     if (!PL_madskills || !svp)
1797 	return s;
1798     start = SvPVX(PL_linestr) + startoff;
1799     if (!PL_thistoken && PL_realtokenstart >= 0) {
1800 	char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
1801 	PL_thistoken = newSVpvn(tstart, start - tstart);
1802 	PL_realtokenstart = -1;
1803     }
1804     if (PL_skipwhite) {
1805 	if (!*svp)
1806 	    *svp = newSVpvs("");
1807 	sv_setsv(*svp, PL_skipwhite);
1808 	sv_free(PL_skipwhite);
1809 	PL_skipwhite = 0;
1810     }
1811 
1812     return s;
1813 }
1814 #endif
1815 
1816 STATIC void
1817 S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
1818 {
1819     AV *av = CopFILEAVx(PL_curcop);
1820     if (av) {
1821 	SV * const sv = newSV_type(SVt_PVMG);
1822 	if (orig_sv)
1823 	    sv_setsv(sv, orig_sv);
1824 	else
1825 	    sv_setpvn(sv, buf, len);
1826 	(void)SvIOK_on(sv);
1827 	SvIV_set(sv, 0);
1828 	av_store(av, (I32)CopLINE(PL_curcop), sv);
1829     }
1830 }
1831 
1832 /*
1833  * S_skipspace
1834  * Called to gobble the appropriate amount and type of whitespace.
1835  * Skips comments as well.
1836  */
1837 
1838 STATIC char *
1839 S_skipspace(pTHX_ char *s)
1840 {
1841 #ifdef PERL_MAD
1842     char *start = s;
1843 #endif /* PERL_MAD */
1844     PERL_ARGS_ASSERT_SKIPSPACE;
1845 #ifdef PERL_MAD
1846     if (PL_skipwhite) {
1847 	sv_free(PL_skipwhite);
1848 	PL_skipwhite = NULL;
1849     }
1850 #endif /* PERL_MAD */
1851     if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
1852 	while (s < PL_bufend && SPACE_OR_TAB(*s))
1853 	    s++;
1854     } else {
1855 	STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
1856 	PL_bufptr = s;
1857 	lex_read_space(LEX_KEEP_PREVIOUS |
1858 		(PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
1859 		    LEX_NO_NEXT_CHUNK : 0));
1860 	s = PL_bufptr;
1861 	PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
1862 	if (PL_linestart > PL_bufptr)
1863 	    PL_bufptr = PL_linestart;
1864 	return s;
1865     }
1866 #ifdef PERL_MAD
1867     if (PL_madskills)
1868 	PL_skipwhite = newSVpvn(start, s-start);
1869 #endif /* PERL_MAD */
1870     return s;
1871 }
1872 
1873 /*
1874  * S_check_uni
1875  * Check the unary operators to ensure there's no ambiguity in how they're
1876  * used.  An ambiguous piece of code would be:
1877  *     rand + 5
1878  * This doesn't mean rand() + 5.  Because rand() is a unary operator,
1879  * the +5 is its argument.
1880  */
1881 
1882 STATIC void
1883 S_check_uni(pTHX)
1884 {
1885     dVAR;
1886     const char *s;
1887     const char *t;
1888 
1889     if (PL_oldoldbufptr != PL_last_uni)
1890 	return;
1891     while (isSPACE(*PL_last_uni))
1892 	PL_last_uni++;
1893     s = PL_last_uni;
1894     while (isWORDCHAR_lazy_if(s,UTF) || *s == '-')
1895 	s++;
1896     if ((t = strchr(s, '(')) && t < PL_bufptr)
1897 	return;
1898 
1899     Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
1900 		     "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1901 		     (int)(s - PL_last_uni), PL_last_uni);
1902 }
1903 
1904 /*
1905  * LOP : macro to build a list operator.  Its behaviour has been replaced
1906  * with a subroutine, S_lop() for which LOP is just another name.
1907  */
1908 
1909 #define LOP(f,x) return lop(f,x,s)
1910 
1911 /*
1912  * S_lop
1913  * Build a list operator (or something that might be one).  The rules:
1914  *  - if we have a next token, then it's a list operator [why?]
1915  *  - if the next thing is an opening paren, then it's a function
1916  *  - else it's a list operator
1917  */
1918 
1919 STATIC I32
1920 S_lop(pTHX_ I32 f, int x, char *s)
1921 {
1922     dVAR;
1923 
1924     PERL_ARGS_ASSERT_LOP;
1925 
1926     pl_yylval.ival = f;
1927     CLINE;
1928     PL_expect = x;
1929     PL_bufptr = s;
1930     PL_last_lop = PL_oldbufptr;
1931     PL_last_lop_op = (OPCODE)f;
1932 #ifdef PERL_MAD
1933     if (PL_lasttoke)
1934 	goto lstop;
1935 #else
1936     if (PL_nexttoke)
1937 	goto lstop;
1938 #endif
1939     if (*s == '(')
1940 	return REPORT(FUNC);
1941     s = PEEKSPACE(s);
1942     if (*s == '(')
1943 	return REPORT(FUNC);
1944     else {
1945 	lstop:
1946 	if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
1947 	    PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
1948 	return REPORT(LSTOP);
1949     }
1950 }
1951 
1952 #ifdef PERL_MAD
1953  /*
1954  * S_start_force
1955  * Sets up for an eventual force_next().  start_force(0) basically does
1956  * an unshift, while start_force(-1) does a push.  yylex removes items
1957  * on the "pop" end.
1958  */
1959 
1960 STATIC void
1961 S_start_force(pTHX_ int where)
1962 {
1963     int i;
1964 
1965     if (where < 0)	/* so people can duplicate start_force(PL_curforce) */
1966 	where = PL_lasttoke;
1967     assert(PL_curforce < 0 || PL_curforce == where);
1968     if (PL_curforce != where) {
1969 	for (i = PL_lasttoke; i > where; --i) {
1970 	    PL_nexttoke[i] = PL_nexttoke[i-1];
1971 	}
1972 	PL_lasttoke++;
1973     }
1974     if (PL_curforce < 0)	/* in case of duplicate start_force() */
1975 	Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1976     PL_curforce = where;
1977     if (PL_nextwhite) {
1978 	if (PL_madskills)
1979 	    curmad('^', newSVpvs(""));
1980 	CURMAD('_', PL_nextwhite);
1981     }
1982 }
1983 
1984 STATIC void
1985 S_curmad(pTHX_ char slot, SV *sv)
1986 {
1987     MADPROP **where;
1988 
1989     if (!sv)
1990 	return;
1991     if (PL_curforce < 0)
1992 	where = &PL_thismad;
1993     else
1994 	where = &PL_nexttoke[PL_curforce].next_mad;
1995 
1996     if (PL_faketokens)
1997 	sv_setpvs(sv, "");
1998     else {
1999 	if (!IN_BYTES) {
2000 	    if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2001 		SvUTF8_on(sv);
2002 	    else if (PL_encoding) {
2003 		sv_recode_to_utf8(sv, PL_encoding);
2004 	    }
2005 	}
2006     }
2007 
2008     /* keep a slot open for the head of the list? */
2009     if (slot != '_' && *where && (*where)->mad_key == '^') {
2010 	(*where)->mad_key = slot;
2011 	sv_free(MUTABLE_SV(((*where)->mad_val)));
2012 	(*where)->mad_val = (void*)sv;
2013     }
2014     else
2015 	addmad(newMADsv(slot, sv), where, 0);
2016 }
2017 #else
2018 #  define start_force(where)    NOOP
2019 #  define curmad(slot, sv)      NOOP
2020 #endif
2021 
2022 /*
2023  * S_force_next
2024  * When the lexer realizes it knows the next token (for instance,
2025  * it is reordering tokens for the parser) then it can call S_force_next
2026  * to know what token to return the next time the lexer is called.  Caller
2027  * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
2028  * and possibly PL_expect to ensure the lexer handles the token correctly.
2029  */
2030 
2031 STATIC void
2032 S_force_next(pTHX_ I32 type)
2033 {
2034     dVAR;
2035 #ifdef DEBUGGING
2036     if (DEBUG_T_TEST) {
2037         PerlIO_printf(Perl_debug_log, "### forced token:\n");
2038 	tokereport(type, &NEXTVAL_NEXTTOKE);
2039     }
2040 #endif
2041 #ifdef PERL_MAD
2042     if (PL_curforce < 0)
2043 	start_force(PL_lasttoke);
2044     PL_nexttoke[PL_curforce].next_type = type;
2045     if (PL_lex_state != LEX_KNOWNEXT)
2046  	PL_lex_defer = PL_lex_state;
2047     PL_lex_state = LEX_KNOWNEXT;
2048     PL_lex_expect = PL_expect;
2049     PL_curforce = -1;
2050 #else
2051     PL_nexttype[PL_nexttoke] = type;
2052     PL_nexttoke++;
2053     if (PL_lex_state != LEX_KNOWNEXT) {
2054 	PL_lex_defer = PL_lex_state;
2055 	PL_lex_expect = PL_expect;
2056 	PL_lex_state = LEX_KNOWNEXT;
2057     }
2058 #endif
2059 }
2060 
2061 void
2062 Perl_yyunlex(pTHX)
2063 {
2064     int yyc = PL_parser->yychar;
2065     if (yyc != YYEMPTY) {
2066 	if (yyc) {
2067 	    start_force(-1);
2068 	    NEXTVAL_NEXTTOKE = PL_parser->yylval;
2069 	    if (yyc == '{'/*}*/ || yyc == HASHBRACK || yyc == '['/*]*/) {
2070 		PL_lex_allbrackets--;
2071 		PL_lex_brackets--;
2072 		yyc |= (3<<24) | (PL_lex_brackstack[PL_lex_brackets] << 16);
2073 	    } else if (yyc == '('/*)*/) {
2074 		PL_lex_allbrackets--;
2075 		yyc |= (2<<24);
2076 	    }
2077 	    force_next(yyc);
2078 	}
2079 	PL_parser->yychar = YYEMPTY;
2080     }
2081 }
2082 
2083 STATIC SV *
2084 S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
2085 {
2086     dVAR;
2087     SV * const sv = newSVpvn_utf8(start, len,
2088 				  !IN_BYTES
2089 				  && UTF
2090 				  && !is_ascii_string((const U8*)start, len)
2091 				  && is_utf8_string((const U8*)start, len));
2092     return sv;
2093 }
2094 
2095 /*
2096  * S_force_word
2097  * When the lexer knows the next thing is a word (for instance, it has
2098  * just seen -> and it knows that the next char is a word char, then
2099  * it calls S_force_word to stick the next word into the PL_nexttoke/val
2100  * lookahead.
2101  *
2102  * Arguments:
2103  *   char *start : buffer position (must be within PL_linestr)
2104  *   int token   : PL_next* will be this type of bare word (e.g., METHOD,WORD)
2105  *   int check_keyword : if true, Perl checks to make sure the word isn't
2106  *       a keyword (do this if the word is a label, e.g. goto FOO)
2107  *   int allow_pack : if true, : characters will also be allowed (require,
2108  *       use, etc. do this)
2109  *   int allow_initial_tick : used by the "sub" lexer only.
2110  */
2111 
2112 STATIC char *
2113 S_force_word(pTHX_ char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
2114 {
2115     dVAR;
2116     char *s;
2117     STRLEN len;
2118 
2119     PERL_ARGS_ASSERT_FORCE_WORD;
2120 
2121     start = SKIPSPACE1(start);
2122     s = start;
2123     if (isIDFIRST_lazy_if(s,UTF) ||
2124 	(allow_pack && *s == ':') ||
2125 	(allow_initial_tick && *s == '\'') )
2126     {
2127 	s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
2128 	if (check_keyword && keyword(PL_tokenbuf, len, 0))
2129 	    return start;
2130 	start_force(PL_curforce);
2131 	if (PL_madskills)
2132 	    curmad('X', newSVpvn(start,s-start));
2133 	if (token == METHOD) {
2134 	    s = SKIPSPACE1(s);
2135 	    if (*s == '(')
2136 		PL_expect = XTERM;
2137 	    else {
2138 		PL_expect = XOPERATOR;
2139 	    }
2140 	}
2141 	if (PL_madskills)
2142 	    curmad('g', newSVpvs( "forced" ));
2143 	NEXTVAL_NEXTTOKE.opval
2144 	    = (OP*)newSVOP(OP_CONST,0,
2145 			   S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
2146 	NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
2147 	force_next(token);
2148     }
2149     return s;
2150 }
2151 
2152 /*
2153  * S_force_ident
2154  * Called when the lexer wants $foo *foo &foo etc, but the program
2155  * text only contains the "foo" portion.  The first argument is a pointer
2156  * to the "foo", and the second argument is the type symbol to prefix.
2157  * Forces the next token to be a "WORD".
2158  * Creates the symbol if it didn't already exist (via gv_fetchpv()).
2159  */
2160 
2161 STATIC void
2162 S_force_ident(pTHX_ const char *s, int kind)
2163 {
2164     dVAR;
2165 
2166     PERL_ARGS_ASSERT_FORCE_IDENT;
2167 
2168     if (s[0]) {
2169 	const STRLEN len = s[1] ? strlen(s) : 1; /* s = "\"" see yylex */
2170 	OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn_flags(s, len,
2171                                                                 UTF ? SVf_UTF8 : 0));
2172 	start_force(PL_curforce);
2173 	NEXTVAL_NEXTTOKE.opval = o;
2174 	force_next(WORD);
2175 	if (kind) {
2176 	    o->op_private = OPpCONST_ENTERED;
2177 	    /* XXX see note in pp_entereval() for why we forgo typo
2178 	       warnings if the symbol must be introduced in an eval.
2179 	       GSAR 96-10-12 */
2180 	    gv_fetchpvn_flags(s, len,
2181 			      (PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
2182 			      : GV_ADD) | ( UTF ? SVf_UTF8 : 0 ),
2183 			      kind == '$' ? SVt_PV :
2184 			      kind == '@' ? SVt_PVAV :
2185 			      kind == '%' ? SVt_PVHV :
2186 			      SVt_PVGV
2187 			      );
2188 	}
2189     }
2190 }
2191 
2192 static void
2193 S_force_ident_maybe_lex(pTHX_ char pit)
2194 {
2195     start_force(PL_curforce);
2196     NEXTVAL_NEXTTOKE.ival = pit;
2197     force_next('p');
2198 }
2199 
2200 NV
2201 Perl_str_to_version(pTHX_ SV *sv)
2202 {
2203     NV retval = 0.0;
2204     NV nshift = 1.0;
2205     STRLEN len;
2206     const char *start = SvPV_const(sv,len);
2207     const char * const end = start + len;
2208     const bool utf = SvUTF8(sv) ? TRUE : FALSE;
2209 
2210     PERL_ARGS_ASSERT_STR_TO_VERSION;
2211 
2212     while (start < end) {
2213 	STRLEN skip;
2214 	UV n;
2215 	if (utf)
2216 	    n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
2217 	else {
2218 	    n = *(U8*)start;
2219 	    skip = 1;
2220 	}
2221 	retval += ((NV)n)/nshift;
2222 	start += skip;
2223 	nshift *= 1000;
2224     }
2225     return retval;
2226 }
2227 
2228 /*
2229  * S_force_version
2230  * Forces the next token to be a version number.
2231  * If the next token appears to be an invalid version number, (e.g. "v2b"),
2232  * and if "guessing" is TRUE, then no new token is created (and the caller
2233  * must use an alternative parsing method).
2234  */
2235 
2236 STATIC char *
2237 S_force_version(pTHX_ char *s, int guessing)
2238 {
2239     dVAR;
2240     OP *version = NULL;
2241     char *d;
2242 #ifdef PERL_MAD
2243     I32 startoff = s - SvPVX(PL_linestr);
2244 #endif
2245 
2246     PERL_ARGS_ASSERT_FORCE_VERSION;
2247 
2248     s = SKIPSPACE1(s);
2249 
2250     d = s;
2251     if (*d == 'v')
2252 	d++;
2253     if (isDIGIT(*d)) {
2254 	while (isDIGIT(*d) || *d == '_' || *d == '.')
2255 	    d++;
2256 #ifdef PERL_MAD
2257 	if (PL_madskills) {
2258 	    start_force(PL_curforce);
2259 	    curmad('X', newSVpvn(s,d-s));
2260 	}
2261 #endif
2262         if (*d == ';' || isSPACE(*d) || *d == '{' || *d == '}' || !*d) {
2263 	    SV *ver;
2264 #ifdef USE_LOCALE_NUMERIC
2265 	    char *loc = savepv(setlocale(LC_NUMERIC, NULL));
2266 	    setlocale(LC_NUMERIC, "C");
2267 #endif
2268             s = scan_num(s, &pl_yylval);
2269 #ifdef USE_LOCALE_NUMERIC
2270 	    setlocale(LC_NUMERIC, loc);
2271 	    Safefree(loc);
2272 #endif
2273             version = pl_yylval.opval;
2274 	    ver = cSVOPx(version)->op_sv;
2275 	    if (SvPOK(ver) && !SvNIOK(ver)) {
2276 		SvUPGRADE(ver, SVt_PVNV);
2277 		SvNV_set(ver, str_to_version(ver));
2278 		SvNOK_on(ver);		/* hint that it is a version */
2279 	    }
2280         }
2281 	else if (guessing) {
2282 #ifdef PERL_MAD
2283 	    if (PL_madskills) {
2284 		sv_free(PL_nextwhite);	/* let next token collect whitespace */
2285 		PL_nextwhite = 0;
2286 		s = SvPVX(PL_linestr) + startoff;
2287 	    }
2288 #endif
2289 	    return s;
2290 	}
2291     }
2292 
2293 #ifdef PERL_MAD
2294     if (PL_madskills && !version) {
2295 	sv_free(PL_nextwhite);	/* let next token collect whitespace */
2296 	PL_nextwhite = 0;
2297 	s = SvPVX(PL_linestr) + startoff;
2298     }
2299 #endif
2300     /* NOTE: The parser sees the package name and the VERSION swapped */
2301     start_force(PL_curforce);
2302     NEXTVAL_NEXTTOKE.opval = version;
2303     force_next(WORD);
2304 
2305     return s;
2306 }
2307 
2308 /*
2309  * S_force_strict_version
2310  * Forces the next token to be a version number using strict syntax rules.
2311  */
2312 
2313 STATIC char *
2314 S_force_strict_version(pTHX_ char *s)
2315 {
2316     dVAR;
2317     OP *version = NULL;
2318 #ifdef PERL_MAD
2319     I32 startoff = s - SvPVX(PL_linestr);
2320 #endif
2321     const char *errstr = NULL;
2322 
2323     PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2324 
2325     while (isSPACE(*s)) /* leading whitespace */
2326 	s++;
2327 
2328     if (is_STRICT_VERSION(s,&errstr)) {
2329 	SV *ver = newSV(0);
2330 	s = (char *)scan_version(s, ver, 0);
2331 	version = newSVOP(OP_CONST, 0, ver);
2332     }
2333     else if ( (*s != ';' && *s != '{' && *s != '}' ) &&
2334 	    (s = SKIPSPACE1(s), (*s != ';' && *s != '{' && *s != '}' )))
2335     {
2336 	PL_bufptr = s;
2337 	if (errstr)
2338 	    yyerror(errstr); /* version required */
2339 	return s;
2340     }
2341 
2342 #ifdef PERL_MAD
2343     if (PL_madskills && !version) {
2344 	sv_free(PL_nextwhite);	/* let next token collect whitespace */
2345 	PL_nextwhite = 0;
2346 	s = SvPVX(PL_linestr) + startoff;
2347     }
2348 #endif
2349     /* NOTE: The parser sees the package name and the VERSION swapped */
2350     start_force(PL_curforce);
2351     NEXTVAL_NEXTTOKE.opval = version;
2352     force_next(WORD);
2353 
2354     return s;
2355 }
2356 
2357 /*
2358  * S_tokeq
2359  * Tokenize a quoted string passed in as an SV.  It finds the next
2360  * chunk, up to end of string or a backslash.  It may make a new
2361  * SV containing that chunk (if HINT_NEW_STRING is on).  It also
2362  * turns \\ into \.
2363  */
2364 
2365 STATIC SV *
2366 S_tokeq(pTHX_ SV *sv)
2367 {
2368     dVAR;
2369     char *s;
2370     char *send;
2371     char *d;
2372     STRLEN len = 0;
2373     SV *pv = sv;
2374 
2375     PERL_ARGS_ASSERT_TOKEQ;
2376 
2377     if (!SvLEN(sv))
2378 	goto finish;
2379 
2380     s = SvPV_force(sv, len);
2381     if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
2382 	goto finish;
2383     send = s + len;
2384     /* This is relying on the SV being "well formed" with a trailing '\0'  */
2385     while (s < send && !(*s == '\\' && s[1] == '\\'))
2386 	s++;
2387     if (s == send)
2388 	goto finish;
2389     d = s;
2390     if ( PL_hints & HINT_NEW_STRING ) {
2391 	pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
2392     }
2393     while (s < send) {
2394 	if (*s == '\\') {
2395 	    if (s + 1 < send && (s[1] == '\\'))
2396 		s++;		/* all that, just for this */
2397 	}
2398 	*d++ = *s++;
2399     }
2400     *d = '\0';
2401     SvCUR_set(sv, d - SvPVX_const(sv));
2402   finish:
2403     if ( PL_hints & HINT_NEW_STRING )
2404        return new_constant(NULL, 0, "q", sv, pv, "q", 1);
2405     return sv;
2406 }
2407 
2408 /*
2409  * Now come three functions related to double-quote context,
2410  * S_sublex_start, S_sublex_push, and S_sublex_done.  They're used when
2411  * converting things like "\u\Lgnat" into ucfirst(lc("gnat")).  They
2412  * interact with PL_lex_state, and create fake ( ... ) argument lists
2413  * to handle functions and concatenation.
2414  * For example,
2415  *   "foo\lbar"
2416  * is tokenised as
2417  *    stringify ( const[foo] concat lcfirst ( const[bar] ) )
2418  */
2419 
2420 /*
2421  * S_sublex_start
2422  * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
2423  *
2424  * Pattern matching will set PL_lex_op to the pattern-matching op to
2425  * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
2426  *
2427  * OP_CONST and OP_READLINE are easy--just make the new op and return.
2428  *
2429  * Everything else becomes a FUNC.
2430  *
2431  * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2432  * had an OP_CONST or OP_READLINE).  This just sets us up for a
2433  * call to S_sublex_push().
2434  */
2435 
2436 STATIC I32
2437 S_sublex_start(pTHX)
2438 {
2439     dVAR;
2440     const I32 op_type = pl_yylval.ival;
2441 
2442     if (op_type == OP_NULL) {
2443 	pl_yylval.opval = PL_lex_op;
2444 	PL_lex_op = NULL;
2445 	return THING;
2446     }
2447     if (op_type == OP_CONST || op_type == OP_READLINE) {
2448 	SV *sv = tokeq(PL_lex_stuff);
2449 
2450 	if (SvTYPE(sv) == SVt_PVIV) {
2451 	    /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2452 	    STRLEN len;
2453 	    const char * const p = SvPV_const(sv, len);
2454 	    SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
2455 	    SvREFCNT_dec(sv);
2456 	    sv = nsv;
2457 	}
2458 	pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
2459 	PL_lex_stuff = NULL;
2460 	/* Allow <FH> // "foo" */
2461 	if (op_type == OP_READLINE)
2462 	    PL_expect = XTERMORDORDOR;
2463 	return THING;
2464     }
2465     else if (op_type == OP_BACKTICK && PL_lex_op) {
2466 	/* readpipe() vas overriden */
2467 	cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
2468 	pl_yylval.opval = PL_lex_op;
2469 	PL_lex_op = NULL;
2470 	PL_lex_stuff = NULL;
2471 	return THING;
2472     }
2473 
2474     PL_sublex_info.super_state = PL_lex_state;
2475     PL_sublex_info.sub_inwhat = (U16)op_type;
2476     PL_sublex_info.sub_op = PL_lex_op;
2477     PL_lex_state = LEX_INTERPPUSH;
2478 
2479     PL_expect = XTERM;
2480     if (PL_lex_op) {
2481 	pl_yylval.opval = PL_lex_op;
2482 	PL_lex_op = NULL;
2483 	return PMFUNC;
2484     }
2485     else
2486 	return FUNC;
2487 }
2488 
2489 /*
2490  * S_sublex_push
2491  * Create a new scope to save the lexing state.  The scope will be
2492  * ended in S_sublex_done.  Returns a '(', starting the function arguments
2493  * to the uc, lc, etc. found before.
2494  * Sets PL_lex_state to LEX_INTERPCONCAT.
2495  */
2496 
2497 STATIC I32
2498 S_sublex_push(pTHX)
2499 {
2500     dVAR;
2501     LEXSHARED *shared;
2502     ENTER;
2503 
2504     PL_lex_state = PL_sublex_info.super_state;
2505     SAVEBOOL(PL_lex_dojoin);
2506     SAVEI32(PL_lex_brackets);
2507     SAVEI32(PL_lex_allbrackets);
2508     SAVEI32(PL_lex_formbrack);
2509     SAVEI8(PL_lex_fakeeof);
2510     SAVEI32(PL_lex_casemods);
2511     SAVEI32(PL_lex_starts);
2512     SAVEI8(PL_lex_state);
2513     SAVESPTR(PL_lex_repl);
2514     SAVEVPTR(PL_lex_inpat);
2515     SAVEI16(PL_lex_inwhat);
2516     SAVECOPLINE(PL_curcop);
2517     SAVEPPTR(PL_bufptr);
2518     SAVEPPTR(PL_bufend);
2519     SAVEPPTR(PL_oldbufptr);
2520     SAVEPPTR(PL_oldoldbufptr);
2521     SAVEPPTR(PL_last_lop);
2522     SAVEPPTR(PL_last_uni);
2523     SAVEPPTR(PL_linestart);
2524     SAVESPTR(PL_linestr);
2525     SAVEGENERICPV(PL_lex_brackstack);
2526     SAVEGENERICPV(PL_lex_casestack);
2527     SAVEGENERICPV(PL_parser->lex_shared);
2528     SAVEBOOL(PL_parser->lex_re_reparsing);
2529 
2530     /* The here-doc parser needs to be able to peek into outer lexing
2531        scopes to find the body of the here-doc.  So we put PL_linestr and
2532        PL_bufptr into lex_shared, to ‘share’ those values.
2533      */
2534     PL_parser->lex_shared->ls_linestr = PL_linestr;
2535     PL_parser->lex_shared->ls_bufptr  = PL_bufptr;
2536 
2537     PL_linestr = PL_lex_stuff;
2538     PL_lex_repl = PL_sublex_info.repl;
2539     PL_lex_stuff = NULL;
2540     PL_sublex_info.repl = NULL;
2541 
2542     PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2543 	= SvPVX(PL_linestr);
2544     PL_bufend += SvCUR(PL_linestr);
2545     PL_last_lop = PL_last_uni = NULL;
2546     SAVEFREESV(PL_linestr);
2547     if (PL_lex_repl) SAVEFREESV(PL_lex_repl);
2548 
2549     PL_lex_dojoin = FALSE;
2550     PL_lex_brackets = PL_lex_formbrack = 0;
2551     PL_lex_allbrackets = 0;
2552     PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
2553     Newx(PL_lex_brackstack, 120, char);
2554     Newx(PL_lex_casestack, 12, char);
2555     PL_lex_casemods = 0;
2556     *PL_lex_casestack = '\0';
2557     PL_lex_starts = 0;
2558     PL_lex_state = LEX_INTERPCONCAT;
2559     CopLINE_set(PL_curcop, (line_t)PL_multi_start);
2560 
2561     Newxz(shared, 1, LEXSHARED);
2562     shared->ls_prev = PL_parser->lex_shared;
2563     PL_parser->lex_shared = shared;
2564 
2565     PL_lex_inwhat = PL_sublex_info.sub_inwhat;
2566     if (PL_lex_inwhat == OP_TRANSR) PL_lex_inwhat = OP_TRANS;
2567     if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2568 	PL_lex_inpat = PL_sublex_info.sub_op;
2569     else
2570 	PL_lex_inpat = NULL;
2571 
2572     PL_parser->lex_re_reparsing = cBOOL(PL_in_eval & EVAL_RE_REPARSING);
2573     PL_in_eval &= ~EVAL_RE_REPARSING;
2574 
2575     return '(';
2576 }
2577 
2578 /*
2579  * S_sublex_done
2580  * Restores lexer state after a S_sublex_push.
2581  */
2582 
2583 STATIC I32
2584 S_sublex_done(pTHX)
2585 {
2586     dVAR;
2587     if (!PL_lex_starts++) {
2588 	SV * const sv = newSVpvs("");
2589 	if (SvUTF8(PL_linestr))
2590 	    SvUTF8_on(sv);
2591 	PL_expect = XOPERATOR;
2592 	pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2593 	return THING;
2594     }
2595 
2596     if (PL_lex_casemods) {		/* oops, we've got some unbalanced parens */
2597 	PL_lex_state = LEX_INTERPCASEMOD;
2598 	return yylex();
2599     }
2600 
2601     /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
2602     assert(PL_lex_inwhat != OP_TRANSR);
2603     if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2604 	PL_linestr = PL_lex_repl;
2605 	PL_lex_inpat = 0;
2606 	PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2607 	PL_bufend += SvCUR(PL_linestr);
2608 	PL_last_lop = PL_last_uni = NULL;
2609 	PL_lex_dojoin = FALSE;
2610 	PL_lex_brackets = 0;
2611 	PL_lex_allbrackets = 0;
2612 	PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
2613 	PL_lex_casemods = 0;
2614 	*PL_lex_casestack = '\0';
2615 	PL_lex_starts = 0;
2616 	if (SvEVALED(PL_lex_repl)) {
2617 	    PL_lex_state = LEX_INTERPNORMAL;
2618 	    PL_lex_starts++;
2619 	    /*	we don't clear PL_lex_repl here, so that we can check later
2620 		whether this is an evalled subst; that means we rely on the
2621 		logic to ensure sublex_done() is called again only via the
2622 		branch (in yylex()) that clears PL_lex_repl, else we'll loop */
2623 	}
2624 	else {
2625 	    PL_lex_state = LEX_INTERPCONCAT;
2626 	    PL_lex_repl = NULL;
2627 	}
2628 	return ',';
2629     }
2630     else {
2631 #ifdef PERL_MAD
2632 	if (PL_madskills) {
2633 	    if (PL_thiswhite) {
2634 		if (!PL_endwhite)
2635 		    PL_endwhite = newSVpvs("");
2636 		sv_catsv(PL_endwhite, PL_thiswhite);
2637 		PL_thiswhite = 0;
2638 	    }
2639 	    if (PL_thistoken)
2640 		sv_setpvs(PL_thistoken,"");
2641 	    else
2642 		PL_realtokenstart = -1;
2643 	}
2644 #endif
2645 	LEAVE;
2646 	PL_bufend = SvPVX(PL_linestr);
2647 	PL_bufend += SvCUR(PL_linestr);
2648 	PL_expect = XOPERATOR;
2649 	PL_sublex_info.sub_inwhat = 0;
2650 	return ')';
2651     }
2652 }
2653 
2654 PERL_STATIC_INLINE SV*
2655 S_get_and_check_backslash_N_name(pTHX_ const char* s, const char* const e)
2656 {
2657     /* <s> points to first character of interior of \N{}, <e> to one beyond the
2658      * interior, hence to the "}".  Finds what the name resolves to, returning
2659      * an SV* containing it; NULL if no valid one found */
2660 
2661     SV* res = newSVpvn_flags(s, e - s, UTF ? SVf_UTF8 : 0);
2662 
2663     HV * table;
2664     SV **cvp;
2665     SV *cv;
2666     SV *rv;
2667     HV *stash;
2668     const U8* first_bad_char_loc;
2669     const char* backslash_ptr = s - 3; /* Points to the <\> of \N{... */
2670 
2671     PERL_ARGS_ASSERT_GET_AND_CHECK_BACKSLASH_N_NAME;
2672 
2673     if (UTF && ! is_utf8_string_loc((U8 *) backslash_ptr,
2674                                      e - backslash_ptr,
2675                                      &first_bad_char_loc))
2676     {
2677         /* If warnings are on, this will print a more detailed analysis of what
2678          * is wrong than the error message below */
2679         utf8n_to_uvuni(first_bad_char_loc,
2680                        e - ((char *) first_bad_char_loc),
2681                        NULL, 0);
2682 
2683         /* We deliberately don't try to print the malformed character, which
2684          * might not print very well; it also may be just the first of many
2685          * malformations, so don't print what comes after it */
2686         yyerror(Perl_form(aTHX_
2687             "Malformed UTF-8 character immediately after '%.*s'",
2688             (int) (first_bad_char_loc - (U8 *) backslash_ptr), backslash_ptr));
2689 	return NULL;
2690     }
2691 
2692     res = new_constant( NULL, 0, "charnames", res, NULL, backslash_ptr,
2693                         /* include the <}> */
2694                         e - backslash_ptr + 1);
2695     if (! SvPOK(res)) {
2696         SvREFCNT_dec_NN(res);
2697         return NULL;
2698     }
2699 
2700     /* See if the charnames handler is the Perl core's, and if so, we can skip
2701      * the validation needed for a user-supplied one, as Perl's does its own
2702      * validation. */
2703     table = GvHV(PL_hintgv);		 /* ^H */
2704     cvp = hv_fetchs(table, "charnames", FALSE);
2705     if (cvp && (cv = *cvp) && SvROK(cv) && ((rv = SvRV(cv)) != NULL)
2706         && SvTYPE(rv) == SVt_PVCV && ((stash = CvSTASH(rv)) != NULL))
2707     {
2708         const char * const name = HvNAME(stash);
2709         if strEQ(name, "_charnames") {
2710            return res;
2711        }
2712     }
2713 
2714     /* Here, it isn't Perl's charname handler.  We can't rely on a
2715      * user-supplied handler to validate the input name.  For non-ut8 input,
2716      * look to see that the first character is legal.  Then loop through the
2717      * rest checking that each is a continuation */
2718 
2719     /* This code needs to be sync'ed with a regex in _charnames.pm which does
2720      * the same thing */
2721 
2722     if (! UTF) {
2723         if (! isALPHAU(*s)) {
2724             goto bad_charname;
2725         }
2726         s++;
2727         while (s < e) {
2728             if (! isCHARNAME_CONT(*s)) {
2729                 goto bad_charname;
2730             }
2731 	    if (*s == ' ' && *(s-1) == ' ' && ckWARN(WARN_DEPRECATED)) {
2732                 Perl_warn(aTHX_ "A sequence of multiple spaces in a charnames alias definition is deprecated");
2733             }
2734             s++;
2735         }
2736         if (*(s-1) == ' ' && ckWARN(WARN_DEPRECATED)) {
2737             Perl_warn(aTHX_ "Trailing white-space in a charnames alias definition is deprecated");
2738         }
2739     }
2740     else {
2741         /* Similarly for utf8.  For invariants can check directly; for other
2742          * Latin1, can calculate their code point and check; otherwise  use a
2743          * swash */
2744         if (UTF8_IS_INVARIANT(*s)) {
2745             if (! isALPHAU(*s)) {
2746                 goto bad_charname;
2747             }
2748             s++;
2749         } else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
2750             if (! isALPHAU(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*s, *(s+1))))) {
2751                 goto bad_charname;
2752             }
2753             s += 2;
2754         }
2755         else {
2756             if (! PL_utf8_charname_begin) {
2757                 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2758                 PL_utf8_charname_begin = _core_swash_init("utf8",
2759                                                         "_Perl_Charname_Begin",
2760                                                         &PL_sv_undef,
2761                                                         1, 0, NULL, &flags);
2762             }
2763             if (! swash_fetch(PL_utf8_charname_begin, (U8 *) s, TRUE)) {
2764                 goto bad_charname;
2765             }
2766             s += UTF8SKIP(s);
2767         }
2768 
2769         while (s < e) {
2770             if (UTF8_IS_INVARIANT(*s)) {
2771                 if (! isCHARNAME_CONT(*s)) {
2772                     goto bad_charname;
2773                 }
2774                 if (*s == ' ' && *(s-1) == ' ' && ckWARN(WARN_DEPRECATED)) {
2775                     Perl_warn(aTHX_ "A sequence of multiple spaces in a charnames alias definition is deprecated");
2776                 }
2777                 s++;
2778             }
2779             else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
2780                 if (! isCHARNAME_CONT(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*s,
2781                                                                     *(s+1)))))
2782                 {
2783                     goto bad_charname;
2784                 }
2785                 s += 2;
2786             }
2787             else {
2788                 if (! PL_utf8_charname_continue) {
2789                     U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2790                     PL_utf8_charname_continue = _core_swash_init("utf8",
2791                                                 "_Perl_Charname_Continue",
2792                                                 &PL_sv_undef,
2793                                                 1, 0, NULL, &flags);
2794                 }
2795                 if (! swash_fetch(PL_utf8_charname_continue, (U8 *) s, TRUE)) {
2796                     goto bad_charname;
2797                 }
2798                 s += UTF8SKIP(s);
2799             }
2800         }
2801         if (*(s-1) == ' ' && ckWARN(WARN_DEPRECATED)) {
2802             Perl_warn(aTHX_ "Trailing white-space in a charnames alias definition is deprecated");
2803         }
2804     }
2805 
2806     if (SvUTF8(res)) { /* Don't accept malformed input */
2807         const U8* first_bad_char_loc;
2808         STRLEN len;
2809         const char* const str = SvPV_const(res, len);
2810         if (! is_utf8_string_loc((U8 *) str, len, &first_bad_char_loc)) {
2811             /* If warnings are on, this will print a more detailed analysis of
2812              * what is wrong than the error message below */
2813             utf8n_to_uvuni(first_bad_char_loc,
2814                            (char *) first_bad_char_loc - str,
2815                            NULL, 0);
2816 
2817             /* We deliberately don't try to print the malformed character,
2818              * which might not print very well; it also may be just the first
2819              * of many malformations, so don't print what comes after it */
2820             yyerror_pv(
2821               Perl_form(aTHX_
2822                 "Malformed UTF-8 returned by %.*s immediately after '%.*s'",
2823                  (int) (e - backslash_ptr + 1), backslash_ptr,
2824                  (int) ((char *) first_bad_char_loc - str), str
2825               ),
2826               SVf_UTF8);
2827             return NULL;
2828         }
2829     }
2830 
2831     return res;
2832 
2833   bad_charname: {
2834         int bad_char_size = ((UTF) ? UTF8SKIP(s) : 1);
2835 
2836         /* The final %.*s makes sure that should the trailing NUL be missing
2837          * that this print won't run off the end of the string */
2838         yyerror_pv(
2839           Perl_form(aTHX_
2840             "Invalid character in \\N{...}; marked by <-- HERE in %.*s<-- HERE %.*s",
2841             (int)(s - backslash_ptr + bad_char_size), backslash_ptr,
2842             (int)(e - s + bad_char_size), s + bad_char_size
2843           ),
2844           UTF ? SVf_UTF8 : 0);
2845         return NULL;
2846     }
2847 }
2848 
2849 /*
2850   scan_const
2851 
2852   Extracts the next constant part of a pattern, double-quoted string,
2853   or transliteration.  This is terrifying code.
2854 
2855   For example, in parsing the double-quoted string "ab\x63$d", it would
2856   stop at the '$' and return an OP_CONST containing 'abc'.
2857 
2858   It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
2859   processing a pattern (PL_lex_inpat is true), a transliteration
2860   (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
2861 
2862   Returns a pointer to the character scanned up to. If this is
2863   advanced from the start pointer supplied (i.e. if anything was
2864   successfully parsed), will leave an OP_CONST for the substring scanned
2865   in pl_yylval. Caller must intuit reason for not parsing further
2866   by looking at the next characters herself.
2867 
2868   In patterns:
2869     expand:
2870       \N{FOO}  => \N{U+hex_for_character_FOO}
2871       (if FOO expands to multiple characters, expands to \N{U+xx.XX.yy ...})
2872 
2873     pass through:
2874 	all other \-char, including \N and \N{ apart from \N{ABC}
2875 
2876     stops on:
2877 	@ and $ where it appears to be a var, but not for $ as tail anchor
2878         \l \L \u \U \Q \E
2879 	(?{  or  (??{
2880 
2881 
2882   In transliterations:
2883     characters are VERY literal, except for - not at the start or end
2884     of the string, which indicates a range. If the range is in bytes,
2885     scan_const expands the range to the full set of intermediate
2886     characters. If the range is in utf8, the hyphen is replaced with
2887     a certain range mark which will be handled by pmtrans() in op.c.
2888 
2889   In double-quoted strings:
2890     backslashes:
2891       double-quoted style: \r and \n
2892       constants: \x31, etc.
2893       deprecated backrefs: \1 (in substitution replacements)
2894       case and quoting: \U \Q \E
2895     stops on @ and $
2896 
2897   scan_const does *not* construct ops to handle interpolated strings.
2898   It stops processing as soon as it finds an embedded $ or @ variable
2899   and leaves it to the caller to work out what's going on.
2900 
2901   embedded arrays (whether in pattern or not) could be:
2902       @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
2903 
2904   $ in double-quoted strings must be the symbol of an embedded scalar.
2905 
2906   $ in pattern could be $foo or could be tail anchor.  Assumption:
2907   it's a tail anchor if $ is the last thing in the string, or if it's
2908   followed by one of "()| \r\n\t"
2909 
2910   \1 (backreferences) are turned into $1 in substitutions
2911 
2912   The structure of the code is
2913       while (there's a character to process) {
2914 	  handle transliteration ranges
2915 	  skip regexp comments /(?#comment)/ and codes /(?{code})/
2916 	  skip #-initiated comments in //x patterns
2917 	  check for embedded arrays
2918 	  check for embedded scalars
2919 	  if (backslash) {
2920 	      deprecate \1 in substitution replacements
2921 	      handle string-changing backslashes \l \U \Q \E, etc.
2922 	      switch (what was escaped) {
2923 		  handle \- in a transliteration (becomes a literal -)
2924 		  if a pattern and not \N{, go treat as regular character
2925 		  handle \132 (octal characters)
2926 		  handle \x15 and \x{1234} (hex characters)
2927 		  handle \N{name} (named characters, also \N{3,5} in a pattern)
2928 		  handle \cV (control characters)
2929 		  handle printf-style backslashes (\f, \r, \n, etc)
2930 	      } (end switch)
2931 	      continue
2932 	  } (end if backslash)
2933           handle regular character
2934     } (end while character to read)
2935 
2936 */
2937 
2938 STATIC char *
2939 S_scan_const(pTHX_ char *start)
2940 {
2941     dVAR;
2942     char *send = PL_bufend;		/* end of the constant */
2943     SV *sv = newSV(send - start);		/* sv for the constant.  See
2944 						   note below on sizing. */
2945     char *s = start;			/* start of the constant */
2946     char *d = SvPVX(sv);		/* destination for copies */
2947     bool dorange = FALSE;			/* are we in a translit range? */
2948     bool didrange = FALSE;		        /* did we just finish a range? */
2949     bool in_charclass = FALSE;			/* within /[...]/ */
2950     bool has_utf8 = FALSE;			/* Output constant is UTF8 */
2951     bool  this_utf8 = cBOOL(UTF);		/* Is the source string assumed
2952 						   to be UTF8?  But, this can
2953 						   show as true when the source
2954 						   isn't utf8, as for example
2955 						   when it is entirely composed
2956 						   of hex constants */
2957     SV *res;		                /* result from charnames */
2958 
2959     /* Note on sizing:  The scanned constant is placed into sv, which is
2960      * initialized by newSV() assuming one byte of output for every byte of
2961      * input.  This routine expects newSV() to allocate an extra byte for a
2962      * trailing NUL, which this routine will append if it gets to the end of
2963      * the input.  There may be more bytes of input than output (eg., \N{LATIN
2964      * CAPITAL LETTER A}), or more output than input if the constant ends up
2965      * recoded to utf8, but each time a construct is found that might increase
2966      * the needed size, SvGROW() is called.  Its size parameter each time is
2967      * based on the best guess estimate at the time, namely the length used so
2968      * far, plus the length the current construct will occupy, plus room for
2969      * the trailing NUL, plus one byte for every input byte still unscanned */
2970 
2971     UV uv = UV_MAX; /* Initialize to weird value to try to catch any uses
2972                        before set */
2973 #ifdef EBCDIC
2974     UV literal_endpoint = 0;
2975     bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
2976 #endif
2977 
2978     PERL_ARGS_ASSERT_SCAN_CONST;
2979 
2980     assert(PL_lex_inwhat != OP_TRANSR);
2981     if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2982 	/* If we are doing a trans and we know we want UTF8 set expectation */
2983 	has_utf8   = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
2984 	this_utf8  = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2985     }
2986 
2987     /* Protect sv from errors and fatal warnings. */
2988     ENTER_with_name("scan_const");
2989     SAVEFREESV(sv);
2990 
2991     while (s < send || dorange) {
2992 
2993         /* get transliterations out of the way (they're most literal) */
2994 	if (PL_lex_inwhat == OP_TRANS) {
2995 	    /* expand a range A-Z to the full set of characters.  AIE! */
2996 	    if (dorange) {
2997 		I32 i;				/* current expanded character */
2998 		I32 min;			/* first character in range */
2999 		I32 max;			/* last character in range */
3000 
3001 #ifdef EBCDIC
3002 		UV uvmax = 0;
3003 #endif
3004 
3005 		if (has_utf8
3006 #ifdef EBCDIC
3007 		    && !native_range
3008 #endif
3009                 ) {
3010 		    char * const c = (char*)utf8_hop((U8*)d, -1);
3011 		    char *e = d++;
3012 		    while (e-- > c)
3013 			*(e + 1) = *e;
3014 		    *c = (char)UTF_TO_NATIVE(0xff);
3015 		    /* mark the range as done, and continue */
3016 		    dorange = FALSE;
3017 		    didrange = TRUE;
3018 		    continue;
3019 		}
3020 
3021 		i = d - SvPVX_const(sv);		/* remember current offset */
3022 #ifdef EBCDIC
3023                 SvGROW(sv,
3024 		       SvLEN(sv) + (has_utf8 ?
3025 				    (512 - UTF_CONTINUATION_MARK +
3026 				     UNISKIP(0x100))
3027 				    : 256));
3028                 /* How many two-byte within 0..255: 128 in UTF-8,
3029 		 * 96 in UTF-8-mod. */
3030 #else
3031 		SvGROW(sv, SvLEN(sv) + 256);	/* never more than 256 chars in a range */
3032 #endif
3033 		d = SvPVX(sv) + i;		/* refresh d after realloc */
3034 #ifdef EBCDIC
3035                 if (has_utf8) {
3036                     int j;
3037                     for (j = 0; j <= 1; j++) {
3038                         char * const c = (char*)utf8_hop((U8*)d, -1);
3039                         const UV uv    = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
3040                         if (j)
3041                             min = (U8)uv;
3042                         else if (uv < 256)
3043                             max = (U8)uv;
3044                         else {
3045                             max = (U8)0xff; /* only to \xff */
3046                             uvmax = uv; /* \x{100} to uvmax */
3047                         }
3048                         d = c; /* eat endpoint chars */
3049                      }
3050                 }
3051                else {
3052 #endif
3053 		   d -= 2;		/* eat the first char and the - */
3054 		   min = (U8)*d;	/* first char in range */
3055 		   max = (U8)d[1];	/* last char in range  */
3056 #ifdef EBCDIC
3057 	       }
3058 #endif
3059 
3060                 if (min > max) {
3061 		    Perl_croak(aTHX_
3062 			       "Invalid range \"%c-%c\" in transliteration operator",
3063 			       (char)min, (char)max);
3064                 }
3065 
3066 #ifdef EBCDIC
3067 		if (literal_endpoint == 2 &&
3068 		    ((isLOWER(min) && isLOWER(max)) ||
3069 		     (isUPPER(min) && isUPPER(max)))) {
3070 		    if (isLOWER(min)) {
3071 			for (i = min; i <= max; i++)
3072 			    if (isLOWER(i))
3073 				*d++ = NATIVE_TO_NEED(has_utf8,i);
3074 		    } else {
3075 			for (i = min; i <= max; i++)
3076 			    if (isUPPER(i))
3077 				*d++ = NATIVE_TO_NEED(has_utf8,i);
3078 		    }
3079 		}
3080 		else
3081 #endif
3082 		    for (i = min; i <= max; i++)
3083 #ifdef EBCDIC
3084                         if (has_utf8) {
3085                             const U8 ch = (U8)NATIVE_TO_UTF(i);
3086                             if (UNI_IS_INVARIANT(ch))
3087                                 *d++ = (U8)i;
3088                             else {
3089                                 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
3090                                 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
3091                             }
3092                         }
3093                         else
3094 #endif
3095                             *d++ = (char)i;
3096 
3097 #ifdef EBCDIC
3098                 if (uvmax) {
3099                     d = (char*)uvchr_to_utf8((U8*)d, 0x100);
3100                     if (uvmax > 0x101)
3101                         *d++ = (char)UTF_TO_NATIVE(0xff);
3102                     if (uvmax > 0x100)
3103                         d = (char*)uvchr_to_utf8((U8*)d, uvmax);
3104                 }
3105 #endif
3106 
3107 		/* mark the range as done, and continue */
3108 		dorange = FALSE;
3109 		didrange = TRUE;
3110 #ifdef EBCDIC
3111 		literal_endpoint = 0;
3112 #endif
3113 		continue;
3114 	    }
3115 
3116 	    /* range begins (ignore - as first or last char) */
3117 	    else if (*s == '-' && s+1 < send  && s != start) {
3118 		if (didrange) {
3119 		    Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
3120 		}
3121 		if (has_utf8
3122 #ifdef EBCDIC
3123 		    && !native_range
3124 #endif
3125 		    ) {
3126 		    *d++ = (char)UTF_TO_NATIVE(0xff);	/* use illegal utf8 byte--see pmtrans */
3127 		    s++;
3128 		    continue;
3129 		}
3130 		dorange = TRUE;
3131 		s++;
3132 	    }
3133 	    else {
3134 		didrange = FALSE;
3135 #ifdef EBCDIC
3136 		literal_endpoint = 0;
3137 		native_range = TRUE;
3138 #endif
3139 	    }
3140 	}
3141 
3142 	/* if we get here, we're not doing a transliteration */
3143 
3144 	else if (*s == '[' && PL_lex_inpat && !in_charclass) {
3145 	    char *s1 = s-1;
3146 	    int esc = 0;
3147 	    while (s1 >= start && *s1-- == '\\')
3148 		esc = !esc;
3149 	    if (!esc)
3150 		in_charclass = TRUE;
3151 	}
3152 
3153 	else if (*s == ']' && PL_lex_inpat &&  in_charclass) {
3154 	    char *s1 = s-1;
3155 	    int esc = 0;
3156 	    while (s1 >= start && *s1-- == '\\')
3157 		esc = !esc;
3158 	    if (!esc)
3159 		in_charclass = FALSE;
3160 	}
3161 
3162 	/* skip for regexp comments /(?#comment)/, except for the last
3163 	 * char, which will be done separately.
3164 	 * Stop on (?{..}) and friends */
3165 
3166 	else if (*s == '(' && PL_lex_inpat && s[1] == '?' && !in_charclass) {
3167 	    if (s[2] == '#') {
3168 		while (s+1 < send && *s != ')')
3169 		    *d++ = NATIVE_TO_NEED(has_utf8,*s++);
3170 	    }
3171 	    else if (!PL_lex_casemods &&
3172 		     (    s[2] == '{' /* This should match regcomp.c */
3173 		      || (s[2] == '?' && s[3] == '{')))
3174 	    {
3175 		break;
3176 	    }
3177 	}
3178 
3179 	/* likewise skip #-initiated comments in //x patterns */
3180 	else if (*s == '#' && PL_lex_inpat &&
3181 	  ((PMOP*)PL_lex_inpat)->op_pmflags & RXf_PMf_EXTENDED) {
3182 	    while (s+1 < send && *s != '\n') {
3183                 /* for maint-5.18, half-fix #-in-charclass bug:
3184                  *   *do* recognise codeblocks: /[#](?{})/
3185                  *   *don't* recognise interpolated vars: /[#$x]/
3186                  */
3187                 if (in_charclass && !PL_lex_casemods && s+3 < send &&
3188 		     s[0] == '(' &&
3189 		     s[1] == '?' &&
3190 		     (    s[2] == '{'
3191 		      || (s[2] == '?' && s[3] == '{')))
3192                     break;
3193 		*d++ = NATIVE_TO_NEED(has_utf8,*s++);
3194             }
3195             if (s+ 1 < send && *s != '\n')
3196                 break; /* we stopped on (?{}), not EOL */
3197 	}
3198 
3199 	/* no further processing of single-quoted regex */
3200 	else if (PL_lex_inpat && SvIVX(PL_linestr) == '\'')
3201 	    goto default_action;
3202 
3203 	/* check for embedded arrays
3204 	   (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
3205 	   */
3206 	else if (*s == '@' && s[1]) {
3207 	    if (isWORDCHAR_lazy_if(s+1,UTF))
3208 		break;
3209 	    if (strchr(":'{$", s[1]))
3210 		break;
3211 	    if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
3212 		break; /* in regexp, neither @+ nor @- are interpolated */
3213 	}
3214 
3215 	/* check for embedded scalars.  only stop if we're sure it's a
3216 	   variable.
3217         */
3218 	else if (*s == '$') {
3219 	    if (!PL_lex_inpat)	/* not a regexp, so $ must be var */
3220 		break;
3221 	    if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
3222 		if (s[1] == '\\') {
3223 		    Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3224 				   "Possible unintended interpolation of $\\ in regex");
3225 		}
3226 		break;		/* in regexp, $ might be tail anchor */
3227             }
3228 	}
3229 
3230 	/* End of else if chain - OP_TRANS rejoin rest */
3231 
3232 	/* backslashes */
3233 	if (*s == '\\' && s+1 < send) {
3234 	    char* e;	/* Can be used for ending '}', etc. */
3235 
3236 	    s++;
3237 
3238 	    /* warn on \1 - \9 in substitution replacements, but note that \11
3239 	     * is an octal; and \19 is \1 followed by '9' */
3240 	    if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
3241 		isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
3242 	    {
3243 		Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
3244 		*--s = '$';
3245 		break;
3246 	    }
3247 
3248 	    /* string-change backslash escapes */
3249 	    if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQF", *s)) {
3250 		--s;
3251 		break;
3252 	    }
3253 	    /* In a pattern, process \N, but skip any other backslash escapes.
3254 	     * This is because we don't want to translate an escape sequence
3255 	     * into a meta symbol and have the regex compiler use the meta
3256 	     * symbol meaning, e.g. \x{2E} would be confused with a dot.  But
3257 	     * in spite of this, we do have to process \N here while the proper
3258 	     * charnames handler is in scope.  See bugs #56444 and #62056.
3259 	     * There is a complication because \N in a pattern may also stand
3260 	     * for 'match a non-nl', and not mean a charname, in which case its
3261 	     * processing should be deferred to the regex compiler.  To be a
3262 	     * charname it must be followed immediately by a '{', and not look
3263 	     * like \N followed by a curly quantifier, i.e., not something like
3264 	     * \N{3,}.  regcurly returns a boolean indicating if it is a legal
3265 	     * quantifier */
3266 	    else if (PL_lex_inpat
3267 		    && (*s != 'N'
3268 			|| s[1] != '{'
3269 			|| regcurly(s + 1, FALSE)))
3270 	    {
3271 		*d++ = NATIVE_TO_NEED(has_utf8,'\\');
3272 		goto default_action;
3273 	    }
3274 
3275 	    switch (*s) {
3276 
3277 	    /* quoted - in transliterations */
3278 	    case '-':
3279 		if (PL_lex_inwhat == OP_TRANS) {
3280 		    *d++ = *s++;
3281 		    continue;
3282 		}
3283 		/* FALL THROUGH */
3284 	    default:
3285 	        {
3286 		    if ((isALPHANUMERIC(*s)))
3287 			Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
3288 				       "Unrecognized escape \\%c passed through",
3289 				       *s);
3290 		    /* default action is to copy the quoted character */
3291 		    goto default_action;
3292 		}
3293 
3294 	    /* eg. \132 indicates the octal constant 0132 */
3295 	    case '0': case '1': case '2': case '3':
3296 	    case '4': case '5': case '6': case '7':
3297 		{
3298                     I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
3299                     STRLEN len = 3;
3300 		    uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL));
3301 		    s += len;
3302                     if (len < 3 && s < send && isDIGIT(*s)
3303                         && ckWARN(WARN_MISC))
3304                     {
3305                         Perl_warner(aTHX_ packWARN(WARN_MISC),
3306                                     "%s", form_short_octal_warning(s, len));
3307                     }
3308 		}
3309 		goto NUM_ESCAPE_INSERT;
3310 
3311 	    /* eg. \o{24} indicates the octal constant \024 */
3312 	    case 'o':
3313 		{
3314 		    const char* error;
3315 
3316 		    bool valid = grok_bslash_o(&s, &uv, &error,
3317                                                TRUE, /* Output warning */
3318                                                FALSE, /* Not strict */
3319                                                TRUE, /* Output warnings for
3320                                                          non-portables */
3321                                                UTF);
3322 		    if (! valid) {
3323 			yyerror(error);
3324 			continue;
3325 		    }
3326 		    goto NUM_ESCAPE_INSERT;
3327 		}
3328 
3329 	    /* eg. \x24 indicates the hex constant 0x24 */
3330 	    case 'x':
3331 		{
3332 		    const char* error;
3333 
3334 		    bool valid = grok_bslash_x(&s, &uv, &error,
3335                                                TRUE, /* Output warning */
3336                                                FALSE, /* Not strict */
3337                                                TRUE,  /* Output warnings for
3338                                                          non-portables */
3339                                                UTF);
3340 		    if (! valid) {
3341 			yyerror(error);
3342 			continue;
3343 		    }
3344 		}
3345 
3346 	      NUM_ESCAPE_INSERT:
3347 		/* Insert oct or hex escaped character.  There will always be
3348 		 * enough room in sv since such escapes will be longer than any
3349 		 * UTF-8 sequence they can end up as, except if they force us
3350 		 * to recode the rest of the string into utf8 */
3351 
3352 		/* Here uv is the ordinal of the next character being added in
3353 		 * unicode (converted from native). */
3354 		if (!UNI_IS_INVARIANT(uv)) {
3355 		    if (!has_utf8 && uv > 255) {
3356 			/* Might need to recode whatever we have accumulated so
3357 			 * far if it contains any chars variant in utf8 or
3358 			 * utf-ebcdic. */
3359 
3360 			SvCUR_set(sv, d - SvPVX_const(sv));
3361 			SvPOK_on(sv);
3362 			*d = '\0';
3363 			/* See Note on sizing above.  */
3364 			sv_utf8_upgrade_flags_grow(sv,
3365 					SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3366 					UNISKIP(uv) + (STRLEN)(send - s) + 1);
3367 			d = SvPVX(sv) + SvCUR(sv);
3368 			has_utf8 = TRUE;
3369                     }
3370 
3371                     if (has_utf8) {
3372 		        d = (char*)uvuni_to_utf8((U8*)d, uv);
3373 			if (PL_lex_inwhat == OP_TRANS &&
3374 			    PL_sublex_info.sub_op) {
3375 			    PL_sublex_info.sub_op->op_private |=
3376 				(PL_lex_repl ? OPpTRANS_FROM_UTF
3377 					     : OPpTRANS_TO_UTF);
3378 			}
3379 #ifdef EBCDIC
3380 			if (uv > 255 && !dorange)
3381 			    native_range = FALSE;
3382 #endif
3383                     }
3384 		    else {
3385 		        *d++ = (char)uv;
3386 		    }
3387 		}
3388 		else {
3389 		    *d++ = (char) uv;
3390 		}
3391 		continue;
3392 
3393  	    case 'N':
3394 		/* In a non-pattern \N must be a named character, like \N{LATIN
3395 		 * SMALL LETTER A} or \N{U+0041}.  For patterns, it also can
3396 		 * mean to match a non-newline.  For non-patterns, named
3397 		 * characters are converted to their string equivalents. In
3398 		 * patterns, named characters are not converted to their
3399 		 * ultimate forms for the same reasons that other escapes
3400 		 * aren't.  Instead, they are converted to the \N{U+...} form
3401 		 * to get the value from the charnames that is in effect right
3402 		 * now, while preserving the fact that it was a named character
3403 		 * so that the regex compiler knows this */
3404 
3405 		/* This section of code doesn't generally use the
3406 		 * NATIVE_TO_NEED() macro to transform the input.  I (khw) did
3407 		 * a close examination of this macro and determined it is a
3408 		 * no-op except on utfebcdic variant characters.  Every
3409 		 * character generated by this that would normally need to be
3410 		 * enclosed by this macro is invariant, so the macro is not
3411 		 * needed, and would complicate use of copy().  XXX There are
3412 		 * other parts of this file where the macro is used
3413 		 * inconsistently, but are saved by it being a no-op */
3414 
3415 		/* The structure of this section of code (besides checking for
3416 		 * errors and upgrading to utf8) is:
3417 		 *  Further disambiguate between the two meanings of \N, and if
3418 		 *	not a charname, go process it elsewhere
3419 		 *  If of form \N{U+...}, pass it through if a pattern;
3420 		 *	otherwise convert to utf8
3421 		 *  Otherwise must be \N{NAME}: convert to \N{U+c1.c2...} if a
3422 		 *  pattern; otherwise convert to utf8 */
3423 
3424 		/* Here, s points to the 'N'; the test below is guaranteed to
3425 		 * succeed if we are being called on a pattern as we already
3426 		 * know from a test above that the next character is a '{'.
3427 		 * On a non-pattern \N must mean 'named sequence, which
3428 		 * requires braces */
3429 		s++;
3430 		if (*s != '{') {
3431 		    yyerror("Missing braces on \\N{}");
3432 		    continue;
3433 		}
3434 		s++;
3435 
3436 		/* If there is no matching '}', it is an error. */
3437 		if (! (e = strchr(s, '}'))) {
3438 		    if (! PL_lex_inpat) {
3439 			yyerror("Missing right brace on \\N{}");
3440 		    } else {
3441 			yyerror("Missing right brace on \\N{} or unescaped left brace after \\N.");
3442 		    }
3443 		    continue;
3444 		}
3445 
3446 		/* Here it looks like a named character */
3447 
3448 		if (*s == 'U' && s[1] == '+') { /* \N{U+...} */
3449 		    I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3450 				| PERL_SCAN_DISALLOW_PREFIX;
3451 		    STRLEN len;
3452 
3453 		    /* For \N{U+...}, the '...' is a unicode value even on
3454 		     * EBCDIC machines */
3455 		    s += 2;	    /* Skip to next char after the 'U+' */
3456 		    len = e - s;
3457 		    uv = grok_hex(s, &len, &flags, NULL);
3458 		    if (len == 0 || len != (STRLEN)(e - s)) {
3459 			yyerror("Invalid hexadecimal number in \\N{U+...}");
3460 			s = e + 1;
3461 			continue;
3462 		    }
3463 
3464 		    if (PL_lex_inpat) {
3465 
3466 			/* On non-EBCDIC platforms, pass through to the regex
3467 			 * compiler unchanged.  The reason we evaluated the
3468 			 * number above is to make sure there wasn't a syntax
3469 			 * error.  But on EBCDIC we convert to native so
3470 			 * downstream code can continue to assume it's native
3471 			 */
3472 			s -= 5;	    /* Include the '\N{U+' */
3473 #ifdef EBCDIC
3474 			d += my_snprintf(d, e - s + 1 + 1,  /* includes the }
3475 							       and the \0 */
3476 				    "\\N{U+%X}",
3477 				    (unsigned int) UNI_TO_NATIVE(uv));
3478 #else
3479 			Copy(s, d, e - s + 1, char);	/* 1 = include the } */
3480 			d += e - s + 1;
3481 #endif
3482 		    }
3483 		    else {  /* Not a pattern: convert the hex to string */
3484 
3485 			 /* If destination is not in utf8, unconditionally
3486 			  * recode it to be so.  This is because \N{} implies
3487 			  * Unicode semantics, and scalars have to be in utf8
3488 			  * to guarantee those semantics */
3489 			if (! has_utf8) {
3490 			    SvCUR_set(sv, d - SvPVX_const(sv));
3491 			    SvPOK_on(sv);
3492 			    *d = '\0';
3493 			    /* See Note on sizing above.  */
3494 			    sv_utf8_upgrade_flags_grow(
3495 					sv,
3496 					SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3497 					UNISKIP(uv) + (STRLEN)(send - e) + 1);
3498 			    d = SvPVX(sv) + SvCUR(sv);
3499 			    has_utf8 = TRUE;
3500 			}
3501 
3502 			/* Add the string to the output */
3503 			if (UNI_IS_INVARIANT(uv)) {
3504 			    *d++ = (char) uv;
3505 			}
3506 			else d = (char*)uvuni_to_utf8((U8*)d, uv);
3507 		    }
3508 		}
3509 		else /* Here is \N{NAME} but not \N{U+...}. */
3510                      if ((res = get_and_check_backslash_N_name(s, e)))
3511                 {
3512                     STRLEN len;
3513                     const char *str = SvPV_const(res, len);
3514                     if (PL_lex_inpat) {
3515 
3516 			if (! len) { /* The name resolved to an empty string */
3517 			    Copy("\\N{}", d, 4, char);
3518 			    d += 4;
3519 			}
3520 			else {
3521 			    /* In order to not lose information for the regex
3522 			    * compiler, pass the result in the specially made
3523 			    * syntax: \N{U+c1.c2.c3...}, where c1 etc. are
3524 			    * the code points in hex of each character
3525 			    * returned by charnames */
3526 
3527 			    const char *str_end = str + len;
3528 			    const STRLEN off = d - SvPVX_const(sv);
3529 
3530                             if (! SvUTF8(res)) {
3531                                 /* For the non-UTF-8 case, we can determine the
3532                                  * exact length needed without having to parse
3533                                  * through the string.  Each character takes up
3534                                  * 2 hex digits plus either a trailing dot or
3535                                  * the "}" */
3536                                 d = off + SvGROW(sv, off
3537                                                     + 3 * len
3538                                                     + 6 /* For the "\N{U+", and
3539                                                            trailing NUL */
3540                                                     + (STRLEN)(send - e));
3541                                 Copy("\\N{U+", d, 5, char);
3542                                 d += 5;
3543                                 while (str < str_end) {
3544                                     char hex_string[4];
3545                                     my_snprintf(hex_string, sizeof(hex_string),
3546                                                 "%02X.", (U8) *str);
3547                                     Copy(hex_string, d, 3, char);
3548                                     d += 3;
3549                                     str++;
3550                                 }
3551                                 d--;    /* We will overwrite below the final
3552                                            dot with a right brace */
3553                             }
3554                             else {
3555                                 STRLEN char_length; /* cur char's byte length */
3556 
3557                                 /* and the number of bytes after this is
3558                                  * translated into hex digits */
3559                                 STRLEN output_length;
3560 
3561                                 /* 2 hex per byte; 2 chars for '\N'; 2 chars
3562                                  * for max('U+', '.'); and 1 for NUL */
3563                                 char hex_string[2 * UTF8_MAXBYTES + 5];
3564 
3565                                 /* Get the first character of the result. */
3566                                 U32 uv = utf8n_to_uvuni((U8 *) str,
3567                                                         len,
3568                                                         &char_length,
3569                                                         UTF8_ALLOW_ANYUV);
3570                                 /* Convert first code point to hex, including
3571                                  * the boiler plate before it.  For all these,
3572                                  * we convert to native format so that
3573                                  * downstream code can continue to assume the
3574                                  * input is native */
3575                                 output_length =
3576                                     my_snprintf(hex_string, sizeof(hex_string),
3577                                             "\\N{U+%X",
3578                                             (unsigned int) UNI_TO_NATIVE(uv));
3579 
3580                                 /* Make sure there is enough space to hold it */
3581                                 d = off + SvGROW(sv, off
3582                                                     + output_length
3583                                                     + (STRLEN)(send - e)
3584                                                     + 2);	/* '}' + NUL */
3585                                 /* And output it */
3586                                 Copy(hex_string, d, output_length, char);
3587                                 d += output_length;
3588 
3589                                 /* For each subsequent character, append dot and
3590                                 * its ordinal in hex */
3591                                 while ((str += char_length) < str_end) {
3592                                     const STRLEN off = d - SvPVX_const(sv);
3593                                     U32 uv = utf8n_to_uvuni((U8 *) str,
3594                                                             str_end - str,
3595                                                             &char_length,
3596                                                             UTF8_ALLOW_ANYUV);
3597                                     output_length =
3598                                         my_snprintf(hex_string,
3599                                             sizeof(hex_string),
3600                                             ".%X",
3601                                             (unsigned int) UNI_TO_NATIVE(uv));
3602 
3603                                     d = off + SvGROW(sv, off
3604                                                         + output_length
3605                                                         + (STRLEN)(send - e)
3606                                                         + 2);	/* '}' +  NUL */
3607                                     Copy(hex_string, d, output_length, char);
3608                                     d += output_length;
3609                                 }
3610 			    }
3611 
3612 			    *d++ = '}';	/* Done.  Add the trailing brace */
3613 			}
3614 		    }
3615 		    else { /* Here, not in a pattern.  Convert the name to a
3616 			    * string. */
3617 
3618 			 /* If destination is not in utf8, unconditionally
3619 			  * recode it to be so.  This is because \N{} implies
3620 			  * Unicode semantics, and scalars have to be in utf8
3621 			  * to guarantee those semantics */
3622 			if (! has_utf8) {
3623 			    SvCUR_set(sv, d - SvPVX_const(sv));
3624 			    SvPOK_on(sv);
3625 			    *d = '\0';
3626 			    /* See Note on sizing above.  */
3627 			    sv_utf8_upgrade_flags_grow(sv,
3628 						SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3629 						len + (STRLEN)(send - s) + 1);
3630 			    d = SvPVX(sv) + SvCUR(sv);
3631 			    has_utf8 = TRUE;
3632 			} else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
3633 
3634 			    /* See Note on sizing above.  (NOTE: SvCUR() is not
3635 			     * set correctly here). */
3636 			    const STRLEN off = d - SvPVX_const(sv);
3637 			    d = off + SvGROW(sv, off + len + (STRLEN)(send - s) + 1);
3638 			}
3639 			Copy(str, d, len, char);
3640 			d += len;
3641 		    }
3642 
3643 		    SvREFCNT_dec(res);
3644 
3645 		} /* End \N{NAME} */
3646 #ifdef EBCDIC
3647 		if (!dorange)
3648 		    native_range = FALSE; /* \N{} is defined to be Unicode */
3649 #endif
3650 		s = e + 1;  /* Point to just after the '}' */
3651 		continue;
3652 
3653 	    /* \c is a control character */
3654 	    case 'c':
3655 		s++;
3656 		if (s < send) {
3657 		    *d++ = grok_bslash_c(*s++, has_utf8, 1);
3658 		}
3659 		else {
3660 		    yyerror("Missing control char name in \\c");
3661 		}
3662 		continue;
3663 
3664 	    /* printf-style backslashes, formfeeds, newlines, etc */
3665 	    case 'b':
3666 		*d++ = NATIVE_TO_NEED(has_utf8,'\b');
3667 		break;
3668 	    case 'n':
3669 		*d++ = NATIVE_TO_NEED(has_utf8,'\n');
3670 		break;
3671 	    case 'r':
3672 		*d++ = NATIVE_TO_NEED(has_utf8,'\r');
3673 		break;
3674 	    case 'f':
3675 		*d++ = NATIVE_TO_NEED(has_utf8,'\f');
3676 		break;
3677 	    case 't':
3678 		*d++ = NATIVE_TO_NEED(has_utf8,'\t');
3679 		break;
3680 	    case 'e':
3681 		*d++ = ASCII_TO_NEED(has_utf8,'\033');
3682 		break;
3683 	    case 'a':
3684 		*d++ = ASCII_TO_NEED(has_utf8,'\007');
3685 		break;
3686 	    } /* end switch */
3687 
3688 	    s++;
3689 	    continue;
3690 	} /* end if (backslash) */
3691 #ifdef EBCDIC
3692 	else
3693 	    literal_endpoint++;
3694 #endif
3695 
3696     default_action:
3697 	/* If we started with encoded form, or already know we want it,
3698 	   then encode the next character */
3699 	if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) {
3700 	    STRLEN len  = 1;
3701 
3702 
3703 	    /* One might think that it is wasted effort in the case of the
3704 	     * source being utf8 (this_utf8 == TRUE) to take the next character
3705 	     * in the source, convert it to an unsigned value, and then convert
3706 	     * it back again.  But the source has not been validated here.  The
3707 	     * routine that does the conversion checks for errors like
3708 	     * malformed utf8 */
3709 
3710 	    const UV nextuv   = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
3711 	    const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
3712 	    if (!has_utf8) {
3713 		SvCUR_set(sv, d - SvPVX_const(sv));
3714 		SvPOK_on(sv);
3715 		*d = '\0';
3716 		/* See Note on sizing above.  */
3717 		sv_utf8_upgrade_flags_grow(sv,
3718 					SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3719 					need + (STRLEN)(send - s) + 1);
3720 		d = SvPVX(sv) + SvCUR(sv);
3721 		has_utf8 = TRUE;
3722 	    } else if (need > len) {
3723 		/* encoded value larger than old, may need extra space (NOTE:
3724 		 * SvCUR() is not set correctly here).   See Note on sizing
3725 		 * above.  */
3726 		const STRLEN off = d - SvPVX_const(sv);
3727 		d = SvGROW(sv, off + need + (STRLEN)(send - s) + 1) + off;
3728 	    }
3729 	    s += len;
3730 
3731 	    d = (char*)uvchr_to_utf8((U8*)d, nextuv);
3732 #ifdef EBCDIC
3733 	    if (uv > 255 && !dorange)
3734 		native_range = FALSE;
3735 #endif
3736 	}
3737 	else {
3738 	    *d++ = NATIVE_TO_NEED(has_utf8,*s++);
3739 	}
3740     } /* while loop to process each character */
3741 
3742     /* terminate the string and set up the sv */
3743     *d = '\0';
3744     SvCUR_set(sv, d - SvPVX_const(sv));
3745     if (SvCUR(sv) >= SvLEN(sv))
3746 	Perl_croak(aTHX_ "panic: constant overflowed allocated space, %"UVuf
3747 		   " >= %"UVuf, (UV)SvCUR(sv), (UV)SvLEN(sv));
3748 
3749     SvPOK_on(sv);
3750     if (PL_encoding && !has_utf8) {
3751 	sv_recode_to_utf8(sv, PL_encoding);
3752 	if (SvUTF8(sv))
3753 	    has_utf8 = TRUE;
3754     }
3755     if (has_utf8) {
3756 	SvUTF8_on(sv);
3757 	if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
3758 	    PL_sublex_info.sub_op->op_private |=
3759 		    (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3760 	}
3761     }
3762 
3763     /* shrink the sv if we allocated more than we used */
3764     if (SvCUR(sv) + 5 < SvLEN(sv)) {
3765 	SvPV_shrink_to_cur(sv);
3766     }
3767 
3768     /* return the substring (via pl_yylval) only if we parsed anything */
3769     if (s > PL_bufptr) {
3770 	SvREFCNT_inc_simple_void_NN(sv);
3771 	if (   (PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ))
3772             && ! PL_parser->lex_re_reparsing)
3773         {
3774 	    const char *const key = PL_lex_inpat ? "qr" : "q";
3775 	    const STRLEN keylen = PL_lex_inpat ? 2 : 1;
3776 	    const char *type;
3777 	    STRLEN typelen;
3778 
3779 	    if (PL_lex_inwhat == OP_TRANS) {
3780 		type = "tr";
3781 		typelen = 2;
3782 	    } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
3783 		type = "s";
3784 		typelen = 1;
3785 	    } else if (PL_lex_inpat && SvIVX(PL_linestr) == '\'') {
3786 		type = "q";
3787 		typelen = 1;
3788 	    } else  {
3789 		type = "qq";
3790 		typelen = 2;
3791 	    }
3792 
3793 	    sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
3794 				type, typelen);
3795 	}
3796 	pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3797     }
3798     LEAVE_with_name("scan_const");
3799     return s;
3800 }
3801 
3802 /* S_intuit_more
3803  * Returns TRUE if there's more to the expression (e.g., a subscript),
3804  * FALSE otherwise.
3805  *
3806  * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
3807  *
3808  * ->[ and ->{ return TRUE
3809  * { and [ outside a pattern are always subscripts, so return TRUE
3810  * if we're outside a pattern and it's not { or [, then return FALSE
3811  * if we're in a pattern and the first char is a {
3812  *   {4,5} (any digits around the comma) returns FALSE
3813  * if we're in a pattern and the first char is a [
3814  *   [] returns FALSE
3815  *   [SOMETHING] has a funky algorithm to decide whether it's a
3816  *      character class or not.  It has to deal with things like
3817  *      /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
3818  * anything else returns TRUE
3819  */
3820 
3821 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
3822 
3823 STATIC int
3824 S_intuit_more(pTHX_ char *s)
3825 {
3826     dVAR;
3827 
3828     PERL_ARGS_ASSERT_INTUIT_MORE;
3829 
3830     if (PL_lex_brackets)
3831 	return TRUE;
3832     if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
3833 	return TRUE;
3834     if (*s != '{' && *s != '[')
3835 	return FALSE;
3836     if (!PL_lex_inpat)
3837 	return TRUE;
3838 
3839     /* In a pattern, so maybe we have {n,m}. */
3840     if (*s == '{') {
3841 	if (regcurly(s, FALSE)) {
3842 	    return FALSE;
3843 	}
3844 	return TRUE;
3845     }
3846 
3847     /* On the other hand, maybe we have a character class */
3848 
3849     s++;
3850     if (*s == ']' || *s == '^')
3851 	return FALSE;
3852     else {
3853         /* this is terrifying, and it works */
3854 	int weight;
3855 	char seen[256];
3856 	const char * const send = strchr(s,']');
3857 	unsigned char un_char, last_un_char;
3858 	char tmpbuf[sizeof PL_tokenbuf * 4];
3859 
3860 	if (!send)		/* has to be an expression */
3861 	    return TRUE;
3862 	weight = 2;		/* let's weigh the evidence */
3863 
3864 	if (*s == '$')
3865 	    weight -= 3;
3866 	else if (isDIGIT(*s)) {
3867 	    if (s[1] != ']') {
3868 		if (isDIGIT(s[1]) && s[2] == ']')
3869 		    weight -= 10;
3870 	    }
3871 	    else
3872 		weight -= 100;
3873 	}
3874 	Zero(seen,256,char);
3875 	un_char = 255;
3876 	for (; s < send; s++) {
3877 	    last_un_char = un_char;
3878 	    un_char = (unsigned char)*s;
3879 	    switch (*s) {
3880 	    case '@':
3881 	    case '&':
3882 	    case '$':
3883 		weight -= seen[un_char] * 10;
3884 		if (isWORDCHAR_lazy_if(s+1,UTF)) {
3885 		    int len;
3886 		    scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
3887 		    len = (int)strlen(tmpbuf);
3888 		    if (len > 1 && gv_fetchpvn_flags(tmpbuf, len,
3889                                                     UTF ? SVf_UTF8 : 0, SVt_PV))
3890 			weight -= 100;
3891 		    else
3892 			weight -= 10;
3893 		}
3894 		else if (*s == '$' && s[1] &&
3895 		  strchr("[#!%*<>()-=",s[1])) {
3896 		    if (/*{*/ strchr("])} =",s[2]))
3897 			weight -= 10;
3898 		    else
3899 			weight -= 1;
3900 		}
3901 		break;
3902 	    case '\\':
3903 		un_char = 254;
3904 		if (s[1]) {
3905 		    if (strchr("wds]",s[1]))
3906 			weight += 100;
3907 		    else if (seen[(U8)'\''] || seen[(U8)'"'])
3908 			weight += 1;
3909 		    else if (strchr("rnftbxcav",s[1]))
3910 			weight += 40;
3911 		    else if (isDIGIT(s[1])) {
3912 			weight += 40;
3913 			while (s[1] && isDIGIT(s[1]))
3914 			    s++;
3915 		    }
3916 		}
3917 		else
3918 		    weight += 100;
3919 		break;
3920 	    case '-':
3921 		if (s[1] == '\\')
3922 		    weight += 50;
3923 		if (strchr("aA01! ",last_un_char))
3924 		    weight += 30;
3925 		if (strchr("zZ79~",s[1]))
3926 		    weight += 30;
3927 		if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
3928 		    weight -= 5;	/* cope with negative subscript */
3929 		break;
3930 	    default:
3931 		if (!isWORDCHAR(last_un_char)
3932 		    && !(last_un_char == '$' || last_un_char == '@'
3933 			 || last_un_char == '&')
3934 		    && isALPHA(*s) && s[1] && isALPHA(s[1])) {
3935 		    char *d = tmpbuf;
3936 		    while (isALPHA(*s))
3937 			*d++ = *s++;
3938 		    *d = '\0';
3939 		    if (keyword(tmpbuf, d - tmpbuf, 0))
3940 			weight -= 150;
3941 		}
3942 		if (un_char == last_un_char + 1)
3943 		    weight += 5;
3944 		weight -= seen[un_char];
3945 		break;
3946 	    }
3947 	    seen[un_char]++;
3948 	}
3949 	if (weight >= 0)	/* probably a character class */
3950 	    return FALSE;
3951     }
3952 
3953     return TRUE;
3954 }
3955 
3956 /*
3957  * S_intuit_method
3958  *
3959  * Does all the checking to disambiguate
3960  *   foo bar
3961  * between foo(bar) and bar->foo.  Returns 0 if not a method, otherwise
3962  * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
3963  *
3964  * First argument is the stuff after the first token, e.g. "bar".
3965  *
3966  * Not a method if foo is a filehandle.
3967  * Not a method if foo is a subroutine prototyped to take a filehandle.
3968  * Not a method if it's really "Foo $bar"
3969  * Method if it's "foo $bar"
3970  * Not a method if it's really "print foo $bar"
3971  * Method if it's really "foo package::" (interpreted as package->foo)
3972  * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
3973  * Not a method if bar is a filehandle or package, but is quoted with
3974  *   =>
3975  */
3976 
3977 STATIC int
3978 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
3979 {
3980     dVAR;
3981     char *s = start + (*start == '$');
3982     char tmpbuf[sizeof PL_tokenbuf];
3983     STRLEN len;
3984     GV* indirgv;
3985 #ifdef PERL_MAD
3986     int soff;
3987 #endif
3988 
3989     PERL_ARGS_ASSERT_INTUIT_METHOD;
3990 
3991     if (gv && SvTYPE(gv) == SVt_PVGV && GvIO(gv))
3992 	    return 0;
3993     if (cv && SvPOK(cv)) {
3994 		const char *proto = CvPROTO(cv);
3995 		if (proto) {
3996 		    if (*proto == ';')
3997 			proto++;
3998 		    if (*proto == '*')
3999 			return 0;
4000 		}
4001     }
4002     s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4003     /* start is the beginning of the possible filehandle/object,
4004      * and s is the end of it
4005      * tmpbuf is a copy of it
4006      */
4007 
4008     if (*start == '$') {
4009 	if (cv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
4010 		isUPPER(*PL_tokenbuf))
4011 	    return 0;
4012 #ifdef PERL_MAD
4013 	len = start - SvPVX(PL_linestr);
4014 #endif
4015 	s = PEEKSPACE(s);
4016 #ifdef PERL_MAD
4017 	start = SvPVX(PL_linestr) + len;
4018 #endif
4019 	PL_bufptr = start;
4020 	PL_expect = XREF;
4021 	return *s == '(' ? FUNCMETH : METHOD;
4022     }
4023     if (!keyword(tmpbuf, len, 0)) {
4024 	if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
4025 	    len -= 2;
4026 	    tmpbuf[len] = '\0';
4027 #ifdef PERL_MAD
4028 	    soff = s - SvPVX(PL_linestr);
4029 #endif
4030 	    goto bare_package;
4031 	}
4032 	indirgv = gv_fetchpvn_flags(tmpbuf, len, ( UTF ? SVf_UTF8 : 0 ), SVt_PVCV);
4033 	if (indirgv && GvCVu(indirgv))
4034 	    return 0;
4035 	/* filehandle or package name makes it a method */
4036 	if (!cv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, UTF ? SVf_UTF8 : 0)) {
4037 #ifdef PERL_MAD
4038 	    soff = s - SvPVX(PL_linestr);
4039 #endif
4040 	    s = PEEKSPACE(s);
4041 	    if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
4042 		return 0;	/* no assumptions -- "=>" quotes bareword */
4043       bare_package:
4044 	    start_force(PL_curforce);
4045 	    NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
4046 						  S_newSV_maybe_utf8(aTHX_ tmpbuf, len));
4047 	    NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
4048 	    if (PL_madskills)
4049 		curmad('X', newSVpvn_flags(start,SvPVX(PL_linestr) + soff - start,
4050                                                             ( UTF ? SVf_UTF8 : 0 )));
4051 	    PL_expect = XTERM;
4052 	    force_next(WORD);
4053 	    PL_bufptr = s;
4054 #ifdef PERL_MAD
4055 	    PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
4056 #endif
4057 	    return *s == '(' ? FUNCMETH : METHOD;
4058 	}
4059     }
4060     return 0;
4061 }
4062 
4063 /* Encoded script support. filter_add() effectively inserts a
4064  * 'pre-processing' function into the current source input stream.
4065  * Note that the filter function only applies to the current source file
4066  * (e.g., it will not affect files 'require'd or 'use'd by this one).
4067  *
4068  * The datasv parameter (which may be NULL) can be used to pass
4069  * private data to this instance of the filter. The filter function
4070  * can recover the SV using the FILTER_DATA macro and use it to
4071  * store private buffers and state information.
4072  *
4073  * The supplied datasv parameter is upgraded to a PVIO type
4074  * and the IoDIRP/IoANY field is used to store the function pointer,
4075  * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
4076  * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
4077  * private use must be set using malloc'd pointers.
4078  */
4079 
4080 SV *
4081 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
4082 {
4083     dVAR;
4084     if (!funcp)
4085 	return NULL;
4086 
4087     if (!PL_parser)
4088 	return NULL;
4089 
4090     if (PL_parser->lex_flags & LEX_IGNORE_UTF8_HINTS)
4091 	Perl_croak(aTHX_ "Source filters apply only to byte streams");
4092 
4093     if (!PL_rsfp_filters)
4094 	PL_rsfp_filters = newAV();
4095     if (!datasv)
4096 	datasv = newSV(0);
4097     SvUPGRADE(datasv, SVt_PVIO);
4098     IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
4099     IoFLAGS(datasv) |= IOf_FAKE_DIRP;
4100     DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
4101 			  FPTR2DPTR(void *, IoANY(datasv)),
4102 			  SvPV_nolen(datasv)));
4103     av_unshift(PL_rsfp_filters, 1);
4104     av_store(PL_rsfp_filters, 0, datasv) ;
4105     if (
4106 	!PL_parser->filtered
4107      && PL_parser->lex_flags & LEX_EVALBYTES
4108      && PL_bufptr < PL_bufend
4109     ) {
4110 	const char *s = PL_bufptr;
4111 	while (s < PL_bufend) {
4112 	    if (*s == '\n') {
4113 		SV *linestr = PL_parser->linestr;
4114 		char *buf = SvPVX(linestr);
4115 		STRLEN const bufptr_pos = PL_parser->bufptr - buf;
4116 		STRLEN const oldbufptr_pos = PL_parser->oldbufptr - buf;
4117 		STRLEN const oldoldbufptr_pos=PL_parser->oldoldbufptr-buf;
4118 		STRLEN const linestart_pos = PL_parser->linestart - buf;
4119 		STRLEN const last_uni_pos =
4120 		    PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
4121 		STRLEN const last_lop_pos =
4122 		    PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
4123 		av_push(PL_rsfp_filters, linestr);
4124 		PL_parser->linestr =
4125 		    newSVpvn(SvPVX(linestr), ++s-SvPVX(linestr));
4126 		buf = SvPVX(PL_parser->linestr);
4127 		PL_parser->bufend = buf + SvCUR(PL_parser->linestr);
4128 		PL_parser->bufptr = buf + bufptr_pos;
4129 		PL_parser->oldbufptr = buf + oldbufptr_pos;
4130 		PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
4131 		PL_parser->linestart = buf + linestart_pos;
4132 		if (PL_parser->last_uni)
4133 		    PL_parser->last_uni = buf + last_uni_pos;
4134 		if (PL_parser->last_lop)
4135 		    PL_parser->last_lop = buf + last_lop_pos;
4136 		SvLEN(linestr) = SvCUR(linestr);
4137 		SvCUR(linestr) = s-SvPVX(linestr);
4138 		PL_parser->filtered = 1;
4139 		break;
4140 	    }
4141 	    s++;
4142 	}
4143     }
4144     return(datasv);
4145 }
4146 
4147 
4148 /* Delete most recently added instance of this filter function.	*/
4149 void
4150 Perl_filter_del(pTHX_ filter_t funcp)
4151 {
4152     dVAR;
4153     SV *datasv;
4154 
4155     PERL_ARGS_ASSERT_FILTER_DEL;
4156 
4157 #ifdef DEBUGGING
4158     DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
4159 			  FPTR2DPTR(void*, funcp)));
4160 #endif
4161     if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
4162 	return;
4163     /* if filter is on top of stack (usual case) just pop it off */
4164     datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
4165     if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
4166 	sv_free(av_pop(PL_rsfp_filters));
4167 
4168         return;
4169     }
4170     /* we need to search for the correct entry and clear it	*/
4171     Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
4172 }
4173 
4174 
4175 /* Invoke the idxth filter function for the current rsfp.	 */
4176 /* maxlen 0 = read one text line */
4177 I32
4178 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
4179 {
4180     dVAR;
4181     filter_t funcp;
4182     SV *datasv = NULL;
4183     /* This API is bad. It should have been using unsigned int for maxlen.
4184        Not sure if we want to change the API, but if not we should sanity
4185        check the value here.  */
4186     unsigned int correct_length
4187 	= maxlen < 0 ?
4188 #ifdef PERL_MICRO
4189 	0x7FFFFFFF
4190 #else
4191 	INT_MAX
4192 #endif
4193 	: maxlen;
4194 
4195     PERL_ARGS_ASSERT_FILTER_READ;
4196 
4197     if (!PL_parser || !PL_rsfp_filters)
4198 	return -1;
4199     if (idx > AvFILLp(PL_rsfp_filters)) {       /* Any more filters?	*/
4200 	/* Provide a default input filter to make life easy.	*/
4201 	/* Note that we append to the line. This is handy.	*/
4202 	DEBUG_P(PerlIO_printf(Perl_debug_log,
4203 			      "filter_read %d: from rsfp\n", idx));
4204 	if (correct_length) {
4205  	    /* Want a block */
4206 	    int len ;
4207 	    const int old_len = SvCUR(buf_sv);
4208 
4209 	    /* ensure buf_sv is large enough */
4210 	    SvGROW(buf_sv, (STRLEN)(old_len + correct_length + 1)) ;
4211 	    if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
4212 				   correct_length)) <= 0) {
4213 		if (PerlIO_error(PL_rsfp))
4214 	            return -1;		/* error */
4215 	        else
4216 		    return 0 ;		/* end of file */
4217 	    }
4218 	    SvCUR_set(buf_sv, old_len + len) ;
4219 	    SvPVX(buf_sv)[old_len + len] = '\0';
4220 	} else {
4221 	    /* Want a line */
4222             if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
4223 		if (PerlIO_error(PL_rsfp))
4224 	            return -1;		/* error */
4225 	        else
4226 		    return 0 ;		/* end of file */
4227 	    }
4228 	}
4229 	return SvCUR(buf_sv);
4230     }
4231     /* Skip this filter slot if filter has been deleted	*/
4232     if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
4233 	DEBUG_P(PerlIO_printf(Perl_debug_log,
4234 			      "filter_read %d: skipped (filter deleted)\n",
4235 			      idx));
4236 	return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
4237     }
4238     if (SvTYPE(datasv) != SVt_PVIO) {
4239 	if (correct_length) {
4240  	    /* Want a block */
4241 	    const STRLEN remainder = SvLEN(datasv) - SvCUR(datasv);
4242 	    if (!remainder) return 0; /* eof */
4243 	    if (correct_length > remainder) correct_length = remainder;
4244 	    sv_catpvn(buf_sv, SvEND(datasv), correct_length);
4245 	    SvCUR_set(datasv, SvCUR(datasv) + correct_length);
4246 	} else {
4247 	    /* Want a line */
4248 	    const char *s = SvEND(datasv);
4249 	    const char *send = SvPVX(datasv) + SvLEN(datasv);
4250 	    while (s < send) {
4251 		if (*s == '\n') {
4252 		    s++;
4253 		    break;
4254 		}
4255 		s++;
4256 	    }
4257 	    if (s == send) return 0; /* eof */
4258 	    sv_catpvn(buf_sv, SvEND(datasv), s-SvEND(datasv));
4259 	    SvCUR_set(datasv, s-SvPVX(datasv));
4260 	}
4261 	return SvCUR(buf_sv);
4262     }
4263     /* Get function pointer hidden within datasv	*/
4264     funcp = DPTR2FPTR(filter_t, IoANY(datasv));
4265     DEBUG_P(PerlIO_printf(Perl_debug_log,
4266 			  "filter_read %d: via function %p (%s)\n",
4267 			  idx, (void*)datasv, SvPV_nolen_const(datasv)));
4268     /* Call function. The function is expected to 	*/
4269     /* call "FILTER_READ(idx+1, buf_sv)" first.		*/
4270     /* Return: <0:error, =0:eof, >0:not eof 		*/
4271     return (*funcp)(aTHX_ idx, buf_sv, correct_length);
4272 }
4273 
4274 STATIC char *
4275 S_filter_gets(pTHX_ SV *sv, STRLEN append)
4276 {
4277     dVAR;
4278 
4279     PERL_ARGS_ASSERT_FILTER_GETS;
4280 
4281 #ifdef PERL_CR_FILTER
4282     if (!PL_rsfp_filters) {
4283 	filter_add(S_cr_textfilter,NULL);
4284     }
4285 #endif
4286     if (PL_rsfp_filters) {
4287 	if (!append)
4288             SvCUR_set(sv, 0);	/* start with empty line	*/
4289         if (FILTER_READ(0, sv, 0) > 0)
4290             return ( SvPVX(sv) ) ;
4291         else
4292 	    return NULL ;
4293     }
4294     else
4295         return (sv_gets(sv, PL_rsfp, append));
4296 }
4297 
4298 STATIC HV *
4299 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
4300 {
4301     dVAR;
4302     GV *gv;
4303 
4304     PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
4305 
4306     if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
4307         return PL_curstash;
4308 
4309     if (len > 2 &&
4310         (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
4311         (gv = gv_fetchpvn_flags(pkgname, len, ( UTF ? SVf_UTF8 : 0 ), SVt_PVHV)))
4312     {
4313         return GvHV(gv);			/* Foo:: */
4314     }
4315 
4316     /* use constant CLASS => 'MyClass' */
4317     gv = gv_fetchpvn_flags(pkgname, len, UTF ? SVf_UTF8 : 0, SVt_PVCV);
4318     if (gv && GvCV(gv)) {
4319 	SV * const sv = cv_const_sv(GvCV(gv));
4320 	if (sv)
4321             pkgname = SvPV_const(sv, len);
4322     }
4323 
4324     return gv_stashpvn(pkgname, len, UTF ? SVf_UTF8 : 0);
4325 }
4326 
4327 /*
4328  * S_readpipe_override
4329  * Check whether readpipe() is overridden, and generates the appropriate
4330  * optree, provided sublex_start() is called afterwards.
4331  */
4332 STATIC void
4333 S_readpipe_override(pTHX)
4334 {
4335     GV **gvp;
4336     GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
4337     pl_yylval.ival = OP_BACKTICK;
4338     if ((gv_readpipe
4339 		&& GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
4340 	    ||
4341 	    ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
4342 	     && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
4343 	     && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
4344     {
4345 	PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
4346 	    op_append_elem(OP_LIST,
4347 		newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
4348 		newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
4349     }
4350 }
4351 
4352 #ifdef PERL_MAD
4353  /*
4354  * Perl_madlex
4355  * The intent of this yylex wrapper is to minimize the changes to the
4356  * tokener when we aren't interested in collecting madprops.  It remains
4357  * to be seen how successful this strategy will be...
4358  */
4359 
4360 int
4361 Perl_madlex(pTHX)
4362 {
4363     int optype;
4364     char *s = PL_bufptr;
4365 
4366     /* make sure PL_thiswhite is initialized */
4367     PL_thiswhite = 0;
4368     PL_thismad = 0;
4369 
4370     /* previous token ate up our whitespace? */
4371     if (!PL_lasttoke && PL_nextwhite) {
4372 	PL_thiswhite = PL_nextwhite;
4373 	PL_nextwhite = 0;
4374     }
4375 
4376     /* isolate the token, and figure out where it is without whitespace */
4377     PL_realtokenstart = -1;
4378     PL_thistoken = 0;
4379     optype = yylex();
4380     s = PL_bufptr;
4381     assert(PL_curforce < 0);
4382 
4383     if (!PL_thismad || PL_thismad->mad_key == '^') {	/* not forced already? */
4384 	if (!PL_thistoken) {
4385 	    if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
4386 		PL_thistoken = newSVpvs("");
4387 	    else {
4388 		char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
4389 		PL_thistoken = newSVpvn(tstart, s - tstart);
4390 	    }
4391 	}
4392 	if (PL_thismad)	/* install head */
4393 	    CURMAD('X', PL_thistoken);
4394     }
4395 
4396     /* last whitespace of a sublex? */
4397     if (optype == ')' && PL_endwhite) {
4398 	CURMAD('X', PL_endwhite);
4399     }
4400 
4401     if (!PL_thismad) {
4402 
4403 	/* if no whitespace and we're at EOF, bail.  Otherwise fake EOF below. */
4404 	if (!PL_thiswhite && !PL_endwhite && !optype) {
4405 	    sv_free(PL_thistoken);
4406 	    PL_thistoken = 0;
4407 	    return 0;
4408 	}
4409 
4410 	/* put off final whitespace till peg */
4411 	if (optype == ';' && !PL_rsfp && !PL_parser->filtered) {
4412 	    PL_nextwhite = PL_thiswhite;
4413 	    PL_thiswhite = 0;
4414 	}
4415 	else if (PL_thisopen) {
4416 	    CURMAD('q', PL_thisopen);
4417 	    if (PL_thistoken)
4418 		sv_free(PL_thistoken);
4419 	    PL_thistoken = 0;
4420 	}
4421 	else {
4422 	    /* Store actual token text as madprop X */
4423 	    CURMAD('X', PL_thistoken);
4424 	}
4425 
4426 	if (PL_thiswhite) {
4427 	    /* add preceding whitespace as madprop _ */
4428 	    CURMAD('_', PL_thiswhite);
4429 	}
4430 
4431 	if (PL_thisstuff) {
4432 	    /* add quoted material as madprop = */
4433 	    CURMAD('=', PL_thisstuff);
4434 	}
4435 
4436 	if (PL_thisclose) {
4437 	    /* add terminating quote as madprop Q */
4438 	    CURMAD('Q', PL_thisclose);
4439 	}
4440     }
4441 
4442     /* special processing based on optype */
4443 
4444     switch (optype) {
4445 
4446     /* opval doesn't need a TOKEN since it can already store mp */
4447     case WORD:
4448     case METHOD:
4449     case FUNCMETH:
4450     case THING:
4451     case PMFUNC:
4452     case PRIVATEREF:
4453     case FUNC0SUB:
4454     case UNIOPSUB:
4455     case LSTOPSUB:
4456 	if (pl_yylval.opval)
4457 	    append_madprops(PL_thismad, pl_yylval.opval, 0);
4458 	PL_thismad = 0;
4459 	return optype;
4460 
4461     /* fake EOF */
4462     case 0:
4463 	optype = PEG;
4464 	if (PL_endwhite) {
4465 	    addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
4466 	    PL_endwhite = 0;
4467 	}
4468 	break;
4469 
4470     /* pval */
4471     case LABEL:
4472 	break;
4473 
4474     case ']':
4475     case '}':
4476 	if (PL_faketokens)
4477 	    break;
4478 	/* remember any fake bracket that lexer is about to discard */
4479 	if (PL_lex_brackets == 1 &&
4480 	    ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
4481 	{
4482 	    s = PL_bufptr;
4483 	    while (s < PL_bufend && (*s == ' ' || *s == '\t'))
4484 		s++;
4485 	    if (*s == '}') {
4486 		PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
4487 		addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
4488 		PL_thiswhite = 0;
4489 		PL_bufptr = s - 1;
4490 		break;	/* don't bother looking for trailing comment */
4491 	    }
4492 	    else
4493 		s = PL_bufptr;
4494 	}
4495 	if (optype == ']')
4496 	    break;
4497 	/* FALLTHROUGH */
4498 
4499     /* attach a trailing comment to its statement instead of next token */
4500     case ';':
4501 	if (PL_faketokens)
4502 	    break;
4503 	if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
4504 	    s = PL_bufptr;
4505 	    while (s < PL_bufend && (*s == ' ' || *s == '\t'))
4506 		s++;
4507 	    if (*s == '\n' || *s == '#') {
4508 		while (s < PL_bufend && *s != '\n')
4509 		    s++;
4510 		if (s < PL_bufend)
4511 		    s++;
4512 		PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
4513 		addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
4514 		PL_thiswhite = 0;
4515 		PL_bufptr = s;
4516 	    }
4517 	}
4518 	break;
4519 
4520     /* ival */
4521     default:
4522 	break;
4523 
4524     }
4525 
4526     /* Create new token struct.  Note: opvals return early above. */
4527     pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
4528     PL_thismad = 0;
4529     return optype;
4530 }
4531 #endif
4532 
4533 STATIC char *
4534 S_tokenize_use(pTHX_ int is_use, char *s) {
4535     dVAR;
4536 
4537     PERL_ARGS_ASSERT_TOKENIZE_USE;
4538 
4539     if (PL_expect != XSTATE)
4540 	yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
4541 		    is_use ? "use" : "no"));
4542     PL_expect = XTERM;
4543     s = SKIPSPACE1(s);
4544     if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4545 	s = force_version(s, TRUE);
4546 	if (*s == ';' || *s == '}'
4547 		|| (s = SKIPSPACE1(s), (*s == ';' || *s == '}'))) {
4548 	    start_force(PL_curforce);
4549 	    NEXTVAL_NEXTTOKE.opval = NULL;
4550 	    force_next(WORD);
4551 	}
4552 	else if (*s == 'v') {
4553 	    s = force_word(s,WORD,FALSE,TRUE,FALSE);
4554 	    s = force_version(s, FALSE);
4555 	}
4556     }
4557     else {
4558 	s = force_word(s,WORD,FALSE,TRUE,FALSE);
4559 	s = force_version(s, FALSE);
4560     }
4561     pl_yylval.ival = is_use;
4562     return s;
4563 }
4564 #ifdef DEBUGGING
4565     static const char* const exp_name[] =
4566 	{ "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
4567 	  "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
4568 	};
4569 #endif
4570 
4571 #define word_takes_any_delimeter(p,l) S_word_takes_any_delimeter(p,l)
4572 STATIC bool
4573 S_word_takes_any_delimeter(char *p, STRLEN len)
4574 {
4575     return (len == 1 && strchr("msyq", p[0])) ||
4576 	   (len == 2 && (
4577 	    (p[0] == 't' && p[1] == 'r') ||
4578 	    (p[0] == 'q' && strchr("qwxr", p[1]))));
4579 }
4580 
4581 /*
4582   yylex
4583 
4584   Works out what to call the token just pulled out of the input
4585   stream.  The yacc parser takes care of taking the ops we return and
4586   stitching them into a tree.
4587 
4588   Returns:
4589     The type of the next token
4590 
4591   Structure:
4592       Switch based on the current state:
4593 	  - if we already built the token before, use it
4594 	  - if we have a case modifier in a string, deal with that
4595 	  - handle other cases of interpolation inside a string
4596 	  - scan the next line if we are inside a format
4597       In the normal state switch on the next character:
4598 	  - default:
4599 	    if alphabetic, go to key lookup
4600 	    unrecoginized character - croak
4601 	  - 0/4/26: handle end-of-line or EOF
4602 	  - cases for whitespace
4603 	  - \n and #: handle comments and line numbers
4604 	  - various operators, brackets and sigils
4605 	  - numbers
4606 	  - quotes
4607 	  - 'v': vstrings (or go to key lookup)
4608 	  - 'x' repetition operator (or go to key lookup)
4609 	  - other ASCII alphanumerics (key lookup begins here):
4610 	      word before => ?
4611 	      keyword plugin
4612 	      scan built-in keyword (but do nothing with it yet)
4613 	      check for statement label
4614 	      check for lexical subs
4615 		  goto just_a_word if there is one
4616 	      see whether built-in keyword is overridden
4617 	      switch on keyword number:
4618 		  - default: just_a_word:
4619 		      not a built-in keyword; handle bareword lookup
4620 		      disambiguate between method and sub call
4621 		      fall back to bareword
4622 		  - cases for built-in keywords
4623 */
4624 
4625 
4626 #ifdef __SC__
4627 #pragma segment Perl_yylex
4628 #endif
4629 int
4630 Perl_yylex(pTHX)
4631 {
4632     dVAR;
4633     char *s = PL_bufptr;
4634     char *d;
4635     STRLEN len;
4636     bool bof = FALSE;
4637     U8 formbrack = 0;
4638     U32 fake_eof = 0;
4639 
4640     /* orig_keyword, gvp, and gv are initialized here because
4641      * jump to the label just_a_word_zero can bypass their
4642      * initialization later. */
4643     I32 orig_keyword = 0;
4644     GV *gv = NULL;
4645     GV **gvp = NULL;
4646 
4647     DEBUG_T( {
4648 	SV* tmp = newSVpvs("");
4649 	PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
4650 	    (IV)CopLINE(PL_curcop),
4651 	    lex_state_names[PL_lex_state],
4652 	    exp_name[PL_expect],
4653 	    pv_display(tmp, s, strlen(s), 0, 60));
4654 	SvREFCNT_dec(tmp);
4655     } );
4656 
4657     switch (PL_lex_state) {
4658 #ifdef COMMENTARY
4659     case LEX_NORMAL:		/* Some compilers will produce faster */
4660     case LEX_INTERPNORMAL:	/* code if we comment these out. */
4661 	break;
4662 #endif
4663 
4664     /* when we've already built the next token, just pull it out of the queue */
4665     case LEX_KNOWNEXT:
4666 #ifdef PERL_MAD
4667 	PL_lasttoke--;
4668 	pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
4669 	if (PL_madskills) {
4670 	    PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
4671 	    PL_nexttoke[PL_lasttoke].next_mad = 0;
4672 	    if (PL_thismad && PL_thismad->mad_key == '_') {
4673 		PL_thiswhite = MUTABLE_SV(PL_thismad->mad_val);
4674 		PL_thismad->mad_val = 0;
4675 		mad_free(PL_thismad);
4676 		PL_thismad = 0;
4677 	    }
4678 	}
4679 	if (!PL_lasttoke) {
4680 	    PL_lex_state = PL_lex_defer;
4681   	    PL_expect = PL_lex_expect;
4682   	    PL_lex_defer = LEX_NORMAL;
4683 	    if (!PL_nexttoke[PL_lasttoke].next_type)
4684 		return yylex();
4685   	}
4686 #else
4687 	PL_nexttoke--;
4688 	pl_yylval = PL_nextval[PL_nexttoke];
4689 	if (!PL_nexttoke) {
4690 	    PL_lex_state = PL_lex_defer;
4691 	    PL_expect = PL_lex_expect;
4692 	    PL_lex_defer = LEX_NORMAL;
4693 	}
4694 #endif
4695 	{
4696 	    I32 next_type;
4697 #ifdef PERL_MAD
4698 	    next_type = PL_nexttoke[PL_lasttoke].next_type;
4699 #else
4700 	    next_type = PL_nexttype[PL_nexttoke];
4701 #endif
4702 	    if (next_type & (7<<24)) {
4703 		if (next_type & (1<<24)) {
4704 		    if (PL_lex_brackets > 100)
4705 			Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4706 		    PL_lex_brackstack[PL_lex_brackets++] =
4707 			(char) ((next_type >> 16) & 0xff);
4708 		}
4709 		if (next_type & (2<<24))
4710 		    PL_lex_allbrackets++;
4711 		if (next_type & (4<<24))
4712 		    PL_lex_allbrackets--;
4713 		next_type &= 0xffff;
4714 	    }
4715 	    return REPORT(next_type == 'p' ? pending_ident() : next_type);
4716 	}
4717 
4718     /* interpolated case modifiers like \L \U, including \Q and \E.
4719        when we get here, PL_bufptr is at the \
4720     */
4721     case LEX_INTERPCASEMOD:
4722 #ifdef DEBUGGING
4723 	if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
4724 	    Perl_croak(aTHX_
4725 		       "panic: INTERPCASEMOD bufptr=%p, bufend=%p, *bufptr=%u",
4726 		       PL_bufptr, PL_bufend, *PL_bufptr);
4727 #endif
4728 	/* handle \E or end of string */
4729        	if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
4730 	    /* if at a \E */
4731 	    if (PL_lex_casemods) {
4732 		const char oldmod = PL_lex_casestack[--PL_lex_casemods];
4733 		PL_lex_casestack[PL_lex_casemods] = '\0';
4734 
4735 		if (PL_bufptr != PL_bufend
4736 		    && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q'
4737                         || oldmod == 'F')) {
4738 		    PL_bufptr += 2;
4739 		    PL_lex_state = LEX_INTERPCONCAT;
4740 #ifdef PERL_MAD
4741 		    if (PL_madskills)
4742 			PL_thistoken = newSVpvs("\\E");
4743 #endif
4744 		}
4745 		PL_lex_allbrackets--;
4746 		return REPORT(')');
4747 	    }
4748             else if ( PL_bufptr != PL_bufend && PL_bufptr[1] == 'E' ) {
4749                /* Got an unpaired \E */
4750                Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
4751                         "Useless use of \\E");
4752             }
4753 #ifdef PERL_MAD
4754 	    while (PL_bufptr != PL_bufend &&
4755 	      PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
4756 		if (PL_madskills) {
4757 		  if (!PL_thiswhite)
4758 		    PL_thiswhite = newSVpvs("");
4759 		  sv_catpvn(PL_thiswhite, PL_bufptr, 2);
4760 		}
4761 		PL_bufptr += 2;
4762 	    }
4763 #else
4764 	    if (PL_bufptr != PL_bufend)
4765 		PL_bufptr += 2;
4766 #endif
4767 	    PL_lex_state = LEX_INTERPCONCAT;
4768 	    return yylex();
4769 	}
4770 	else {
4771 	    DEBUG_T({ PerlIO_printf(Perl_debug_log,
4772               "### Saw case modifier\n"); });
4773 	    s = PL_bufptr + 1;
4774 	    if (s[1] == '\\' && s[2] == 'E') {
4775 #ifdef PERL_MAD
4776 		if (PL_madskills) {
4777 		  if (!PL_thiswhite)
4778 		    PL_thiswhite = newSVpvs("");
4779 		  sv_catpvn(PL_thiswhite, PL_bufptr, 4);
4780 		}
4781 #endif
4782 	        PL_bufptr = s + 3;
4783 		PL_lex_state = LEX_INTERPCONCAT;
4784 		return yylex();
4785 	    }
4786 	    else {
4787 		I32 tmp;
4788 		if (!PL_madskills) /* when just compiling don't need correct */
4789 		    if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
4790 			tmp = *s, *s = s[2], s[2] = (char)tmp;	/* misordered... */
4791 		if ((*s == 'L' || *s == 'U' || *s == 'F') &&
4792 		    (strchr(PL_lex_casestack, 'L')
4793                         || strchr(PL_lex_casestack, 'U')
4794                         || strchr(PL_lex_casestack, 'F'))) {
4795 		    PL_lex_casestack[--PL_lex_casemods] = '\0';
4796 		    PL_lex_allbrackets--;
4797 		    return REPORT(')');
4798 		}
4799 		if (PL_lex_casemods > 10)
4800 		    Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
4801 		PL_lex_casestack[PL_lex_casemods++] = *s;
4802 		PL_lex_casestack[PL_lex_casemods] = '\0';
4803 		PL_lex_state = LEX_INTERPCONCAT;
4804 		start_force(PL_curforce);
4805 		NEXTVAL_NEXTTOKE.ival = 0;
4806 		force_next((2<<24)|'(');
4807 		start_force(PL_curforce);
4808 		if (*s == 'l')
4809 		    NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
4810 		else if (*s == 'u')
4811 		    NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
4812 		else if (*s == 'L')
4813 		    NEXTVAL_NEXTTOKE.ival = OP_LC;
4814 		else if (*s == 'U')
4815 		    NEXTVAL_NEXTTOKE.ival = OP_UC;
4816 		else if (*s == 'Q')
4817 		    NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
4818                 else if (*s == 'F')
4819 		    NEXTVAL_NEXTTOKE.ival = OP_FC;
4820 		else
4821 		    Perl_croak(aTHX_ "panic: yylex, *s=%u", *s);
4822 		if (PL_madskills) {
4823 		    SV* const tmpsv = newSVpvs("\\ ");
4824 		    /* replace the space with the character we want to escape
4825 		     */
4826 		    SvPVX(tmpsv)[1] = *s;
4827 		    curmad('_', tmpsv);
4828 		}
4829 		PL_bufptr = s + 1;
4830 	    }
4831 	    force_next(FUNC);
4832 	    if (PL_lex_starts) {
4833 		s = PL_bufptr;
4834 		PL_lex_starts = 0;
4835 #ifdef PERL_MAD
4836 		if (PL_madskills) {
4837 		    if (PL_thistoken)
4838 			sv_free(PL_thistoken);
4839 		    PL_thistoken = newSVpvs("");
4840 		}
4841 #endif
4842 		/* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
4843 		if (PL_lex_casemods == 1 && PL_lex_inpat)
4844 		    OPERATOR(',');
4845 		else
4846 		    Aop(OP_CONCAT);
4847 	    }
4848 	    else
4849 		return yylex();
4850 	}
4851 
4852     case LEX_INTERPPUSH:
4853         return REPORT(sublex_push());
4854 
4855     case LEX_INTERPSTART:
4856 	if (PL_bufptr == PL_bufend)
4857 	    return REPORT(sublex_done());
4858 	DEBUG_T({ if(*PL_bufptr != '(') PerlIO_printf(Perl_debug_log,
4859               "### Interpolated variable\n"); });
4860 	PL_expect = XTERM;
4861         /* for /@a/, we leave the joining for the regex engine to do
4862          * (unless we're within \Q etc) */
4863 	PL_lex_dojoin = (*PL_bufptr == '@'
4864                             && (!PL_lex_inpat || PL_lex_casemods));
4865 	PL_lex_state = LEX_INTERPNORMAL;
4866 	if (PL_lex_dojoin) {
4867 	    start_force(PL_curforce);
4868 	    NEXTVAL_NEXTTOKE.ival = 0;
4869 	    force_next(',');
4870 	    start_force(PL_curforce);
4871 	    force_ident("\"", '$');
4872 	    start_force(PL_curforce);
4873 	    NEXTVAL_NEXTTOKE.ival = 0;
4874 	    force_next('$');
4875 	    start_force(PL_curforce);
4876 	    NEXTVAL_NEXTTOKE.ival = 0;
4877 	    force_next((2<<24)|'(');
4878 	    start_force(PL_curforce);
4879 	    NEXTVAL_NEXTTOKE.ival = OP_JOIN;	/* emulate join($", ...) */
4880 	    force_next(FUNC);
4881 	}
4882 	/* Convert (?{...}) and friends to 'do {...}' */
4883 	if (PL_lex_inpat && *PL_bufptr == '(') {
4884 	    PL_parser->lex_shared->re_eval_start = PL_bufptr;
4885 	    PL_bufptr += 2;
4886 	    if (*PL_bufptr != '{')
4887 		PL_bufptr++;
4888 	    start_force(PL_curforce);
4889 	    /* XXX probably need a CURMAD(something) here */
4890 	    PL_expect = XTERMBLOCK;
4891 	    force_next(DO);
4892 	}
4893 
4894 	if (PL_lex_starts++) {
4895 	    s = PL_bufptr;
4896 #ifdef PERL_MAD
4897 	    if (PL_madskills) {
4898 		if (PL_thistoken)
4899 		    sv_free(PL_thistoken);
4900 		PL_thistoken = newSVpvs("");
4901 	    }
4902 #endif
4903 	    /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
4904 	    if (!PL_lex_casemods && PL_lex_inpat)
4905 		OPERATOR(',');
4906 	    else
4907 		Aop(OP_CONCAT);
4908 	}
4909 	return yylex();
4910 
4911     case LEX_INTERPENDMAYBE:
4912 	if (intuit_more(PL_bufptr)) {
4913 	    PL_lex_state = LEX_INTERPNORMAL;	/* false alarm, more expr */
4914 	    break;
4915 	}
4916 	/* FALL THROUGH */
4917 
4918     case LEX_INTERPEND:
4919 	if (PL_lex_dojoin) {
4920 	    PL_lex_dojoin = FALSE;
4921 	    PL_lex_state = LEX_INTERPCONCAT;
4922 #ifdef PERL_MAD
4923 	    if (PL_madskills) {
4924 		if (PL_thistoken)
4925 		    sv_free(PL_thistoken);
4926 		PL_thistoken = newSVpvs("");
4927 	    }
4928 #endif
4929 	    PL_lex_allbrackets--;
4930 	    return REPORT(')');
4931 	}
4932 	if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
4933 	    && SvEVALED(PL_lex_repl))
4934 	{
4935 	    if (PL_bufptr != PL_bufend)
4936 		Perl_croak(aTHX_ "Bad evalled substitution pattern");
4937 	    PL_lex_repl = NULL;
4938 	}
4939 	/* Paranoia.  re_eval_start is adjusted when S_scan_heredoc sets
4940 	   re_eval_str.  If the here-doc body’s length equals the previous
4941 	   value of re_eval_start, re_eval_start will now be null.  So
4942 	   check re_eval_str as well. */
4943 	if (PL_parser->lex_shared->re_eval_start
4944 	 || PL_parser->lex_shared->re_eval_str) {
4945 	    SV *sv;
4946 	    if (*PL_bufptr != ')')
4947 		Perl_croak(aTHX_ "Sequence (?{...}) not terminated with ')'");
4948 	    PL_bufptr++;
4949 	    /* having compiled a (?{..}) expression, return the original
4950 	     * text too, as a const */
4951 	    if (PL_parser->lex_shared->re_eval_str) {
4952 		sv = PL_parser->lex_shared->re_eval_str;
4953 		PL_parser->lex_shared->re_eval_str = NULL;
4954 		SvCUR_set(sv,
4955 			 PL_bufptr - PL_parser->lex_shared->re_eval_start);
4956 		SvPV_shrink_to_cur(sv);
4957 	    }
4958 	    else sv = newSVpvn(PL_parser->lex_shared->re_eval_start,
4959 			 PL_bufptr - PL_parser->lex_shared->re_eval_start);
4960 	    start_force(PL_curforce);
4961 	    /* XXX probably need a CURMAD(something) here */
4962 	    NEXTVAL_NEXTTOKE.opval =
4963 		    (OP*)newSVOP(OP_CONST, 0,
4964 				 sv);
4965 	    force_next(THING);
4966 	    PL_parser->lex_shared->re_eval_start = NULL;
4967 	    PL_expect = XTERM;
4968 	    return REPORT(',');
4969 	}
4970 
4971 	/* FALLTHROUGH */
4972     case LEX_INTERPCONCAT:
4973 #ifdef DEBUGGING
4974 	if (PL_lex_brackets)
4975 	    Perl_croak(aTHX_ "panic: INTERPCONCAT, lex_brackets=%ld",
4976 		       (long) PL_lex_brackets);
4977 #endif
4978 	if (PL_bufptr == PL_bufend)
4979 	    return REPORT(sublex_done());
4980 
4981 	/* m'foo' still needs to be parsed for possible (?{...}) */
4982 	if (SvIVX(PL_linestr) == '\'' && !PL_lex_inpat) {
4983 	    SV *sv = newSVsv(PL_linestr);
4984 	    sv = tokeq(sv);
4985 	    pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4986 	    s = PL_bufend;
4987 	}
4988 	else {
4989 	    s = scan_const(PL_bufptr);
4990 	    if (*s == '\\')
4991 		PL_lex_state = LEX_INTERPCASEMOD;
4992 	    else
4993 		PL_lex_state = LEX_INTERPSTART;
4994 	}
4995 
4996 	if (s != PL_bufptr) {
4997 	    start_force(PL_curforce);
4998 	    if (PL_madskills) {
4999 		curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
5000 	    }
5001 	    NEXTVAL_NEXTTOKE = pl_yylval;
5002 	    PL_expect = XTERM;
5003 	    force_next(THING);
5004 	    if (PL_lex_starts++) {
5005 #ifdef PERL_MAD
5006 		if (PL_madskills) {
5007 		    if (PL_thistoken)
5008 			sv_free(PL_thistoken);
5009 		    PL_thistoken = newSVpvs("");
5010 		}
5011 #endif
5012 		/* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
5013 		if (!PL_lex_casemods && PL_lex_inpat)
5014 		    OPERATOR(',');
5015 		else
5016 		    Aop(OP_CONCAT);
5017 	    }
5018 	    else {
5019 		PL_bufptr = s;
5020 		return yylex();
5021 	    }
5022 	}
5023 
5024 	return yylex();
5025     case LEX_FORMLINE:
5026 	s = scan_formline(PL_bufptr);
5027 	if (!PL_lex_formbrack)
5028 	{
5029 	    formbrack = 1;
5030 	    goto rightbracket;
5031 	}
5032 	PL_bufptr = s;
5033 	return yylex();
5034     }
5035 
5036     s = PL_bufptr;
5037     PL_oldoldbufptr = PL_oldbufptr;
5038     PL_oldbufptr = s;
5039 
5040   retry:
5041 #ifdef PERL_MAD
5042     if (PL_thistoken) {
5043 	sv_free(PL_thistoken);
5044 	PL_thistoken = 0;
5045     }
5046     PL_realtokenstart = s - SvPVX(PL_linestr);	/* assume but undo on ws */
5047 #endif
5048     switch (*s) {
5049     default:
5050 	if (UTF ? isIDFIRST_utf8((U8*)s) : isALNUMC(*s))
5051 	    goto keylookup;
5052 	{
5053         SV *dsv = newSVpvs_flags("", SVs_TEMP);
5054         const char *c = UTF ? savepv(sv_uni_display(dsv, newSVpvn_flags(s,
5055                                                     UTF8SKIP(s),
5056                                                     SVs_TEMP | SVf_UTF8),
5057                                             10, UNI_DISPLAY_ISPRINT))
5058                             : Perl_form(aTHX_ "\\x%02X", (unsigned char)*s);
5059         len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
5060         if (len > UNRECOGNIZED_PRECEDE_COUNT) {
5061             d = UTF ? (char *) Perl_utf8_hop(aTHX_ (U8 *) s, -UNRECOGNIZED_PRECEDE_COUNT) : s - UNRECOGNIZED_PRECEDE_COUNT;
5062         } else {
5063             d = PL_linestart;
5064         }
5065         *s = '\0';
5066         sv_setpv(dsv, d);
5067         if (UTF)
5068             SvUTF8_on(dsv);
5069         Perl_croak(aTHX_  "Unrecognized character %s; marked by <-- HERE after %"SVf"<-- HERE near column %d", c, SVfARG(dsv), (int) len + 1);
5070     }
5071     case 4:
5072     case 26:
5073 	goto fake_eof;			/* emulate EOF on ^D or ^Z */
5074     case 0:
5075 #ifdef PERL_MAD
5076 	if (PL_madskills)
5077 	    PL_faketokens = 0;
5078 #endif
5079 	if (!PL_rsfp && (!PL_parser->filtered || s+1 < PL_bufend)) {
5080 	    PL_last_uni = 0;
5081 	    PL_last_lop = 0;
5082 	    if (PL_lex_brackets &&
5083 		    PL_lex_brackstack[PL_lex_brackets-1] != XFAKEEOF) {
5084 		yyerror((const char *)
5085 			(PL_lex_formbrack
5086 			 ? "Format not terminated"
5087 			 : "Missing right curly or square bracket"));
5088 	    }
5089             DEBUG_T( { PerlIO_printf(Perl_debug_log,
5090                         "### Tokener got EOF\n");
5091             } );
5092 	    TOKEN(0);
5093 	}
5094 	if (s++ < PL_bufend)
5095 	    goto retry;			/* ignore stray nulls */
5096 	PL_last_uni = 0;
5097 	PL_last_lop = 0;
5098 	if (!PL_in_eval && !PL_preambled) {
5099 	    PL_preambled = TRUE;
5100 #ifdef PERL_MAD
5101 	    if (PL_madskills)
5102 		PL_faketokens = 1;
5103 #endif
5104 	    if (PL_perldb) {
5105 		/* Generate a string of Perl code to load the debugger.
5106 		 * If PERL5DB is set, it will return the contents of that,
5107 		 * otherwise a compile-time require of perl5db.pl.  */
5108 
5109 		const char * const pdb = PerlEnv_getenv("PERL5DB");
5110 
5111 		if (pdb) {
5112 		    sv_setpv(PL_linestr, pdb);
5113 		    sv_catpvs(PL_linestr,";");
5114 		} else {
5115 		    SETERRNO(0,SS_NORMAL);
5116 		    sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
5117 		}
5118 	    } else
5119 		sv_setpvs(PL_linestr,"");
5120 	    if (PL_preambleav) {
5121 		SV **svp = AvARRAY(PL_preambleav);
5122 		SV **const end = svp + AvFILLp(PL_preambleav);
5123 		while(svp <= end) {
5124 		    sv_catsv(PL_linestr, *svp);
5125 		    ++svp;
5126 		    sv_catpvs(PL_linestr, ";");
5127 		}
5128 		sv_free(MUTABLE_SV(PL_preambleav));
5129 		PL_preambleav = NULL;
5130 	    }
5131 	    if (PL_minus_E)
5132 		sv_catpvs(PL_linestr,
5133 			  "use feature ':5." STRINGIFY(PERL_VERSION) "';");
5134 	    if (PL_minus_n || PL_minus_p) {
5135 		sv_catpvs(PL_linestr, "LINE: while (<>) {"/*}*/);
5136 		if (PL_minus_l)
5137 		    sv_catpvs(PL_linestr,"chomp;");
5138 		if (PL_minus_a) {
5139 		    if (PL_minus_F) {
5140 			if ((*PL_splitstr == '/' || *PL_splitstr == '\''
5141 			     || *PL_splitstr == '"')
5142 			      && strchr(PL_splitstr + 1, *PL_splitstr))
5143 			    Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
5144 			else {
5145 			    /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
5146 			       bytes can be used as quoting characters.  :-) */
5147 			    const char *splits = PL_splitstr;
5148 			    sv_catpvs(PL_linestr, "our @F=split(q\0");
5149 			    do {
5150 				/* Need to \ \s  */
5151 				if (*splits == '\\')
5152 				    sv_catpvn(PL_linestr, splits, 1);
5153 				sv_catpvn(PL_linestr, splits, 1);
5154 			    } while (*splits++);
5155 			    /* This loop will embed the trailing NUL of
5156 			       PL_linestr as the last thing it does before
5157 			       terminating.  */
5158 			    sv_catpvs(PL_linestr, ");");
5159 			}
5160 		    }
5161 		    else
5162 		        sv_catpvs(PL_linestr,"our @F=split(' ');");
5163 		}
5164 	    }
5165 	    sv_catpvs(PL_linestr, "\n");
5166 	    PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
5167 	    PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
5168 	    PL_last_lop = PL_last_uni = NULL;
5169 	    if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
5170 		update_debugger_info(PL_linestr, NULL, 0);
5171 	    goto retry;
5172 	}
5173 	do {
5174 	    fake_eof = 0;
5175 	    bof = PL_rsfp ? TRUE : FALSE;
5176 	    if (0) {
5177 	      fake_eof:
5178 		fake_eof = LEX_FAKE_EOF;
5179 	    }
5180 	    PL_bufptr = PL_bufend;
5181 	    COPLINE_INC_WITH_HERELINES;
5182 	    if (!lex_next_chunk(fake_eof)) {
5183 		CopLINE_dec(PL_curcop);
5184 		s = PL_bufptr;
5185 		TOKEN(';');	/* not infinite loop because rsfp is NULL now */
5186 	    }
5187 	    CopLINE_dec(PL_curcop);
5188 #ifdef PERL_MAD
5189 	    if (!PL_rsfp)
5190 		PL_realtokenstart = -1;
5191 #endif
5192 	    s = PL_bufptr;
5193 	    /* If it looks like the start of a BOM or raw UTF-16,
5194 	     * check if it in fact is. */
5195 	    if (bof && PL_rsfp &&
5196 		     (*s == 0 ||
5197 		      *(U8*)s == 0xEF ||
5198 		      *(U8*)s >= 0xFE ||
5199 		      s[1] == 0)) {
5200 		Off_t offset = (IV)PerlIO_tell(PL_rsfp);
5201 		bof = (offset == (Off_t)SvCUR(PL_linestr));
5202 #if defined(PERLIO_USING_CRLF) && defined(PERL_TEXTMODE_SCRIPTS)
5203 		/* offset may include swallowed CR */
5204 		if (!bof)
5205 		    bof = (offset == (Off_t)SvCUR(PL_linestr)+1);
5206 #endif
5207 		if (bof) {
5208 		    PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
5209 		    s = swallow_bom((U8*)s);
5210 		}
5211 	    }
5212 	    if (PL_parser->in_pod) {
5213 		/* Incest with pod. */
5214 #ifdef PERL_MAD
5215 		if (PL_madskills)
5216 		    sv_catsv(PL_thiswhite, PL_linestr);
5217 #endif
5218 		if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
5219 		    sv_setpvs(PL_linestr, "");
5220 		    PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
5221 		    PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
5222 		    PL_last_lop = PL_last_uni = NULL;
5223 		    PL_parser->in_pod = 0;
5224 		}
5225 	    }
5226 	    if (PL_rsfp || PL_parser->filtered)
5227 		incline(s);
5228 	} while (PL_parser->in_pod);
5229 	PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
5230 	PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
5231 	PL_last_lop = PL_last_uni = NULL;
5232 	if (CopLINE(PL_curcop) == 1) {
5233 	    while (s < PL_bufend && isSPACE(*s))
5234 		s++;
5235 	    if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
5236 		s++;
5237 #ifdef PERL_MAD
5238 	    if (PL_madskills)
5239 		PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
5240 #endif
5241 	    d = NULL;
5242 	    if (!PL_in_eval) {
5243 		if (*s == '#' && *(s+1) == '!')
5244 		    d = s + 2;
5245 #ifdef ALTERNATE_SHEBANG
5246 		else {
5247 		    static char const as[] = ALTERNATE_SHEBANG;
5248 		    if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
5249 			d = s + (sizeof(as) - 1);
5250 		}
5251 #endif /* ALTERNATE_SHEBANG */
5252 	    }
5253 	    if (d) {
5254 		char *ipath;
5255 		char *ipathend;
5256 
5257 		while (isSPACE(*d))
5258 		    d++;
5259 		ipath = d;
5260 		while (*d && !isSPACE(*d))
5261 		    d++;
5262 		ipathend = d;
5263 
5264 #ifdef ARG_ZERO_IS_SCRIPT
5265 		if (ipathend > ipath) {
5266 		    /*
5267 		     * HP-UX (at least) sets argv[0] to the script name,
5268 		     * which makes $^X incorrect.  And Digital UNIX and Linux,
5269 		     * at least, set argv[0] to the basename of the Perl
5270 		     * interpreter. So, having found "#!", we'll set it right.
5271 		     */
5272 		    SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
5273 						    SVt_PV)); /* $^X */
5274 		    assert(SvPOK(x) || SvGMAGICAL(x));
5275 		    if (sv_eq(x, CopFILESV(PL_curcop))) {
5276 			sv_setpvn(x, ipath, ipathend - ipath);
5277 			SvSETMAGIC(x);
5278 		    }
5279 		    else {
5280 			STRLEN blen;
5281 			STRLEN llen;
5282 			const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
5283 			const char * const lstart = SvPV_const(x,llen);
5284 			if (llen < blen) {
5285 			    bstart += blen - llen;
5286 			    if (strnEQ(bstart, lstart, llen) &&	bstart[-1] == '/') {
5287 				sv_setpvn(x, ipath, ipathend - ipath);
5288 				SvSETMAGIC(x);
5289 			    }
5290 			}
5291 		    }
5292 		    TAINT_NOT;	/* $^X is always tainted, but that's OK */
5293 		}
5294 #endif /* ARG_ZERO_IS_SCRIPT */
5295 
5296 		/*
5297 		 * Look for options.
5298 		 */
5299 		d = instr(s,"perl -");
5300 		if (!d) {
5301 		    d = instr(s,"perl");
5302 #if defined(DOSISH)
5303 		    /* avoid getting into infinite loops when shebang
5304 		     * line contains "Perl" rather than "perl" */
5305 		    if (!d) {
5306 			for (d = ipathend-4; d >= ipath; --d) {
5307 			    if ((*d == 'p' || *d == 'P')
5308 				&& !ibcmp(d, "perl", 4))
5309 			    {
5310 				break;
5311 			    }
5312 			}
5313 			if (d < ipath)
5314 			    d = NULL;
5315 		    }
5316 #endif
5317 		}
5318 #ifdef ALTERNATE_SHEBANG
5319 		/*
5320 		 * If the ALTERNATE_SHEBANG on this system starts with a
5321 		 * character that can be part of a Perl expression, then if
5322 		 * we see it but not "perl", we're probably looking at the
5323 		 * start of Perl code, not a request to hand off to some
5324 		 * other interpreter.  Similarly, if "perl" is there, but
5325 		 * not in the first 'word' of the line, we assume the line
5326 		 * contains the start of the Perl program.
5327 		 */
5328 		if (d && *s != '#') {
5329 		    const char *c = ipath;
5330 		    while (*c && !strchr("; \t\r\n\f\v#", *c))
5331 			c++;
5332 		    if (c < d)
5333 			d = NULL;	/* "perl" not in first word; ignore */
5334 		    else
5335 			*s = '#';	/* Don't try to parse shebang line */
5336 		}
5337 #endif /* ALTERNATE_SHEBANG */
5338 		if (!d &&
5339 		    *s == '#' &&
5340 		    ipathend > ipath &&
5341 		    !PL_minus_c &&
5342 		    !instr(s,"indir") &&
5343 		    instr(PL_origargv[0],"perl"))
5344 		{
5345 		    dVAR;
5346 		    char **newargv;
5347 
5348 		    *ipathend = '\0';
5349 		    s = ipathend + 1;
5350 		    while (s < PL_bufend && isSPACE(*s))
5351 			s++;
5352 		    if (s < PL_bufend) {
5353 			Newx(newargv,PL_origargc+3,char*);
5354 			newargv[1] = s;
5355 			while (s < PL_bufend && !isSPACE(*s))
5356 			    s++;
5357 			*s = '\0';
5358 			Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
5359 		    }
5360 		    else
5361 			newargv = PL_origargv;
5362 		    newargv[0] = ipath;
5363 		    PERL_FPU_PRE_EXEC
5364 		    PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
5365 		    PERL_FPU_POST_EXEC
5366 		    Perl_croak(aTHX_ "Can't exec %s", ipath);
5367 		}
5368 		if (d) {
5369 		    while (*d && !isSPACE(*d))
5370 			d++;
5371 		    while (SPACE_OR_TAB(*d))
5372 			d++;
5373 
5374 		    if (*d++ == '-') {
5375 			const bool switches_done = PL_doswitches;
5376 			const U32 oldpdb = PL_perldb;
5377 			const bool oldn = PL_minus_n;
5378 			const bool oldp = PL_minus_p;
5379 			const char *d1 = d;
5380 
5381 			do {
5382 			    bool baduni = FALSE;
5383 			    if (*d1 == 'C') {
5384 				const char *d2 = d1 + 1;
5385 				if (parse_unicode_opts((const char **)&d2)
5386 				    != PL_unicode)
5387 				    baduni = TRUE;
5388 			    }
5389 			    if (baduni || *d1 == 'M' || *d1 == 'm') {
5390 				const char * const m = d1;
5391 				while (*d1 && !isSPACE(*d1))
5392 				    d1++;
5393 				Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
5394 				      (int)(d1 - m), m);
5395 			    }
5396 			    d1 = moreswitches(d1);
5397 			} while (d1);
5398 			if (PL_doswitches && !switches_done) {
5399 			    int argc = PL_origargc;
5400 			    char **argv = PL_origargv;
5401 			    do {
5402 				argc--,argv++;
5403 			    } while (argc && argv[0][0] == '-' && argv[0][1]);
5404 			    init_argv_symbols(argc,argv);
5405 			}
5406 			if (((PERLDB_LINE || PERLDB_SAVESRC) && !oldpdb) ||
5407 			    ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
5408 			      /* if we have already added "LINE: while (<>) {",
5409 			         we must not do it again */
5410 			{
5411 			    sv_setpvs(PL_linestr, "");
5412 			    PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
5413 			    PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
5414 			    PL_last_lop = PL_last_uni = NULL;
5415 			    PL_preambled = FALSE;
5416 			    if (PERLDB_LINE || PERLDB_SAVESRC)
5417 				(void)gv_fetchfile(PL_origfilename);
5418 			    goto retry;
5419 			}
5420 		    }
5421 		}
5422 	    }
5423 	}
5424 	if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
5425 	    PL_lex_state = LEX_FORMLINE;
5426 	    start_force(PL_curforce);
5427 	    NEXTVAL_NEXTTOKE.ival = 0;
5428 	    force_next(FORMRBRACK);
5429 	    TOKEN(';');
5430 	}
5431 	goto retry;
5432     case '\r':
5433 #ifdef PERL_STRICT_CR
5434 	Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
5435 	Perl_croak(aTHX_
5436       "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
5437 #endif
5438     case ' ': case '\t': case '\f': case 013:
5439 #ifdef PERL_MAD
5440 	PL_realtokenstart = -1;
5441 	if (PL_madskills) {
5442 	  if (!PL_thiswhite)
5443 	    PL_thiswhite = newSVpvs("");
5444 	  sv_catpvn(PL_thiswhite, s, 1);
5445 	}
5446 #endif
5447 	s++;
5448 	goto retry;
5449     case '#':
5450     case '\n':
5451 #ifdef PERL_MAD
5452 	PL_realtokenstart = -1;
5453 	if (PL_madskills)
5454 	    PL_faketokens = 0;
5455 #endif
5456 	if (PL_lex_state != LEX_NORMAL ||
5457 	     (PL_in_eval && !PL_rsfp && !PL_parser->filtered)) {
5458 	    if (*s == '#' && s == PL_linestart && PL_in_eval
5459 	     && !PL_rsfp && !PL_parser->filtered) {
5460 		/* handle eval qq[#line 1 "foo"\n ...] */
5461 		CopLINE_dec(PL_curcop);
5462 		incline(s);
5463 	    }
5464 	    if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
5465 		s = SKIPSPACE0(s);
5466 		if (!PL_in_eval || PL_rsfp || PL_parser->filtered)
5467 		    incline(s);
5468 	    }
5469 	    else {
5470 		const bool in_comment = *s == '#';
5471 		d = s;
5472 		while (d < PL_bufend && *d != '\n')
5473 		    d++;
5474 		if (d < PL_bufend)
5475 		    d++;
5476 		else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
5477 		    Perl_croak(aTHX_ "panic: input overflow, %p > %p",
5478 			       d, PL_bufend);
5479 #ifdef PERL_MAD
5480 		if (PL_madskills)
5481 		    PL_thiswhite = newSVpvn(s, d - s);
5482 #endif
5483 		s = d;
5484 		if (in_comment && d == PL_bufend
5485 		 && PL_lex_state == LEX_INTERPNORMAL
5486 		 && PL_lex_inwhat == OP_SUBST && PL_lex_repl == PL_linestr
5487 		 && SvEVALED(PL_lex_repl) && d[-1] == '}') s--;
5488 		else incline(s);
5489 	    }
5490 	    if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
5491 		PL_lex_state = LEX_FORMLINE;
5492 		start_force(PL_curforce);
5493 		NEXTVAL_NEXTTOKE.ival = 0;
5494 		force_next(FORMRBRACK);
5495 		TOKEN(';');
5496 	    }
5497 	}
5498 	else {
5499 #ifdef PERL_MAD
5500 	    if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
5501 		if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
5502 		    PL_faketokens = 0;
5503 		    s = SKIPSPACE0(s);
5504 		    TOKEN(PEG);	/* make sure any #! line is accessible */
5505 		}
5506 		s = SKIPSPACE0(s);
5507 	    }
5508 	    else {
5509 /*		if (PL_madskills && PL_lex_formbrack) { */
5510 		    d = s;
5511 		    while (d < PL_bufend && *d != '\n')
5512 			d++;
5513 		    if (d < PL_bufend)
5514 			d++;
5515 		    else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
5516 		      Perl_croak(aTHX_ "panic: input overflow");
5517 		    if (PL_madskills && CopLINE(PL_curcop) >= 1) {
5518 			if (!PL_thiswhite)
5519 			    PL_thiswhite = newSVpvs("");
5520 			if (CopLINE(PL_curcop) == 1) {
5521 			    sv_setpvs(PL_thiswhite, "");
5522 			    PL_faketokens = 0;
5523 			}
5524 			sv_catpvn(PL_thiswhite, s, d - s);
5525 		    }
5526 		    s = d;
5527 /*		}
5528 		*s = '\0';
5529 		PL_bufend = s; */
5530 	    }
5531 #else
5532 	    *s = '\0';
5533 	    PL_bufend = s;
5534 #endif
5535 	}
5536 	goto retry;
5537     case '-':
5538 	if (s[1] && isALPHA(s[1]) && !isWORDCHAR(s[2])) {
5539 	    I32 ftst = 0;
5540 	    char tmp;
5541 
5542 	    s++;
5543 	    PL_bufptr = s;
5544 	    tmp = *s++;
5545 
5546 	    while (s < PL_bufend && SPACE_OR_TAB(*s))
5547 		s++;
5548 
5549 	    if (strnEQ(s,"=>",2)) {
5550 		s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
5551 		DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
5552 		OPERATOR('-');		/* unary minus */
5553 	    }
5554 	    PL_last_uni = PL_oldbufptr;
5555 	    switch (tmp) {
5556 	    case 'r': ftst = OP_FTEREAD;	break;
5557 	    case 'w': ftst = OP_FTEWRITE;	break;
5558 	    case 'x': ftst = OP_FTEEXEC;	break;
5559 	    case 'o': ftst = OP_FTEOWNED;	break;
5560 	    case 'R': ftst = OP_FTRREAD;	break;
5561 	    case 'W': ftst = OP_FTRWRITE;	break;
5562 	    case 'X': ftst = OP_FTREXEC;	break;
5563 	    case 'O': ftst = OP_FTROWNED;	break;
5564 	    case 'e': ftst = OP_FTIS;		break;
5565 	    case 'z': ftst = OP_FTZERO;		break;
5566 	    case 's': ftst = OP_FTSIZE;		break;
5567 	    case 'f': ftst = OP_FTFILE;		break;
5568 	    case 'd': ftst = OP_FTDIR;		break;
5569 	    case 'l': ftst = OP_FTLINK;		break;
5570 	    case 'p': ftst = OP_FTPIPE;		break;
5571 	    case 'S': ftst = OP_FTSOCK;		break;
5572 	    case 'u': ftst = OP_FTSUID;		break;
5573 	    case 'g': ftst = OP_FTSGID;		break;
5574 	    case 'k': ftst = OP_FTSVTX;		break;
5575 	    case 'b': ftst = OP_FTBLK;		break;
5576 	    case 'c': ftst = OP_FTCHR;		break;
5577 	    case 't': ftst = OP_FTTTY;		break;
5578 	    case 'T': ftst = OP_FTTEXT;		break;
5579 	    case 'B': ftst = OP_FTBINARY;	break;
5580 	    case 'M': case 'A': case 'C':
5581 		gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
5582 		switch (tmp) {
5583 		case 'M': ftst = OP_FTMTIME;	break;
5584 		case 'A': ftst = OP_FTATIME;	break;
5585 		case 'C': ftst = OP_FTCTIME;	break;
5586 		default:			break;
5587 		}
5588 		break;
5589 	    default:
5590 		break;
5591 	    }
5592 	    if (ftst) {
5593 		PL_last_lop_op = (OPCODE)ftst;
5594 		DEBUG_T( { PerlIO_printf(Perl_debug_log,
5595                         "### Saw file test %c\n", (int)tmp);
5596 		} );
5597 		FTST(ftst);
5598 	    }
5599 	    else {
5600 		/* Assume it was a minus followed by a one-letter named
5601 		 * subroutine call (or a -bareword), then. */
5602 		DEBUG_T( { PerlIO_printf(Perl_debug_log,
5603 			"### '-%c' looked like a file test but was not\n",
5604 			(int) tmp);
5605 		} );
5606 		s = --PL_bufptr;
5607 	    }
5608 	}
5609 	{
5610 	    const char tmp = *s++;
5611 	    if (*s == tmp) {
5612 		s++;
5613 		if (PL_expect == XOPERATOR)
5614 		    TERM(POSTDEC);
5615 		else
5616 		    OPERATOR(PREDEC);
5617 	    }
5618 	    else if (*s == '>') {
5619 		s++;
5620 		s = SKIPSPACE1(s);
5621 		if (isIDFIRST_lazy_if(s,UTF)) {
5622 		    s = force_word(s,METHOD,FALSE,TRUE,FALSE);
5623 		    TOKEN(ARROW);
5624 		}
5625 		else if (*s == '$')
5626 		    OPERATOR(ARROW);
5627 		else
5628 		    TERM(ARROW);
5629 	    }
5630 	    if (PL_expect == XOPERATOR) {
5631 		if (*s == '=' && !PL_lex_allbrackets &&
5632 			PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
5633 		    s--;
5634 		    TOKEN(0);
5635 		}
5636 		Aop(OP_SUBTRACT);
5637 	    }
5638 	    else {
5639 		if (isSPACE(*s) || !isSPACE(*PL_bufptr))
5640 		    check_uni();
5641 		OPERATOR('-');		/* unary minus */
5642 	    }
5643 	}
5644 
5645     case '+':
5646 	{
5647 	    const char tmp = *s++;
5648 	    if (*s == tmp) {
5649 		s++;
5650 		if (PL_expect == XOPERATOR)
5651 		    TERM(POSTINC);
5652 		else
5653 		    OPERATOR(PREINC);
5654 	    }
5655 	    if (PL_expect == XOPERATOR) {
5656 		if (*s == '=' && !PL_lex_allbrackets &&
5657 			PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
5658 		    s--;
5659 		    TOKEN(0);
5660 		}
5661 		Aop(OP_ADD);
5662 	    }
5663 	    else {
5664 		if (isSPACE(*s) || !isSPACE(*PL_bufptr))
5665 		    check_uni();
5666 		OPERATOR('+');
5667 	    }
5668 	}
5669 
5670     case '*':
5671 	if (PL_expect != XOPERATOR) {
5672 	    s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
5673 	    PL_expect = XOPERATOR;
5674 	    force_ident(PL_tokenbuf, '*');
5675 	    if (!*PL_tokenbuf)
5676 		PREREF('*');
5677 	    TERM('*');
5678 	}
5679 	s++;
5680 	if (*s == '*') {
5681 	    s++;
5682 	    if (*s == '=' && !PL_lex_allbrackets &&
5683 		    PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
5684 		s -= 2;
5685 		TOKEN(0);
5686 	    }
5687 	    PWop(OP_POW);
5688 	}
5689 	if (*s == '=' && !PL_lex_allbrackets &&
5690 		PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
5691 	    s--;
5692 	    TOKEN(0);
5693 	}
5694 	Mop(OP_MULTIPLY);
5695 
5696     case '%':
5697 	if (PL_expect == XOPERATOR) {
5698 	    if (s[1] == '=' && !PL_lex_allbrackets &&
5699 		    PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN)
5700 		TOKEN(0);
5701 	    ++s;
5702 	    Mop(OP_MODULO);
5703 	}
5704 	PL_tokenbuf[0] = '%';
5705 	s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
5706 		sizeof PL_tokenbuf - 1, FALSE);
5707 	if (!PL_tokenbuf[1]) {
5708 	    PREREF('%');
5709 	}
5710 	PL_expect = XOPERATOR;
5711 	force_ident_maybe_lex('%');
5712 	TERM('%');
5713 
5714     case '^':
5715 	if (!PL_lex_allbrackets && PL_lex_fakeeof >=
5716 		(s[1] == '=' ? LEX_FAKEEOF_ASSIGN : LEX_FAKEEOF_BITWISE))
5717 	    TOKEN(0);
5718 	s++;
5719 	BOop(OP_BIT_XOR);
5720     case '[':
5721 	if (PL_lex_brackets > 100)
5722 	    Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
5723 	PL_lex_brackstack[PL_lex_brackets++] = 0;
5724 	PL_lex_allbrackets++;
5725 	{
5726 	    const char tmp = *s++;
5727 	    OPERATOR(tmp);
5728 	}
5729     case '~':
5730 	if (s[1] == '~'
5731 	    && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
5732 	{
5733 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE)
5734 		TOKEN(0);
5735 	    s += 2;
5736             Perl_ck_warner_d(aTHX_
5737                 packWARN(WARN_EXPERIMENTAL__SMARTMATCH),
5738                 "Smartmatch is experimental");
5739 	    Eop(OP_SMARTMATCH);
5740 	}
5741 	s++;
5742 	OPERATOR('~');
5743     case ',':
5744 	if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMMA)
5745 	    TOKEN(0);
5746 	s++;
5747 	OPERATOR(',');
5748     case ':':
5749 	if (s[1] == ':') {
5750 	    len = 0;
5751 	    goto just_a_word_zero_gv;
5752 	}
5753 	s++;
5754 	switch (PL_expect) {
5755 	    OP *attrs;
5756 #ifdef PERL_MAD
5757 	    I32 stuffstart;
5758 #endif
5759 	case XOPERATOR:
5760 	    if (!PL_in_my || PL_lex_state != LEX_NORMAL)
5761 		break;
5762 	    PL_bufptr = s;	/* update in case we back off */
5763 	    if (*s == '=') {
5764 		Perl_croak(aTHX_
5765 			   "Use of := for an empty attribute list is not allowed");
5766 	    }
5767 	    goto grabattrs;
5768 	case XATTRBLOCK:
5769 	    PL_expect = XBLOCK;
5770 	    goto grabattrs;
5771 	case XATTRTERM:
5772 	    PL_expect = XTERMBLOCK;
5773 	 grabattrs:
5774 #ifdef PERL_MAD
5775 	    stuffstart = s - SvPVX(PL_linestr) - 1;
5776 #endif
5777 	    s = PEEKSPACE(s);
5778 	    attrs = NULL;
5779 	    while (isIDFIRST_lazy_if(s,UTF)) {
5780 		I32 tmp;
5781 		SV *sv;
5782 		d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5783 		if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
5784 		    if (tmp < 0) tmp = -tmp;
5785 		    switch (tmp) {
5786 		    case KEY_or:
5787 		    case KEY_and:
5788 		    case KEY_for:
5789 		    case KEY_foreach:
5790 		    case KEY_unless:
5791 		    case KEY_if:
5792 		    case KEY_while:
5793 		    case KEY_until:
5794 			goto got_attrs;
5795 		    default:
5796 			break;
5797 		    }
5798 		}
5799 		sv = newSVpvn_flags(s, len, UTF ? SVf_UTF8 : 0);
5800 		if (*d == '(') {
5801 		    d = scan_str(d,TRUE,TRUE,FALSE, FALSE);
5802 		    if (!d) {
5803 			/* MUST advance bufptr here to avoid bogus
5804 			   "at end of line" context messages from yyerror().
5805 			 */
5806 			PL_bufptr = s + len;
5807 			yyerror("Unterminated attribute parameter in attribute list");
5808 			if (attrs)
5809 			    op_free(attrs);
5810 			sv_free(sv);
5811 			return REPORT(0);	/* EOF indicator */
5812 		    }
5813 		}
5814 		if (PL_lex_stuff) {
5815 		    sv_catsv(sv, PL_lex_stuff);
5816 		    attrs = op_append_elem(OP_LIST, attrs,
5817 					newSVOP(OP_CONST, 0, sv));
5818 		    SvREFCNT_dec(PL_lex_stuff);
5819 		    PL_lex_stuff = NULL;
5820 		}
5821 		else {
5822 		    if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
5823 			sv_free(sv);
5824 			if (PL_in_my == KEY_our) {
5825 			    deprecate(":unique");
5826 			}
5827 			else
5828 			    Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
5829 		    }
5830 
5831 		    /* NOTE: any CV attrs applied here need to be part of
5832 		       the CVf_BUILTIN_ATTRS define in cv.h! */
5833 		    else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
5834 			sv_free(sv);
5835 			CvLVALUE_on(PL_compcv);
5836 		    }
5837 		    else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
5838 			sv_free(sv);
5839 			deprecate(":locked");
5840 		    }
5841 		    else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
5842 			sv_free(sv);
5843 			CvMETHOD_on(PL_compcv);
5844 		    }
5845 		    /* After we've set the flags, it could be argued that
5846 		       we don't need to do the attributes.pm-based setting
5847 		       process, and shouldn't bother appending recognized
5848 		       flags.  To experiment with that, uncomment the
5849 		       following "else".  (Note that's already been
5850 		       uncommented.  That keeps the above-applied built-in
5851 		       attributes from being intercepted (and possibly
5852 		       rejected) by a package's attribute routines, but is
5853 		       justified by the performance win for the common case
5854 		       of applying only built-in attributes.) */
5855 		    else
5856 		        attrs = op_append_elem(OP_LIST, attrs,
5857 					    newSVOP(OP_CONST, 0,
5858 					      	    sv));
5859 		}
5860 		s = PEEKSPACE(d);
5861 		if (*s == ':' && s[1] != ':')
5862 		    s = PEEKSPACE(s+1);
5863 		else if (s == d)
5864 		    break;	/* require real whitespace or :'s */
5865 		/* XXX losing whitespace on sequential attributes here */
5866 	    }
5867 	    {
5868 		const char tmp
5869 		    = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
5870 		if (*s != ';' && *s != '}' && *s != tmp
5871 		    && (tmp != '=' || *s != ')')) {
5872 		    const char q = ((*s == '\'') ? '"' : '\'');
5873 		    /* If here for an expression, and parsed no attrs, back
5874 		       off. */
5875 		    if (tmp == '=' && !attrs) {
5876 			s = PL_bufptr;
5877 			break;
5878 		    }
5879 		    /* MUST advance bufptr here to avoid bogus "at end of line"
5880 		       context messages from yyerror().
5881 		    */
5882 		    PL_bufptr = s;
5883 		    yyerror( (const char *)
5884 			     (*s
5885 			      ? Perl_form(aTHX_ "Invalid separator character "
5886 					  "%c%c%c in attribute list", q, *s, q)
5887 			      : "Unterminated attribute list" ) );
5888 		    if (attrs)
5889 			op_free(attrs);
5890 		    OPERATOR(':');
5891 		}
5892 	    }
5893 	got_attrs:
5894 	    if (attrs) {
5895 		start_force(PL_curforce);
5896 		NEXTVAL_NEXTTOKE.opval = attrs;
5897 		CURMAD('_', PL_nextwhite);
5898 		force_next(THING);
5899 	    }
5900 #ifdef PERL_MAD
5901 	    if (PL_madskills) {
5902 		PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
5903 				     (s - SvPVX(PL_linestr)) - stuffstart);
5904 	    }
5905 #endif
5906 	    TOKEN(COLONATTR);
5907 	}
5908 	if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_CLOSING) {
5909 	    s--;
5910 	    TOKEN(0);
5911 	}
5912 	PL_lex_allbrackets--;
5913 	OPERATOR(':');
5914     case '(':
5915 	s++;
5916 	if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
5917 	    PL_oldbufptr = PL_oldoldbufptr;		/* allow print(STDOUT 123) */
5918 	else
5919 	    PL_expect = XTERM;
5920 	s = SKIPSPACE1(s);
5921 	PL_lex_allbrackets++;
5922 	TOKEN('(');
5923     case ';':
5924 	if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_NONEXPR)
5925 	    TOKEN(0);
5926 	CLINE;
5927 	s++;
5928 	OPERATOR(';');
5929     case ')':
5930 	if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_CLOSING)
5931 	    TOKEN(0);
5932 	s++;
5933 	PL_lex_allbrackets--;
5934 	s = SKIPSPACE1(s);
5935 	if (*s == '{')
5936 	    PREBLOCK(')');
5937 	TERM(')');
5938     case ']':
5939 	if (PL_lex_brackets && PL_lex_brackstack[PL_lex_brackets-1] == XFAKEEOF)
5940 	    TOKEN(0);
5941 	s++;
5942 	if (PL_lex_brackets <= 0)
5943 	    yyerror("Unmatched right square bracket");
5944 	else
5945 	    --PL_lex_brackets;
5946 	PL_lex_allbrackets--;
5947 	if (PL_lex_state == LEX_INTERPNORMAL) {
5948 	    if (PL_lex_brackets == 0) {
5949 		if (*s == '-' && s[1] == '>')
5950 		    PL_lex_state = LEX_INTERPENDMAYBE;
5951 		else if (*s != '[' && *s != '{')
5952 		    PL_lex_state = LEX_INTERPEND;
5953 	    }
5954 	}
5955 	TERM(']');
5956     case '{':
5957 	s++;
5958       leftbracket:
5959 	if (PL_lex_brackets > 100) {
5960 	    Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
5961 	}
5962 	switch (PL_expect) {
5963 	case XTERM:
5964 	    PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5965 	    PL_lex_allbrackets++;
5966 	    OPERATOR(HASHBRACK);
5967 	case XOPERATOR:
5968 	    while (s < PL_bufend && SPACE_OR_TAB(*s))
5969 		s++;
5970 	    d = s;
5971 	    PL_tokenbuf[0] = '\0';
5972 	    if (d < PL_bufend && *d == '-') {
5973 		PL_tokenbuf[0] = '-';
5974 		d++;
5975 		while (d < PL_bufend && SPACE_OR_TAB(*d))
5976 		    d++;
5977 	    }
5978 	    if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
5979 		d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
5980 			      FALSE, &len);
5981 		while (d < PL_bufend && SPACE_OR_TAB(*d))
5982 		    d++;
5983 		if (*d == '}') {
5984 		    const char minus = (PL_tokenbuf[0] == '-');
5985 		    s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
5986 		    if (minus)
5987 			force_next('-');
5988 		}
5989 	    }
5990 	    /* FALL THROUGH */
5991 	case XATTRBLOCK:
5992 	case XBLOCK:
5993 	    PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
5994 	    PL_lex_allbrackets++;
5995 	    PL_expect = XSTATE;
5996 	    break;
5997 	case XATTRTERM:
5998 	case XTERMBLOCK:
5999 	    PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
6000 	    PL_lex_allbrackets++;
6001 	    PL_expect = XSTATE;
6002 	    break;
6003 	default: {
6004 		const char *t;
6005 		if (PL_oldoldbufptr == PL_last_lop)
6006 		    PL_lex_brackstack[PL_lex_brackets++] = XTERM;
6007 		else
6008 		    PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
6009 		PL_lex_allbrackets++;
6010 		s = SKIPSPACE1(s);
6011 		if (*s == '}') {
6012 		    if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
6013 			PL_expect = XTERM;
6014 			/* This hack is to get the ${} in the message. */
6015 			PL_bufptr = s+1;
6016 			yyerror("syntax error");
6017 			break;
6018 		    }
6019 		    OPERATOR(HASHBRACK);
6020 		}
6021 		/* This hack serves to disambiguate a pair of curlies
6022 		 * as being a block or an anon hash.  Normally, expectation
6023 		 * determines that, but in cases where we're not in a
6024 		 * position to expect anything in particular (like inside
6025 		 * eval"") we have to resolve the ambiguity.  This code
6026 		 * covers the case where the first term in the curlies is a
6027 		 * quoted string.  Most other cases need to be explicitly
6028 		 * disambiguated by prepending a "+" before the opening
6029 		 * curly in order to force resolution as an anon hash.
6030 		 *
6031 		 * XXX should probably propagate the outer expectation
6032 		 * into eval"" to rely less on this hack, but that could
6033 		 * potentially break current behavior of eval"".
6034 		 * GSAR 97-07-21
6035 		 */
6036 		t = s;
6037 		if (*s == '\'' || *s == '"' || *s == '`') {
6038 		    /* common case: get past first string, handling escapes */
6039 		    for (t++; t < PL_bufend && *t != *s;)
6040 			if (*t++ == '\\' && (*t == '\\' || *t == *s))
6041 			    t++;
6042 		    t++;
6043 		}
6044 		else if (*s == 'q') {
6045 		    if (++t < PL_bufend
6046 			&& (!isWORDCHAR(*t)
6047 			    || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
6048 				&& !isWORDCHAR(*t))))
6049 		    {
6050 			/* skip q//-like construct */
6051 			const char *tmps;
6052 			char open, close, term;
6053 			I32 brackets = 1;
6054 
6055 			while (t < PL_bufend && isSPACE(*t))
6056 			    t++;
6057 			/* check for q => */
6058 			if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
6059 			    OPERATOR(HASHBRACK);
6060 			}
6061 			term = *t;
6062 			open = term;
6063 			if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6064 			    term = tmps[5];
6065 			close = term;
6066 			if (open == close)
6067 			    for (t++; t < PL_bufend; t++) {
6068 				if (*t == '\\' && t+1 < PL_bufend && open != '\\')
6069 				    t++;
6070 				else if (*t == open)
6071 				    break;
6072 			    }
6073 			else {
6074 			    for (t++; t < PL_bufend; t++) {
6075 				if (*t == '\\' && t+1 < PL_bufend)
6076 				    t++;
6077 				else if (*t == close && --brackets <= 0)
6078 				    break;
6079 				else if (*t == open)
6080 				    brackets++;
6081 			    }
6082 			}
6083 			t++;
6084 		    }
6085 		    else
6086 			/* skip plain q word */
6087 			while (t < PL_bufend && isWORDCHAR_lazy_if(t,UTF))
6088 			     t += UTF8SKIP(t);
6089 		}
6090 		else if (isWORDCHAR_lazy_if(t,UTF)) {
6091 		    t += UTF8SKIP(t);
6092 		    while (t < PL_bufend && isWORDCHAR_lazy_if(t,UTF))
6093 			 t += UTF8SKIP(t);
6094 		}
6095 		while (t < PL_bufend && isSPACE(*t))
6096 		    t++;
6097 		/* if comma follows first term, call it an anon hash */
6098 		/* XXX it could be a comma expression with loop modifiers */
6099 		if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
6100 				   || (*t == '=' && t[1] == '>')))
6101 		    OPERATOR(HASHBRACK);
6102 		if (PL_expect == XREF)
6103 		    PL_expect = XTERM;
6104 		else {
6105 		    PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
6106 		    PL_expect = XSTATE;
6107 		}
6108 	    }
6109 	    break;
6110 	}
6111 	pl_yylval.ival = CopLINE(PL_curcop);
6112 	if (isSPACE(*s) || *s == '#')
6113 	    PL_copline = NOLINE;   /* invalidate current command line number */
6114 	TOKEN(formbrack ? '=' : '{');
6115     case '}':
6116 	if (PL_lex_brackets && PL_lex_brackstack[PL_lex_brackets-1] == XFAKEEOF)
6117 	    TOKEN(0);
6118       rightbracket:
6119 	s++;
6120 	if (PL_lex_brackets <= 0)
6121 	    yyerror("Unmatched right curly bracket");
6122 	else
6123 	    PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
6124 	PL_lex_allbrackets--;
6125 	if (PL_lex_state == LEX_INTERPNORMAL) {
6126 	    if (PL_lex_brackets == 0) {
6127 		if (PL_expect & XFAKEBRACK) {
6128 		    PL_expect &= XENUMMASK;
6129 		    PL_lex_state = LEX_INTERPEND;
6130 		    PL_bufptr = s;
6131 #if 0
6132 		    if (PL_madskills) {
6133 			if (!PL_thiswhite)
6134 			    PL_thiswhite = newSVpvs("");
6135 			sv_catpvs(PL_thiswhite,"}");
6136 		    }
6137 #endif
6138 		    return yylex();	/* ignore fake brackets */
6139 		}
6140 		if (PL_lex_inwhat == OP_SUBST && PL_lex_repl == PL_linestr
6141 		 && SvEVALED(PL_lex_repl))
6142 		    PL_lex_state = LEX_INTERPEND;
6143 		else if (*s == '-' && s[1] == '>')
6144 		    PL_lex_state = LEX_INTERPENDMAYBE;
6145 		else if (*s != '[' && *s != '{')
6146 		    PL_lex_state = LEX_INTERPEND;
6147 	    }
6148 	}
6149 	if (PL_expect & XFAKEBRACK) {
6150 	    PL_expect &= XENUMMASK;
6151 	    PL_bufptr = s;
6152 	    return yylex();		/* ignore fake brackets */
6153 	}
6154 	start_force(PL_curforce);
6155 	if (PL_madskills) {
6156 	    curmad('X', newSVpvn(s-1,1));
6157 	    CURMAD('_', PL_thiswhite);
6158 	}
6159 	force_next(formbrack ? '.' : '}');
6160 	if (formbrack) LEAVE;
6161 #ifdef PERL_MAD
6162 	if (PL_madskills && !PL_thistoken)
6163 	    PL_thistoken = newSVpvs("");
6164 #endif
6165 	if (formbrack == 2) { /* means . where arguments were expected */
6166 	    start_force(PL_curforce);
6167 	    force_next(';');
6168 	    TOKEN(FORMRBRACK);
6169 	}
6170 	TOKEN(';');
6171     case '&':
6172 	s++;
6173 	if (*s++ == '&') {
6174 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >=
6175 		    (*s == '=' ? LEX_FAKEEOF_ASSIGN : LEX_FAKEEOF_LOGIC)) {
6176 		s -= 2;
6177 		TOKEN(0);
6178 	    }
6179 	    AOPERATOR(ANDAND);
6180 	}
6181 	s--;
6182 	if (PL_expect == XOPERATOR) {
6183 	    if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
6184 		&& isIDFIRST_lazy_if(s,UTF))
6185 	    {
6186 		CopLINE_dec(PL_curcop);
6187 		Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
6188 		CopLINE_inc(PL_curcop);
6189 	    }
6190 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >=
6191 		    (*s == '=' ? LEX_FAKEEOF_ASSIGN : LEX_FAKEEOF_BITWISE)) {
6192 		s--;
6193 		TOKEN(0);
6194 	    }
6195 	    BAop(OP_BIT_AND);
6196 	}
6197 
6198 	PL_tokenbuf[0] = '&';
6199 	s = scan_ident(s - 1, PL_bufend, PL_tokenbuf + 1,
6200 		       sizeof PL_tokenbuf - 1, TRUE);
6201 	if (PL_tokenbuf[1]) {
6202 	    PL_expect = XOPERATOR;
6203 	    force_ident_maybe_lex('&');
6204 	}
6205 	else
6206 	    PREREF('&');
6207 	pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
6208 	TERM('&');
6209 
6210     case '|':
6211 	s++;
6212 	if (*s++ == '|') {
6213 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >=
6214 		    (*s == '=' ? LEX_FAKEEOF_ASSIGN : LEX_FAKEEOF_LOGIC)) {
6215 		s -= 2;
6216 		TOKEN(0);
6217 	    }
6218 	    AOPERATOR(OROR);
6219 	}
6220 	s--;
6221 	if (!PL_lex_allbrackets && PL_lex_fakeeof >=
6222 		(*s == '=' ? LEX_FAKEEOF_ASSIGN : LEX_FAKEEOF_BITWISE)) {
6223 	    s--;
6224 	    TOKEN(0);
6225 	}
6226 	BOop(OP_BIT_OR);
6227     case '=':
6228 	s++;
6229 	{
6230 	    const char tmp = *s++;
6231 	    if (tmp == '=') {
6232 		if (!PL_lex_allbrackets &&
6233 			PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE) {
6234 		    s -= 2;
6235 		    TOKEN(0);
6236 		}
6237 		Eop(OP_EQ);
6238 	    }
6239 	    if (tmp == '>') {
6240 		if (!PL_lex_allbrackets &&
6241 			PL_lex_fakeeof >= LEX_FAKEEOF_COMMA) {
6242 		    s -= 2;
6243 		    TOKEN(0);
6244 		}
6245 		OPERATOR(',');
6246 	    }
6247 	    if (tmp == '~')
6248 		PMop(OP_MATCH);
6249 	    if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
6250 		&& strchr("+-*/%.^&|<",tmp))
6251 		Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6252 			    "Reversed %c= operator",(int)tmp);
6253 	    s--;
6254 	    if (PL_expect == XSTATE && isALPHA(tmp) &&
6255 		(s == PL_linestart+1 || s[-2] == '\n') )
6256 		{
6257 		    if ((PL_in_eval && !PL_rsfp && !PL_parser->filtered)
6258 			|| PL_lex_state != LEX_NORMAL) {
6259 			d = PL_bufend;
6260 			while (s < d) {
6261 			    if (*s++ == '\n') {
6262 				incline(s);
6263 				if (strnEQ(s,"=cut",4)) {
6264 				    s = strchr(s,'\n');
6265 				    if (s)
6266 					s++;
6267 				    else
6268 					s = d;
6269 				    incline(s);
6270 				    goto retry;
6271 				}
6272 			    }
6273 			}
6274 			goto retry;
6275 		    }
6276 #ifdef PERL_MAD
6277 		    if (PL_madskills) {
6278 			if (!PL_thiswhite)
6279 			    PL_thiswhite = newSVpvs("");
6280 			sv_catpvn(PL_thiswhite, PL_linestart,
6281 				  PL_bufend - PL_linestart);
6282 		    }
6283 #endif
6284 		    s = PL_bufend;
6285 		    PL_parser->in_pod = 1;
6286 		    goto retry;
6287 		}
6288 	}
6289 	if (PL_expect == XBLOCK) {
6290 	    const char *t = s;
6291 #ifdef PERL_STRICT_CR
6292 	    while (SPACE_OR_TAB(*t))
6293 #else
6294 	    while (SPACE_OR_TAB(*t) || *t == '\r')
6295 #endif
6296 		t++;
6297 	    if (*t == '\n' || *t == '#') {
6298 		formbrack = 1;
6299 		ENTER;
6300 		SAVEI8(PL_parser->form_lex_state);
6301 		SAVEI32(PL_lex_formbrack);
6302 		PL_parser->form_lex_state = PL_lex_state;
6303 		PL_lex_formbrack = PL_lex_brackets + 1;
6304 		goto leftbracket;
6305 	    }
6306 	}
6307 	if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
6308 	    s--;
6309 	    TOKEN(0);
6310 	}
6311 	pl_yylval.ival = 0;
6312 	OPERATOR(ASSIGNOP);
6313     case '!':
6314 	s++;
6315 	{
6316 	    const char tmp = *s++;
6317 	    if (tmp == '=') {
6318 		/* was this !=~ where !~ was meant?
6319 		 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
6320 
6321 		if (*s == '~' && ckWARN(WARN_SYNTAX)) {
6322 		    const char *t = s+1;
6323 
6324 		    while (t < PL_bufend && isSPACE(*t))
6325 			++t;
6326 
6327 		    if (*t == '/' || *t == '?' ||
6328 			((*t == 'm' || *t == 's' || *t == 'y')
6329 			 && !isWORDCHAR(t[1])) ||
6330 			(*t == 't' && t[1] == 'r' && !isWORDCHAR(t[2])))
6331 			Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6332 				    "!=~ should be !~");
6333 		}
6334 		if (!PL_lex_allbrackets &&
6335 			PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE) {
6336 		    s -= 2;
6337 		    TOKEN(0);
6338 		}
6339 		Eop(OP_NE);
6340 	    }
6341 	    if (tmp == '~')
6342 		PMop(OP_NOT);
6343 	}
6344 	s--;
6345 	OPERATOR('!');
6346     case '<':
6347 	if (PL_expect != XOPERATOR) {
6348 	    if (s[1] != '<' && !strchr(s,'>'))
6349 		check_uni();
6350 	    if (s[1] == '<')
6351 		s = scan_heredoc(s);
6352 	    else
6353 		s = scan_inputsymbol(s);
6354 	    PL_expect = XOPERATOR;
6355 	    TOKEN(sublex_start());
6356 	}
6357 	s++;
6358 	{
6359 	    char tmp = *s++;
6360 	    if (tmp == '<') {
6361 		if (*s == '=' && !PL_lex_allbrackets &&
6362 			PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
6363 		    s -= 2;
6364 		    TOKEN(0);
6365 		}
6366 		SHop(OP_LEFT_SHIFT);
6367 	    }
6368 	    if (tmp == '=') {
6369 		tmp = *s++;
6370 		if (tmp == '>') {
6371 		    if (!PL_lex_allbrackets &&
6372 			    PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE) {
6373 			s -= 3;
6374 			TOKEN(0);
6375 		    }
6376 		    Eop(OP_NCMP);
6377 		}
6378 		s--;
6379 		if (!PL_lex_allbrackets &&
6380 			PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE) {
6381 		    s -= 2;
6382 		    TOKEN(0);
6383 		}
6384 		Rop(OP_LE);
6385 	    }
6386 	}
6387 	s--;
6388 	if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE) {
6389 	    s--;
6390 	    TOKEN(0);
6391 	}
6392 	Rop(OP_LT);
6393     case '>':
6394 	s++;
6395 	{
6396 	    const char tmp = *s++;
6397 	    if (tmp == '>') {
6398 		if (*s == '=' && !PL_lex_allbrackets &&
6399 			PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
6400 		    s -= 2;
6401 		    TOKEN(0);
6402 		}
6403 		SHop(OP_RIGHT_SHIFT);
6404 	    }
6405 	    else if (tmp == '=') {
6406 		if (!PL_lex_allbrackets &&
6407 			PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE) {
6408 		    s -= 2;
6409 		    TOKEN(0);
6410 		}
6411 		Rop(OP_GE);
6412 	    }
6413 	}
6414 	s--;
6415 	if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE) {
6416 	    s--;
6417 	    TOKEN(0);
6418 	}
6419 	Rop(OP_GT);
6420 
6421     case '$':
6422 	CLINE;
6423 
6424 	if (PL_expect == XOPERATOR) {
6425 	    if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
6426 		return deprecate_commaless_var_list();
6427 	    }
6428 	}
6429 
6430 	if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-@", s[2]))) {
6431 	    PL_tokenbuf[0] = '@';
6432 	    s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
6433 			   sizeof PL_tokenbuf - 1, FALSE);
6434 	    if (PL_expect == XOPERATOR)
6435 		no_op("Array length", s);
6436 	    if (!PL_tokenbuf[1])
6437 		PREREF(DOLSHARP);
6438 	    PL_expect = XOPERATOR;
6439 	    force_ident_maybe_lex('#');
6440 	    TOKEN(DOLSHARP);
6441 	}
6442 
6443 	PL_tokenbuf[0] = '$';
6444 	s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
6445 		       sizeof PL_tokenbuf - 1, FALSE);
6446 	if (PL_expect == XOPERATOR)
6447 	    no_op("Scalar", s);
6448 	if (!PL_tokenbuf[1]) {
6449 	    if (s == PL_bufend)
6450 		yyerror("Final $ should be \\$ or $name");
6451 	    PREREF('$');
6452 	}
6453 
6454 	d = s;
6455 	{
6456 	    const char tmp = *s;
6457 	    if (PL_lex_state == LEX_NORMAL || PL_lex_brackets)
6458 		s = SKIPSPACE1(s);
6459 
6460 	    if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
6461 		&& intuit_more(s)) {
6462 		if (*s == '[') {
6463 		    PL_tokenbuf[0] = '@';
6464 		    if (ckWARN(WARN_SYNTAX)) {
6465 			char *t = s+1;
6466 
6467 			while (isSPACE(*t) || isWORDCHAR_lazy_if(t,UTF) || *t == '$')
6468 			    t++;
6469 			if (*t++ == ',') {
6470 			    PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
6471 			    while (t < PL_bufend && *t != ']')
6472 				t++;
6473 			    Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6474 					"Multidimensional syntax %.*s not supported",
6475 				    (int)((t - PL_bufptr) + 1), PL_bufptr);
6476 			}
6477 		    }
6478 		}
6479 		else if (*s == '{') {
6480 		    char *t;
6481 		    PL_tokenbuf[0] = '%';
6482 		    if (strEQ(PL_tokenbuf+1, "SIG")  && ckWARN(WARN_SYNTAX)
6483 			&& (t = strchr(s, '}')) && (t = strchr(t, '=')))
6484 			{
6485 			    char tmpbuf[sizeof PL_tokenbuf];
6486 			    do {
6487 				t++;
6488 			    } while (isSPACE(*t));
6489 			    if (isIDFIRST_lazy_if(t,UTF)) {
6490 				STRLEN len;
6491 				t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
6492 					      &len);
6493 				while (isSPACE(*t))
6494 				    t++;
6495 				if (*t == ';'
6496                                        && get_cvn_flags(tmpbuf, len, UTF ? SVf_UTF8 : 0))
6497 				    Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6498 						"You need to quote \"%"SVf"\"",
6499 						  SVfARG(newSVpvn_flags(tmpbuf, len,
6500                                                     SVs_TEMP | (UTF ? SVf_UTF8 : 0))));
6501 			    }
6502 			}
6503 		}
6504 	    }
6505 
6506 	    PL_expect = XOPERATOR;
6507 	    if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
6508 		const bool islop = (PL_last_lop == PL_oldoldbufptr);
6509 		if (!islop || PL_last_lop_op == OP_GREPSTART)
6510 		    PL_expect = XOPERATOR;
6511 		else if (strchr("$@\"'`q", *s))
6512 		    PL_expect = XTERM;		/* e.g. print $fh "foo" */
6513 		else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
6514 		    PL_expect = XTERM;		/* e.g. print $fh &sub */
6515 		else if (isIDFIRST_lazy_if(s,UTF)) {
6516 		    char tmpbuf[sizeof PL_tokenbuf];
6517 		    int t2;
6518 		    scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6519 		    if ((t2 = keyword(tmpbuf, len, 0))) {
6520 			/* binary operators exclude handle interpretations */
6521 			switch (t2) {
6522 			case -KEY_x:
6523 			case -KEY_eq:
6524 			case -KEY_ne:
6525 			case -KEY_gt:
6526 			case -KEY_lt:
6527 			case -KEY_ge:
6528 			case -KEY_le:
6529 			case -KEY_cmp:
6530 			    break;
6531 			default:
6532 			    PL_expect = XTERM;	/* e.g. print $fh length() */
6533 			    break;
6534 			}
6535 		    }
6536 		    else {
6537 			PL_expect = XTERM;	/* e.g. print $fh subr() */
6538 		    }
6539 		}
6540 		else if (isDIGIT(*s))
6541 		    PL_expect = XTERM;		/* e.g. print $fh 3 */
6542 		else if (*s == '.' && isDIGIT(s[1]))
6543 		    PL_expect = XTERM;		/* e.g. print $fh .3 */
6544 		else if ((*s == '?' || *s == '-' || *s == '+')
6545 			 && !isSPACE(s[1]) && s[1] != '=')
6546 		    PL_expect = XTERM;		/* e.g. print $fh -1 */
6547 		else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
6548 			 && s[1] != '/')
6549 		    PL_expect = XTERM;		/* e.g. print $fh /.../
6550 						   XXX except DORDOR operator
6551 						*/
6552 		else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
6553 			 && s[2] != '=')
6554 		    PL_expect = XTERM;		/* print $fh <<"EOF" */
6555 	    }
6556 	}
6557 	force_ident_maybe_lex('$');
6558 	TOKEN('$');
6559 
6560     case '@':
6561 	if (PL_expect == XOPERATOR)
6562 	    no_op("Array", s);
6563 	PL_tokenbuf[0] = '@';
6564 	s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
6565 	if (!PL_tokenbuf[1]) {
6566 	    PREREF('@');
6567 	}
6568 	if (PL_lex_state == LEX_NORMAL)
6569 	    s = SKIPSPACE1(s);
6570 	if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
6571 	    if (*s == '{')
6572 		PL_tokenbuf[0] = '%';
6573 
6574 	    /* Warn about @ where they meant $. */
6575 	    if (*s == '[' || *s == '{') {
6576 		if (ckWARN(WARN_SYNTAX)) {
6577 		    const char *t = s + 1;
6578 		    while (*t && (isWORDCHAR_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
6579 			t += UTF ? UTF8SKIP(t) : 1;
6580 		    if (*t == '}' || *t == ']') {
6581 			t++;
6582 			PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
6583        /* diag_listed_as: Scalar value @%s[%s] better written as $%s[%s] */
6584 			Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6585 			    "Scalar value %"SVf" better written as $%"SVf,
6586 			    SVfARG(newSVpvn_flags(PL_bufptr, (STRLEN)(t-PL_bufptr),
6587                                                 SVs_TEMP | (UTF ? SVf_UTF8 : 0 ))),
6588                             SVfARG(newSVpvn_flags(PL_bufptr+1, (STRLEN)(t-PL_bufptr-1),
6589                                                 SVs_TEMP | (UTF ? SVf_UTF8 : 0 ))));
6590 		    }
6591 		}
6592 	    }
6593 	}
6594 	PL_expect = XOPERATOR;
6595 	force_ident_maybe_lex('@');
6596 	TERM('@');
6597 
6598      case '/':			/* may be division, defined-or, or pattern */
6599 	if (PL_expect == XTERMORDORDOR && s[1] == '/') {
6600 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >=
6601 		    (s[2] == '=' ? LEX_FAKEEOF_ASSIGN : LEX_FAKEEOF_LOGIC))
6602 		TOKEN(0);
6603 	    s += 2;
6604 	    AOPERATOR(DORDOR);
6605 	}
6606      case '?':			/* may either be conditional or pattern */
6607 	if (PL_expect == XOPERATOR) {
6608 	     char tmp = *s++;
6609 	     if(tmp == '?') {
6610 		if (!PL_lex_allbrackets &&
6611 			PL_lex_fakeeof >= LEX_FAKEEOF_IFELSE) {
6612 		    s--;
6613 		    TOKEN(0);
6614 		}
6615 		PL_lex_allbrackets++;
6616 		OPERATOR('?');
6617 	     }
6618              else {
6619 	         tmp = *s++;
6620 	         if(tmp == '/') {
6621 	             /* A // operator. */
6622 		    if (!PL_lex_allbrackets && PL_lex_fakeeof >=
6623 			    (*s == '=' ? LEX_FAKEEOF_ASSIGN :
6624 					    LEX_FAKEEOF_LOGIC)) {
6625 			s -= 2;
6626 			TOKEN(0);
6627 		    }
6628 	            AOPERATOR(DORDOR);
6629 	         }
6630 	         else {
6631 	             s--;
6632 		     if (*s == '=' && !PL_lex_allbrackets &&
6633 			     PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
6634 			 s--;
6635 			 TOKEN(0);
6636 		     }
6637 	             Mop(OP_DIVIDE);
6638 	         }
6639 	     }
6640 	 }
6641 	 else {
6642 	     /* Disable warning on "study /blah/" */
6643 	     if (PL_oldoldbufptr == PL_last_uni
6644 	      && (*PL_last_uni != 's' || s - PL_last_uni < 5
6645 	          || memNE(PL_last_uni, "study", 5)
6646 	          || isWORDCHAR_lazy_if(PL_last_uni+5,UTF)
6647 	      ))
6648 	         check_uni();
6649 	     if (*s == '?')
6650 		 deprecate("?PATTERN? without explicit operator");
6651 	     s = scan_pat(s,OP_MATCH);
6652 	     TERM(sublex_start());
6653 	 }
6654 
6655     case '.':
6656 	if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
6657 #ifdef PERL_STRICT_CR
6658 	    && s[1] == '\n'
6659 #else
6660 	    && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
6661 #endif
6662 	    && (s == PL_linestart || s[-1] == '\n') )
6663 	{
6664 	    PL_expect = XSTATE;
6665 	    formbrack = 2; /* dot seen where arguments expected */
6666 	    goto rightbracket;
6667 	}
6668 	if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
6669 	    s += 3;
6670 	    OPERATOR(YADAYADA);
6671 	}
6672 	if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
6673 	    char tmp = *s++;
6674 	    if (*s == tmp) {
6675 		if (!PL_lex_allbrackets &&
6676 			PL_lex_fakeeof >= LEX_FAKEEOF_RANGE) {
6677 		    s--;
6678 		    TOKEN(0);
6679 		}
6680 		s++;
6681 		if (*s == tmp) {
6682 		    s++;
6683 		    pl_yylval.ival = OPf_SPECIAL;
6684 		}
6685 		else
6686 		    pl_yylval.ival = 0;
6687 		OPERATOR(DOTDOT);
6688 	    }
6689 	    if (*s == '=' && !PL_lex_allbrackets &&
6690 		    PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN) {
6691 		s--;
6692 		TOKEN(0);
6693 	    }
6694 	    Aop(OP_CONCAT);
6695 	}
6696 	/* FALL THROUGH */
6697     case '0': case '1': case '2': case '3': case '4':
6698     case '5': case '6': case '7': case '8': case '9':
6699 	s = scan_num(s, &pl_yylval);
6700 	DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
6701 	if (PL_expect == XOPERATOR)
6702 	    no_op("Number",s);
6703 	TERM(THING);
6704 
6705     case '\'':
6706 	s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
6707 	DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
6708 	if (PL_expect == XOPERATOR) {
6709 	    if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
6710 		return deprecate_commaless_var_list();
6711 	    }
6712 	    else
6713 		no_op("String",s);
6714 	}
6715 	if (!s)
6716 	    missingterm(NULL);
6717 	pl_yylval.ival = OP_CONST;
6718 	TERM(sublex_start());
6719 
6720     case '"':
6721 	s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
6722 	DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
6723 	if (PL_expect == XOPERATOR) {
6724 	    if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
6725 		return deprecate_commaless_var_list();
6726 	    }
6727 	    else
6728 		no_op("String",s);
6729 	}
6730 	if (!s)
6731 	    missingterm(NULL);
6732 	pl_yylval.ival = OP_CONST;
6733 	/* FIXME. I think that this can be const if char *d is replaced by
6734 	   more localised variables.  */
6735 	for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
6736 	    if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
6737 		pl_yylval.ival = OP_STRINGIFY;
6738 		break;
6739 	    }
6740 	}
6741 	TERM(sublex_start());
6742 
6743     case '`':
6744 	s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
6745 	DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
6746 	if (PL_expect == XOPERATOR)
6747 	    no_op("Backticks",s);
6748 	if (!s)
6749 	    missingterm(NULL);
6750 	readpipe_override();
6751 	TERM(sublex_start());
6752 
6753     case '\\':
6754 	s++;
6755 	if (PL_lex_inwhat && isDIGIT(*s))
6756 	    Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
6757 			   *s, *s);
6758 	if (PL_expect == XOPERATOR)
6759 	    no_op("Backslash",s);
6760 	OPERATOR(REFGEN);
6761 
6762     case 'v':
6763 	if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
6764 	    char *start = s + 2;
6765 	    while (isDIGIT(*start) || *start == '_')
6766 		start++;
6767 	    if (*start == '.' && isDIGIT(start[1])) {
6768 		s = scan_num(s, &pl_yylval);
6769 		TERM(THING);
6770 	    }
6771 	    else if ((*start == ':' && start[1] == ':')
6772 		  || (PL_expect == XSTATE && *start == ':'))
6773 		goto keylookup;
6774 	    else if (PL_expect == XSTATE) {
6775 		d = start;
6776 		while (d < PL_bufend && isSPACE(*d)) d++;
6777 		if (*d == ':') goto keylookup;
6778 	    }
6779 	    /* avoid v123abc() or $h{v1}, allow C<print v10;> */
6780 	    if (!isALPHA(*start) && (PL_expect == XTERM
6781 			|| PL_expect == XREF || PL_expect == XSTATE
6782 			|| PL_expect == XTERMORDORDOR)) {
6783 		GV *const gv = gv_fetchpvn_flags(s, start - s,
6784                                                     UTF ? SVf_UTF8 : 0, SVt_PVCV);
6785 		if (!gv) {
6786 		    s = scan_num(s, &pl_yylval);
6787 		    TERM(THING);
6788 		}
6789 	    }
6790 	}
6791 	goto keylookup;
6792     case 'x':
6793 	if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
6794 	    s++;
6795 	    Mop(OP_REPEAT);
6796 	}
6797 	goto keylookup;
6798 
6799     case '_':
6800     case 'a': case 'A':
6801     case 'b': case 'B':
6802     case 'c': case 'C':
6803     case 'd': case 'D':
6804     case 'e': case 'E':
6805     case 'f': case 'F':
6806     case 'g': case 'G':
6807     case 'h': case 'H':
6808     case 'i': case 'I':
6809     case 'j': case 'J':
6810     case 'k': case 'K':
6811     case 'l': case 'L':
6812     case 'm': case 'M':
6813     case 'n': case 'N':
6814     case 'o': case 'O':
6815     case 'p': case 'P':
6816     case 'q': case 'Q':
6817     case 'r': case 'R':
6818     case 's': case 'S':
6819     case 't': case 'T':
6820     case 'u': case 'U':
6821 	      case 'V':
6822     case 'w': case 'W':
6823 	      case 'X':
6824     case 'y': case 'Y':
6825     case 'z': case 'Z':
6826 
6827       keylookup: {
6828 	bool anydelim;
6829 	bool lex;
6830 	I32 tmp;
6831 	SV *sv;
6832 	CV *cv;
6833 	PADOFFSET off;
6834 	OP *rv2cv_op;
6835 
6836 	lex = FALSE;
6837 	orig_keyword = 0;
6838 	off = 0;
6839 	sv = NULL;
6840 	cv = NULL;
6841 	gv = NULL;
6842 	gvp = NULL;
6843 	rv2cv_op = NULL;
6844 
6845 	PL_bufptr = s;
6846 	s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
6847 
6848 	/* Some keywords can be followed by any delimiter, including ':' */
6849 	anydelim = word_takes_any_delimeter(PL_tokenbuf, len);
6850 
6851 	/* x::* is just a word, unless x is "CORE" */
6852 	if (!anydelim && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
6853 	    goto just_a_word;
6854 
6855 	d = s;
6856 	while (d < PL_bufend && isSPACE(*d))
6857 		d++;	/* no comments skipped here, or s### is misparsed */
6858 
6859 	/* Is this a word before a => operator? */
6860 	if (*d == '=' && d[1] == '>') {
6861 	    CLINE;
6862 	    pl_yylval.opval
6863 		= (OP*)newSVOP(OP_CONST, 0,
6864 			       S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
6865 	    pl_yylval.opval->op_private = OPpCONST_BARE;
6866 	    TERM(WORD);
6867 	}
6868 
6869 	/* Check for plugged-in keyword */
6870 	{
6871 	    OP *o;
6872 	    int result;
6873 	    char *saved_bufptr = PL_bufptr;
6874 	    PL_bufptr = s;
6875 	    result = PL_keyword_plugin(aTHX_ PL_tokenbuf, len, &o);
6876 	    s = PL_bufptr;
6877 	    if (result == KEYWORD_PLUGIN_DECLINE) {
6878 		/* not a plugged-in keyword */
6879 		PL_bufptr = saved_bufptr;
6880 	    } else if (result == KEYWORD_PLUGIN_STMT) {
6881 		pl_yylval.opval = o;
6882 		CLINE;
6883 		PL_expect = XSTATE;
6884 		return REPORT(PLUGSTMT);
6885 	    } else if (result == KEYWORD_PLUGIN_EXPR) {
6886 		pl_yylval.opval = o;
6887 		CLINE;
6888 		PL_expect = XOPERATOR;
6889 		return REPORT(PLUGEXPR);
6890 	    } else {
6891 		Perl_croak(aTHX_ "Bad plugin affecting keyword '%s'",
6892 					PL_tokenbuf);
6893 	    }
6894 	}
6895 
6896 	/* Check for built-in keyword */
6897 	tmp = keyword(PL_tokenbuf, len, 0);
6898 
6899 	/* Is this a label? */
6900 	if (!anydelim && PL_expect == XSTATE
6901 	      && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
6902 	    s = d + 1;
6903 	    pl_yylval.pval = savepvn(PL_tokenbuf, len+1);
6904 	    pl_yylval.pval[len] = '\0';
6905 	    pl_yylval.pval[len+1] = UTF ? 1 : 0;
6906 	    CLINE;
6907 	    TOKEN(LABEL);
6908 	}
6909 
6910 	/* Check for lexical sub */
6911 	if (PL_expect != XOPERATOR) {
6912 	    char tmpbuf[sizeof PL_tokenbuf + 1];
6913 	    *tmpbuf = '&';
6914 	    Copy(PL_tokenbuf, tmpbuf+1, len, char);
6915 	    off = pad_findmy_pvn(tmpbuf, len+1, UTF ? SVf_UTF8 : 0);
6916 	    if (off != NOT_IN_PAD) {
6917 		assert(off); /* we assume this is boolean-true below */
6918 		if (PAD_COMPNAME_FLAGS_isOUR(off)) {
6919 		    HV *  const stash = PAD_COMPNAME_OURSTASH(off);
6920 		    HEK * const stashname = HvNAME_HEK(stash);
6921 		    sv = newSVhek(stashname);
6922                     sv_catpvs(sv, "::");
6923                     sv_catpvn_flags(sv, PL_tokenbuf, len,
6924 				    (UTF ? SV_CATUTF8 : SV_CATBYTES));
6925 		    gv = gv_fetchsv(sv, GV_NOADD_NOINIT | SvUTF8(sv),
6926 				    SVt_PVCV);
6927 		    off = 0;
6928 		    if (!gv) {
6929 			sv_free(sv);
6930 			sv = NULL;
6931 			goto just_a_word;
6932 		    }
6933 		}
6934 		else {
6935 		    rv2cv_op = newOP(OP_PADANY, 0);
6936 		    rv2cv_op->op_targ = off;
6937 		    cv = find_lexical_cv(off);
6938 		}
6939 		lex = TRUE;
6940 		goto just_a_word;
6941 	    }
6942 	    off = 0;
6943 	}
6944 
6945 	if (tmp < 0) {			/* second-class keyword? */
6946 	    GV *ogv = NULL;	/* override (winner) */
6947 	    GV *hgv = NULL;	/* hidden (loser) */
6948 	    if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
6949 		CV *cv;
6950 		if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len,
6951                                             UTF ? SVf_UTF8 : 0, SVt_PVCV)) &&
6952 		    (cv = GvCVu(gv)))
6953 		{
6954 		    if (GvIMPORTED_CV(gv))
6955 			ogv = gv;
6956 		    else if (! CvMETHOD(cv))
6957 			hgv = gv;
6958 		}
6959 		if (!ogv &&
6960 		    (gvp = (GV**)hv_fetch(PL_globalstash, PL_tokenbuf,
6961                                             UTF ? -(I32)len : (I32)len, FALSE)) &&
6962 		    (gv = *gvp) && isGV_with_GP(gv) &&
6963 		    GvCVu(gv) && GvIMPORTED_CV(gv))
6964 		{
6965 		    ogv = gv;
6966 		}
6967 	    }
6968 	    if (ogv) {
6969 		orig_keyword = tmp;
6970 		tmp = 0;		/* overridden by import or by GLOBAL */
6971 	    }
6972 	    else if (gv && !gvp
6973 		     && -tmp==KEY_lock	/* XXX generalizable kludge */
6974 		     && GvCVu(gv))
6975 	    {
6976 		tmp = 0;		/* any sub overrides "weak" keyword */
6977 	    }
6978 	    else {			/* no override */
6979 		tmp = -tmp;
6980 		if (tmp == KEY_dump) {
6981 		    Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
6982 				   "dump() better written as CORE::dump()");
6983 		}
6984 		gv = NULL;
6985 		gvp = 0;
6986 		if (hgv && tmp != KEY_x && tmp != KEY_CORE)	/* never ambiguous */
6987 		    Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6988 				   "Ambiguous call resolved as CORE::%s(), "
6989 				   "qualify as such or use &",
6990 				   GvENAME(hgv));
6991 	    }
6992 	}
6993 
6994       reserved_word:
6995 	switch (tmp) {
6996 
6997 	default:			/* not a keyword */
6998 	    /* Trade off - by using this evil construction we can pull the
6999 	       variable gv into the block labelled keylookup. If not, then
7000 	       we have to give it function scope so that the goto from the
7001 	       earlier ':' case doesn't bypass the initialisation.  */
7002 	    if (0) {
7003 	    just_a_word_zero_gv:
7004 		sv = NULL;
7005 		cv = NULL;
7006 		gv = NULL;
7007 		gvp = NULL;
7008 		rv2cv_op = NULL;
7009 		orig_keyword = 0;
7010 		lex = 0;
7011 		off = 0;
7012 	    }
7013 	  just_a_word: {
7014 		int pkgname = 0;
7015 		const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
7016 		const char penultchar =
7017 		    lastchar && PL_bufptr - 2 >= PL_linestart
7018 			 ? PL_bufptr[-2]
7019 			 : 0;
7020 #ifdef PERL_MAD
7021 		SV *nextPL_nextwhite = 0;
7022 #endif
7023 
7024 
7025 		/* Get the rest if it looks like a package qualifier */
7026 
7027 		if (*s == '\'' || (*s == ':' && s[1] == ':')) {
7028 		    STRLEN morelen;
7029 		    s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
7030 				  TRUE, &morelen);
7031 		    if (!morelen)
7032 			Perl_croak(aTHX_ "Bad name after %"SVf"%s",
7033                                         SVfARG(newSVpvn_flags(PL_tokenbuf, len,
7034                                             (UTF ? SVf_UTF8 : 0) | SVs_TEMP )),
7035 				*s == '\'' ? "'" : "::");
7036 		    len += morelen;
7037 		    pkgname = 1;
7038 		}
7039 
7040 		if (PL_expect == XOPERATOR) {
7041 		    if (PL_bufptr == PL_linestart) {
7042 			CopLINE_dec(PL_curcop);
7043 			Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
7044 			CopLINE_inc(PL_curcop);
7045 		    }
7046 		    else
7047 			no_op("Bareword",s);
7048 		}
7049 
7050 		/* Look for a subroutine with this name in current package,
7051 		   unless this is a lexical sub, or name is "Foo::",
7052 		   in which case Foo is a bareword
7053 		   (and a package name). */
7054 
7055 		if (len > 2 && !PL_madskills &&
7056 		    PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
7057 		{
7058 		    if (ckWARN(WARN_BAREWORD)
7059 			&& ! gv_fetchpvn_flags(PL_tokenbuf, len, UTF ? SVf_UTF8 : 0, SVt_PVHV))
7060 			Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
7061 		  	    "Bareword \"%"SVf"\" refers to nonexistent package",
7062 			     SVfARG(newSVpvn_flags(PL_tokenbuf, len,
7063                                         (UTF ? SVf_UTF8 : 0) | SVs_TEMP)));
7064 		    len -= 2;
7065 		    PL_tokenbuf[len] = '\0';
7066 		    gv = NULL;
7067 		    gvp = 0;
7068 		}
7069 		else {
7070 		    if (!lex && !gv) {
7071 			/* Mustn't actually add anything to a symbol table.
7072 			   But also don't want to "initialise" any placeholder
7073 			   constants that might already be there into full
7074 			   blown PVGVs with attached PVCV.  */
7075 			gv = gv_fetchpvn_flags(PL_tokenbuf, len,
7076 					       GV_NOADD_NOINIT | ( UTF ? SVf_UTF8 : 0 ),
7077 					       SVt_PVCV);
7078 		    }
7079 		    len = 0;
7080 		}
7081 
7082 		/* if we saw a global override before, get the right name */
7083 
7084 		if (!sv)
7085 		  sv = S_newSV_maybe_utf8(aTHX_ PL_tokenbuf,
7086 		    len ? len : strlen(PL_tokenbuf));
7087 		if (gvp) {
7088 		    SV * const tmp_sv = sv;
7089 		    sv = newSVpvs("CORE::GLOBAL::");
7090 		    sv_catsv(sv, tmp_sv);
7091 		    SvREFCNT_dec(tmp_sv);
7092 		}
7093 
7094 #ifdef PERL_MAD
7095 		if (PL_madskills && !PL_thistoken) {
7096 		    char *start = SvPVX(PL_linestr) + PL_realtokenstart;
7097 		    PL_thistoken = newSVpvn(start,s - start);
7098 		    PL_realtokenstart = s - SvPVX(PL_linestr);
7099 		}
7100 #endif
7101 
7102 		/* Presume this is going to be a bareword of some sort. */
7103 		CLINE;
7104 		pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
7105 		pl_yylval.opval->op_private = OPpCONST_BARE;
7106 
7107 		/* And if "Foo::", then that's what it certainly is. */
7108 		if (len)
7109 		    goto safe_bareword;
7110 
7111 		if (!off)
7112 		{
7113 		    OP *const_op = newSVOP(OP_CONST, 0, SvREFCNT_inc_NN(sv));
7114 		    const_op->op_private = OPpCONST_BARE;
7115 		    rv2cv_op = newCVREF(0, const_op);
7116 		    cv = lex ? GvCV(gv) : rv2cv_op_cv(rv2cv_op, 0);
7117 		}
7118 
7119 		/* See if it's the indirect object for a list operator. */
7120 
7121 		if (PL_oldoldbufptr &&
7122 		    PL_oldoldbufptr < PL_bufptr &&
7123 		    (PL_oldoldbufptr == PL_last_lop
7124 		     || PL_oldoldbufptr == PL_last_uni) &&
7125 		    /* NO SKIPSPACE BEFORE HERE! */
7126 		    (PL_expect == XREF ||
7127 		     ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
7128 		{
7129 		    bool immediate_paren = *s == '(';
7130 
7131 		    /* (Now we can afford to cross potential line boundary.) */
7132 		    s = SKIPSPACE2(s,nextPL_nextwhite);
7133 #ifdef PERL_MAD
7134 		    PL_nextwhite = nextPL_nextwhite;	/* assume no & deception */
7135 #endif
7136 
7137 		    /* Two barewords in a row may indicate method call. */
7138 
7139 		    if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
7140 			(tmp = intuit_method(s, gv, cv))) {
7141 			op_free(rv2cv_op);
7142 			if (tmp == METHOD && !PL_lex_allbrackets &&
7143 				PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
7144 			    PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
7145 			return REPORT(tmp);
7146 		    }
7147 
7148 		    /* If not a declared subroutine, it's an indirect object. */
7149 		    /* (But it's an indir obj regardless for sort.) */
7150 		    /* Also, if "_" follows a filetest operator, it's a bareword */
7151 
7152 		    if (
7153 			( !immediate_paren && (PL_last_lop_op == OP_SORT ||
7154                          (!cv &&
7155                         (PL_last_lop_op != OP_MAPSTART &&
7156 			 PL_last_lop_op != OP_GREPSTART))))
7157 		       || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
7158 			    && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
7159 		       )
7160 		    {
7161 			PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
7162 			goto bareword;
7163 		    }
7164 		}
7165 
7166 		PL_expect = XOPERATOR;
7167 #ifdef PERL_MAD
7168 		if (isSPACE(*s))
7169 		    s = SKIPSPACE2(s,nextPL_nextwhite);
7170 		PL_nextwhite = nextPL_nextwhite;
7171 #else
7172 		s = skipspace(s);
7173 #endif
7174 
7175 		/* Is this a word before a => operator? */
7176 		if (*s == '=' && s[1] == '>' && !pkgname) {
7177 		    op_free(rv2cv_op);
7178 		    CLINE;
7179 		    sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
7180 		    if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
7181 		      SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
7182 		    TERM(WORD);
7183 		}
7184 
7185 		/* If followed by a paren, it's certainly a subroutine. */
7186 		if (*s == '(') {
7187 		    CLINE;
7188 		    if (cv) {
7189 			d = s + 1;
7190 			while (SPACE_OR_TAB(*d))
7191 			    d++;
7192 			if (*d == ')' && (sv = cv_const_sv(cv))) {
7193 			    s = d + 1;
7194 			    goto its_constant;
7195 			}
7196 		    }
7197 #ifdef PERL_MAD
7198 		    if (PL_madskills) {
7199 			PL_nextwhite = PL_thiswhite;
7200 			PL_thiswhite = 0;
7201 		    }
7202 		    start_force(PL_curforce);
7203 #endif
7204 		    NEXTVAL_NEXTTOKE.opval =
7205 			off ? rv2cv_op : pl_yylval.opval;
7206 		    PL_expect = XOPERATOR;
7207 #ifdef PERL_MAD
7208 		    if (PL_madskills) {
7209 			PL_nextwhite = nextPL_nextwhite;
7210 			curmad('X', PL_thistoken);
7211 			PL_thistoken = newSVpvs("");
7212 		    }
7213 #endif
7214 		    if (off)
7215 			 op_free(pl_yylval.opval), force_next(PRIVATEREF);
7216 		    else op_free(rv2cv_op),	   force_next(WORD);
7217 		    pl_yylval.ival = 0;
7218 		    TOKEN('&');
7219 		}
7220 
7221 		/* If followed by var or block, call it a method (unless sub) */
7222 
7223 		if ((*s == '$' || *s == '{') && !cv) {
7224 		    op_free(rv2cv_op);
7225 		    PL_last_lop = PL_oldbufptr;
7226 		    PL_last_lop_op = OP_METHOD;
7227 		    if (!PL_lex_allbrackets &&
7228 			    PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
7229 			PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
7230 		    PREBLOCK(METHOD);
7231 		}
7232 
7233 		/* If followed by a bareword, see if it looks like indir obj. */
7234 
7235 		if (!orig_keyword
7236 			&& (isIDFIRST_lazy_if(s,UTF) || *s == '$')
7237 			&& (tmp = intuit_method(s, gv, cv))) {
7238 		    op_free(rv2cv_op);
7239 		    if (tmp == METHOD && !PL_lex_allbrackets &&
7240 			    PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
7241 			PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
7242 		    return REPORT(tmp);
7243 		}
7244 
7245 		/* Not a method, so call it a subroutine (if defined) */
7246 
7247 		if (cv) {
7248 		    if (lastchar == '-' && penultchar != '-') {
7249                         const SV *tmpsv = newSVpvn_flags( PL_tokenbuf, len ? len : strlen(PL_tokenbuf), (UTF ? SVf_UTF8 : 0) | SVs_TEMP );
7250  			Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
7251 				"Ambiguous use of -%"SVf" resolved as -&%"SVf"()",
7252 				SVfARG(tmpsv), SVfARG(tmpsv));
7253                     }
7254 		    /* Check for a constant sub */
7255 		    if ((sv = cv_const_sv(cv))) {
7256 		  its_constant:
7257 			op_free(rv2cv_op);
7258 			SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
7259 			((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
7260 			pl_yylval.opval->op_private = OPpCONST_FOLDED;
7261 			pl_yylval.opval->op_flags |= OPf_SPECIAL;
7262 			TOKEN(WORD);
7263 		    }
7264 
7265 		    op_free(pl_yylval.opval);
7266 		    pl_yylval.opval =
7267 			off ? (OP *)newCVREF(0, rv2cv_op) : rv2cv_op;
7268 		    pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
7269 		    PL_last_lop = PL_oldbufptr;
7270 		    PL_last_lop_op = OP_ENTERSUB;
7271 		    /* Is there a prototype? */
7272 		    if (
7273 #ifdef PERL_MAD
7274 			cv &&
7275 #endif
7276 			SvPOK(cv))
7277 		    {
7278 			STRLEN protolen = CvPROTOLEN(cv);
7279 			const char *proto = CvPROTO(cv);
7280 			bool optional;
7281 			if (!protolen)
7282 			    TERM(FUNC0SUB);
7283 			if ((optional = *proto == ';'))
7284 			  do
7285 			    proto++;
7286 			  while (*proto == ';');
7287 			if (
7288 			    (
7289 			        (
7290 			            *proto == '$' || *proto == '_'
7291 			         || *proto == '*' || *proto == '+'
7292 			        )
7293 			     && proto[1] == '\0'
7294 			    )
7295 			 || (
7296 			     *proto == '\\' && proto[1] && proto[2] == '\0'
7297 			    )
7298 			)
7299 			    UNIPROTO(UNIOPSUB,optional);
7300 			if (*proto == '\\' && proto[1] == '[') {
7301 			    const char *p = proto + 2;
7302 			    while(*p && *p != ']')
7303 				++p;
7304 			    if(*p == ']' && !p[1])
7305 				UNIPROTO(UNIOPSUB,optional);
7306 			}
7307 			if (*proto == '&' && *s == '{') {
7308 			    if (PL_curstash)
7309 				sv_setpvs(PL_subname, "__ANON__");
7310 			    else
7311 				sv_setpvs(PL_subname, "__ANON__::__ANON__");
7312 			    if (!PL_lex_allbrackets &&
7313 				    PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
7314 				PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
7315 			    PREBLOCK(LSTOPSUB);
7316 			}
7317 		    }
7318 #ifdef PERL_MAD
7319 		    {
7320 			if (PL_madskills) {
7321 			    PL_nextwhite = PL_thiswhite;
7322 			    PL_thiswhite = 0;
7323 			}
7324 			start_force(PL_curforce);
7325 			NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
7326 			PL_expect = XTERM;
7327 			if (PL_madskills) {
7328 			    PL_nextwhite = nextPL_nextwhite;
7329 			    curmad('X', PL_thistoken);
7330 			    PL_thistoken = newSVpvs("");
7331 			}
7332 			force_next(off ? PRIVATEREF : WORD);
7333 			if (!PL_lex_allbrackets &&
7334 				PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
7335 			    PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
7336 			TOKEN(NOAMP);
7337 		    }
7338 		}
7339 
7340 		/* Guess harder when madskills require "best effort". */
7341 		if (PL_madskills && (!gv || !GvCVu(gv))) {
7342 		    int probable_sub = 0;
7343 		    if (strchr("\"'`$@%0123456789!*+{[<", *s))
7344 			probable_sub = 1;
7345 		    else if (isALPHA(*s)) {
7346 			char tmpbuf[1024];
7347 			STRLEN tmplen;
7348 			d = s;
7349 			d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
7350 			if (!keyword(tmpbuf, tmplen, 0))
7351 			    probable_sub = 1;
7352 			else {
7353 			    while (d < PL_bufend && isSPACE(*d))
7354 				d++;
7355 			    if (*d == '=' && d[1] == '>')
7356 				probable_sub = 1;
7357 			}
7358 		    }
7359 		    if (probable_sub) {
7360 			gv = gv_fetchpv(PL_tokenbuf, GV_ADD | ( UTF ? SVf_UTF8 : 0 ),
7361                                         SVt_PVCV);
7362 			op_free(pl_yylval.opval);
7363 			pl_yylval.opval =
7364 			    off ? (OP *)newCVREF(0, rv2cv_op) : rv2cv_op;
7365 			pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
7366 			PL_last_lop = PL_oldbufptr;
7367 			PL_last_lop_op = OP_ENTERSUB;
7368 			PL_nextwhite = PL_thiswhite;
7369 			PL_thiswhite = 0;
7370 			start_force(PL_curforce);
7371 			NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
7372 			PL_expect = XTERM;
7373 			PL_nextwhite = nextPL_nextwhite;
7374 			curmad('X', PL_thistoken);
7375 			PL_thistoken = newSVpvs("");
7376 			force_next(off ? PRIVATEREF : WORD);
7377 			if (!PL_lex_allbrackets &&
7378 				PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
7379 			    PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
7380 			TOKEN(NOAMP);
7381 		    }
7382 #else
7383 		    NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
7384 		    PL_expect = XTERM;
7385 		    force_next(off ? PRIVATEREF : WORD);
7386 		    if (!PL_lex_allbrackets &&
7387 			    PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
7388 			PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
7389 		    TOKEN(NOAMP);
7390 #endif
7391 		}
7392 
7393 		/* Call it a bare word */
7394 
7395 		if (PL_hints & HINT_STRICT_SUBS)
7396 		    pl_yylval.opval->op_private |= OPpCONST_STRICT;
7397 		else {
7398 		bareword:
7399 		    /* after "print" and similar functions (corresponding to
7400 		     * "F? L" in opcode.pl), whatever wasn't already parsed as
7401 		     * a filehandle should be subject to "strict subs".
7402 		     * Likewise for the optional indirect-object argument to system
7403 		     * or exec, which can't be a bareword */
7404 		    if ((PL_last_lop_op == OP_PRINT
7405 			    || PL_last_lop_op == OP_PRTF
7406 			    || PL_last_lop_op == OP_SAY
7407 			    || PL_last_lop_op == OP_SYSTEM
7408 			    || PL_last_lop_op == OP_EXEC)
7409 			    && (PL_hints & HINT_STRICT_SUBS))
7410 			pl_yylval.opval->op_private |= OPpCONST_STRICT;
7411 		    if (lastchar != '-') {
7412 			if (ckWARN(WARN_RESERVED)) {
7413 			    d = PL_tokenbuf;
7414 			    while (isLOWER(*d))
7415 				d++;
7416 			    if (!*d && !gv_stashpv(PL_tokenbuf, UTF ? SVf_UTF8 : 0))
7417 				Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
7418 				       PL_tokenbuf);
7419 			}
7420 		    }
7421 		}
7422 		op_free(rv2cv_op);
7423 
7424 	    safe_bareword:
7425 		if ((lastchar == '*' || lastchar == '%' || lastchar == '&')) {
7426 		    Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
7427 				     "Operator or semicolon missing before %c%"SVf,
7428 				     lastchar, SVfARG(newSVpvn_flags(PL_tokenbuf,
7429                                                     strlen(PL_tokenbuf),
7430                                                     SVs_TEMP | (UTF ? SVf_UTF8 : 0))));
7431 		    Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
7432 				     "Ambiguous use of %c resolved as operator %c",
7433 				     lastchar, lastchar);
7434 		}
7435 		TOKEN(WORD);
7436 	    }
7437 
7438 	case KEY___FILE__:
7439 	    FUN0OP(
7440 		(OP*)newSVOP(OP_CONST, 0, newSVpv(CopFILE(PL_curcop),0))
7441 	    );
7442 
7443 	case KEY___LINE__:
7444 	    FUN0OP(
7445         	(OP*)newSVOP(OP_CONST, 0,
7446 		    Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)))
7447 	    );
7448 
7449 	case KEY___PACKAGE__:
7450 	    FUN0OP(
7451 		(OP*)newSVOP(OP_CONST, 0,
7452 					(PL_curstash
7453 					 ? newSVhek(HvNAME_HEK(PL_curstash))
7454 					 : &PL_sv_undef))
7455 	    );
7456 
7457 	case KEY___DATA__:
7458 	case KEY___END__: {
7459 	    GV *gv;
7460 	    if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
7461 		const char *pname = "main";
7462 		STRLEN plen = 4;
7463 		U32 putf8 = 0;
7464 		if (PL_tokenbuf[2] == 'D')
7465 		{
7466 		    HV * const stash =
7467 			PL_curstash ? PL_curstash : PL_defstash;
7468 		    pname = HvNAME_get(stash);
7469 		    plen  = HvNAMELEN (stash);
7470 		    if(HvNAMEUTF8(stash)) putf8 = SVf_UTF8;
7471 		}
7472 		gv = gv_fetchpvn_flags(
7473 			Perl_form(aTHX_ "%*s::DATA", (int)plen, pname),
7474 			plen+6, GV_ADD|putf8, SVt_PVIO
7475 		);
7476 		GvMULTI_on(gv);
7477 		if (!GvIO(gv))
7478 		    GvIOp(gv) = newIO();
7479 		IoIFP(GvIOp(gv)) = PL_rsfp;
7480 #if defined(HAS_FCNTL) && defined(F_SETFD)
7481 		{
7482 		    const int fd = PerlIO_fileno(PL_rsfp);
7483 		    fcntl(fd,F_SETFD,fd >= 3);
7484 		}
7485 #endif
7486 		/* Mark this internal pseudo-handle as clean */
7487 		IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
7488 		if ((PerlIO*)PL_rsfp == PerlIO_stdin())
7489 		    IoTYPE(GvIOp(gv)) = IoTYPE_STD;
7490 		else
7491 		    IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
7492 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
7493 		/* if the script was opened in binmode, we need to revert
7494 		 * it to text mode for compatibility; but only iff it has CRs
7495 		 * XXX this is a questionable hack at best. */
7496 		if (PL_bufend-PL_bufptr > 2
7497 		    && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
7498 		{
7499 		    Off_t loc = 0;
7500 		    if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
7501 			loc = PerlIO_tell(PL_rsfp);
7502 			(void)PerlIO_seek(PL_rsfp, 0L, 0);
7503 		    }
7504 #ifdef NETWARE
7505 			if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
7506 #else
7507 		    if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
7508 #endif	/* NETWARE */
7509 			if (loc > 0)
7510 			    PerlIO_seek(PL_rsfp, loc, 0);
7511 		    }
7512 		}
7513 #endif
7514 #ifdef PERLIO_LAYERS
7515 		if (!IN_BYTES) {
7516 		    if (UTF)
7517 			PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
7518 		    else if (PL_encoding) {
7519 			SV *name;
7520 			dSP;
7521 			ENTER;
7522 			SAVETMPS;
7523 			PUSHMARK(sp);
7524 			EXTEND(SP, 1);
7525 			XPUSHs(PL_encoding);
7526 			PUTBACK;
7527 			call_method("name", G_SCALAR);
7528 			SPAGAIN;
7529 			name = POPs;
7530 			PUTBACK;
7531 			PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
7532 					    Perl_form(aTHX_ ":encoding(%"SVf")",
7533 						      SVfARG(name)));
7534 			FREETMPS;
7535 			LEAVE;
7536 		    }
7537 		}
7538 #endif
7539 #ifdef PERL_MAD
7540 		if (PL_madskills) {
7541 		    if (PL_realtokenstart >= 0) {
7542 			char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
7543 			if (!PL_endwhite)
7544 			    PL_endwhite = newSVpvs("");
7545 			sv_catsv(PL_endwhite, PL_thiswhite);
7546 			PL_thiswhite = 0;
7547 			sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
7548 			PL_realtokenstart = -1;
7549 		    }
7550 		    while ((s = filter_gets(PL_endwhite, SvCUR(PL_endwhite)))
7551 			   != NULL) ;
7552 		}
7553 #endif
7554 		PL_rsfp = NULL;
7555 	    }
7556 	    goto fake_eof;
7557 	}
7558 
7559 	case KEY___SUB__:
7560 	    FUN0OP(newPVOP(OP_RUNCV,0,NULL));
7561 
7562 	case KEY_AUTOLOAD:
7563 	case KEY_DESTROY:
7564 	case KEY_BEGIN:
7565 	case KEY_UNITCHECK:
7566 	case KEY_CHECK:
7567 	case KEY_INIT:
7568 	case KEY_END:
7569 	    if (PL_expect == XSTATE) {
7570 		s = PL_bufptr;
7571 		goto really_sub;
7572 	    }
7573 	    goto just_a_word;
7574 
7575 	case KEY_CORE:
7576 	    if (*s == ':' && s[1] == ':') {
7577 		STRLEN olen = len;
7578 		d = s;
7579 		s += 2;
7580 		s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
7581 		if ((*s == ':' && s[1] == ':')
7582 		 || (!(tmp = keyword(PL_tokenbuf, len, 1)) && *s == '\''))
7583 		{
7584 		    s = d;
7585 		    len = olen;
7586 		    Copy(PL_bufptr, PL_tokenbuf, olen, char);
7587 		    goto just_a_word;
7588 		}
7589 		if (!tmp)
7590 		    Perl_croak(aTHX_ "CORE::%"SVf" is not a keyword",
7591                                     SVfARG(newSVpvn_flags(PL_tokenbuf, len,
7592                                                 (UTF ? SVf_UTF8 : 0) | SVs_TEMP)));
7593 		if (tmp < 0)
7594 		    tmp = -tmp;
7595 		else if (tmp == KEY_require || tmp == KEY_do
7596 		      || tmp == KEY_glob)
7597 		    /* that's a way to remember we saw "CORE::" */
7598 		    orig_keyword = tmp;
7599 		goto reserved_word;
7600 	    }
7601 	    goto just_a_word;
7602 
7603 	case KEY_abs:
7604 	    UNI(OP_ABS);
7605 
7606 	case KEY_alarm:
7607 	    UNI(OP_ALARM);
7608 
7609 	case KEY_accept:
7610 	    LOP(OP_ACCEPT,XTERM);
7611 
7612 	case KEY_and:
7613 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_LOWLOGIC)
7614 		return REPORT(0);
7615 	    OPERATOR(ANDOP);
7616 
7617 	case KEY_atan2:
7618 	    LOP(OP_ATAN2,XTERM);
7619 
7620 	case KEY_bind:
7621 	    LOP(OP_BIND,XTERM);
7622 
7623 	case KEY_binmode:
7624 	    LOP(OP_BINMODE,XTERM);
7625 
7626 	case KEY_bless:
7627 	    LOP(OP_BLESS,XTERM);
7628 
7629 	case KEY_break:
7630 	    FUN0(OP_BREAK);
7631 
7632 	case KEY_chop:
7633 	    UNI(OP_CHOP);
7634 
7635 	case KEY_continue:
7636 		    /* We have to disambiguate the two senses of
7637 		      "continue". If the next token is a '{' then
7638 		      treat it as the start of a continue block;
7639 		      otherwise treat it as a control operator.
7640 		     */
7641 		    s = skipspace(s);
7642 		    if (*s == '{')
7643 	    PREBLOCK(CONTINUE);
7644 		    else
7645 			FUN0(OP_CONTINUE);
7646 
7647 	case KEY_chdir:
7648 	    /* may use HOME */
7649 	    (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
7650 	    UNI(OP_CHDIR);
7651 
7652 	case KEY_close:
7653 	    UNI(OP_CLOSE);
7654 
7655 	case KEY_closedir:
7656 	    UNI(OP_CLOSEDIR);
7657 
7658 	case KEY_cmp:
7659 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE)
7660 		return REPORT(0);
7661 	    Eop(OP_SCMP);
7662 
7663 	case KEY_caller:
7664 	    UNI(OP_CALLER);
7665 
7666 	case KEY_crypt:
7667 #ifdef FCRYPT
7668 	    if (!PL_cryptseen) {
7669 		PL_cryptseen = TRUE;
7670 		init_des();
7671 	    }
7672 #endif
7673 	    LOP(OP_CRYPT,XTERM);
7674 
7675 	case KEY_chmod:
7676 	    LOP(OP_CHMOD,XTERM);
7677 
7678 	case KEY_chown:
7679 	    LOP(OP_CHOWN,XTERM);
7680 
7681 	case KEY_connect:
7682 	    LOP(OP_CONNECT,XTERM);
7683 
7684 	case KEY_chr:
7685 	    UNI(OP_CHR);
7686 
7687 	case KEY_cos:
7688 	    UNI(OP_COS);
7689 
7690 	case KEY_chroot:
7691 	    UNI(OP_CHROOT);
7692 
7693 	case KEY_default:
7694 	    PREBLOCK(DEFAULT);
7695 
7696 	case KEY_do:
7697 	    s = SKIPSPACE1(s);
7698 	    if (*s == '{')
7699 		PRETERMBLOCK(DO);
7700 	    if (*s != '\'') {
7701 		*PL_tokenbuf = '&';
7702 		d = scan_word(s, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
7703 			      1, &len);
7704 		if (len && !keyword(PL_tokenbuf + 1, len, 0)) {
7705 		    d = SKIPSPACE1(d);
7706 		    if (*d == '(') {
7707 			force_ident_maybe_lex('&');
7708 			s = d;
7709 		    }
7710 		}
7711 	    }
7712 	    if (orig_keyword == KEY_do) {
7713 		orig_keyword = 0;
7714 		pl_yylval.ival = 1;
7715 	    }
7716 	    else
7717 		pl_yylval.ival = 0;
7718 	    OPERATOR(DO);
7719 
7720 	case KEY_die:
7721 	    PL_hints |= HINT_BLOCK_SCOPE;
7722 	    LOP(OP_DIE,XTERM);
7723 
7724 	case KEY_defined:
7725 	    UNI(OP_DEFINED);
7726 
7727 	case KEY_delete:
7728 	    UNI(OP_DELETE);
7729 
7730 	case KEY_dbmopen:
7731 	    Perl_populate_isa(aTHX_ STR_WITH_LEN("AnyDBM_File::ISA"),
7732 			      STR_WITH_LEN("NDBM_File::"),
7733 			      STR_WITH_LEN("DB_File::"),
7734 			      STR_WITH_LEN("GDBM_File::"),
7735 			      STR_WITH_LEN("SDBM_File::"),
7736 			      STR_WITH_LEN("ODBM_File::"),
7737 			      NULL);
7738 	    LOP(OP_DBMOPEN,XTERM);
7739 
7740 	case KEY_dbmclose:
7741 	    UNI(OP_DBMCLOSE);
7742 
7743 	case KEY_dump:
7744 	    PL_expect = XOPERATOR;
7745 	    s = force_word(s,WORD,TRUE,FALSE,FALSE);
7746 	    LOOPX(OP_DUMP);
7747 
7748 	case KEY_else:
7749 	    PREBLOCK(ELSE);
7750 
7751 	case KEY_elsif:
7752 	    pl_yylval.ival = CopLINE(PL_curcop);
7753 	    OPERATOR(ELSIF);
7754 
7755 	case KEY_eq:
7756 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE)
7757 		return REPORT(0);
7758 	    Eop(OP_SEQ);
7759 
7760 	case KEY_exists:
7761 	    UNI(OP_EXISTS);
7762 
7763 	case KEY_exit:
7764 	    if (PL_madskills)
7765 		UNI(OP_INT);
7766 	    UNI(OP_EXIT);
7767 
7768 	case KEY_eval:
7769 	    s = SKIPSPACE1(s);
7770 	    if (*s == '{') { /* block eval */
7771 		PL_expect = XTERMBLOCK;
7772 		UNIBRACK(OP_ENTERTRY);
7773 	    }
7774 	    else { /* string eval */
7775 		PL_expect = XTERM;
7776 		UNIBRACK(OP_ENTEREVAL);
7777 	    }
7778 
7779 	case KEY_evalbytes:
7780 	    PL_expect = XTERM;
7781 	    UNIBRACK(-OP_ENTEREVAL);
7782 
7783 	case KEY_eof:
7784 	    UNI(OP_EOF);
7785 
7786 	case KEY_exp:
7787 	    UNI(OP_EXP);
7788 
7789 	case KEY_each:
7790 	    UNI(OP_EACH);
7791 
7792 	case KEY_exec:
7793 	    LOP(OP_EXEC,XREF);
7794 
7795 	case KEY_endhostent:
7796 	    FUN0(OP_EHOSTENT);
7797 
7798 	case KEY_endnetent:
7799 	    FUN0(OP_ENETENT);
7800 
7801 	case KEY_endservent:
7802 	    FUN0(OP_ESERVENT);
7803 
7804 	case KEY_endprotoent:
7805 	    FUN0(OP_EPROTOENT);
7806 
7807 	case KEY_endpwent:
7808 	    FUN0(OP_EPWENT);
7809 
7810 	case KEY_endgrent:
7811 	    FUN0(OP_EGRENT);
7812 
7813 	case KEY_for:
7814 	case KEY_foreach:
7815 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_NONEXPR)
7816 		return REPORT(0);
7817 	    pl_yylval.ival = CopLINE(PL_curcop);
7818 	    s = SKIPSPACE1(s);
7819 	    if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
7820 		char *p = s;
7821 #ifdef PERL_MAD
7822 		int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
7823 #endif
7824 
7825 		if ((PL_bufend - p) >= 3 &&
7826 		    strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
7827 		    p += 2;
7828 		else if ((PL_bufend - p) >= 4 &&
7829 		    strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
7830 		    p += 3;
7831 		p = PEEKSPACE(p);
7832 		if (isIDFIRST_lazy_if(p,UTF)) {
7833 		    p = scan_ident(p, PL_bufend,
7834 			PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
7835 		    p = PEEKSPACE(p);
7836 		}
7837 		if (*p != '$')
7838 		    Perl_croak(aTHX_ "Missing $ on loop variable");
7839 #ifdef PERL_MAD
7840 		s = SvPVX(PL_linestr) + soff;
7841 #endif
7842 	    }
7843 	    OPERATOR(FOR);
7844 
7845 	case KEY_formline:
7846 	    LOP(OP_FORMLINE,XTERM);
7847 
7848 	case KEY_fork:
7849 	    FUN0(OP_FORK);
7850 
7851 	case KEY_fc:
7852 	    UNI(OP_FC);
7853 
7854 	case KEY_fcntl:
7855 	    LOP(OP_FCNTL,XTERM);
7856 
7857 	case KEY_fileno:
7858 	    UNI(OP_FILENO);
7859 
7860 	case KEY_flock:
7861 	    LOP(OP_FLOCK,XTERM);
7862 
7863 	case KEY_gt:
7864 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE)
7865 		return REPORT(0);
7866 	    Rop(OP_SGT);
7867 
7868 	case KEY_ge:
7869 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE)
7870 		return REPORT(0);
7871 	    Rop(OP_SGE);
7872 
7873 	case KEY_grep:
7874 	    LOP(OP_GREPSTART, XREF);
7875 
7876 	case KEY_goto:
7877 	    PL_expect = XOPERATOR;
7878 	    s = force_word(s,WORD,TRUE,FALSE,FALSE);
7879 	    LOOPX(OP_GOTO);
7880 
7881 	case KEY_gmtime:
7882 	    UNI(OP_GMTIME);
7883 
7884 	case KEY_getc:
7885 	    UNIDOR(OP_GETC);
7886 
7887 	case KEY_getppid:
7888 	    FUN0(OP_GETPPID);
7889 
7890 	case KEY_getpgrp:
7891 	    UNI(OP_GETPGRP);
7892 
7893 	case KEY_getpriority:
7894 	    LOP(OP_GETPRIORITY,XTERM);
7895 
7896 	case KEY_getprotobyname:
7897 	    UNI(OP_GPBYNAME);
7898 
7899 	case KEY_getprotobynumber:
7900 	    LOP(OP_GPBYNUMBER,XTERM);
7901 
7902 	case KEY_getprotoent:
7903 	    FUN0(OP_GPROTOENT);
7904 
7905 	case KEY_getpwent:
7906 	    FUN0(OP_GPWENT);
7907 
7908 	case KEY_getpwnam:
7909 	    UNI(OP_GPWNAM);
7910 
7911 	case KEY_getpwuid:
7912 	    UNI(OP_GPWUID);
7913 
7914 	case KEY_getpeername:
7915 	    UNI(OP_GETPEERNAME);
7916 
7917 	case KEY_gethostbyname:
7918 	    UNI(OP_GHBYNAME);
7919 
7920 	case KEY_gethostbyaddr:
7921 	    LOP(OP_GHBYADDR,XTERM);
7922 
7923 	case KEY_gethostent:
7924 	    FUN0(OP_GHOSTENT);
7925 
7926 	case KEY_getnetbyname:
7927 	    UNI(OP_GNBYNAME);
7928 
7929 	case KEY_getnetbyaddr:
7930 	    LOP(OP_GNBYADDR,XTERM);
7931 
7932 	case KEY_getnetent:
7933 	    FUN0(OP_GNETENT);
7934 
7935 	case KEY_getservbyname:
7936 	    LOP(OP_GSBYNAME,XTERM);
7937 
7938 	case KEY_getservbyport:
7939 	    LOP(OP_GSBYPORT,XTERM);
7940 
7941 	case KEY_getservent:
7942 	    FUN0(OP_GSERVENT);
7943 
7944 	case KEY_getsockname:
7945 	    UNI(OP_GETSOCKNAME);
7946 
7947 	case KEY_getsockopt:
7948 	    LOP(OP_GSOCKOPT,XTERM);
7949 
7950 	case KEY_getgrent:
7951 	    FUN0(OP_GGRENT);
7952 
7953 	case KEY_getgrnam:
7954 	    UNI(OP_GGRNAM);
7955 
7956 	case KEY_getgrgid:
7957 	    UNI(OP_GGRGID);
7958 
7959 	case KEY_getlogin:
7960 	    FUN0(OP_GETLOGIN);
7961 
7962 	case KEY_given:
7963 	    pl_yylval.ival = CopLINE(PL_curcop);
7964             Perl_ck_warner_d(aTHX_
7965                 packWARN(WARN_EXPERIMENTAL__SMARTMATCH),
7966                 "given is experimental");
7967 	    OPERATOR(GIVEN);
7968 
7969 	case KEY_glob:
7970 	    LOP(
7971 	     orig_keyword==KEY_glob ? (orig_keyword=0, -OP_GLOB) : OP_GLOB,
7972 	     XTERM
7973 	    );
7974 
7975 	case KEY_hex:
7976 	    UNI(OP_HEX);
7977 
7978 	case KEY_if:
7979 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_NONEXPR)
7980 		return REPORT(0);
7981 	    pl_yylval.ival = CopLINE(PL_curcop);
7982 	    OPERATOR(IF);
7983 
7984 	case KEY_index:
7985 	    LOP(OP_INDEX,XTERM);
7986 
7987 	case KEY_int:
7988 	    UNI(OP_INT);
7989 
7990 	case KEY_ioctl:
7991 	    LOP(OP_IOCTL,XTERM);
7992 
7993 	case KEY_join:
7994 	    LOP(OP_JOIN,XTERM);
7995 
7996 	case KEY_keys:
7997 	    UNI(OP_KEYS);
7998 
7999 	case KEY_kill:
8000 	    LOP(OP_KILL,XTERM);
8001 
8002 	case KEY_last:
8003 	    PL_expect = XOPERATOR;
8004 	    s = force_word(s,WORD,TRUE,FALSE,FALSE);
8005 	    LOOPX(OP_LAST);
8006 
8007 	case KEY_lc:
8008 	    UNI(OP_LC);
8009 
8010 	case KEY_lcfirst:
8011 	    UNI(OP_LCFIRST);
8012 
8013 	case KEY_local:
8014 	    pl_yylval.ival = 0;
8015 	    OPERATOR(LOCAL);
8016 
8017 	case KEY_length:
8018 	    UNI(OP_LENGTH);
8019 
8020 	case KEY_lt:
8021 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE)
8022 		return REPORT(0);
8023 	    Rop(OP_SLT);
8024 
8025 	case KEY_le:
8026 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE)
8027 		return REPORT(0);
8028 	    Rop(OP_SLE);
8029 
8030 	case KEY_localtime:
8031 	    UNI(OP_LOCALTIME);
8032 
8033 	case KEY_log:
8034 	    UNI(OP_LOG);
8035 
8036 	case KEY_link:
8037 	    LOP(OP_LINK,XTERM);
8038 
8039 	case KEY_listen:
8040 	    LOP(OP_LISTEN,XTERM);
8041 
8042 	case KEY_lock:
8043 	    UNI(OP_LOCK);
8044 
8045 	case KEY_lstat:
8046 	    UNI(OP_LSTAT);
8047 
8048 	case KEY_m:
8049 	    s = scan_pat(s,OP_MATCH);
8050 	    TERM(sublex_start());
8051 
8052 	case KEY_map:
8053 	    LOP(OP_MAPSTART, XREF);
8054 
8055 	case KEY_mkdir:
8056 	    LOP(OP_MKDIR,XTERM);
8057 
8058 	case KEY_msgctl:
8059 	    LOP(OP_MSGCTL,XTERM);
8060 
8061 	case KEY_msgget:
8062 	    LOP(OP_MSGGET,XTERM);
8063 
8064 	case KEY_msgrcv:
8065 	    LOP(OP_MSGRCV,XTERM);
8066 
8067 	case KEY_msgsnd:
8068 	    LOP(OP_MSGSND,XTERM);
8069 
8070 	case KEY_our:
8071 	case KEY_my:
8072 	case KEY_state:
8073 	    PL_in_my = (U16)tmp;
8074 	    s = SKIPSPACE1(s);
8075 	    if (isIDFIRST_lazy_if(s,UTF)) {
8076 #ifdef PERL_MAD
8077 		char* start = s;
8078 #endif
8079 		s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
8080 		if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
8081 		{
8082 		    if (!FEATURE_LEXSUBS_IS_ENABLED)
8083 			Perl_croak(aTHX_
8084 				  "Experimental \"%s\" subs not enabled",
8085 				   tmp == KEY_my    ? "my"    :
8086 				   tmp == KEY_state ? "state" : "our");
8087 		    Perl_ck_warner_d(aTHX_
8088 			packWARN(WARN_EXPERIMENTAL__LEXICAL_SUBS),
8089 			"The lexical_subs feature is experimental");
8090 		    goto really_sub;
8091 		}
8092 		PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
8093 		if (!PL_in_my_stash) {
8094 		    char tmpbuf[1024];
8095 		    PL_bufptr = s;
8096 		    my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
8097 		    yyerror_pv(tmpbuf, UTF ? SVf_UTF8 : 0);
8098 		}
8099 #ifdef PERL_MAD
8100 		if (PL_madskills) {	/* just add type to declarator token */
8101 		    sv_catsv(PL_thistoken, PL_nextwhite);
8102 		    PL_nextwhite = 0;
8103 		    sv_catpvn(PL_thistoken, start, s - start);
8104 		}
8105 #endif
8106 	    }
8107 	    pl_yylval.ival = 1;
8108 	    OPERATOR(MY);
8109 
8110 	case KEY_next:
8111 	    PL_expect = XOPERATOR;
8112 	    s = force_word(s,WORD,TRUE,FALSE,FALSE);
8113 	    LOOPX(OP_NEXT);
8114 
8115 	case KEY_ne:
8116 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_COMPARE)
8117 		return REPORT(0);
8118 	    Eop(OP_SNE);
8119 
8120 	case KEY_no:
8121 	    s = tokenize_use(0, s);
8122 	    TERM(USE);
8123 
8124 	case KEY_not:
8125 	    if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
8126 		FUN1(OP_NOT);
8127 	    else {
8128 		if (!PL_lex_allbrackets &&
8129 			PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
8130 		    PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
8131 		OPERATOR(NOTOP);
8132 	    }
8133 
8134 	case KEY_open:
8135 	    s = SKIPSPACE1(s);
8136 	    if (isIDFIRST_lazy_if(s,UTF)) {
8137           const char *t;
8138           d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE,
8139               &len);
8140 		for (t=d; isSPACE(*t);)
8141 		    t++;
8142 		if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
8143 		    /* [perl #16184] */
8144 		    && !(t[0] == '=' && t[1] == '>')
8145 		    && !(t[0] == ':' && t[1] == ':')
8146 		    && !keyword(s, d-s, 0)
8147 		) {
8148 		    SV *tmpsv = newSVpvn_flags(s, (STRLEN)(d-s),
8149                                                 SVs_TEMP | (UTF ? SVf_UTF8 : 0));
8150 		    Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
8151 			   "Precedence problem: open %"SVf" should be open(%"SVf")",
8152 			    SVfARG(tmpsv), SVfARG(tmpsv));
8153 		}
8154 	    }
8155 	    LOP(OP_OPEN,XTERM);
8156 
8157 	case KEY_or:
8158 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_LOWLOGIC)
8159 		return REPORT(0);
8160 	    pl_yylval.ival = OP_OR;
8161 	    OPERATOR(OROP);
8162 
8163 	case KEY_ord:
8164 	    UNI(OP_ORD);
8165 
8166 	case KEY_oct:
8167 	    UNI(OP_OCT);
8168 
8169 	case KEY_opendir:
8170 	    LOP(OP_OPEN_DIR,XTERM);
8171 
8172 	case KEY_print:
8173 	    checkcomma(s,PL_tokenbuf,"filehandle");
8174 	    LOP(OP_PRINT,XREF);
8175 
8176 	case KEY_printf:
8177 	    checkcomma(s,PL_tokenbuf,"filehandle");
8178 	    LOP(OP_PRTF,XREF);
8179 
8180 	case KEY_prototype:
8181 	    UNI(OP_PROTOTYPE);
8182 
8183 	case KEY_push:
8184 	    LOP(OP_PUSH,XTERM);
8185 
8186 	case KEY_pop:
8187 	    UNIDOR(OP_POP);
8188 
8189 	case KEY_pos:
8190 	    UNIDOR(OP_POS);
8191 
8192 	case KEY_pack:
8193 	    LOP(OP_PACK,XTERM);
8194 
8195 	case KEY_package:
8196 	    s = force_word(s,WORD,FALSE,TRUE,FALSE);
8197 	    s = SKIPSPACE1(s);
8198 	    s = force_strict_version(s);
8199 	    PL_lex_expect = XBLOCK;
8200 	    OPERATOR(PACKAGE);
8201 
8202 	case KEY_pipe:
8203 	    LOP(OP_PIPE_OP,XTERM);
8204 
8205 	case KEY_q:
8206 	    s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
8207 	    if (!s)
8208 		missingterm(NULL);
8209 	    pl_yylval.ival = OP_CONST;
8210 	    TERM(sublex_start());
8211 
8212 	case KEY_quotemeta:
8213 	    UNI(OP_QUOTEMETA);
8214 
8215 	case KEY_qw: {
8216 	    OP *words = NULL;
8217 	    s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
8218 	    if (!s)
8219 		missingterm(NULL);
8220 	    PL_expect = XOPERATOR;
8221 	    if (SvCUR(PL_lex_stuff)) {
8222 		int warned_comma = !ckWARN(WARN_QW);
8223 		int warned_comment = warned_comma;
8224 		d = SvPV_force(PL_lex_stuff, len);
8225 		while (len) {
8226 		    for (; isSPACE(*d) && len; --len, ++d)
8227 			/**/;
8228 		    if (len) {
8229 			SV *sv;
8230 			const char *b = d;
8231 			if (!warned_comma || !warned_comment) {
8232 			    for (; !isSPACE(*d) && len; --len, ++d) {
8233 				if (!warned_comma && *d == ',') {
8234 				    Perl_warner(aTHX_ packWARN(WARN_QW),
8235 					"Possible attempt to separate words with commas");
8236 				    ++warned_comma;
8237 				}
8238 				else if (!warned_comment && *d == '#') {
8239 				    Perl_warner(aTHX_ packWARN(WARN_QW),
8240 					"Possible attempt to put comments in qw() list");
8241 				    ++warned_comment;
8242 				}
8243 			    }
8244 			}
8245 			else {
8246 			    for (; !isSPACE(*d) && len; --len, ++d)
8247 				/**/;
8248 			}
8249 			sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
8250 			words = op_append_elem(OP_LIST, words,
8251 					    newSVOP(OP_CONST, 0, tokeq(sv)));
8252 		    }
8253 		}
8254 	    }
8255 	    if (!words)
8256 		words = newNULLLIST();
8257 	    if (PL_lex_stuff) {
8258 		SvREFCNT_dec(PL_lex_stuff);
8259 		PL_lex_stuff = NULL;
8260 	    }
8261 	    PL_expect = XOPERATOR;
8262 	    pl_yylval.opval = sawparens(words);
8263 	    TOKEN(QWLIST);
8264 	}
8265 
8266 	case KEY_qq:
8267 	    s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
8268 	    if (!s)
8269 		missingterm(NULL);
8270 	    pl_yylval.ival = OP_STRINGIFY;
8271 	    if (SvIVX(PL_lex_stuff) == '\'')
8272 		SvIV_set(PL_lex_stuff, 0);	/* qq'$foo' should interpolate */
8273 	    TERM(sublex_start());
8274 
8275 	case KEY_qr:
8276 	    s = scan_pat(s,OP_QR);
8277 	    TERM(sublex_start());
8278 
8279 	case KEY_qx:
8280 	    s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
8281 	    if (!s)
8282 		missingterm(NULL);
8283 	    readpipe_override();
8284 	    TERM(sublex_start());
8285 
8286 	case KEY_return:
8287 	    OLDLOP(OP_RETURN);
8288 
8289 	case KEY_require:
8290 	    s = SKIPSPACE1(s);
8291 	    PL_expect = XOPERATOR;
8292 	    if (isDIGIT(*s)) {
8293 		s = force_version(s, FALSE);
8294 	    }
8295 	    else if (*s != 'v' || !isDIGIT(s[1])
8296 		    || (s = force_version(s, TRUE), *s == 'v'))
8297 	    {
8298 		*PL_tokenbuf = '\0';
8299 		s = force_word(s,WORD,TRUE,TRUE,FALSE);
8300 		if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
8301 		    gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf),
8302                                 GV_ADD | (UTF ? SVf_UTF8 : 0));
8303 		else if (*s == '<')
8304 		    yyerror("<> should be quotes");
8305 	    }
8306 	    if (orig_keyword == KEY_require) {
8307 		orig_keyword = 0;
8308 		pl_yylval.ival = 1;
8309 	    }
8310 	    else
8311 		pl_yylval.ival = 0;
8312 	    PL_expect = XTERM;
8313 	    PL_bufptr = s;
8314 	    PL_last_uni = PL_oldbufptr;
8315 	    PL_last_lop_op = OP_REQUIRE;
8316 	    s = skipspace(s);
8317 	    return REPORT( (int)REQUIRE );
8318 
8319 	case KEY_reset:
8320 	    UNI(OP_RESET);
8321 
8322 	case KEY_redo:
8323 	    PL_expect = XOPERATOR;
8324 	    s = force_word(s,WORD,TRUE,FALSE,FALSE);
8325 	    LOOPX(OP_REDO);
8326 
8327 	case KEY_rename:
8328 	    LOP(OP_RENAME,XTERM);
8329 
8330 	case KEY_rand:
8331 	    UNI(OP_RAND);
8332 
8333 	case KEY_rmdir:
8334 	    UNI(OP_RMDIR);
8335 
8336 	case KEY_rindex:
8337 	    LOP(OP_RINDEX,XTERM);
8338 
8339 	case KEY_read:
8340 	    LOP(OP_READ,XTERM);
8341 
8342 	case KEY_readdir:
8343 	    UNI(OP_READDIR);
8344 
8345 	case KEY_readline:
8346 	    UNIDOR(OP_READLINE);
8347 
8348 	case KEY_readpipe:
8349 	    UNIDOR(OP_BACKTICK);
8350 
8351 	case KEY_rewinddir:
8352 	    UNI(OP_REWINDDIR);
8353 
8354 	case KEY_recv:
8355 	    LOP(OP_RECV,XTERM);
8356 
8357 	case KEY_reverse:
8358 	    LOP(OP_REVERSE,XTERM);
8359 
8360 	case KEY_readlink:
8361 	    UNIDOR(OP_READLINK);
8362 
8363 	case KEY_ref:
8364 	    UNI(OP_REF);
8365 
8366 	case KEY_s:
8367 	    s = scan_subst(s);
8368 	    if (pl_yylval.opval)
8369 		TERM(sublex_start());
8370 	    else
8371 		TOKEN(1);	/* force error */
8372 
8373 	case KEY_say:
8374 	    checkcomma(s,PL_tokenbuf,"filehandle");
8375 	    LOP(OP_SAY,XREF);
8376 
8377 	case KEY_chomp:
8378 	    UNI(OP_CHOMP);
8379 
8380 	case KEY_scalar:
8381 	    UNI(OP_SCALAR);
8382 
8383 	case KEY_select:
8384 	    LOP(OP_SELECT,XTERM);
8385 
8386 	case KEY_seek:
8387 	    LOP(OP_SEEK,XTERM);
8388 
8389 	case KEY_semctl:
8390 	    LOP(OP_SEMCTL,XTERM);
8391 
8392 	case KEY_semget:
8393 	    LOP(OP_SEMGET,XTERM);
8394 
8395 	case KEY_semop:
8396 	    LOP(OP_SEMOP,XTERM);
8397 
8398 	case KEY_send:
8399 	    LOP(OP_SEND,XTERM);
8400 
8401 	case KEY_setpgrp:
8402 	    LOP(OP_SETPGRP,XTERM);
8403 
8404 	case KEY_setpriority:
8405 	    LOP(OP_SETPRIORITY,XTERM);
8406 
8407 	case KEY_sethostent:
8408 	    UNI(OP_SHOSTENT);
8409 
8410 	case KEY_setnetent:
8411 	    UNI(OP_SNETENT);
8412 
8413 	case KEY_setservent:
8414 	    UNI(OP_SSERVENT);
8415 
8416 	case KEY_setprotoent:
8417 	    UNI(OP_SPROTOENT);
8418 
8419 	case KEY_setpwent:
8420 	    FUN0(OP_SPWENT);
8421 
8422 	case KEY_setgrent:
8423 	    FUN0(OP_SGRENT);
8424 
8425 	case KEY_seekdir:
8426 	    LOP(OP_SEEKDIR,XTERM);
8427 
8428 	case KEY_setsockopt:
8429 	    LOP(OP_SSOCKOPT,XTERM);
8430 
8431 	case KEY_shift:
8432 	    UNIDOR(OP_SHIFT);
8433 
8434 	case KEY_shmctl:
8435 	    LOP(OP_SHMCTL,XTERM);
8436 
8437 	case KEY_shmget:
8438 	    LOP(OP_SHMGET,XTERM);
8439 
8440 	case KEY_shmread:
8441 	    LOP(OP_SHMREAD,XTERM);
8442 
8443 	case KEY_shmwrite:
8444 	    LOP(OP_SHMWRITE,XTERM);
8445 
8446 	case KEY_shutdown:
8447 	    LOP(OP_SHUTDOWN,XTERM);
8448 
8449 	case KEY_sin:
8450 	    UNI(OP_SIN);
8451 
8452 	case KEY_sleep:
8453 	    UNI(OP_SLEEP);
8454 
8455 	case KEY_socket:
8456 	    LOP(OP_SOCKET,XTERM);
8457 
8458 	case KEY_socketpair:
8459 	    LOP(OP_SOCKPAIR,XTERM);
8460 
8461 	case KEY_sort:
8462 	    checkcomma(s,PL_tokenbuf,"subroutine name");
8463 	    s = SKIPSPACE1(s);
8464 	    PL_expect = XTERM;
8465 	    s = force_word(s,WORD,TRUE,TRUE,FALSE);
8466 	    LOP(OP_SORT,XREF);
8467 
8468 	case KEY_split:
8469 	    LOP(OP_SPLIT,XTERM);
8470 
8471 	case KEY_sprintf:
8472 	    LOP(OP_SPRINTF,XTERM);
8473 
8474 	case KEY_splice:
8475 	    LOP(OP_SPLICE,XTERM);
8476 
8477 	case KEY_sqrt:
8478 	    UNI(OP_SQRT);
8479 
8480 	case KEY_srand:
8481 	    UNI(OP_SRAND);
8482 
8483 	case KEY_stat:
8484 	    UNI(OP_STAT);
8485 
8486 	case KEY_study:
8487 	    UNI(OP_STUDY);
8488 
8489 	case KEY_substr:
8490 	    LOP(OP_SUBSTR,XTERM);
8491 
8492 	case KEY_format:
8493 	case KEY_sub:
8494 	  really_sub:
8495 	    {
8496 		char * const tmpbuf = PL_tokenbuf + 1;
8497 		SSize_t tboffset = 0;
8498 		expectation attrful;
8499 		bool have_name, have_proto;
8500 		const int key = tmp;
8501 
8502 #ifdef PERL_MAD
8503 		SV *tmpwhite = 0;
8504 
8505 		char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
8506 		SV *subtoken = PL_madskills
8507 		   ? newSVpvn_flags(tstart, s - tstart, SvUTF8(PL_linestr))
8508 		   : NULL;
8509 		PL_thistoken = 0;
8510 
8511 		d = s;
8512 		s = SKIPSPACE2(s,tmpwhite);
8513 #else
8514 		d = s;
8515 		s = skipspace(s);
8516 #endif
8517 
8518 		if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
8519 		    (*s == ':' && s[1] == ':'))
8520 		{
8521 #ifdef PERL_MAD
8522 		    SV *nametoke = NULL;
8523 #endif
8524 
8525 		    PL_expect = XBLOCK;
8526 		    attrful = XATTRBLOCK;
8527 		    /* remember buffer pos'n for later force_word */
8528 		    tboffset = s - PL_oldbufptr;
8529 		    d = scan_word(s, tmpbuf, sizeof PL_tokenbuf - 1, TRUE,
8530 				  &len);
8531 #ifdef PERL_MAD
8532 		    if (PL_madskills)
8533 			nametoke = newSVpvn_flags(s, d - s, SvUTF8(PL_linestr));
8534 #endif
8535 		    *PL_tokenbuf = '&';
8536 		    if (memchr(tmpbuf, ':', len) || key != KEY_sub
8537 		     || pad_findmy_pvn(
8538 			    PL_tokenbuf, len + 1, UTF ? SVf_UTF8 : 0
8539 			) != NOT_IN_PAD)
8540 			sv_setpvn(PL_subname, tmpbuf, len);
8541 		    else {
8542 			sv_setsv(PL_subname,PL_curstname);
8543 			sv_catpvs(PL_subname,"::");
8544 			sv_catpvn(PL_subname,tmpbuf,len);
8545 		    }
8546                     if (SvUTF8(PL_linestr))
8547                         SvUTF8_on(PL_subname);
8548 		    have_name = TRUE;
8549 
8550 
8551 #ifdef PERL_MAD
8552 		    start_force(0);
8553 		    CURMAD('X', nametoke);
8554 		    CURMAD('_', tmpwhite);
8555 		    force_ident_maybe_lex('&');
8556 
8557 		    s = SKIPSPACE2(d,tmpwhite);
8558 #else
8559 		    s = skipspace(d);
8560 #endif
8561 		}
8562 		else {
8563 		    if (key == KEY_my || key == KEY_our || key==KEY_state)
8564 		    {
8565 			*d = '\0';
8566 			/* diag_listed_as: Missing name in "%s sub" */
8567 			Perl_croak(aTHX_
8568 				  "Missing name in \"%s\"", PL_bufptr);
8569 		    }
8570 		    PL_expect = XTERMBLOCK;
8571 		    attrful = XATTRTERM;
8572 		    sv_setpvs(PL_subname,"?");
8573 		    have_name = FALSE;
8574 		}
8575 
8576 		if (key == KEY_format) {
8577 #ifdef PERL_MAD
8578 		    PL_thistoken = subtoken;
8579 		    s = d;
8580                     PERL_UNUSED_VAR(tboffset);
8581 #else
8582 		    if (have_name)
8583 			(void) force_word(PL_oldbufptr + tboffset, WORD,
8584 					  FALSE, TRUE, TRUE);
8585 #endif
8586 		    PREBLOCK(FORMAT);
8587 		}
8588 
8589 		/* Look for a prototype */
8590 		if (*s == '(') {
8591 		    char *p;
8592 		    bool bad_proto = FALSE;
8593 		    bool in_brackets = FALSE;
8594 		    char greedy_proto = ' ';
8595 		    bool proto_after_greedy_proto = FALSE;
8596 		    bool must_be_last = FALSE;
8597 		    bool underscore = FALSE;
8598 		    bool seen_underscore = FALSE;
8599 		    const bool warnillegalproto = ckWARN(WARN_ILLEGALPROTO);
8600                     STRLEN tmplen;
8601 
8602 		    s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
8603 		    if (!s)
8604 			Perl_croak(aTHX_ "Prototype not terminated");
8605 		    /* strip spaces and check for bad characters */
8606 		    d = SvPV(PL_lex_stuff, tmplen);
8607 		    tmp = 0;
8608 		    for (p = d; tmplen; tmplen--, ++p) {
8609 			if (!isSPACE(*p)) {
8610                             d[tmp++] = *p;
8611 
8612 			    if (warnillegalproto) {
8613 				if (must_be_last)
8614 				    proto_after_greedy_proto = TRUE;
8615 				if (!strchr("$@%*;[]&\\_+", *p) || *p == '\0') {
8616 				    bad_proto = TRUE;
8617 				}
8618 				else {
8619 				    if ( underscore ) {
8620 					if ( !strchr(";@%", *p) )
8621 					    bad_proto = TRUE;
8622 					underscore = FALSE;
8623 				    }
8624 				    if ( *p == '[' ) {
8625 					in_brackets = TRUE;
8626 				    }
8627 				    else if ( *p == ']' ) {
8628 					in_brackets = FALSE;
8629 				    }
8630 				    else if ( (*p == '@' || *p == '%') &&
8631 					 ( tmp < 2 || d[tmp-2] != '\\' ) &&
8632 					 !in_brackets ) {
8633 					must_be_last = TRUE;
8634 					greedy_proto = *p;
8635 				    }
8636 				    else if ( *p == '_' ) {
8637 					underscore = seen_underscore = TRUE;
8638 				    }
8639 				}
8640 			    }
8641 			}
8642 		    }
8643                     d[tmp] = '\0';
8644 		    if (proto_after_greedy_proto)
8645 			Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
8646 				    "Prototype after '%c' for %"SVf" : %s",
8647 				    greedy_proto, SVfARG(PL_subname), d);
8648 		    if (bad_proto) {
8649                         SV *dsv = newSVpvs_flags("", SVs_TEMP);
8650 			Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
8651 				    "Illegal character %sin prototype for %"SVf" : %s",
8652 				    seen_underscore ? "after '_' " : "",
8653 				    SVfARG(PL_subname),
8654                                     SvUTF8(PL_lex_stuff)
8655                                         ? sv_uni_display(dsv,
8656                                             newSVpvn_flags(d, tmp, SVs_TEMP | SVf_UTF8),
8657                                             tmp,
8658                                             UNI_DISPLAY_ISPRINT)
8659                                         : pv_pretty(dsv, d, tmp, 60, NULL, NULL,
8660                                             PERL_PV_ESCAPE_NONASCII));
8661                     }
8662                     SvCUR_set(PL_lex_stuff, tmp);
8663 		    have_proto = TRUE;
8664 
8665 #ifdef PERL_MAD
8666 		    start_force(0);
8667 		    CURMAD('q', PL_thisopen);
8668 		    CURMAD('_', tmpwhite);
8669 		    CURMAD('=', PL_thisstuff);
8670 		    CURMAD('Q', PL_thisclose);
8671 		    NEXTVAL_NEXTTOKE.opval =
8672 			(OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
8673 		    PL_lex_stuff = NULL;
8674 		    force_next(THING);
8675 
8676 		    s = SKIPSPACE2(s,tmpwhite);
8677 #else
8678 		    s = skipspace(s);
8679 #endif
8680 		}
8681 		else
8682 		    have_proto = FALSE;
8683 
8684 		if (*s == ':' && s[1] != ':')
8685 		    PL_expect = attrful;
8686 		else if (*s != '{' && key == KEY_sub) {
8687 		    if (!have_name)
8688 			Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
8689 		    else if (*s != ';' && *s != '}')
8690 			Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
8691 		}
8692 
8693 #ifdef PERL_MAD
8694 		start_force(0);
8695 		if (tmpwhite) {
8696 		    if (PL_madskills)
8697 			curmad('^', newSVpvs(""));
8698 		    CURMAD('_', tmpwhite);
8699 		}
8700 		force_next(0);
8701 
8702 		PL_thistoken = subtoken;
8703                 PERL_UNUSED_VAR(have_proto);
8704 #else
8705 		if (have_proto) {
8706 		    NEXTVAL_NEXTTOKE.opval =
8707 			(OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
8708 		    PL_lex_stuff = NULL;
8709 		    force_next(THING);
8710 		}
8711 #endif
8712 		if (!have_name) {
8713 		    if (PL_curstash)
8714 			sv_setpvs(PL_subname, "__ANON__");
8715 		    else
8716 			sv_setpvs(PL_subname, "__ANON__::__ANON__");
8717 		    TOKEN(ANONSUB);
8718 		}
8719 #ifndef PERL_MAD
8720 		force_ident_maybe_lex('&');
8721 #endif
8722 		TOKEN(SUB);
8723 	    }
8724 
8725 	case KEY_system:
8726 	    LOP(OP_SYSTEM,XREF);
8727 
8728 	case KEY_symlink:
8729 	    LOP(OP_SYMLINK,XTERM);
8730 
8731 	case KEY_syscall:
8732 	    LOP(OP_SYSCALL,XTERM);
8733 
8734 	case KEY_sysopen:
8735 	    LOP(OP_SYSOPEN,XTERM);
8736 
8737 	case KEY_sysseek:
8738 	    LOP(OP_SYSSEEK,XTERM);
8739 
8740 	case KEY_sysread:
8741 	    LOP(OP_SYSREAD,XTERM);
8742 
8743 	case KEY_syswrite:
8744 	    LOP(OP_SYSWRITE,XTERM);
8745 
8746 	case KEY_tr:
8747 	case KEY_y:
8748 	    s = scan_trans(s);
8749 	    TERM(sublex_start());
8750 
8751 	case KEY_tell:
8752 	    UNI(OP_TELL);
8753 
8754 	case KEY_telldir:
8755 	    UNI(OP_TELLDIR);
8756 
8757 	case KEY_tie:
8758 	    LOP(OP_TIE,XTERM);
8759 
8760 	case KEY_tied:
8761 	    UNI(OP_TIED);
8762 
8763 	case KEY_time:
8764 	    FUN0(OP_TIME);
8765 
8766 	case KEY_times:
8767 	    FUN0(OP_TMS);
8768 
8769 	case KEY_truncate:
8770 	    LOP(OP_TRUNCATE,XTERM);
8771 
8772 	case KEY_uc:
8773 	    UNI(OP_UC);
8774 
8775 	case KEY_ucfirst:
8776 	    UNI(OP_UCFIRST);
8777 
8778 	case KEY_untie:
8779 	    UNI(OP_UNTIE);
8780 
8781 	case KEY_until:
8782 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_NONEXPR)
8783 		return REPORT(0);
8784 	    pl_yylval.ival = CopLINE(PL_curcop);
8785 	    OPERATOR(UNTIL);
8786 
8787 	case KEY_unless:
8788 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_NONEXPR)
8789 		return REPORT(0);
8790 	    pl_yylval.ival = CopLINE(PL_curcop);
8791 	    OPERATOR(UNLESS);
8792 
8793 	case KEY_unlink:
8794 	    LOP(OP_UNLINK,XTERM);
8795 
8796 	case KEY_undef:
8797 	    UNIDOR(OP_UNDEF);
8798 
8799 	case KEY_unpack:
8800 	    LOP(OP_UNPACK,XTERM);
8801 
8802 	case KEY_utime:
8803 	    LOP(OP_UTIME,XTERM);
8804 
8805 	case KEY_umask:
8806 	    UNIDOR(OP_UMASK);
8807 
8808 	case KEY_unshift:
8809 	    LOP(OP_UNSHIFT,XTERM);
8810 
8811 	case KEY_use:
8812 	    s = tokenize_use(1, s);
8813 	    OPERATOR(USE);
8814 
8815 	case KEY_values:
8816 	    UNI(OP_VALUES);
8817 
8818 	case KEY_vec:
8819 	    LOP(OP_VEC,XTERM);
8820 
8821 	case KEY_when:
8822 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_NONEXPR)
8823 		return REPORT(0);
8824 	    pl_yylval.ival = CopLINE(PL_curcop);
8825             Perl_ck_warner_d(aTHX_
8826                 packWARN(WARN_EXPERIMENTAL__SMARTMATCH),
8827                 "when is experimental");
8828 	    OPERATOR(WHEN);
8829 
8830 	case KEY_while:
8831 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_NONEXPR)
8832 		return REPORT(0);
8833 	    pl_yylval.ival = CopLINE(PL_curcop);
8834 	    OPERATOR(WHILE);
8835 
8836 	case KEY_warn:
8837 	    PL_hints |= HINT_BLOCK_SCOPE;
8838 	    LOP(OP_WARN,XTERM);
8839 
8840 	case KEY_wait:
8841 	    FUN0(OP_WAIT);
8842 
8843 	case KEY_waitpid:
8844 	    LOP(OP_WAITPID,XTERM);
8845 
8846 	case KEY_wantarray:
8847 	    FUN0(OP_WANTARRAY);
8848 
8849 	case KEY_write:
8850 #ifdef EBCDIC
8851 	{
8852 	    char ctl_l[2];
8853 	    ctl_l[0] = toCTRL('L');
8854 	    ctl_l[1] = '\0';
8855 	    gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
8856 	}
8857 #else
8858 	    /* Make sure $^L is defined */
8859 	    gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
8860 #endif
8861 	    UNI(OP_ENTERWRITE);
8862 
8863 	case KEY_x:
8864 	    if (PL_expect == XOPERATOR) {
8865 		if (*s == '=' && !PL_lex_allbrackets &&
8866 			PL_lex_fakeeof >= LEX_FAKEEOF_ASSIGN)
8867 		    return REPORT(0);
8868 		Mop(OP_REPEAT);
8869 	    }
8870 	    check_uni();
8871 	    goto just_a_word;
8872 
8873 	case KEY_xor:
8874 	    if (!PL_lex_allbrackets && PL_lex_fakeeof >= LEX_FAKEEOF_LOWLOGIC)
8875 		return REPORT(0);
8876 	    pl_yylval.ival = OP_XOR;
8877 	    OPERATOR(OROP);
8878 	}
8879     }}
8880 }
8881 #ifdef __SC__
8882 #pragma segment Main
8883 #endif
8884 
8885 /*
8886   S_pending_ident
8887 
8888   Looks up an identifier in the pad or in a package
8889 
8890   Returns:
8891     PRIVATEREF if this is a lexical name.
8892     WORD       if this belongs to a package.
8893 
8894   Structure:
8895       if we're in a my declaration
8896 	  croak if they tried to say my($foo::bar)
8897 	  build the ops for a my() declaration
8898       if it's an access to a my() variable
8899 	  build ops for access to a my() variable
8900       if in a dq string, and they've said @foo and we can't find @foo
8901 	  warn
8902       build ops for a bareword
8903 */
8904 
8905 static int
8906 S_pending_ident(pTHX)
8907 {
8908     dVAR;
8909     PADOFFSET tmp = 0;
8910     const char pit = (char)pl_yylval.ival;
8911     const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
8912     /* All routes through this function want to know if there is a colon.  */
8913     const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
8914 
8915     DEBUG_T({ PerlIO_printf(Perl_debug_log,
8916           "### Pending identifier '%s'\n", PL_tokenbuf); });
8917 
8918     /* if we're in a my(), we can't allow dynamics here.
8919        $foo'bar has already been turned into $foo::bar, so
8920        just check for colons.
8921 
8922        if it's a legal name, the OP is a PADANY.
8923     */
8924     if (PL_in_my) {
8925         if (PL_in_my == KEY_our) {	/* "our" is merely analogous to "my" */
8926             if (has_colon)
8927                 yyerror_pv(Perl_form(aTHX_ "No package name allowed for "
8928                                   "variable %s in \"our\"",
8929                                   PL_tokenbuf), UTF ? SVf_UTF8 : 0);
8930             tmp = allocmy(PL_tokenbuf, tokenbuf_len, UTF ? SVf_UTF8 : 0);
8931         }
8932         else {
8933             if (has_colon)
8934                 yyerror_pv(Perl_form(aTHX_ PL_no_myglob,
8935 			    PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf),
8936                             UTF ? SVf_UTF8 : 0);
8937 
8938             pl_yylval.opval = newOP(OP_PADANY, 0);
8939             pl_yylval.opval->op_targ = allocmy(PL_tokenbuf, tokenbuf_len,
8940                                                         UTF ? SVf_UTF8 : 0);
8941 	    return PRIVATEREF;
8942         }
8943     }
8944 
8945     /*
8946        build the ops for accesses to a my() variable.
8947     */
8948 
8949     if (!has_colon) {
8950 	if (!PL_in_my)
8951 	    tmp = pad_findmy_pvn(PL_tokenbuf, tokenbuf_len,
8952                                     UTF ? SVf_UTF8 : 0);
8953         if (tmp != NOT_IN_PAD) {
8954             /* might be an "our" variable" */
8955             if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
8956                 /* build ops for a bareword */
8957 		HV *  const stash = PAD_COMPNAME_OURSTASH(tmp);
8958 		HEK * const stashname = HvNAME_HEK(stash);
8959 		SV *  const sym = newSVhek(stashname);
8960                 sv_catpvs(sym, "::");
8961                 sv_catpvn_flags(sym, PL_tokenbuf+1, tokenbuf_len - 1, (UTF ? SV_CATUTF8 : SV_CATBYTES ));
8962                 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
8963                 pl_yylval.opval->op_private = OPpCONST_ENTERED;
8964                 if (pit != '&')
8965                   gv_fetchsv(sym,
8966                     (PL_in_eval
8967                         ? (GV_ADDMULTI | GV_ADDINEVAL)
8968                         : GV_ADDMULTI
8969                     ),
8970                     ((PL_tokenbuf[0] == '$') ? SVt_PV
8971                      : (PL_tokenbuf[0] == '@') ? SVt_PVAV
8972                      : SVt_PVHV));
8973                 return WORD;
8974             }
8975 
8976             pl_yylval.opval = newOP(OP_PADANY, 0);
8977             pl_yylval.opval->op_targ = tmp;
8978             return PRIVATEREF;
8979         }
8980     }
8981 
8982     /*
8983        Whine if they've said @foo in a doublequoted string,
8984        and @foo isn't a variable we can find in the symbol
8985        table.
8986     */
8987     if (ckWARN(WARN_AMBIGUOUS) &&
8988 	pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
8989         GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1,
8990                                         ( UTF ? SVf_UTF8 : 0 ), SVt_PVAV);
8991         if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
8992 		/* DO NOT warn for @- and @+ */
8993 		&& !( PL_tokenbuf[2] == '\0' &&
8994 		    ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
8995 	   )
8996         {
8997             /* Downgraded from fatal to warning 20000522 mjd */
8998             Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
8999 			"Possible unintended interpolation of %"SVf" in string",
9000 			SVfARG(newSVpvn_flags(PL_tokenbuf, tokenbuf_len,
9001                                         SVs_TEMP | ( UTF ? SVf_UTF8 : 0 ))));
9002         }
9003     }
9004 
9005     /* build ops for a bareword */
9006     pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
9007 				   newSVpvn_flags(PL_tokenbuf + 1,
9008 						      tokenbuf_len - 1,
9009                                                       UTF ? SVf_UTF8 : 0 ));
9010     pl_yylval.opval->op_private = OPpCONST_ENTERED;
9011     if (pit != '&')
9012 	gv_fetchpvn_flags(PL_tokenbuf+1, tokenbuf_len - 1,
9013 		     (PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD)
9014                      | ( UTF ? SVf_UTF8 : 0 ),
9015 		     ((PL_tokenbuf[0] == '$') ? SVt_PV
9016 		      : (PL_tokenbuf[0] == '@') ? SVt_PVAV
9017 		      : SVt_PVHV));
9018     return WORD;
9019 }
9020 
9021 STATIC void
9022 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
9023 {
9024     dVAR;
9025 
9026     PERL_ARGS_ASSERT_CHECKCOMMA;
9027 
9028     if (*s == ' ' && s[1] == '(') {	/* XXX gotta be a better way */
9029 	if (ckWARN(WARN_SYNTAX)) {
9030 	    int level = 1;
9031 	    const char *w;
9032 	    for (w = s+2; *w && level; w++) {
9033 		if (*w == '(')
9034 		    ++level;
9035 		else if (*w == ')')
9036 		    --level;
9037 	    }
9038 	    while (isSPACE(*w))
9039 		++w;
9040 	    /* the list of chars below is for end of statements or
9041 	     * block / parens, boolean operators (&&, ||, //) and branch
9042 	     * constructs (or, and, if, until, unless, while, err, for).
9043 	     * Not a very solid hack... */
9044 	    if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
9045 		Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
9046 			    "%s (...) interpreted as function",name);
9047 	}
9048     }
9049     while (s < PL_bufend && isSPACE(*s))
9050 	s++;
9051     if (*s == '(')
9052 	s++;
9053     while (s < PL_bufend && isSPACE(*s))
9054 	s++;
9055     if (isIDFIRST_lazy_if(s,UTF)) {
9056 	const char * const w = s;
9057         s += UTF ? UTF8SKIP(s) : 1;
9058 	while (isWORDCHAR_lazy_if(s,UTF))
9059 	    s += UTF ? UTF8SKIP(s) : 1;
9060 	while (s < PL_bufend && isSPACE(*s))
9061 	    s++;
9062 	if (*s == ',') {
9063 	    GV* gv;
9064 	    if (keyword(w, s - w, 0))
9065 		return;
9066 
9067 	    gv = gv_fetchpvn_flags(w, s - w, ( UTF ? SVf_UTF8 : 0 ), SVt_PVCV);
9068 	    if (gv && GvCVu(gv))
9069 		return;
9070 	    Perl_croak(aTHX_ "No comma allowed after %s", what);
9071 	}
9072     }
9073 }
9074 
9075 /* S_new_constant(): do any overload::constant lookup.
9076 
9077    Either returns sv, or mortalizes/frees sv and returns a new SV*.
9078    Best used as sv=new_constant(..., sv, ...).
9079    If s, pv are NULL, calls subroutine with one argument,
9080    and <type> is used with error messages only.
9081    <type> is assumed to be well formed UTF-8 */
9082 
9083 STATIC SV *
9084 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
9085 	       SV *sv, SV *pv, const char *type, STRLEN typelen)
9086 {
9087     dVAR; dSP;
9088     HV * table = GvHV(PL_hintgv);		 /* ^H */
9089     SV *res;
9090     SV *errsv = NULL;
9091     SV **cvp;
9092     SV *cv, *typesv;
9093     const char *why1 = "", *why2 = "", *why3 = "";
9094 
9095     PERL_ARGS_ASSERT_NEW_CONSTANT;
9096     /* We assume that this is true: */
9097     if (*key == 'c') { assert (strEQ(key, "charnames")); }
9098     assert(type || s);
9099 
9100     /* charnames doesn't work well if there have been errors found */
9101     if (PL_error_count > 0 && *key == 'c')
9102     {
9103 	SvREFCNT_dec_NN(sv);
9104 	return &PL_sv_undef;
9105     }
9106 
9107     sv_2mortal(sv);			/* Parent created it permanently */
9108     if (!table
9109 	|| ! (PL_hints & HINT_LOCALIZE_HH)
9110 	|| ! (cvp = hv_fetch(table, key, keylen, FALSE))
9111 	|| ! SvOK(*cvp))
9112     {
9113 	char *msg;
9114 
9115 	/* Here haven't found what we're looking for.  If it is charnames,
9116 	 * perhaps it needs to be loaded.  Try doing that before giving up */
9117 	if (*key == 'c') {
9118 	    Perl_load_module(aTHX_
9119 		            0,
9120 			    newSVpvs("_charnames"),
9121 			     /* version parameter; no need to specify it, as if
9122 			      * we get too early a version, will fail anyway,
9123 			      * not being able to find '_charnames' */
9124 			    NULL,
9125 			    newSVpvs(":full"),
9126 			    newSVpvs(":short"),
9127 			    NULL);
9128 	    SPAGAIN;
9129 	    table = GvHV(PL_hintgv);
9130 	    if (table
9131 		&& (PL_hints & HINT_LOCALIZE_HH)
9132 		&& (cvp = hv_fetch(table, key, keylen, FALSE))
9133 		&& SvOK(*cvp))
9134 	    {
9135 		goto now_ok;
9136 	    }
9137 	}
9138 	if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
9139 	    msg = Perl_form(aTHX_
9140 			       "Constant(%.*s) unknown",
9141 				(int)(type ? typelen : len),
9142 				(type ? type: s));
9143 	}
9144 	else {
9145             why1 = "$^H{";
9146             why2 = key;
9147             why3 = "} is not defined";
9148         report:
9149             if (*key == 'c') {
9150                 msg = Perl_form(aTHX_
9151                             /* The +3 is for '\N{'; -4 for that, plus '}' */
9152                             "Unknown charname '%.*s'", (int)typelen - 4, type + 3
9153                       );
9154             }
9155             else {
9156                 msg = Perl_form(aTHX_ "Constant(%.*s): %s%s%s",
9157                                     (int)(type ? typelen : len),
9158                                     (type ? type: s), why1, why2, why3);
9159             }
9160         }
9161 	yyerror_pv(msg, UTF ? SVf_UTF8 : 0);
9162   	return SvREFCNT_inc_simple_NN(sv);
9163     }
9164 now_ok:
9165     cv = *cvp;
9166     if (!pv && s)
9167   	pv = newSVpvn_flags(s, len, SVs_TEMP);
9168     if (type && pv)
9169   	typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
9170     else
9171   	typesv = &PL_sv_undef;
9172 
9173     PUSHSTACKi(PERLSI_OVERLOAD);
9174     ENTER ;
9175     SAVETMPS;
9176 
9177     PUSHMARK(SP) ;
9178     EXTEND(sp, 3);
9179     if (pv)
9180  	PUSHs(pv);
9181     PUSHs(sv);
9182     if (pv)
9183  	PUSHs(typesv);
9184     PUTBACK;
9185     call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
9186 
9187     SPAGAIN ;
9188 
9189     /* Check the eval first */
9190     if (!PL_in_eval && ((errsv = ERRSV), SvTRUE_NN(errsv))) {
9191 	STRLEN errlen;
9192 	const char * errstr;
9193 	sv_catpvs(errsv, "Propagated");
9194 	errstr = SvPV_const(errsv, errlen);
9195 	yyerror_pvn(errstr, errlen, 0); /* Duplicates the message inside eval */
9196 	(void)POPs;
9197 	res = SvREFCNT_inc_simple_NN(sv);
9198     }
9199     else {
9200  	res = POPs;
9201 	SvREFCNT_inc_simple_void_NN(res);
9202     }
9203 
9204     PUTBACK ;
9205     FREETMPS ;
9206     LEAVE ;
9207     POPSTACK;
9208 
9209     if (!SvOK(res)) {
9210  	why1 = "Call to &{$^H{";
9211  	why2 = key;
9212  	why3 = "}} did not return a defined value";
9213  	sv = res;
9214 	(void)sv_2mortal(sv);
9215  	goto report;
9216     }
9217 
9218     return res;
9219 }
9220 
9221 PERL_STATIC_INLINE void
9222 S_parse_ident(pTHX_ char **s, char **d, char * const e, int allow_package, bool is_utf8) {
9223     dVAR;
9224     PERL_ARGS_ASSERT_PARSE_IDENT;
9225 
9226     for (;;) {
9227         if (*d >= e)
9228             Perl_croak(aTHX_ "%s", ident_too_long);
9229         if (is_utf8 && isIDFIRST_utf8((U8*)*s)) {
9230              /* The UTF-8 case must come first, otherwise things
9231              * like c\N{COMBINING TILDE} would start failing, as the
9232              * isWORDCHAR_A case below would gobble the 'c' up.
9233              */
9234 
9235             char *t = *s + UTF8SKIP(*s);
9236             while (isIDCONT_utf8((U8*)t))
9237                 t += UTF8SKIP(t);
9238             if (*d + (t - *s) > e)
9239                 Perl_croak(aTHX_ "%s", ident_too_long);
9240             Copy(*s, *d, t - *s, char);
9241             *d += t - *s;
9242             *s = t;
9243         }
9244         else if ( isWORDCHAR_A(**s) ) {
9245             do {
9246                 *(*d)++ = *(*s)++;
9247             } while (isWORDCHAR_A(**s) && *d < e);
9248         }
9249         else if (allow_package && **s == '\'' && isIDFIRST_lazy_if(*s+1,is_utf8)) {
9250             *(*d)++ = ':';
9251             *(*d)++ = ':';
9252             (*s)++;
9253         }
9254         else if (allow_package && **s == ':' && (*s)[1] == ':'
9255            /* Disallow things like Foo::$bar. For the curious, this is
9256             * the code path that triggers the "Bad name after" warning
9257             * when looking for barewords.
9258             */
9259            && (*s)[2] != '$') {
9260             *(*d)++ = *(*s)++;
9261             *(*d)++ = *(*s)++;
9262         }
9263         else
9264             break;
9265     }
9266     return;
9267 }
9268 
9269 /* Returns a NUL terminated string, with the length of the string written to
9270    *slp
9271    */
9272 STATIC char *
9273 S_scan_word(pTHX_ char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
9274 {
9275     dVAR;
9276     char *d = dest;
9277     char * const e = d + destlen - 3;  /* two-character token, ending NUL */
9278     bool is_utf8 = cBOOL(UTF);
9279 
9280     PERL_ARGS_ASSERT_SCAN_WORD;
9281 
9282     parse_ident(&s, &d, e, allow_package, is_utf8);
9283     *d = '\0';
9284     *slp = d - dest;
9285     return s;
9286 }
9287 
9288 STATIC char *
9289 S_scan_ident(pTHX_ char *s, const char *send, char *dest, STRLEN destlen, I32 ck_uni)
9290 {
9291     dVAR;
9292     char *bracket = NULL;
9293     char funny = *s++;
9294     char *d = dest;
9295     char * const e = d + destlen - 3;    /* two-character token, ending NUL */
9296     bool is_utf8 = cBOOL(UTF);
9297 
9298     PERL_ARGS_ASSERT_SCAN_IDENT;
9299 
9300     if (isSPACE(*s))
9301 	s = PEEKSPACE(s);
9302     if (isDIGIT(*s)) {
9303 	while (isDIGIT(*s)) {
9304 	    if (d >= e)
9305 		Perl_croak(aTHX_ "%s", ident_too_long);
9306 	    *d++ = *s++;
9307 	}
9308     }
9309     else {
9310         parse_ident(&s, &d, e, 1, is_utf8);
9311     }
9312     *d = '\0';
9313     d = dest;
9314     if (*d) {
9315 	if (PL_lex_state != LEX_NORMAL)
9316 	    PL_lex_state = LEX_INTERPENDMAYBE;
9317 	return s;
9318     }
9319     if (*s == '$' && s[1] &&
9320       (isIDFIRST_lazy_if(s+1,is_utf8)
9321          || isDIGIT_A((U8)s[1])
9322          || s[1] == '$'
9323          || s[1] == '{'
9324          || strnEQ(s+1,"::",2)) )
9325     {
9326 	return s;
9327     }
9328     if (*s == '{') {
9329 	bracket = s;
9330 	s++;
9331 	while (s < send && SPACE_OR_TAB(*s))
9332 	   s++;
9333     }
9334 
9335 #define VALID_LEN_ONE_IDENT(d, u)     (isPUNCT_A((U8)*(d))     \
9336                                         || isCNTRL_A((U8)*(d)) \
9337                                         || isDIGIT_A((U8)*(d)) \
9338                                         || (!(u) && !UTF8_IS_INVARIANT((U8)*(d))))
9339     if (s < send
9340         && (isIDFIRST_lazy_if(s, is_utf8) || VALID_LEN_ONE_IDENT(s, is_utf8)))
9341     {
9342         if (is_utf8) {
9343             const STRLEN skip = UTF8SKIP(s);
9344             STRLEN i;
9345             d[skip] = '\0';
9346             for ( i = 0; i < skip; i++ )
9347                 d[i] = *s++;
9348         }
9349         else {
9350             *d = *s++;
9351             d[1] = '\0';
9352         }
9353     }
9354     if (*d == '^' && *s && isCONTROLVAR(*s)) {
9355 	*d = toCTRL(*s);
9356 	s++;
9357     }
9358     else if (ck_uni && !bracket)
9359 	check_uni();
9360     if (bracket) {
9361 	if (isIDFIRST_lazy_if(d,is_utf8)) {
9362         d += is_utf8 ? UTF8SKIP(d) : 1;
9363         parse_ident(&s, &d, e, 1, is_utf8);
9364 	    *d = '\0';
9365 	    while (s < send && SPACE_OR_TAB(*s))
9366 		s++;
9367 	    if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
9368 		if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
9369 		    const char * const brack =
9370 			(const char *)
9371 			((*s == '[') ? "[...]" : "{...}");
9372    /* diag_listed_as: Ambiguous use of %c{%s[...]} resolved to %c%s[...] */
9373 		    Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9374 			"Ambiguous use of %c{%s%s} resolved to %c%s%s",
9375 			funny, dest, brack, funny, dest, brack);
9376 		}
9377 		bracket++;
9378 		PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
9379 		PL_lex_allbrackets++;
9380 		return s;
9381 	    }
9382 	}
9383 	/* Handle extended ${^Foo} variables
9384 	 * 1999-02-27 mjd-perl-patch@plover.com */
9385 	else if (!isWORDCHAR(*d) && !isPRINT(*d) /* isCTRL(d) */
9386 		 && isWORDCHAR(*s))
9387 	{
9388 	    d++;
9389 	    while (isWORDCHAR(*s) && d < e) {
9390 		*d++ = *s++;
9391 	    }
9392 	    if (d >= e)
9393 		Perl_croak(aTHX_ "%s", ident_too_long);
9394 	    *d = '\0';
9395 	}
9396 
9397         while (s < send && SPACE_OR_TAB(*s))
9398 	    s++;
9399 
9400 	if (*s == '}') {
9401 	    s++;
9402 	    if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
9403 		PL_lex_state = LEX_INTERPEND;
9404 		PL_expect = XREF;
9405 	    }
9406 	    if (PL_lex_state == LEX_NORMAL) {
9407 		if (ckWARN(WARN_AMBIGUOUS) &&
9408 		    (keyword(dest, d - dest, 0)
9409 		     || get_cvn_flags(dest, d - dest, is_utf8 ? SVf_UTF8 : 0)))
9410 		{
9411                     SV *tmp = newSVpvn_flags( dest, d - dest,
9412                                             SVs_TEMP | (is_utf8 ? SVf_UTF8 : 0) );
9413 		    if (funny == '#')
9414 			funny = '@';
9415 		    Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
9416 			"Ambiguous use of %c{%"SVf"} resolved to %c%"SVf,
9417 			funny, tmp, funny, tmp);
9418 		}
9419 	    }
9420 	}
9421 	else {
9422 	    s = bracket;		/* let the parser handle it */
9423 	    *dest = '\0';
9424 	}
9425     }
9426     else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
9427 	PL_lex_state = LEX_INTERPEND;
9428     return s;
9429 }
9430 
9431 static bool
9432 S_pmflag(pTHX_ const char* const valid_flags, U32 * pmfl, char** s, char* charset) {
9433 
9434     /* Adds, subtracts to/from 'pmfl' based on regex modifier flags found in
9435      * the parse starting at 's', based on the subset that are valid in this
9436      * context input to this routine in 'valid_flags'. Advances s.  Returns
9437      * TRUE if the input should be treated as a valid flag, so the next char
9438      * may be as well; otherwise FALSE. 'charset' should point to a NUL upon
9439      * first call on the current regex.  This routine will set it to any
9440      * charset modifier found.  The caller shouldn't change it.  This way,
9441      * another charset modifier encountered in the parse can be detected as an
9442      * error, as we have decided to allow only one */
9443 
9444     const char c = **s;
9445     STRLEN charlen = UTF ? UTF8SKIP(*s) : 1;
9446 
9447     if ( charlen != 1 || ! strchr(valid_flags, c) ) {
9448         if (isWORDCHAR_lazy_if(*s, UTF)) {
9449             yyerror_pv(Perl_form(aTHX_ "Unknown regexp modifier \"/%.*s\"", (int)charlen, *s),
9450                        UTF ? SVf_UTF8 : 0);
9451             (*s) += charlen;
9452             /* Pretend that it worked, so will continue processing before
9453              * dieing */
9454             return TRUE;
9455         }
9456         return FALSE;
9457     }
9458 
9459     switch (c) {
9460 
9461         CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
9462         case GLOBAL_PAT_MOD:      *pmfl |= PMf_GLOBAL; break;
9463         case CONTINUE_PAT_MOD:    *pmfl |= PMf_CONTINUE; break;
9464         case ONCE_PAT_MOD:        *pmfl |= PMf_KEEP; break;
9465         case KEEPCOPY_PAT_MOD:    *pmfl |= RXf_PMf_KEEPCOPY; break;
9466         case NONDESTRUCT_PAT_MOD: *pmfl |= PMf_NONDESTRUCT; break;
9467 	case LOCALE_PAT_MOD:
9468 	    if (*charset) {
9469 		goto multiple_charsets;
9470 	    }
9471 	    set_regex_charset(pmfl, REGEX_LOCALE_CHARSET);
9472 	    *charset = c;
9473 	    break;
9474 	case UNICODE_PAT_MOD:
9475 	    if (*charset) {
9476 		goto multiple_charsets;
9477 	    }
9478 	    set_regex_charset(pmfl, REGEX_UNICODE_CHARSET);
9479 	    *charset = c;
9480 	    break;
9481 	case ASCII_RESTRICT_PAT_MOD:
9482 	    if (! *charset) {
9483 		set_regex_charset(pmfl, REGEX_ASCII_RESTRICTED_CHARSET);
9484 	    }
9485 	    else {
9486 
9487 		/* Error if previous modifier wasn't an 'a', but if it was, see
9488 		 * if, and accept, a second occurrence (only) */
9489 		if (*charset != 'a'
9490 		    || get_regex_charset(*pmfl)
9491 			!= REGEX_ASCII_RESTRICTED_CHARSET)
9492 		{
9493 			goto multiple_charsets;
9494 		}
9495 		set_regex_charset(pmfl, REGEX_ASCII_MORE_RESTRICTED_CHARSET);
9496 	    }
9497 	    *charset = c;
9498 	    break;
9499 	case DEPENDS_PAT_MOD:
9500 	    if (*charset) {
9501 		goto multiple_charsets;
9502 	    }
9503 	    set_regex_charset(pmfl, REGEX_DEPENDS_CHARSET);
9504 	    *charset = c;
9505 	    break;
9506     }
9507 
9508     (*s)++;
9509     return TRUE;
9510 
9511     multiple_charsets:
9512 	if (*charset != c) {
9513 	    yyerror(Perl_form(aTHX_ "Regexp modifiers \"/%c\" and \"/%c\" are mutually exclusive", *charset, c));
9514 	}
9515 	else if (c == 'a') {
9516 	    yyerror("Regexp modifier \"/a\" may appear a maximum of twice");
9517 	}
9518 	else {
9519 	    yyerror(Perl_form(aTHX_ "Regexp modifier \"/%c\" may not appear twice", c));
9520 	}
9521 
9522 	/* Pretend that it worked, so will continue processing before dieing */
9523 	(*s)++;
9524 	return TRUE;
9525 }
9526 
9527 STATIC char *
9528 S_scan_pat(pTHX_ char *start, I32 type)
9529 {
9530     dVAR;
9531     PMOP *pm;
9532     char *s;
9533     const char * const valid_flags =
9534 	(const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
9535     char charset = '\0';    /* character set modifier */
9536 #ifdef PERL_MAD
9537     char *modstart;
9538 #endif
9539 
9540     PERL_ARGS_ASSERT_SCAN_PAT;
9541 
9542     s = scan_str(start,!!PL_madskills,FALSE, (PL_in_eval & EVAL_RE_REPARSING),
9543                        TRUE /* look for escaped bracketed metas */ );
9544 
9545     if (!s) {
9546 	const char * const delimiter = skipspace(start);
9547 	Perl_croak(aTHX_
9548 		   (const char *)
9549 		   (*delimiter == '?'
9550 		    ? "Search pattern not terminated or ternary operator parsed as search pattern"
9551 		    : "Search pattern not terminated" ));
9552     }
9553 
9554     pm = (PMOP*)newPMOP(type, 0);
9555     if (PL_multi_open == '?') {
9556 	/* This is the only point in the code that sets PMf_ONCE:  */
9557 	pm->op_pmflags |= PMf_ONCE;
9558 
9559 	/* Hence it's safe to do this bit of PMOP book-keeping here, which
9560 	   allows us to restrict the list needed by reset to just the ??
9561 	   matches.  */
9562 	assert(type != OP_TRANS);
9563 	if (PL_curstash) {
9564 	    MAGIC *mg = mg_find((const SV *)PL_curstash, PERL_MAGIC_symtab);
9565 	    U32 elements;
9566 	    if (!mg) {
9567 		mg = sv_magicext(MUTABLE_SV(PL_curstash), 0, PERL_MAGIC_symtab, 0, 0,
9568 				 0);
9569 	    }
9570 	    elements = mg->mg_len / sizeof(PMOP**);
9571 	    Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
9572 	    ((PMOP**)mg->mg_ptr) [elements++] = pm;
9573 	    mg->mg_len = elements * sizeof(PMOP**);
9574 	    PmopSTASH_set(pm,PL_curstash);
9575 	}
9576     }
9577 #ifdef PERL_MAD
9578     modstart = s;
9579 #endif
9580 
9581     /* if qr/...(?{..}).../, then need to parse the pattern within a new
9582      * anon CV. False positives like qr/[(?{]/ are harmless */
9583 
9584     if (type == OP_QR) {
9585 	STRLEN len;
9586 	char *e, *p = SvPV(PL_lex_stuff, len);
9587 	e = p + len;
9588 	for (; p < e; p++) {
9589 	    if (p[0] == '(' && p[1] == '?'
9590 		&& (p[2] == '{' || (p[2] == '?' && p[3] == '{')))
9591 	    {
9592 		pm->op_pmflags |= PMf_HAS_CV;
9593 		break;
9594 	    }
9595 	}
9596 	pm->op_pmflags |= PMf_IS_QR;
9597     }
9598 
9599     while (*s && S_pmflag(aTHX_ valid_flags, &(pm->op_pmflags), &s, &charset)) {};
9600 #ifdef PERL_MAD
9601     if (PL_madskills && modstart != s) {
9602 	SV* tmptoken = newSVpvn(modstart, s - modstart);
9603 	append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
9604     }
9605 #endif
9606     /* issue a warning if /c is specified,but /g is not */
9607     if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
9608     {
9609         Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP),
9610 		       "Use of /c modifier is meaningless without /g" );
9611     }
9612 
9613     PL_lex_op = (OP*)pm;
9614     pl_yylval.ival = OP_MATCH;
9615     return s;
9616 }
9617 
9618 STATIC char *
9619 S_scan_subst(pTHX_ char *start)
9620 {
9621     dVAR;
9622     char *s;
9623     PMOP *pm;
9624     I32 first_start;
9625     I32 es = 0;
9626     char charset = '\0';    /* character set modifier */
9627 #ifdef PERL_MAD
9628     char *modstart;
9629 #endif
9630 
9631     PERL_ARGS_ASSERT_SCAN_SUBST;
9632 
9633     pl_yylval.ival = OP_NULL;
9634 
9635     s = scan_str(start,!!PL_madskills,FALSE,FALSE,
9636                  TRUE /* look for escaped bracketed metas */ );
9637 
9638     if (!s)
9639 	Perl_croak(aTHX_ "Substitution pattern not terminated");
9640 
9641     if (s[-1] == PL_multi_open)
9642 	s--;
9643 #ifdef PERL_MAD
9644     if (PL_madskills) {
9645 	CURMAD('q', PL_thisopen);
9646 	CURMAD('_', PL_thiswhite);
9647 	CURMAD('E', PL_thisstuff);
9648 	CURMAD('Q', PL_thisclose);
9649 	PL_realtokenstart = s - SvPVX(PL_linestr);
9650     }
9651 #endif
9652 
9653     first_start = PL_multi_start;
9654     s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
9655     if (!s) {
9656 	if (PL_lex_stuff) {
9657 	    SvREFCNT_dec(PL_lex_stuff);
9658 	    PL_lex_stuff = NULL;
9659 	}
9660 	Perl_croak(aTHX_ "Substitution replacement not terminated");
9661     }
9662     PL_multi_start = first_start;	/* so whole substitution is taken together */
9663 
9664     pm = (PMOP*)newPMOP(OP_SUBST, 0);
9665 
9666 #ifdef PERL_MAD
9667     if (PL_madskills) {
9668 	CURMAD('z', PL_thisopen);
9669 	CURMAD('R', PL_thisstuff);
9670 	CURMAD('Z', PL_thisclose);
9671     }
9672     modstart = s;
9673 #endif
9674 
9675     while (*s) {
9676 	if (*s == EXEC_PAT_MOD) {
9677 	    s++;
9678 	    es++;
9679 	}
9680 	else if (! S_pmflag(aTHX_ S_PAT_MODS, &(pm->op_pmflags), &s, &charset))
9681 	{
9682 	    break;
9683 	}
9684     }
9685 
9686 #ifdef PERL_MAD
9687     if (PL_madskills) {
9688 	if (modstart != s)
9689 	    curmad('m', newSVpvn(modstart, s - modstart));
9690 	append_madprops(PL_thismad, (OP*)pm, 0);
9691 	PL_thismad = 0;
9692     }
9693 #endif
9694     if ((pm->op_pmflags & PMf_CONTINUE)) {
9695         Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
9696     }
9697 
9698     if (es) {
9699 	SV * const repl = newSVpvs("");
9700 
9701 	PL_multi_end = 0;
9702 	pm->op_pmflags |= PMf_EVAL;
9703 	while (es-- > 0) {
9704 	    if (es)
9705 		sv_catpvs(repl, "eval ");
9706 	    else
9707 		sv_catpvs(repl, "do ");
9708 	}
9709 	sv_catpvs(repl, "{");
9710 	sv_catsv(repl, PL_sublex_info.repl);
9711 	sv_catpvs(repl, "}");
9712 	SvEVALED_on(repl);
9713 	SvREFCNT_dec(PL_sublex_info.repl);
9714 	PL_sublex_info.repl = repl;
9715     }
9716 
9717     PL_lex_op = (OP*)pm;
9718     pl_yylval.ival = OP_SUBST;
9719     return s;
9720 }
9721 
9722 STATIC char *
9723 S_scan_trans(pTHX_ char *start)
9724 {
9725     dVAR;
9726     char* s;
9727     OP *o;
9728     U8 squash;
9729     U8 del;
9730     U8 complement;
9731     bool nondestruct = 0;
9732 #ifdef PERL_MAD
9733     char *modstart;
9734 #endif
9735 
9736     PERL_ARGS_ASSERT_SCAN_TRANS;
9737 
9738     pl_yylval.ival = OP_NULL;
9739 
9740     s = scan_str(start,!!PL_madskills,FALSE,FALSE, FALSE);
9741     if (!s)
9742 	Perl_croak(aTHX_ "Transliteration pattern not terminated");
9743 
9744     if (s[-1] == PL_multi_open)
9745 	s--;
9746 #ifdef PERL_MAD
9747     if (PL_madskills) {
9748 	CURMAD('q', PL_thisopen);
9749 	CURMAD('_', PL_thiswhite);
9750 	CURMAD('E', PL_thisstuff);
9751 	CURMAD('Q', PL_thisclose);
9752 	PL_realtokenstart = s - SvPVX(PL_linestr);
9753     }
9754 #endif
9755 
9756     s = scan_str(s,!!PL_madskills,FALSE,FALSE, FALSE);
9757     if (!s) {
9758 	if (PL_lex_stuff) {
9759 	    SvREFCNT_dec(PL_lex_stuff);
9760 	    PL_lex_stuff = NULL;
9761 	}
9762 	Perl_croak(aTHX_ "Transliteration replacement not terminated");
9763     }
9764     if (PL_madskills) {
9765 	CURMAD('z', PL_thisopen);
9766 	CURMAD('R', PL_thisstuff);
9767 	CURMAD('Z', PL_thisclose);
9768     }
9769 
9770     complement = del = squash = 0;
9771 #ifdef PERL_MAD
9772     modstart = s;
9773 #endif
9774     while (1) {
9775 	switch (*s) {
9776 	case 'c':
9777 	    complement = OPpTRANS_COMPLEMENT;
9778 	    break;
9779 	case 'd':
9780 	    del = OPpTRANS_DELETE;
9781 	    break;
9782 	case 's':
9783 	    squash = OPpTRANS_SQUASH;
9784 	    break;
9785 	case 'r':
9786 	    nondestruct = 1;
9787 	    break;
9788 	default:
9789 	    goto no_more;
9790 	}
9791 	s++;
9792     }
9793   no_more:
9794 
9795     o = newPVOP(nondestruct ? OP_TRANSR : OP_TRANS, 0, (char*)NULL);
9796     o->op_private &= ~OPpTRANS_ALL;
9797     o->op_private |= del|squash|complement|
9798       (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
9799       (DO_UTF8(PL_sublex_info.repl) ? OPpTRANS_TO_UTF   : 0);
9800 
9801     PL_lex_op = o;
9802     pl_yylval.ival = nondestruct ? OP_TRANSR : OP_TRANS;
9803 
9804 #ifdef PERL_MAD
9805     if (PL_madskills) {
9806 	if (modstart != s)
9807 	    curmad('m', newSVpvn(modstart, s - modstart));
9808 	append_madprops(PL_thismad, o, 0);
9809 	PL_thismad = 0;
9810     }
9811 #endif
9812 
9813     return s;
9814 }
9815 
9816 /* scan_heredoc
9817    Takes a pointer to the first < in <<FOO.
9818    Returns a pointer to the byte following <<FOO.
9819 
9820    This function scans a heredoc, which involves different methods
9821    depending on whether we are in a string eval, quoted construct, etc.
9822    This is because PL_linestr could containing a single line of input, or
9823    a whole string being evalled, or the contents of the current quote-
9824    like operator.
9825 
9826    The two basic methods are:
9827     - Steal lines from the input stream
9828     - Scan the heredoc in PL_linestr and remove it therefrom
9829 
9830    In a file scope or filtered eval, the first method is used; in a
9831    string eval, the second.
9832 
9833    In a quote-like operator, we have to choose between the two,
9834    depending on where we can find a newline.  We peek into outer lex-
9835    ing scopes until we find one with a newline in it.  If we reach the
9836    outermost lexing scope and it is a file, we use the stream method.
9837    Otherwise it is treated as an eval.
9838 */
9839 
9840 STATIC char *
9841 S_scan_heredoc(pTHX_ char *s)
9842 {
9843     dVAR;
9844     I32 op_type = OP_SCALAR;
9845     I32 len;
9846     SV *tmpstr;
9847     char term;
9848     char *d;
9849     char *e;
9850     char *peek;
9851     const bool infile = PL_rsfp || PL_parser->filtered;
9852     LEXSHARED *shared = PL_parser->lex_shared;
9853 #ifdef PERL_MAD
9854     I32 stuffstart = s - SvPVX(PL_linestr);
9855     char *tstart;
9856 
9857     PL_realtokenstart = -1;
9858 #endif
9859 
9860     PERL_ARGS_ASSERT_SCAN_HEREDOC;
9861 
9862     s += 2;
9863     d = PL_tokenbuf + 1;
9864     e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
9865     *PL_tokenbuf = '\n';
9866     peek = s;
9867     while (SPACE_OR_TAB(*peek))
9868 	peek++;
9869     if (*peek == '`' || *peek == '\'' || *peek =='"') {
9870 	s = peek;
9871 	term = *s++;
9872 	s = delimcpy(d, e, s, PL_bufend, term, &len);
9873 	if (s == PL_bufend)
9874 	    Perl_croak(aTHX_ "Unterminated delimiter for here document");
9875 	d += len;
9876 	s++;
9877     }
9878     else {
9879 	if (*s == '\\')
9880             /* <<\FOO is equivalent to <<'FOO' */
9881 	    s++, term = '\'';
9882 	else
9883 	    term = '"';
9884 	if (!isWORDCHAR_lazy_if(s,UTF))
9885 	    deprecate("bare << to mean <<\"\"");
9886 	for (; isWORDCHAR_lazy_if(s,UTF); s++) {
9887 	    if (d < e)
9888 		*d++ = *s;
9889 	}
9890     }
9891     if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
9892 	Perl_croak(aTHX_ "Delimiter for here document is too long");
9893     *d++ = '\n';
9894     *d = '\0';
9895     len = d - PL_tokenbuf;
9896 
9897 #ifdef PERL_MAD
9898     if (PL_madskills) {
9899 	tstart = PL_tokenbuf + 1;
9900 	PL_thisclose = newSVpvn(tstart, len - 1);
9901 	tstart = SvPVX(PL_linestr) + stuffstart;
9902 	PL_thisopen = newSVpvn(tstart, s - tstart);
9903 	stuffstart = s - SvPVX(PL_linestr);
9904     }
9905 #endif
9906 #ifndef PERL_STRICT_CR
9907     d = strchr(s, '\r');
9908     if (d) {
9909 	char * const olds = s;
9910 	s = d;
9911 	while (s < PL_bufend) {
9912 	    if (*s == '\r') {
9913 		*d++ = '\n';
9914 		if (*++s == '\n')
9915 		    s++;
9916 	    }
9917 	    else if (*s == '\n' && s[1] == '\r') {	/* \015\013 on a mac? */
9918 		*d++ = *s++;
9919 		s++;
9920 	    }
9921 	    else
9922 		*d++ = *s++;
9923 	}
9924 	*d = '\0';
9925 	PL_bufend = d;
9926 	SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
9927 	s = olds;
9928     }
9929 #endif
9930 #ifdef PERL_MAD
9931     if (PL_madskills) {
9932 	tstart = SvPVX(PL_linestr) + stuffstart;
9933 	if (PL_thisstuff)
9934 	    sv_catpvn(PL_thisstuff, tstart, s - tstart);
9935 	else
9936 	    PL_thisstuff = newSVpvn(tstart, s - tstart);
9937     }
9938 
9939     stuffstart = s - SvPVX(PL_linestr);
9940 #endif
9941 
9942     tmpstr = newSV_type(SVt_PVIV);
9943     SvGROW(tmpstr, 80);
9944     if (term == '\'') {
9945 	op_type = OP_CONST;
9946 	SvIV_set(tmpstr, -1);
9947     }
9948     else if (term == '`') {
9949 	op_type = OP_BACKTICK;
9950 	SvIV_set(tmpstr, '\\');
9951     }
9952 
9953     PL_multi_start = CopLINE(PL_curcop) + 1;
9954     PL_multi_open = PL_multi_close = '<';
9955     /* inside a string eval or quote-like operator */
9956     if (!infile || PL_lex_inwhat) {
9957 	SV *linestr;
9958 	char *bufend;
9959 	char * const olds = s;
9960 	PERL_CONTEXT * const cx = &cxstack[cxstack_ix];
9961 	/* These two fields are not set until an inner lexing scope is
9962 	   entered.  But we need them set here. */
9963 	shared->ls_bufptr  = s;
9964 	shared->ls_linestr = PL_linestr;
9965 	if (PL_lex_inwhat)
9966 	  /* Look for a newline.  If the current buffer does not have one,
9967 	     peek into the line buffer of the parent lexing scope, going
9968  	     up as many levels as necessary to find one with a newline
9969 	     after bufptr.
9970 	   */
9971 	  while (!(s = (char *)memchr(
9972 		    (void *)shared->ls_bufptr, '\n',
9973 		    SvEND(shared->ls_linestr)-shared->ls_bufptr
9974 		))) {
9975 	    shared = shared->ls_prev;
9976 	    /* shared is only null if we have gone beyond the outermost
9977 	       lexing scope.  In a file, we will have broken out of the
9978 	       loop in the previous iteration.  In an eval, the string buf-
9979 	       fer ends with "\n;", so the while condition below will have
9980 	       evaluated to false.  So shared can never be null. */
9981 	    assert(shared);
9982 	    /* A LEXSHARED struct with a null ls_prev pointer is the outer-
9983 	       most lexing scope.  In a file, shared->ls_linestr at that
9984 	       level is just one line, so there is no body to steal. */
9985 	    if (infile && !shared->ls_prev) {
9986 		s = olds;
9987 		goto streaming;
9988 	    }
9989 	  }
9990 	else {	/* eval */
9991 	    s = (char*)memchr((void*)s, '\n', PL_bufend - s);
9992 	    assert(s);
9993 	}
9994 	linestr = shared->ls_linestr;
9995 	bufend = SvEND(linestr);
9996 	d = s;
9997 	while (s < bufend - len + 1 &&
9998           memNE(s,PL_tokenbuf,len) ) {
9999 	    if (*s++ == '\n')
10000 		++shared->herelines;
10001 	}
10002 	if (s >= bufend - len + 1) {
10003 	    goto interminable;
10004 	}
10005 	sv_setpvn(tmpstr,d+1,s-d);
10006 #ifdef PERL_MAD
10007 	if (PL_madskills) {
10008 	    if (PL_thisstuff)
10009 		sv_catpvn(PL_thisstuff, d + 1, s - d);
10010 	    else
10011 		PL_thisstuff = newSVpvn(d + 1, s - d);
10012 	    stuffstart = s - SvPVX(PL_linestr);
10013 	}
10014 #endif
10015 	s += len - 1;
10016 	/* the preceding stmt passes a newline */
10017 	shared->herelines++;
10018 
10019 	/* s now points to the newline after the heredoc terminator.
10020 	   d points to the newline before the body of the heredoc.
10021 	 */
10022 
10023 	/* We are going to modify linestr in place here, so set
10024 	   aside copies of the string if necessary for re-evals or
10025 	   (caller $n)[6]. */
10026 	/* See the Paranoia note in case LEX_INTERPEND in yylex, for why we
10027 	   check shared->re_eval_str. */
10028 	if (shared->re_eval_start || shared->re_eval_str) {
10029 	    /* Set aside the rest of the regexp */
10030 	    if (!shared->re_eval_str)
10031 		shared->re_eval_str =
10032 		       newSVpvn(shared->re_eval_start,
10033 				bufend - shared->re_eval_start);
10034 	    shared->re_eval_start -= s-d;
10035 	}
10036 	if (cxstack_ix >= 0 && CxTYPE(cx) == CXt_EVAL &&
10037             CxOLD_OP_TYPE(cx) == OP_ENTEREVAL &&
10038             cx->blk_eval.cur_text == linestr)
10039         {
10040 	    cx->blk_eval.cur_text = newSVsv(linestr);
10041 	    SvSCREAM_on(cx->blk_eval.cur_text);
10042 	}
10043 	/* Copy everything from s onwards back to d. */
10044 	Move(s,d,bufend-s + 1,char);
10045 	SvCUR_set(linestr, SvCUR(linestr) - (s-d));
10046 	/* Setting PL_bufend only applies when we have not dug deeper
10047 	   into other scopes, because sublex_done sets PL_bufend to
10048 	   SvEND(PL_linestr). */
10049 	if (shared == PL_parser->lex_shared) PL_bufend = SvEND(linestr);
10050 	s = olds;
10051     }
10052     else
10053     {
10054       SV *linestr_save;
10055      streaming:
10056       sv_setpvs(tmpstr,"");   /* avoid "uninitialized" warning */
10057       term = PL_tokenbuf[1];
10058       len--;
10059       linestr_save = PL_linestr; /* must restore this afterwards */
10060       d = s;			 /* and this */
10061       PL_linestr = newSVpvs("");
10062       PL_bufend = SvPVX(PL_linestr);
10063       while (1) {
10064 #ifdef PERL_MAD
10065 	if (PL_madskills) {
10066 	    tstart = SvPVX(PL_linestr) + stuffstart;
10067 	    if (PL_thisstuff)
10068 		sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
10069 	    else
10070 		PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
10071 	}
10072 #endif
10073 	PL_bufptr = PL_bufend;
10074 	CopLINE_set(PL_curcop,
10075 		    PL_multi_start + shared->herelines);
10076 	if (!lex_next_chunk(LEX_NO_TERM)
10077 	 && (!SvCUR(tmpstr) || SvEND(tmpstr)[-1] != '\n')) {
10078 	    SvREFCNT_dec(linestr_save);
10079 	    goto interminable;
10080 	}
10081 	CopLINE_set(PL_curcop, (line_t)PL_multi_start - 1);
10082 	if (!SvCUR(PL_linestr) || PL_bufend[-1] != '\n') {
10083 	    lex_grow_linestr(SvCUR(PL_linestr) + 2);
10084 	    sv_catpvs(PL_linestr, "\n\0");
10085 	}
10086 	s = PL_bufptr;
10087 #ifdef PERL_MAD
10088 	stuffstart = s - SvPVX(PL_linestr);
10089 #endif
10090 	shared->herelines++;
10091 	PL_last_lop = PL_last_uni = NULL;
10092 #ifndef PERL_STRICT_CR
10093 	if (PL_bufend - PL_linestart >= 2) {
10094 	    if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
10095 		(PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
10096 	    {
10097 		PL_bufend[-2] = '\n';
10098 		PL_bufend--;
10099 		SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
10100 	    }
10101 	    else if (PL_bufend[-1] == '\r')
10102 		PL_bufend[-1] = '\n';
10103 	}
10104 	else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
10105 	    PL_bufend[-1] = '\n';
10106 #endif
10107 	if (*s == term && memEQ(s,PL_tokenbuf + 1,len)) {
10108 	    SvREFCNT_dec(PL_linestr);
10109 	    PL_linestr = linestr_save;
10110 	    PL_linestart = SvPVX(linestr_save);
10111 	    PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
10112 	    s = d;
10113 	    break;
10114 	}
10115 	else {
10116 	    sv_catsv(tmpstr,PL_linestr);
10117 	}
10118       }
10119     }
10120     PL_multi_end = CopLINE(PL_curcop);
10121     if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
10122 	SvPV_shrink_to_cur(tmpstr);
10123     }
10124     if (!IN_BYTES) {
10125 	if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
10126 	    SvUTF8_on(tmpstr);
10127 	else if (PL_encoding)
10128 	    sv_recode_to_utf8(tmpstr, PL_encoding);
10129     }
10130     PL_lex_stuff = tmpstr;
10131     pl_yylval.ival = op_type;
10132     return s;
10133 
10134   interminable:
10135     SvREFCNT_dec(tmpstr);
10136     CopLINE_set(PL_curcop, (line_t)PL_multi_start - 1);
10137     missingterm(PL_tokenbuf + 1);
10138 }
10139 
10140 /* scan_inputsymbol
10141    takes: current position in input buffer
10142    returns: new position in input buffer
10143    side-effects: pl_yylval and lex_op are set.
10144 
10145    This code handles:
10146 
10147    <>		read from ARGV
10148    <FH> 	read from filehandle
10149    <pkg::FH>	read from package qualified filehandle
10150    <pkg'FH>	read from package qualified filehandle
10151    <$fh>	read from filehandle in $fh
10152    <*.h>	filename glob
10153 
10154 */
10155 
10156 STATIC char *
10157 S_scan_inputsymbol(pTHX_ char *start)
10158 {
10159     dVAR;
10160     char *s = start;		/* current position in buffer */
10161     char *end;
10162     I32 len;
10163     char *d = PL_tokenbuf;					/* start of temp holding space */
10164     const char * const e = PL_tokenbuf + sizeof PL_tokenbuf;	/* end of temp holding space */
10165 
10166     PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
10167 
10168     end = strchr(s, '\n');
10169     if (!end)
10170 	end = PL_bufend;
10171     s = delimcpy(d, e, s + 1, end, '>', &len);	/* extract until > */
10172 
10173     /* die if we didn't have space for the contents of the <>,
10174        or if it didn't end, or if we see a newline
10175     */
10176 
10177     if (len >= (I32)sizeof PL_tokenbuf)
10178 	Perl_croak(aTHX_ "Excessively long <> operator");
10179     if (s >= end)
10180 	Perl_croak(aTHX_ "Unterminated <> operator");
10181 
10182     s++;
10183 
10184     /* check for <$fh>
10185        Remember, only scalar variables are interpreted as filehandles by
10186        this code.  Anything more complex (e.g., <$fh{$num}>) will be
10187        treated as a glob() call.
10188        This code makes use of the fact that except for the $ at the front,
10189        a scalar variable and a filehandle look the same.
10190     */
10191     if (*d == '$' && d[1]) d++;
10192 
10193     /* allow <Pkg'VALUE> or <Pkg::VALUE> */
10194     while (*d && (isWORDCHAR_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
10195 	d += UTF ? UTF8SKIP(d) : 1;
10196 
10197     /* If we've tried to read what we allow filehandles to look like, and
10198        there's still text left, then it must be a glob() and not a getline.
10199        Use scan_str to pull out the stuff between the <> and treat it
10200        as nothing more than a string.
10201     */
10202 
10203     if (d - PL_tokenbuf != len) {
10204 	pl_yylval.ival = OP_GLOB;
10205 	s = scan_str(start,!!PL_madskills,FALSE,FALSE, FALSE);
10206 	if (!s)
10207 	   Perl_croak(aTHX_ "Glob not terminated");
10208 	return s;
10209     }
10210     else {
10211 	bool readline_overriden = FALSE;
10212 	GV *gv_readline;
10213 	GV **gvp;
10214     	/* we're in a filehandle read situation */
10215 	d = PL_tokenbuf;
10216 
10217 	/* turn <> into <ARGV> */
10218 	if (!len)
10219 	    Copy("ARGV",d,5,char);
10220 
10221 	/* Check whether readline() is overriden */
10222 	gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
10223 	if ((gv_readline
10224 		&& GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
10225 		||
10226 		((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
10227 		 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
10228 		&& GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
10229 	    readline_overriden = TRUE;
10230 
10231 	/* if <$fh>, create the ops to turn the variable into a
10232 	   filehandle
10233 	*/
10234 	if (*d == '$') {
10235 	    /* try to find it in the pad for this block, otherwise find
10236 	       add symbol table ops
10237 	    */
10238 	    const PADOFFSET tmp = pad_findmy_pvn(d, len, UTF ? SVf_UTF8 : 0);
10239 	    if (tmp != NOT_IN_PAD) {
10240 		if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
10241 		    HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
10242 		    HEK * const stashname = HvNAME_HEK(stash);
10243 		    SV * const sym = sv_2mortal(newSVhek(stashname));
10244 		    sv_catpvs(sym, "::");
10245 		    sv_catpv(sym, d+1);
10246 		    d = SvPVX(sym);
10247 		    goto intro_sym;
10248 		}
10249 		else {
10250 		    OP * const o = newOP(OP_PADSV, 0);
10251 		    o->op_targ = tmp;
10252 		    PL_lex_op = readline_overriden
10253 			? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10254 				op_append_elem(OP_LIST, o,
10255 				    newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
10256 			: (OP*)newUNOP(OP_READLINE, 0, o);
10257 		}
10258 	    }
10259 	    else {
10260 		GV *gv;
10261 		++d;
10262 intro_sym:
10263 		gv = gv_fetchpv(d,
10264 				(PL_in_eval
10265 				 ? (GV_ADDMULTI | GV_ADDINEVAL)
10266 				 : GV_ADDMULTI) | ( UTF ? SVf_UTF8 : 0 ),
10267 				SVt_PV);
10268 		PL_lex_op = readline_overriden
10269 		    ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10270 			    op_append_elem(OP_LIST,
10271 				newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
10272 				newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10273 		    : (OP*)newUNOP(OP_READLINE, 0,
10274 			    newUNOP(OP_RV2SV, 0,
10275 				newGVOP(OP_GV, 0, gv)));
10276 	    }
10277 	    if (!readline_overriden)
10278 		PL_lex_op->op_flags |= OPf_SPECIAL;
10279 	    /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
10280 	    pl_yylval.ival = OP_NULL;
10281 	}
10282 
10283 	/* If it's none of the above, it must be a literal filehandle
10284 	   (<Foo::BAR> or <FOO>) so build a simple readline OP */
10285 	else {
10286 	    GV * const gv = gv_fetchpv(d, GV_ADD | ( UTF ? SVf_UTF8 : 0 ), SVt_PVIO);
10287 	    PL_lex_op = readline_overriden
10288 		? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
10289 			op_append_elem(OP_LIST,
10290 			    newGVOP(OP_GV, 0, gv),
10291 			    newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
10292 		: (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
10293 	    pl_yylval.ival = OP_NULL;
10294 	}
10295     }
10296 
10297     return s;
10298 }
10299 
10300 
10301 /* scan_str
10302    takes: start position in buffer
10303 	  keep_quoted preserve \ on the embedded delimiter(s)
10304 	  keep_delims preserve the delimiters around the string
10305 	  re_reparse  compiling a run-time /(?{})/:
10306 			collapse // to /,  and skip encoding src
10307    returns: position to continue reading from buffer
10308    side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
10309    	updates the read buffer.
10310 
10311    This subroutine pulls a string out of the input.  It is called for:
10312    	q		single quotes		q(literal text)
10313 	'		single quotes		'literal text'
10314 	qq		double quotes		qq(interpolate $here please)
10315 	"		double quotes		"interpolate $here please"
10316 	qx		backticks		qx(/bin/ls -l)
10317 	`		backticks		`/bin/ls -l`
10318 	qw		quote words		@EXPORT_OK = qw( func() $spam )
10319 	m//		regexp match		m/this/
10320 	s///		regexp substitute	s/this/that/
10321 	tr///		string transliterate	tr/this/that/
10322 	y///		string transliterate	y/this/that/
10323 	($*@)		sub prototypes		sub foo ($)
10324 	(stuff)		sub attr parameters	sub foo : attr(stuff)
10325 	<>		readline or globs	<FOO>, <>, <$fh>, or <*.c>
10326 
10327    In most of these cases (all but <>, patterns and transliterate)
10328    yylex() calls scan_str().  m// makes yylex() call scan_pat() which
10329    calls scan_str().  s/// makes yylex() call scan_subst() which calls
10330    scan_str().  tr/// and y/// make yylex() call scan_trans() which
10331    calls scan_str().
10332 
10333    It skips whitespace before the string starts, and treats the first
10334    character as the delimiter.  If the delimiter is one of ([{< then
10335    the corresponding "close" character )]}> is used as the closing
10336    delimiter.  It allows quoting of delimiters, and if the string has
10337    balanced delimiters ([{<>}]) it allows nesting.
10338 
10339    On success, the SV with the resulting string is put into lex_stuff or,
10340    if that is already non-NULL, into lex_repl. The second case occurs only
10341    when parsing the RHS of the special constructs s/// and tr/// (y///).
10342    For convenience, the terminating delimiter character is stuffed into
10343    SvIVX of the SV.
10344 */
10345 
10346 STATIC char *
10347 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims, int re_reparse,
10348         bool deprecate_escaped_meta /* Should we issue a deprecation warning
10349                                        for certain paired metacharacters that
10350                                        appear escaped within it */
10351     )
10352 {
10353     dVAR;
10354     SV *sv;			/* scalar value: string */
10355     const char *tmps;		/* temp string, used for delimiter matching */
10356     char *s = start;		/* current position in the buffer */
10357     char term;			/* terminating character */
10358     char *to;			/* current position in the sv's data */
10359     I32 brackets = 1;		/* bracket nesting level */
10360     bool has_utf8 = FALSE;	/* is there any utf8 content? */
10361     I32 termcode;		/* terminating char. code */
10362     U8 termstr[UTF8_MAXBYTES];	/* terminating string */
10363     STRLEN termlen;		/* length of terminating string */
10364     int last_off = 0;		/* last position for nesting bracket */
10365     char *escaped_open = NULL;
10366 #ifdef PERL_MAD
10367     int stuffstart;
10368     char *tstart;
10369 #endif
10370 
10371     PERL_ARGS_ASSERT_SCAN_STR;
10372 
10373     /* skip space before the delimiter */
10374     if (isSPACE(*s)) {
10375 	s = PEEKSPACE(s);
10376     }
10377 
10378 #ifdef PERL_MAD
10379     if (PL_realtokenstart >= 0) {
10380 	stuffstart = PL_realtokenstart;
10381 	PL_realtokenstart = -1;
10382     }
10383     else
10384 	stuffstart = start - SvPVX(PL_linestr);
10385 #endif
10386     /* mark where we are, in case we need to report errors */
10387     CLINE;
10388 
10389     /* after skipping whitespace, the next character is the terminator */
10390     term = *s;
10391     if (!UTF) {
10392 	termcode = termstr[0] = term;
10393 	termlen = 1;
10394     }
10395     else {
10396 	termcode = utf8_to_uvchr_buf((U8*)s, (U8*)PL_bufend, &termlen);
10397 	Copy(s, termstr, termlen, U8);
10398 	if (!UTF8_IS_INVARIANT(term))
10399 	    has_utf8 = TRUE;
10400     }
10401 
10402     /* mark where we are */
10403     PL_multi_start = CopLINE(PL_curcop);
10404     PL_multi_open = term;
10405 
10406     /* find corresponding closing delimiter */
10407     if (term && (tmps = strchr("([{< )]}> )]}>",term)))
10408 	termcode = termstr[0] = term = tmps[5];
10409 
10410     PL_multi_close = term;
10411 
10412     /* A warning is raised if the input parameter requires it for escaped (by a
10413      * backslash) paired metacharacters {} [] and () when the delimiters are
10414      * those same characters, and the backslash is ineffective.  This doesn't
10415      * happen for <>, as they aren't metas. */
10416     if (deprecate_escaped_meta
10417         && (PL_multi_open == PL_multi_close
10418             || ! ckWARN_d(WARN_DEPRECATED)
10419             || PL_multi_open == '<'))
10420     {
10421         deprecate_escaped_meta = FALSE;
10422     }
10423 
10424     /* create a new SV to hold the contents.  79 is the SV's initial length.
10425        What a random number. */
10426     sv = newSV_type(SVt_PVIV);
10427     SvGROW(sv, 80);
10428     SvIV_set(sv, termcode);
10429     (void)SvPOK_only(sv);		/* validate pointer */
10430 
10431     /* move past delimiter and try to read a complete string */
10432     if (keep_delims)
10433 	sv_catpvn(sv, s, termlen);
10434     s += termlen;
10435 #ifdef PERL_MAD
10436     tstart = SvPVX(PL_linestr) + stuffstart;
10437     if (PL_madskills && !PL_thisopen && !keep_delims) {
10438 	PL_thisopen = newSVpvn(tstart, s - tstart);
10439 	stuffstart = s - SvPVX(PL_linestr);
10440     }
10441 #endif
10442     for (;;) {
10443 	if (PL_encoding && !UTF && !re_reparse) {
10444 	    bool cont = TRUE;
10445 
10446 	    while (cont) {
10447 		int offset = s - SvPVX_const(PL_linestr);
10448 		const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
10449 					   &offset, (char*)termstr, termlen);
10450 		const char * const ns = SvPVX_const(PL_linestr) + offset;
10451 		char * const svlast = SvEND(sv) - 1;
10452 
10453 		for (; s < ns; s++) {
10454 		    if (*s == '\n' && !PL_rsfp && !PL_parser->filtered)
10455 			COPLINE_INC_WITH_HERELINES;
10456 		}
10457 		if (!found)
10458 		    goto read_more_line;
10459 		else {
10460 		    /* handle quoted delimiters */
10461 		    if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
10462 			const char *t;
10463 			for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
10464 			    t--;
10465 			if ((svlast-1 - t) % 2) {
10466 			    if (!keep_quoted) {
10467 				*(svlast-1) = term;
10468 				*svlast = '\0';
10469 				SvCUR_set(sv, SvCUR(sv) - 1);
10470 			    }
10471 			    continue;
10472 			}
10473 		    }
10474 		    if (PL_multi_open == PL_multi_close) {
10475 			cont = FALSE;
10476 		    }
10477 		    else {
10478 			const char *t;
10479 			char *w;
10480 			for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
10481 			    /* At here, all closes are "was quoted" one,
10482 			       so we don't check PL_multi_close. */
10483 			    if (*t == '\\') {
10484 				if (!keep_quoted && *(t+1) == PL_multi_open)
10485 				    t++;
10486 				else
10487 				    *w++ = *t++;
10488 			    }
10489 			    else if (*t == PL_multi_open)
10490 				brackets++;
10491 
10492 			    *w = *t;
10493 			}
10494 			if (w < t) {
10495 			    *w++ = term;
10496 			    *w = '\0';
10497 			    SvCUR_set(sv, w - SvPVX_const(sv));
10498 			}
10499 			last_off = w - SvPVX(sv);
10500 			if (--brackets <= 0)
10501 			    cont = FALSE;
10502 		    }
10503 		}
10504 	    }
10505 	    if (!keep_delims) {
10506 		SvCUR_set(sv, SvCUR(sv) - 1);
10507 		*SvEND(sv) = '\0';
10508 	    }
10509 	    break;
10510 	}
10511 
10512     	/* extend sv if need be */
10513 	SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
10514 	/* set 'to' to the next character in the sv's string */
10515 	to = SvPVX(sv)+SvCUR(sv);
10516 
10517 	/* if open delimiter is the close delimiter read unbridle */
10518 	if (PL_multi_open == PL_multi_close) {
10519 	    for (; s < PL_bufend; s++,to++) {
10520 	    	/* embedded newlines increment the current line number */
10521 		if (*s == '\n' && !PL_rsfp && !PL_parser->filtered)
10522 		    COPLINE_INC_WITH_HERELINES;
10523 		/* handle quoted delimiters */
10524 		if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
10525 		    if (!keep_quoted
10526 		        && (s[1] == term
10527 			    || (re_reparse && s[1] == '\\'))
10528 		    )
10529 			s++;
10530 		    /* any other quotes are simply copied straight through */
10531 		    else
10532 			*to++ = *s++;
10533 		}
10534 		/* terminate when run out of buffer (the for() condition), or
10535 		   have found the terminator */
10536 		else if (*s == term) {
10537 		    if (termlen == 1)
10538 			break;
10539 		    if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
10540 			break;
10541 		}
10542 		else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10543 		    has_utf8 = TRUE;
10544 		*to = *s;
10545 	    }
10546 	}
10547 
10548 	/* if the terminator isn't the same as the start character (e.g.,
10549 	   matched brackets), we have to allow more in the quoting, and
10550 	   be prepared for nested brackets.
10551 	*/
10552 	else {
10553 	    /* read until we run out of string, or we find the terminator */
10554 	    for (; s < PL_bufend; s++,to++) {
10555 	    	/* embedded newlines increment the line count */
10556 		if (*s == '\n' && !PL_rsfp && !PL_parser->filtered)
10557 		    COPLINE_INC_WITH_HERELINES;
10558 		/* backslashes can escape the open or closing characters */
10559 		if (*s == '\\' && s+1 < PL_bufend) {
10560 		    if (!keep_quoted &&
10561 			((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
10562                     {
10563 			s++;
10564 
10565                         /* Here, 'deprecate_escaped_meta' is true iff the
10566                          * delimiters are paired metacharacters, and 's' points
10567                          * to an occurrence of one of them within the string,
10568                          * which was preceded by a backslash.  If this is a
10569                          * context where the delimiter is also a metacharacter,
10570                          * the backslash is useless, and deprecated.  () and []
10571                          * are meta in any context. {} are meta only when
10572                          * appearing in a quantifier or in things like '\p{'.
10573                          * They also aren't meta unless there is a matching
10574                          * closed, escaped char later on within the string.
10575                          * If 's' points to an open, set a flag; if to a close,
10576                          * test that flag, and raise a warning if it was set */
10577 
10578 			if (deprecate_escaped_meta) {
10579                             if (*s == PL_multi_open) {
10580                                 if (*s != '{') {
10581                                     escaped_open = s;
10582                                 }
10583                                 else if (regcurly(s,
10584                                                   TRUE /* Look for a closing
10585                                                           '\}' */)
10586                                          || (s - start > 2  /* Look for e.g.
10587                                                                '\x{' */
10588                                              && _generic_isCC(*(s-2), _CC_BACKSLASH_FOO_LBRACE_IS_META)))
10589                                 {
10590                                     escaped_open = s;
10591                                 }
10592                             }
10593                             else if (escaped_open) {
10594                                 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
10595                                     "Useless use of '\\'; doesn't escape metacharacter '%c'", PL_multi_open);
10596                                 escaped_open = NULL;
10597                             }
10598                         }
10599                     }
10600 		    else
10601 			*to++ = *s++;
10602 		}
10603 		/* allow nested opens and closes */
10604 		else if (*s == PL_multi_close && --brackets <= 0)
10605 		    break;
10606 		else if (*s == PL_multi_open)
10607 		    brackets++;
10608 		else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
10609 		    has_utf8 = TRUE;
10610 		*to = *s;
10611 	    }
10612 	}
10613 	/* terminate the copied string and update the sv's end-of-string */
10614 	*to = '\0';
10615 	SvCUR_set(sv, to - SvPVX_const(sv));
10616 
10617 	/*
10618 	 * this next chunk reads more into the buffer if we're not done yet
10619 	 */
10620 
10621   	if (s < PL_bufend)
10622 	    break;		/* handle case where we are done yet :-) */
10623 
10624 #ifndef PERL_STRICT_CR
10625 	if (to - SvPVX_const(sv) >= 2) {
10626 	    if ((to[-2] == '\r' && to[-1] == '\n') ||
10627 		(to[-2] == '\n' && to[-1] == '\r'))
10628 	    {
10629 		to[-2] = '\n';
10630 		to--;
10631 		SvCUR_set(sv, to - SvPVX_const(sv));
10632 	    }
10633 	    else if (to[-1] == '\r')
10634 		to[-1] = '\n';
10635 	}
10636 	else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
10637 	    to[-1] = '\n';
10638 #endif
10639 
10640      read_more_line:
10641 	/* if we're out of file, or a read fails, bail and reset the current
10642 	   line marker so we can report where the unterminated string began
10643 	*/
10644 #ifdef PERL_MAD
10645 	if (PL_madskills) {
10646 	    char * const tstart = SvPVX(PL_linestr) + stuffstart;
10647 	    if (PL_thisstuff)
10648 		sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
10649 	    else
10650 		PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
10651 	}
10652 #endif
10653 	COPLINE_INC_WITH_HERELINES;
10654 	PL_bufptr = PL_bufend;
10655 	if (!lex_next_chunk(0)) {
10656 	    sv_free(sv);
10657 	    CopLINE_set(PL_curcop, (line_t)PL_multi_start);
10658 	    return NULL;
10659 	}
10660 	s = PL_bufptr;
10661 #ifdef PERL_MAD
10662 	stuffstart = 0;
10663 #endif
10664     }
10665 
10666     /* at this point, we have successfully read the delimited string */
10667 
10668     if (!PL_encoding || UTF || re_reparse) {
10669 #ifdef PERL_MAD
10670 	if (PL_madskills) {
10671 	    char * const tstart = SvPVX(PL_linestr) + stuffstart;
10672 	    const int len = s - tstart;
10673 	    if (PL_thisstuff)
10674 		sv_catpvn(PL_thisstuff, tstart, len);
10675 	    else
10676 		PL_thisstuff = newSVpvn(tstart, len);
10677 	    if (!PL_thisclose && !keep_delims)
10678 		PL_thisclose = newSVpvn(s,termlen);
10679 	}
10680 #endif
10681 
10682 	if (keep_delims)
10683 	    sv_catpvn(sv, s, termlen);
10684 	s += termlen;
10685     }
10686 #ifdef PERL_MAD
10687     else {
10688 	if (PL_madskills) {
10689 	    char * const tstart = SvPVX(PL_linestr) + stuffstart;
10690 	    const int len = s - tstart - termlen;
10691 	    if (PL_thisstuff)
10692 		sv_catpvn(PL_thisstuff, tstart, len);
10693 	    else
10694 		PL_thisstuff = newSVpvn(tstart, len);
10695 	    if (!PL_thisclose && !keep_delims)
10696 		PL_thisclose = newSVpvn(s - termlen,termlen);
10697 	}
10698     }
10699 #endif
10700     if (has_utf8 || (PL_encoding && !re_reparse))
10701 	SvUTF8_on(sv);
10702 
10703     PL_multi_end = CopLINE(PL_curcop);
10704 
10705     /* if we allocated too much space, give some back */
10706     if (SvCUR(sv) + 5 < SvLEN(sv)) {
10707 	SvLEN_set(sv, SvCUR(sv) + 1);
10708 	SvPV_renew(sv, SvLEN(sv));
10709     }
10710 
10711     /* decide whether this is the first or second quoted string we've read
10712        for this op
10713     */
10714 
10715     if (PL_lex_stuff)
10716 	PL_sublex_info.repl = sv;
10717     else
10718 	PL_lex_stuff = sv;
10719     return s;
10720 }
10721 
10722 /*
10723   scan_num
10724   takes: pointer to position in buffer
10725   returns: pointer to new position in buffer
10726   side-effects: builds ops for the constant in pl_yylval.op
10727 
10728   Read a number in any of the formats that Perl accepts:
10729 
10730   \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*)	12 12.34 12.
10731   \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*)			.34
10732   0b[01](_?[01])*
10733   0[0-7](_?[0-7])*
10734   0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
10735 
10736   Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
10737   thing it reads.
10738 
10739   If it reads a number without a decimal point or an exponent, it will
10740   try converting the number to an integer and see if it can do so
10741   without loss of precision.
10742 */
10743 
10744 char *
10745 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
10746 {
10747     dVAR;
10748     const char *s = start;	/* current position in buffer */
10749     char *d;			/* destination in temp buffer */
10750     char *e;			/* end of temp buffer */
10751     NV nv;				/* number read, as a double */
10752     SV *sv = NULL;			/* place to put the converted number */
10753     bool floatit;			/* boolean: int or float? */
10754     const char *lastub = NULL;		/* position of last underbar */
10755     static const char* const number_too_long = "Number too long";
10756 
10757     PERL_ARGS_ASSERT_SCAN_NUM;
10758 
10759     /* We use the first character to decide what type of number this is */
10760 
10761     switch (*s) {
10762     default:
10763 	Perl_croak(aTHX_ "panic: scan_num, *s=%d", *s);
10764 
10765     /* if it starts with a 0, it could be an octal number, a decimal in
10766        0.13 disguise, or a hexadecimal number, or a binary number. */
10767     case '0':
10768 	{
10769 	  /* variables:
10770 	     u		holds the "number so far"
10771 	     shift	the power of 2 of the base
10772 			(hex == 4, octal == 3, binary == 1)
10773 	     overflowed	was the number more than we can hold?
10774 
10775 	     Shift is used when we add a digit.  It also serves as an "are
10776 	     we in octal/hex/binary?" indicator to disallow hex characters
10777 	     when in octal mode.
10778 	   */
10779 	    NV n = 0.0;
10780 	    UV u = 0;
10781 	    I32 shift;
10782 	    bool overflowed = FALSE;
10783 	    bool just_zero  = TRUE;	/* just plain 0 or binary number? */
10784 	    static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
10785 	    static const char* const bases[5] =
10786 	      { "", "binary", "", "octal", "hexadecimal" };
10787 	    static const char* const Bases[5] =
10788 	      { "", "Binary", "", "Octal", "Hexadecimal" };
10789 	    static const char* const maxima[5] =
10790 	      { "",
10791 		"0b11111111111111111111111111111111",
10792 		"",
10793 		"037777777777",
10794 		"0xffffffff" };
10795 	    const char *base, *Base, *max;
10796 
10797 	    /* check for hex */
10798 	    if (s[1] == 'x' || s[1] == 'X') {
10799 		shift = 4;
10800 		s += 2;
10801 		just_zero = FALSE;
10802 	    } else if (s[1] == 'b' || s[1] == 'B') {
10803 		shift = 1;
10804 		s += 2;
10805 		just_zero = FALSE;
10806 	    }
10807 	    /* check for a decimal in disguise */
10808 	    else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
10809 		goto decimal;
10810 	    /* so it must be octal */
10811 	    else {
10812 		shift = 3;
10813 		s++;
10814 	    }
10815 
10816 	    if (*s == '_') {
10817 		Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
10818 			       "Misplaced _ in number");
10819 	       lastub = s++;
10820 	    }
10821 
10822 	    base = bases[shift];
10823 	    Base = Bases[shift];
10824 	    max  = maxima[shift];
10825 
10826 	    /* read the rest of the number */
10827 	    for (;;) {
10828 		/* x is used in the overflow test,
10829 		   b is the digit we're adding on. */
10830 		UV x, b;
10831 
10832 		switch (*s) {
10833 
10834 		/* if we don't mention it, we're done */
10835 		default:
10836 		    goto out;
10837 
10838 		/* _ are ignored -- but warned about if consecutive */
10839 		case '_':
10840 		    if (lastub && s == lastub + 1)
10841 		        Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
10842 				       "Misplaced _ in number");
10843 		    lastub = s++;
10844 		    break;
10845 
10846 		/* 8 and 9 are not octal */
10847 		case '8': case '9':
10848 		    if (shift == 3)
10849 			yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
10850 		    /* FALL THROUGH */
10851 
10852 	        /* octal digits */
10853 		case '2': case '3': case '4':
10854 		case '5': case '6': case '7':
10855 		    if (shift == 1)
10856 			yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
10857 		    /* FALL THROUGH */
10858 
10859 		case '0': case '1':
10860 		    b = *s++ & 15;		/* ASCII digit -> value of digit */
10861 		    goto digit;
10862 
10863 	        /* hex digits */
10864 		case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
10865 		case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
10866 		    /* make sure they said 0x */
10867 		    if (shift != 4)
10868 			goto out;
10869 		    b = (*s++ & 7) + 9;
10870 
10871 		    /* Prepare to put the digit we have onto the end
10872 		       of the number so far.  We check for overflows.
10873 		    */
10874 
10875 		  digit:
10876 		    just_zero = FALSE;
10877 		    if (!overflowed) {
10878 			x = u << shift;	/* make room for the digit */
10879 
10880 			if ((x >> shift) != u
10881 			    && !(PL_hints & HINT_NEW_BINARY)) {
10882 			    overflowed = TRUE;
10883 			    n = (NV) u;
10884 			    Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
10885 					     "Integer overflow in %s number",
10886 					     base);
10887 			} else
10888 			    u = x | b;		/* add the digit to the end */
10889 		    }
10890 		    if (overflowed) {
10891 			n *= nvshift[shift];
10892 			/* If an NV has not enough bits in its
10893 			 * mantissa to represent an UV this summing of
10894 			 * small low-order numbers is a waste of time
10895 			 * (because the NV cannot preserve the
10896 			 * low-order bits anyway): we could just
10897 			 * remember when did we overflow and in the
10898 			 * end just multiply n by the right
10899 			 * amount. */
10900 			n += (NV) b;
10901 		    }
10902 		    break;
10903 		}
10904 	    }
10905 
10906 	  /* if we get here, we had success: make a scalar value from
10907 	     the number.
10908 	  */
10909 	  out:
10910 
10911 	    /* final misplaced underbar check */
10912 	    if (s[-1] == '_') {
10913 		Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10914 	    }
10915 
10916 	    if (overflowed) {
10917 		if (n > 4294967295.0)
10918 		    Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
10919 				   "%s number > %s non-portable",
10920 				   Base, max);
10921 		sv = newSVnv(n);
10922 	    }
10923 	    else {
10924 #if UVSIZE > 4
10925 		if (u > 0xffffffff)
10926 		    Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
10927 				   "%s number > %s non-portable",
10928 				   Base, max);
10929 #endif
10930 		sv = newSVuv(u);
10931 	    }
10932 	    if (just_zero && (PL_hints & HINT_NEW_INTEGER))
10933 		sv = new_constant(start, s - start, "integer",
10934 				  sv, NULL, NULL, 0);
10935 	    else if (PL_hints & HINT_NEW_BINARY)
10936 		sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
10937 	}
10938 	break;
10939 
10940     /*
10941       handle decimal numbers.
10942       we're also sent here when we read a 0 as the first digit
10943     */
10944     case '1': case '2': case '3': case '4': case '5':
10945     case '6': case '7': case '8': case '9': case '.':
10946       decimal:
10947 	d = PL_tokenbuf;
10948 	e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
10949 	floatit = FALSE;
10950 
10951 	/* read next group of digits and _ and copy into d */
10952 	while (isDIGIT(*s) || *s == '_') {
10953 	    /* skip underscores, checking for misplaced ones
10954 	       if -w is on
10955 	    */
10956 	    if (*s == '_') {
10957 		if (lastub && s == lastub + 1)
10958 		    Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
10959 				   "Misplaced _ in number");
10960 		lastub = s++;
10961 	    }
10962 	    else {
10963 	        /* check for end of fixed-length buffer */
10964 		if (d >= e)
10965 		    Perl_croak(aTHX_ "%s", number_too_long);
10966 		/* if we're ok, copy the character */
10967 		*d++ = *s++;
10968 	    }
10969 	}
10970 
10971 	/* final misplaced underbar check */
10972 	if (lastub && s == lastub + 1) {
10973 	    Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
10974 	}
10975 
10976 	/* read a decimal portion if there is one.  avoid
10977 	   3..5 being interpreted as the number 3. followed
10978 	   by .5
10979 	*/
10980 	if (*s == '.' && s[1] != '.') {
10981 	    floatit = TRUE;
10982 	    *d++ = *s++;
10983 
10984 	    if (*s == '_') {
10985 		Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
10986 			       "Misplaced _ in number");
10987 		lastub = s;
10988 	    }
10989 
10990 	    /* copy, ignoring underbars, until we run out of digits.
10991 	    */
10992 	    for (; isDIGIT(*s) || *s == '_'; s++) {
10993 	        /* fixed length buffer check */
10994 		if (d >= e)
10995 		    Perl_croak(aTHX_ "%s", number_too_long);
10996 		if (*s == '_') {
10997 		   if (lastub && s == lastub + 1)
10998 		       Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
10999 				      "Misplaced _ in number");
11000 		   lastub = s;
11001 		}
11002 		else
11003 		    *d++ = *s;
11004 	    }
11005 	    /* fractional part ending in underbar? */
11006 	    if (s[-1] == '_') {
11007 		Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
11008 			       "Misplaced _ in number");
11009 	    }
11010 	    if (*s == '.' && isDIGIT(s[1])) {
11011 		/* oops, it's really a v-string, but without the "v" */
11012 		s = start;
11013 		goto vstring;
11014 	    }
11015 	}
11016 
11017 	/* read exponent part, if present */
11018 	if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
11019 	    floatit = TRUE;
11020 	    s++;
11021 
11022 	    /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
11023 	    *d++ = 'e';		/* At least some Mach atof()s don't grok 'E' */
11024 
11025 	    /* stray preinitial _ */
11026 	    if (*s == '_') {
11027 		Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
11028 			       "Misplaced _ in number");
11029 	        lastub = s++;
11030 	    }
11031 
11032 	    /* allow positive or negative exponent */
11033 	    if (*s == '+' || *s == '-')
11034 		*d++ = *s++;
11035 
11036 	    /* stray initial _ */
11037 	    if (*s == '_') {
11038 		Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
11039 			       "Misplaced _ in number");
11040 	        lastub = s++;
11041 	    }
11042 
11043 	    /* read digits of exponent */
11044 	    while (isDIGIT(*s) || *s == '_') {
11045 	        if (isDIGIT(*s)) {
11046 		    if (d >= e)
11047 		        Perl_croak(aTHX_ "%s", number_too_long);
11048 		    *d++ = *s++;
11049 		}
11050 		else {
11051 		   if (((lastub && s == lastub + 1) ||
11052 			(!isDIGIT(s[1]) && s[1] != '_')))
11053 		       Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
11054 				      "Misplaced _ in number");
11055 		   lastub = s++;
11056 		}
11057 	    }
11058 	}
11059 
11060 
11061 	/*
11062            We try to do an integer conversion first if no characters
11063            indicating "float" have been found.
11064 	 */
11065 
11066 	if (!floatit) {
11067     	    UV uv;
11068 	    const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
11069 
11070             if (flags == IS_NUMBER_IN_UV) {
11071               if (uv <= IV_MAX)
11072 		sv = newSViv(uv); /* Prefer IVs over UVs. */
11073               else
11074 	    	sv = newSVuv(uv);
11075             } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
11076               if (uv <= (UV) IV_MIN)
11077                 sv = newSViv(-(IV)uv);
11078               else
11079 	    	floatit = TRUE;
11080             } else
11081               floatit = TRUE;
11082         }
11083 	if (floatit) {
11084 	    /* terminate the string */
11085 	    *d = '\0';
11086 	    nv = Atof(PL_tokenbuf);
11087 	    sv = newSVnv(nv);
11088 	}
11089 
11090 	if ( floatit
11091 	     ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
11092 	    const char *const key = floatit ? "float" : "integer";
11093 	    const STRLEN keylen = floatit ? 5 : 7;
11094 	    sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
11095 				key, keylen, sv, NULL, NULL, 0);
11096 	}
11097 	break;
11098 
11099     /* if it starts with a v, it could be a v-string */
11100     case 'v':
11101 vstring:
11102 		sv = newSV(5); /* preallocate storage space */
11103 		ENTER_with_name("scan_vstring");
11104 		SAVEFREESV(sv);
11105 		s = scan_vstring(s, PL_bufend, sv);
11106 		SvREFCNT_inc_simple_void_NN(sv);
11107 		LEAVE_with_name("scan_vstring");
11108 	break;
11109     }
11110 
11111     /* make the op for the constant and return */
11112 
11113     if (sv)
11114 	lvalp->opval = newSVOP(OP_CONST, 0, sv);
11115     else
11116 	lvalp->opval = NULL;
11117 
11118     return (char *)s;
11119 }
11120 
11121 STATIC char *
11122 S_scan_formline(pTHX_ char *s)
11123 {
11124     dVAR;
11125     char *eol;
11126     char *t;
11127     SV * const stuff = newSVpvs("");
11128     bool needargs = FALSE;
11129     bool eofmt = FALSE;
11130 #ifdef PERL_MAD
11131     char *tokenstart = s;
11132     SV* savewhite = NULL;
11133 
11134     if (PL_madskills) {
11135 	savewhite = PL_thiswhite;
11136 	PL_thiswhite = 0;
11137     }
11138 #endif
11139 
11140     PERL_ARGS_ASSERT_SCAN_FORMLINE;
11141 
11142     while (!needargs) {
11143 	if (*s == '.') {
11144 	    t = s+1;
11145 #ifdef PERL_STRICT_CR
11146 	    while (SPACE_OR_TAB(*t))
11147 		t++;
11148 #else
11149 	    while (SPACE_OR_TAB(*t) || *t == '\r')
11150 		t++;
11151 #endif
11152 	    if (*t == '\n' || t == PL_bufend) {
11153 	        eofmt = TRUE;
11154 		break;
11155             }
11156 	}
11157 	eol = (char *) memchr(s,'\n',PL_bufend-s);
11158 	if (!eol++)
11159 		eol = PL_bufend;
11160 	if (*s != '#') {
11161 	    for (t = s; t < eol; t++) {
11162 		if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
11163 		    needargs = FALSE;
11164 		    goto enough;	/* ~~ must be first line in formline */
11165 		}
11166 		if (*t == '@' || *t == '^')
11167 		    needargs = TRUE;
11168 	    }
11169 	    if (eol > s) {
11170 	        sv_catpvn(stuff, s, eol-s);
11171 #ifndef PERL_STRICT_CR
11172 		if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
11173 		    char *end = SvPVX(stuff) + SvCUR(stuff);
11174 		    end[-2] = '\n';
11175 		    end[-1] = '\0';
11176 		    SvCUR_set(stuff, SvCUR(stuff) - 1);
11177 		}
11178 #endif
11179 	    }
11180 	    else
11181 	      break;
11182 	}
11183 	s = (char*)eol;
11184 	if ((PL_rsfp || PL_parser->filtered)
11185 	 && PL_parser->form_lex_state == LEX_NORMAL) {
11186 	    bool got_some;
11187 #ifdef PERL_MAD
11188 	    if (PL_madskills) {
11189 		if (PL_thistoken)
11190 		    sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
11191 		else
11192 		    PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
11193 	    }
11194 #endif
11195 	    PL_bufptr = PL_bufend;
11196 	    COPLINE_INC_WITH_HERELINES;
11197 	    got_some = lex_next_chunk(0);
11198 	    CopLINE_dec(PL_curcop);
11199 	    s = PL_bufptr;
11200 #ifdef PERL_MAD
11201 	    tokenstart = PL_bufptr;
11202 #endif
11203 	    if (!got_some)
11204 		break;
11205 	}
11206 	incline(s);
11207     }
11208   enough:
11209     if (!SvCUR(stuff) || needargs)
11210 	PL_lex_state = PL_parser->form_lex_state;
11211     if (SvCUR(stuff)) {
11212 	PL_expect = XSTATE;
11213 	if (needargs) {
11214 	    start_force(PL_curforce);
11215 	    NEXTVAL_NEXTTOKE.ival = 0;
11216 	    force_next(FORMLBRACK);
11217 	}
11218 	if (!IN_BYTES) {
11219 	    if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
11220 		SvUTF8_on(stuff);
11221 	    else if (PL_encoding)
11222 		sv_recode_to_utf8(stuff, PL_encoding);
11223 	}
11224 	start_force(PL_curforce);
11225 	NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
11226 	force_next(THING);
11227     }
11228     else {
11229 	SvREFCNT_dec(stuff);
11230 	if (eofmt)
11231 	    PL_lex_formbrack = 0;
11232     }
11233 #ifdef PERL_MAD
11234     if (PL_madskills) {
11235 	if (PL_thistoken)
11236 	    sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
11237 	else
11238 	    PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
11239 	PL_thiswhite = savewhite;
11240     }
11241 #endif
11242     return s;
11243 }
11244 
11245 I32
11246 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
11247 {
11248     dVAR;
11249     const I32 oldsavestack_ix = PL_savestack_ix;
11250     CV* const outsidecv = PL_compcv;
11251 
11252     SAVEI32(PL_subline);
11253     save_item(PL_subname);
11254     SAVESPTR(PL_compcv);
11255 
11256     PL_compcv = MUTABLE_CV(newSV_type(is_format ? SVt_PVFM : SVt_PVCV));
11257     CvFLAGS(PL_compcv) |= flags;
11258 
11259     PL_subline = CopLINE(PL_curcop);
11260     CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
11261     CvOUTSIDE(PL_compcv) = MUTABLE_CV(SvREFCNT_inc_simple(outsidecv));
11262     CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
11263     if (outsidecv && CvPADLIST(outsidecv))
11264 	CvPADLIST(PL_compcv)->xpadl_outid =
11265 	    PadlistNAMES(CvPADLIST(outsidecv));
11266 
11267     return oldsavestack_ix;
11268 }
11269 
11270 #ifdef __SC__
11271 #pragma segment Perl_yylex
11272 #endif
11273 static int
11274 S_yywarn(pTHX_ const char *const s, U32 flags)
11275 {
11276     dVAR;
11277 
11278     PERL_ARGS_ASSERT_YYWARN;
11279 
11280     PL_in_eval |= EVAL_WARNONLY;
11281     yyerror_pv(s, flags);
11282     PL_in_eval &= ~EVAL_WARNONLY;
11283     return 0;
11284 }
11285 
11286 int
11287 Perl_yyerror(pTHX_ const char *const s)
11288 {
11289     PERL_ARGS_ASSERT_YYERROR;
11290     return yyerror_pvn(s, strlen(s), 0);
11291 }
11292 
11293 int
11294 Perl_yyerror_pv(pTHX_ const char *const s, U32 flags)
11295 {
11296     PERL_ARGS_ASSERT_YYERROR_PV;
11297     return yyerror_pvn(s, strlen(s), flags);
11298 }
11299 
11300 int
11301 Perl_yyerror_pvn(pTHX_ const char *const s, STRLEN len, U32 flags)
11302 {
11303     dVAR;
11304     const char *context = NULL;
11305     int contlen = -1;
11306     SV *msg;
11307     SV * const where_sv = newSVpvs_flags("", SVs_TEMP);
11308     int yychar  = PL_parser->yychar;
11309 
11310     PERL_ARGS_ASSERT_YYERROR_PVN;
11311 
11312     if (!yychar || (yychar == ';' && !PL_rsfp))
11313 	sv_catpvs(where_sv, "at EOF");
11314     else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
11315       PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
11316       PL_oldbufptr != PL_bufptr) {
11317 	/*
11318 		Only for NetWare:
11319 		The code below is removed for NetWare because it abends/crashes on NetWare
11320 		when the script has error such as not having the closing quotes like:
11321 		    if ($var eq "value)
11322 		Checking of white spaces is anyway done in NetWare code.
11323 	*/
11324 #ifndef NETWARE
11325 	while (isSPACE(*PL_oldoldbufptr))
11326 	    PL_oldoldbufptr++;
11327 #endif
11328 	context = PL_oldoldbufptr;
11329 	contlen = PL_bufptr - PL_oldoldbufptr;
11330     }
11331     else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
11332       PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
11333 	/*
11334 		Only for NetWare:
11335 		The code below is removed for NetWare because it abends/crashes on NetWare
11336 		when the script has error such as not having the closing quotes like:
11337 		    if ($var eq "value)
11338 		Checking of white spaces is anyway done in NetWare code.
11339 	*/
11340 #ifndef NETWARE
11341 	while (isSPACE(*PL_oldbufptr))
11342 	    PL_oldbufptr++;
11343 #endif
11344 	context = PL_oldbufptr;
11345 	contlen = PL_bufptr - PL_oldbufptr;
11346     }
11347     else if (yychar > 255)
11348 	sv_catpvs(where_sv, "next token ???");
11349     else if (yychar == -2) { /* YYEMPTY */
11350 	if (PL_lex_state == LEX_NORMAL ||
11351 	   (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
11352 	    sv_catpvs(where_sv, "at end of line");
11353 	else if (PL_lex_inpat)
11354 	    sv_catpvs(where_sv, "within pattern");
11355 	else
11356 	    sv_catpvs(where_sv, "within string");
11357     }
11358     else {
11359 	sv_catpvs(where_sv, "next char ");
11360 	if (yychar < 32)
11361 	    Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
11362 	else if (isPRINT_LC(yychar)) {
11363 	    const char string = yychar;
11364 	    sv_catpvn(where_sv, &string, 1);
11365 	}
11366 	else
11367 	    Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
11368     }
11369     msg = newSVpvn_flags(s, len, (flags & SVf_UTF8) | SVs_TEMP);
11370     Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
11371         OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
11372     if (context)
11373 	Perl_sv_catpvf(aTHX_ msg, "near \"%"SVf"\"\n",
11374                             SVfARG(newSVpvn_flags(context, contlen,
11375                                         SVs_TEMP | (UTF ? SVf_UTF8 : 0))));
11376     else
11377 	Perl_sv_catpvf(aTHX_ msg, "%"SVf"\n", SVfARG(where_sv));
11378     if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
11379         Perl_sv_catpvf(aTHX_ msg,
11380         "  (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
11381                 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
11382         PL_multi_end = 0;
11383     }
11384     if (PL_in_eval & EVAL_WARNONLY) {
11385 	Perl_ck_warner_d(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
11386     }
11387     else
11388 	qerror(msg);
11389     if (PL_error_count >= 10) {
11390 	SV * errsv;
11391 	if (PL_in_eval && ((errsv = ERRSV), SvCUR(errsv)))
11392 	    Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
11393 		       SVfARG(errsv), OutCopFILE(PL_curcop));
11394 	else
11395 	    Perl_croak(aTHX_ "%s has too many errors.\n",
11396             OutCopFILE(PL_curcop));
11397     }
11398     PL_in_my = 0;
11399     PL_in_my_stash = NULL;
11400     return 0;
11401 }
11402 #ifdef __SC__
11403 #pragma segment Main
11404 #endif
11405 
11406 STATIC char*
11407 S_swallow_bom(pTHX_ U8 *s)
11408 {
11409     dVAR;
11410     const STRLEN slen = SvCUR(PL_linestr);
11411 
11412     PERL_ARGS_ASSERT_SWALLOW_BOM;
11413 
11414     switch (s[0]) {
11415     case 0xFF:
11416 	if (s[1] == 0xFE) {
11417 	    /* UTF-16 little-endian? (or UTF-32LE?) */
11418 	    if (s[2] == 0 && s[3] == 0)  /* UTF-32 little-endian */
11419 		/* diag_listed_as: Unsupported script encoding %s */
11420 		Perl_croak(aTHX_ "Unsupported script encoding UTF-32LE");
11421 #ifndef PERL_NO_UTF16_FILTER
11422 	    if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (BOM)\n");
11423 	    s += 2;
11424 	    if (PL_bufend > (char*)s) {
11425 		s = add_utf16_textfilter(s, TRUE);
11426 	    }
11427 #else
11428 	    /* diag_listed_as: Unsupported script encoding %s */
11429 	    Perl_croak(aTHX_ "Unsupported script encoding UTF-16LE");
11430 #endif
11431 	}
11432 	break;
11433     case 0xFE:
11434 	if (s[1] == 0xFF) {   /* UTF-16 big-endian? */
11435 #ifndef PERL_NO_UTF16_FILTER
11436 	    if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
11437 	    s += 2;
11438 	    if (PL_bufend > (char *)s) {
11439 		s = add_utf16_textfilter(s, FALSE);
11440 	    }
11441 #else
11442 	    /* diag_listed_as: Unsupported script encoding %s */
11443 	    Perl_croak(aTHX_ "Unsupported script encoding UTF-16BE");
11444 #endif
11445 	}
11446 	break;
11447     case 0xEF:
11448 	if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
11449 	    if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
11450 	    s += 3;                      /* UTF-8 */
11451 	}
11452 	break;
11453     case 0:
11454 	if (slen > 3) {
11455 	     if (s[1] == 0) {
11456 		  if (s[2] == 0xFE && s[3] == 0xFF) {
11457 		       /* UTF-32 big-endian */
11458 		       /* diag_listed_as: Unsupported script encoding %s */
11459 		       Perl_croak(aTHX_ "Unsupported script encoding UTF-32BE");
11460 		  }
11461 	     }
11462 	     else if (s[2] == 0 && s[3] != 0) {
11463 		  /* Leading bytes
11464 		   * 00 xx 00 xx
11465 		   * are a good indicator of UTF-16BE. */
11466 #ifndef PERL_NO_UTF16_FILTER
11467 		  if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
11468 		  s = add_utf16_textfilter(s, FALSE);
11469 #else
11470 		  /* diag_listed_as: Unsupported script encoding %s */
11471 		  Perl_croak(aTHX_ "Unsupported script encoding UTF-16BE");
11472 #endif
11473 	     }
11474 	}
11475 #ifdef EBCDIC
11476     case 0xDD:
11477         if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
11478             if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
11479             s += 4;                      /* UTF-8 */
11480         }
11481         break;
11482 #endif
11483 
11484     default:
11485 	 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
11486 		  /* Leading bytes
11487 		   * xx 00 xx 00
11488 		   * are a good indicator of UTF-16LE. */
11489 #ifndef PERL_NO_UTF16_FILTER
11490 	      if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
11491 	      s = add_utf16_textfilter(s, TRUE);
11492 #else
11493 	      /* diag_listed_as: Unsupported script encoding %s */
11494 	      Perl_croak(aTHX_ "Unsupported script encoding UTF-16LE");
11495 #endif
11496 	 }
11497     }
11498     return (char*)s;
11499 }
11500 
11501 
11502 #ifndef PERL_NO_UTF16_FILTER
11503 static I32
11504 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
11505 {
11506     dVAR;
11507     SV *const filter = FILTER_DATA(idx);
11508     /* We re-use this each time round, throwing the contents away before we
11509        return.  */
11510     SV *const utf16_buffer = MUTABLE_SV(IoTOP_GV(filter));
11511     SV *const utf8_buffer = filter;
11512     IV status = IoPAGE(filter);
11513     const bool reverse = cBOOL(IoLINES(filter));
11514     I32 retval;
11515 
11516     PERL_ARGS_ASSERT_UTF16_TEXTFILTER;
11517 
11518     /* As we're automatically added, at the lowest level, and hence only called
11519        from this file, we can be sure that we're not called in block mode. Hence
11520        don't bother writing code to deal with block mode.  */
11521     if (maxlen) {
11522 	Perl_croak(aTHX_ "panic: utf16_textfilter called in block mode (for %d characters)", maxlen);
11523     }
11524     if (status < 0) {
11525 	Perl_croak(aTHX_ "panic: utf16_textfilter called after error (status=%"IVdf")", status);
11526     }
11527     DEBUG_P(PerlIO_printf(Perl_debug_log,
11528 			  "utf16_textfilter(%p,%ce): idx=%d maxlen=%d status=%"IVdf" utf16=%"UVuf" utf8=%"UVuf"\n",
11529 			  FPTR2DPTR(void *, S_utf16_textfilter),
11530 			  reverse ? 'l' : 'b', idx, maxlen, status,
11531 			  (UV)SvCUR(utf16_buffer), (UV)SvCUR(utf8_buffer)));
11532 
11533     while (1) {
11534 	STRLEN chars;
11535 	STRLEN have;
11536 	I32 newlen;
11537 	U8 *end;
11538 	/* First, look in our buffer of existing UTF-8 data:  */
11539 	char *nl = (char *)memchr(SvPVX(utf8_buffer), '\n', SvCUR(utf8_buffer));
11540 
11541 	if (nl) {
11542 	    ++nl;
11543 	} else if (status == 0) {
11544 	    /* EOF */
11545 	    IoPAGE(filter) = 0;
11546 	    nl = SvEND(utf8_buffer);
11547 	}
11548 	if (nl) {
11549 	    STRLEN got = nl - SvPVX(utf8_buffer);
11550 	    /* Did we have anything to append?  */
11551 	    retval = got != 0;
11552 	    sv_catpvn(sv, SvPVX(utf8_buffer), got);
11553 	    /* Everything else in this code works just fine if SVp_POK isn't
11554 	       set.  This, however, needs it, and we need it to work, else
11555 	       we loop infinitely because the buffer is never consumed.  */
11556 	    sv_chop(utf8_buffer, nl);
11557 	    break;
11558 	}
11559 
11560 	/* OK, not a complete line there, so need to read some more UTF-16.
11561 	   Read an extra octect if the buffer currently has an odd number. */
11562 	while (1) {
11563 	    if (status <= 0)
11564 		break;
11565 	    if (SvCUR(utf16_buffer) >= 2) {
11566 		/* Location of the high octet of the last complete code point.
11567 		   Gosh, UTF-16 is a pain. All the benefits of variable length,
11568 		   *coupled* with all the benefits of partial reads and
11569 		   endianness.  */
11570 		const U8 *const last_hi = (U8*)SvPVX(utf16_buffer)
11571 		    + ((SvCUR(utf16_buffer) & ~1) - (reverse ? 1 : 2));
11572 
11573 		if (*last_hi < 0xd8 || *last_hi > 0xdb) {
11574 		    break;
11575 		}
11576 
11577 		/* We have the first half of a surrogate. Read more.  */
11578 		DEBUG_P(PerlIO_printf(Perl_debug_log, "utf16_textfilter partial surrogate detected at %p\n", last_hi));
11579 	    }
11580 
11581 	    status = FILTER_READ(idx + 1, utf16_buffer,
11582 				 160 + (SvCUR(utf16_buffer) & 1));
11583 	    DEBUG_P(PerlIO_printf(Perl_debug_log, "utf16_textfilter status=%"IVdf" SvCUR(sv)=%"UVuf"\n", status, (UV)SvCUR(utf16_buffer)));
11584 	    DEBUG_P({ sv_dump(utf16_buffer); sv_dump(utf8_buffer);});
11585 	    if (status < 0) {
11586 		/* Error */
11587 		IoPAGE(filter) = status;
11588 		return status;
11589 	    }
11590 	}
11591 
11592 	chars = SvCUR(utf16_buffer) >> 1;
11593 	have = SvCUR(utf8_buffer);
11594 	SvGROW(utf8_buffer, have + chars * 3 + 1);
11595 
11596 	if (reverse) {
11597 	    end = utf16_to_utf8_reversed((U8*)SvPVX(utf16_buffer),
11598 					 (U8*)SvPVX_const(utf8_buffer) + have,
11599 					 chars * 2, &newlen);
11600 	} else {
11601 	    end = utf16_to_utf8((U8*)SvPVX(utf16_buffer),
11602 				(U8*)SvPVX_const(utf8_buffer) + have,
11603 				chars * 2, &newlen);
11604 	}
11605 	SvCUR_set(utf8_buffer, have + newlen);
11606 	*end = '\0';
11607 
11608 	/* No need to keep this SV "well-formed" with a '\0' after the end, as
11609 	   it's private to us, and utf16_to_utf8{,reversed} take a
11610 	   (pointer,length) pair, rather than a NUL-terminated string.  */
11611 	if(SvCUR(utf16_buffer) & 1) {
11612 	    *SvPVX(utf16_buffer) = SvEND(utf16_buffer)[-1];
11613 	    SvCUR_set(utf16_buffer, 1);
11614 	} else {
11615 	    SvCUR_set(utf16_buffer, 0);
11616 	}
11617     }
11618     DEBUG_P(PerlIO_printf(Perl_debug_log,
11619 			  "utf16_textfilter: returns, status=%"IVdf" utf16=%"UVuf" utf8=%"UVuf"\n",
11620 			  status,
11621 			  (UV)SvCUR(utf16_buffer), (UV)SvCUR(utf8_buffer)));
11622     DEBUG_P({ sv_dump(utf8_buffer); sv_dump(sv);});
11623     return retval;
11624 }
11625 
11626 static U8 *
11627 S_add_utf16_textfilter(pTHX_ U8 *const s, bool reversed)
11628 {
11629     SV *filter = filter_add(S_utf16_textfilter, NULL);
11630 
11631     PERL_ARGS_ASSERT_ADD_UTF16_TEXTFILTER;
11632 
11633     IoTOP_GV(filter) = MUTABLE_GV(newSVpvn((char *)s, PL_bufend - (char*)s));
11634     sv_setpvs(filter, "");
11635     IoLINES(filter) = reversed;
11636     IoPAGE(filter) = 1; /* Not EOF */
11637 
11638     /* Sadly, we have to return a valid pointer, come what may, so we have to
11639        ignore any error return from this.  */
11640     SvCUR_set(PL_linestr, 0);
11641     if (FILTER_READ(0, PL_linestr, 0)) {
11642 	SvUTF8_on(PL_linestr);
11643     } else {
11644 	SvUTF8_on(PL_linestr);
11645     }
11646     PL_bufend = SvEND(PL_linestr);
11647     return (U8*)SvPVX(PL_linestr);
11648 }
11649 #endif
11650 
11651 /*
11652 Returns a pointer to the next character after the parsed
11653 vstring, as well as updating the passed in sv.
11654 
11655 Function must be called like
11656 
11657 	sv = sv_2mortal(newSV(5));
11658 	s = scan_vstring(s,e,sv);
11659 
11660 where s and e are the start and end of the string.
11661 The sv should already be large enough to store the vstring
11662 passed in, for performance reasons.
11663 
11664 This function may croak if fatal warnings are enabled in the
11665 calling scope, hence the sv_2mortal in the example (to prevent
11666 a leak).  Make sure to do SvREFCNT_inc afterwards if you use
11667 sv_2mortal.
11668 
11669 */
11670 
11671 char *
11672 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
11673 {
11674     dVAR;
11675     const char *pos = s;
11676     const char *start = s;
11677 
11678     PERL_ARGS_ASSERT_SCAN_VSTRING;
11679 
11680     if (*pos == 'v') pos++;  /* get past 'v' */
11681     while (pos < e && (isDIGIT(*pos) || *pos == '_'))
11682 	pos++;
11683     if ( *pos != '.') {
11684 	/* this may not be a v-string if followed by => */
11685 	const char *next = pos;
11686 	while (next < e && isSPACE(*next))
11687 	    ++next;
11688 	if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
11689 	    /* return string not v-string */
11690 	    sv_setpvn(sv,(char *)s,pos-s);
11691 	    return (char *)pos;
11692 	}
11693     }
11694 
11695     if (!isALPHA(*pos)) {
11696 	U8 tmpbuf[UTF8_MAXBYTES+1];
11697 
11698 	if (*s == 'v')
11699 	    s++;  /* get past 'v' */
11700 
11701 	sv_setpvs(sv, "");
11702 
11703 	for (;;) {
11704 	    /* this is atoi() that tolerates underscores */
11705 	    U8 *tmpend;
11706 	    UV rev = 0;
11707 	    const char *end = pos;
11708 	    UV mult = 1;
11709 	    while (--end >= s) {
11710 		if (*end != '_') {
11711 		    const UV orev = rev;
11712 		    rev += (*end - '0') * mult;
11713 		    mult *= 10;
11714 		    if (orev > rev)
11715 			/* diag_listed_as: Integer overflow in %s number */
11716 			Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
11717 					 "Integer overflow in decimal number");
11718 		}
11719 	    }
11720 #ifdef EBCDIC
11721 	    if (rev > 0x7FFFFFFF)
11722 		 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
11723 #endif
11724 	    /* Append native character for the rev point */
11725 	    tmpend = uvchr_to_utf8(tmpbuf, rev);
11726 	    sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
11727 	    if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
11728 		 SvUTF8_on(sv);
11729 	    if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
11730 		 s = ++pos;
11731 	    else {
11732 		 s = pos;
11733 		 break;
11734 	    }
11735 	    while (pos < e && (isDIGIT(*pos) || *pos == '_'))
11736 		 pos++;
11737 	}
11738 	SvPOK_on(sv);
11739 	sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
11740 	SvRMAGICAL_on(sv);
11741     }
11742     return (char *)s;
11743 }
11744 
11745 int
11746 Perl_keyword_plugin_standard(pTHX_
11747 	char *keyword_ptr, STRLEN keyword_len, OP **op_ptr)
11748 {
11749     PERL_ARGS_ASSERT_KEYWORD_PLUGIN_STANDARD;
11750     PERL_UNUSED_CONTEXT;
11751     PERL_UNUSED_ARG(keyword_ptr);
11752     PERL_UNUSED_ARG(keyword_len);
11753     PERL_UNUSED_ARG(op_ptr);
11754     return KEYWORD_PLUGIN_DECLINE;
11755 }
11756 
11757 #define parse_recdescent(g,p) S_parse_recdescent(aTHX_ g,p)
11758 static void
11759 S_parse_recdescent(pTHX_ int gramtype, I32 fakeeof)
11760 {
11761     SAVEI32(PL_lex_brackets);
11762     if (PL_lex_brackets > 100)
11763 	Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
11764     PL_lex_brackstack[PL_lex_brackets++] = XFAKEEOF;
11765     SAVEI32(PL_lex_allbrackets);
11766     PL_lex_allbrackets = 0;
11767     SAVEI8(PL_lex_fakeeof);
11768     PL_lex_fakeeof = (U8)fakeeof;
11769     if(yyparse(gramtype) && !PL_parser->error_count)
11770 	qerror(Perl_mess(aTHX_ "Parse error"));
11771 }
11772 
11773 #define parse_recdescent_for_op(g,p) S_parse_recdescent_for_op(aTHX_ g,p)
11774 static OP *
11775 S_parse_recdescent_for_op(pTHX_ int gramtype, I32 fakeeof)
11776 {
11777     OP *o;
11778     ENTER;
11779     SAVEVPTR(PL_eval_root);
11780     PL_eval_root = NULL;
11781     parse_recdescent(gramtype, fakeeof);
11782     o = PL_eval_root;
11783     LEAVE;
11784     return o;
11785 }
11786 
11787 #define parse_expr(p,f) S_parse_expr(aTHX_ p,f)
11788 static OP *
11789 S_parse_expr(pTHX_ I32 fakeeof, U32 flags)
11790 {
11791     OP *exprop;
11792     if (flags & ~PARSE_OPTIONAL)
11793 	Perl_croak(aTHX_ "Parsing code internal error (%s)", "parse_expr");
11794     exprop = parse_recdescent_for_op(GRAMEXPR, fakeeof);
11795     if (!exprop && !(flags & PARSE_OPTIONAL)) {
11796 	if (!PL_parser->error_count)
11797 	    qerror(Perl_mess(aTHX_ "Parse error"));
11798 	exprop = newOP(OP_NULL, 0);
11799     }
11800     return exprop;
11801 }
11802 
11803 /*
11804 =for apidoc Amx|OP *|parse_arithexpr|U32 flags
11805 
11806 Parse a Perl arithmetic expression.  This may contain operators of precedence
11807 down to the bit shift operators.  The expression must be followed (and thus
11808 terminated) either by a comparison or lower-precedence operator or by
11809 something that would normally terminate an expression such as semicolon.
11810 If I<flags> includes C<PARSE_OPTIONAL> then the expression is optional,
11811 otherwise it is mandatory.  It is up to the caller to ensure that the
11812 dynamic parser state (L</PL_parser> et al) is correctly set to reflect
11813 the source of the code to be parsed and the lexical context for the
11814 expression.
11815 
11816 The op tree representing the expression is returned.  If an optional
11817 expression is absent, a null pointer is returned, otherwise the pointer
11818 will be non-null.
11819 
11820 If an error occurs in parsing or compilation, in most cases a valid op
11821 tree is returned anyway.  The error is reflected in the parser state,
11822 normally resulting in a single exception at the top level of parsing
11823 which covers all the compilation errors that occurred.  Some compilation
11824 errors, however, will throw an exception immediately.
11825 
11826 =cut
11827 */
11828 
11829 OP *
11830 Perl_parse_arithexpr(pTHX_ U32 flags)
11831 {
11832     return parse_expr(LEX_FAKEEOF_COMPARE, flags);
11833 }
11834 
11835 /*
11836 =for apidoc Amx|OP *|parse_termexpr|U32 flags
11837 
11838 Parse a Perl term expression.  This may contain operators of precedence
11839 down to the assignment operators.  The expression must be followed (and thus
11840 terminated) either by a comma or lower-precedence operator or by
11841 something that would normally terminate an expression such as semicolon.
11842 If I<flags> includes C<PARSE_OPTIONAL> then the expression is optional,
11843 otherwise it is mandatory.  It is up to the caller to ensure that the
11844 dynamic parser state (L</PL_parser> et al) is correctly set to reflect
11845 the source of the code to be parsed and the lexical context for the
11846 expression.
11847 
11848 The op tree representing the expression is returned.  If an optional
11849 expression is absent, a null pointer is returned, otherwise the pointer
11850 will be non-null.
11851 
11852 If an error occurs in parsing or compilation, in most cases a valid op
11853 tree is returned anyway.  The error is reflected in the parser state,
11854 normally resulting in a single exception at the top level of parsing
11855 which covers all the compilation errors that occurred.  Some compilation
11856 errors, however, will throw an exception immediately.
11857 
11858 =cut
11859 */
11860 
11861 OP *
11862 Perl_parse_termexpr(pTHX_ U32 flags)
11863 {
11864     return parse_expr(LEX_FAKEEOF_COMMA, flags);
11865 }
11866 
11867 /*
11868 =for apidoc Amx|OP *|parse_listexpr|U32 flags
11869 
11870 Parse a Perl list expression.  This may contain operators of precedence
11871 down to the comma operator.  The expression must be followed (and thus
11872 terminated) either by a low-precedence logic operator such as C<or> or by
11873 something that would normally terminate an expression such as semicolon.
11874 If I<flags> includes C<PARSE_OPTIONAL> then the expression is optional,
11875 otherwise it is mandatory.  It is up to the caller to ensure that the
11876 dynamic parser state (L</PL_parser> et al) is correctly set to reflect
11877 the source of the code to be parsed and the lexical context for the
11878 expression.
11879 
11880 The op tree representing the expression is returned.  If an optional
11881 expression is absent, a null pointer is returned, otherwise the pointer
11882 will be non-null.
11883 
11884 If an error occurs in parsing or compilation, in most cases a valid op
11885 tree is returned anyway.  The error is reflected in the parser state,
11886 normally resulting in a single exception at the top level of parsing
11887 which covers all the compilation errors that occurred.  Some compilation
11888 errors, however, will throw an exception immediately.
11889 
11890 =cut
11891 */
11892 
11893 OP *
11894 Perl_parse_listexpr(pTHX_ U32 flags)
11895 {
11896     return parse_expr(LEX_FAKEEOF_LOWLOGIC, flags);
11897 }
11898 
11899 /*
11900 =for apidoc Amx|OP *|parse_fullexpr|U32 flags
11901 
11902 Parse a single complete Perl expression.  This allows the full
11903 expression grammar, including the lowest-precedence operators such
11904 as C<or>.  The expression must be followed (and thus terminated) by a
11905 token that an expression would normally be terminated by: end-of-file,
11906 closing bracketing punctuation, semicolon, or one of the keywords that
11907 signals a postfix expression-statement modifier.  If I<flags> includes
11908 C<PARSE_OPTIONAL> then the expression is optional, otherwise it is
11909 mandatory.  It is up to the caller to ensure that the dynamic parser
11910 state (L</PL_parser> et al) is correctly set to reflect the source of
11911 the code to be parsed and the lexical context for the expression.
11912 
11913 The op tree representing the expression is returned.  If an optional
11914 expression is absent, a null pointer is returned, otherwise the pointer
11915 will be non-null.
11916 
11917 If an error occurs in parsing or compilation, in most cases a valid op
11918 tree is returned anyway.  The error is reflected in the parser state,
11919 normally resulting in a single exception at the top level of parsing
11920 which covers all the compilation errors that occurred.  Some compilation
11921 errors, however, will throw an exception immediately.
11922 
11923 =cut
11924 */
11925 
11926 OP *
11927 Perl_parse_fullexpr(pTHX_ U32 flags)
11928 {
11929     return parse_expr(LEX_FAKEEOF_NONEXPR, flags);
11930 }
11931 
11932 /*
11933 =for apidoc Amx|OP *|parse_block|U32 flags
11934 
11935 Parse a single complete Perl code block.  This consists of an opening
11936 brace, a sequence of statements, and a closing brace.  The block
11937 constitutes a lexical scope, so C<my> variables and various compile-time
11938 effects can be contained within it.  It is up to the caller to ensure
11939 that the dynamic parser state (L</PL_parser> et al) is correctly set to
11940 reflect the source of the code to be parsed and the lexical context for
11941 the statement.
11942 
11943 The op tree representing the code block is returned.  This is always a
11944 real op, never a null pointer.  It will normally be a C<lineseq> list,
11945 including C<nextstate> or equivalent ops.  No ops to construct any kind
11946 of runtime scope are included by virtue of it being a block.
11947 
11948 If an error occurs in parsing or compilation, in most cases a valid op
11949 tree (most likely null) is returned anyway.  The error is reflected in
11950 the parser state, normally resulting in a single exception at the top
11951 level of parsing which covers all the compilation errors that occurred.
11952 Some compilation errors, however, will throw an exception immediately.
11953 
11954 The I<flags> parameter is reserved for future use, and must always
11955 be zero.
11956 
11957 =cut
11958 */
11959 
11960 OP *
11961 Perl_parse_block(pTHX_ U32 flags)
11962 {
11963     if (flags)
11964 	Perl_croak(aTHX_ "Parsing code internal error (%s)", "parse_block");
11965     return parse_recdescent_for_op(GRAMBLOCK, LEX_FAKEEOF_NEVER);
11966 }
11967 
11968 /*
11969 =for apidoc Amx|OP *|parse_barestmt|U32 flags
11970 
11971 Parse a single unadorned Perl statement.  This may be a normal imperative
11972 statement or a declaration that has compile-time effect.  It does not
11973 include any label or other affixture.  It is up to the caller to ensure
11974 that the dynamic parser state (L</PL_parser> et al) is correctly set to
11975 reflect the source of the code to be parsed and the lexical context for
11976 the statement.
11977 
11978 The op tree representing the statement is returned.  This may be a
11979 null pointer if the statement is null, for example if it was actually
11980 a subroutine definition (which has compile-time side effects).  If not
11981 null, it will be ops directly implementing the statement, suitable to
11982 pass to L</newSTATEOP>.  It will not normally include a C<nextstate> or
11983 equivalent op (except for those embedded in a scope contained entirely
11984 within the statement).
11985 
11986 If an error occurs in parsing or compilation, in most cases a valid op
11987 tree (most likely null) is returned anyway.  The error is reflected in
11988 the parser state, normally resulting in a single exception at the top
11989 level of parsing which covers all the compilation errors that occurred.
11990 Some compilation errors, however, will throw an exception immediately.
11991 
11992 The I<flags> parameter is reserved for future use, and must always
11993 be zero.
11994 
11995 =cut
11996 */
11997 
11998 OP *
11999 Perl_parse_barestmt(pTHX_ U32 flags)
12000 {
12001     if (flags)
12002 	Perl_croak(aTHX_ "Parsing code internal error (%s)", "parse_barestmt");
12003     return parse_recdescent_for_op(GRAMBARESTMT, LEX_FAKEEOF_NEVER);
12004 }
12005 
12006 /*
12007 =for apidoc Amx|SV *|parse_label|U32 flags
12008 
12009 Parse a single label, possibly optional, of the type that may prefix a
12010 Perl statement.  It is up to the caller to ensure that the dynamic parser
12011 state (L</PL_parser> et al) is correctly set to reflect the source of
12012 the code to be parsed.  If I<flags> includes C<PARSE_OPTIONAL> then the
12013 label is optional, otherwise it is mandatory.
12014 
12015 The name of the label is returned in the form of a fresh scalar.  If an
12016 optional label is absent, a null pointer is returned.
12017 
12018 If an error occurs in parsing, which can only occur if the label is
12019 mandatory, a valid label is returned anyway.  The error is reflected in
12020 the parser state, normally resulting in a single exception at the top
12021 level of parsing which covers all the compilation errors that occurred.
12022 
12023 =cut
12024 */
12025 
12026 SV *
12027 Perl_parse_label(pTHX_ U32 flags)
12028 {
12029     if (flags & ~PARSE_OPTIONAL)
12030 	Perl_croak(aTHX_ "Parsing code internal error (%s)", "parse_label");
12031     if (PL_lex_state == LEX_KNOWNEXT) {
12032 	PL_parser->yychar = yylex();
12033 	if (PL_parser->yychar == LABEL) {
12034 	    char * const lpv = pl_yylval.pval;
12035 	    STRLEN llen = strlen(lpv);
12036 	    PL_parser->yychar = YYEMPTY;
12037 	    return newSVpvn_flags(lpv, llen, lpv[llen+1] ? SVf_UTF8 : 0);
12038 	} else {
12039 	    yyunlex();
12040 	    goto no_label;
12041 	}
12042     } else {
12043 	char *s, *t;
12044 	STRLEN wlen, bufptr_pos;
12045 	lex_read_space(0);
12046 	t = s = PL_bufptr;
12047         if (!isIDFIRST_lazy_if(s, UTF))
12048 	    goto no_label;
12049 	t = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &wlen);
12050 	if (word_takes_any_delimeter(s, wlen))
12051 	    goto no_label;
12052 	bufptr_pos = s - SvPVX(PL_linestr);
12053 	PL_bufptr = t;
12054 	lex_read_space(LEX_KEEP_PREVIOUS);
12055 	t = PL_bufptr;
12056 	s = SvPVX(PL_linestr) + bufptr_pos;
12057 	if (t[0] == ':' && t[1] != ':') {
12058 	    PL_oldoldbufptr = PL_oldbufptr;
12059 	    PL_oldbufptr = s;
12060 	    PL_bufptr = t+1;
12061 	    return newSVpvn_flags(s, wlen, UTF ? SVf_UTF8 : 0);
12062 	} else {
12063 	    PL_bufptr = s;
12064 	    no_label:
12065 	    if (flags & PARSE_OPTIONAL) {
12066 		return NULL;
12067 	    } else {
12068 		qerror(Perl_mess(aTHX_ "Parse error"));
12069 		return newSVpvs("x");
12070 	    }
12071 	}
12072     }
12073 }
12074 
12075 /*
12076 =for apidoc Amx|OP *|parse_fullstmt|U32 flags
12077 
12078 Parse a single complete Perl statement.  This may be a normal imperative
12079 statement or a declaration that has compile-time effect, and may include
12080 optional labels.  It is up to the caller to ensure that the dynamic
12081 parser state (L</PL_parser> et al) is correctly set to reflect the source
12082 of the code to be parsed and the lexical context for the statement.
12083 
12084 The op tree representing the statement is returned.  This may be a
12085 null pointer if the statement is null, for example if it was actually
12086 a subroutine definition (which has compile-time side effects).  If not
12087 null, it will be the result of a L</newSTATEOP> call, normally including
12088 a C<nextstate> or equivalent op.
12089 
12090 If an error occurs in parsing or compilation, in most cases a valid op
12091 tree (most likely null) is returned anyway.  The error is reflected in
12092 the parser state, normally resulting in a single exception at the top
12093 level of parsing which covers all the compilation errors that occurred.
12094 Some compilation errors, however, will throw an exception immediately.
12095 
12096 The I<flags> parameter is reserved for future use, and must always
12097 be zero.
12098 
12099 =cut
12100 */
12101 
12102 OP *
12103 Perl_parse_fullstmt(pTHX_ U32 flags)
12104 {
12105     if (flags)
12106 	Perl_croak(aTHX_ "Parsing code internal error (%s)", "parse_fullstmt");
12107     return parse_recdescent_for_op(GRAMFULLSTMT, LEX_FAKEEOF_NEVER);
12108 }
12109 
12110 /*
12111 =for apidoc Amx|OP *|parse_stmtseq|U32 flags
12112 
12113 Parse a sequence of zero or more Perl statements.  These may be normal
12114 imperative statements, including optional labels, or declarations
12115 that have compile-time effect, or any mixture thereof.  The statement
12116 sequence ends when a closing brace or end-of-file is encountered in a
12117 place where a new statement could have validly started.  It is up to
12118 the caller to ensure that the dynamic parser state (L</PL_parser> et al)
12119 is correctly set to reflect the source of the code to be parsed and the
12120 lexical context for the statements.
12121 
12122 The op tree representing the statement sequence is returned.  This may
12123 be a null pointer if the statements were all null, for example if there
12124 were no statements or if there were only subroutine definitions (which
12125 have compile-time side effects).  If not null, it will be a C<lineseq>
12126 list, normally including C<nextstate> or equivalent ops.
12127 
12128 If an error occurs in parsing or compilation, in most cases a valid op
12129 tree is returned anyway.  The error is reflected in the parser state,
12130 normally resulting in a single exception at the top level of parsing
12131 which covers all the compilation errors that occurred.  Some compilation
12132 errors, however, will throw an exception immediately.
12133 
12134 The I<flags> parameter is reserved for future use, and must always
12135 be zero.
12136 
12137 =cut
12138 */
12139 
12140 OP *
12141 Perl_parse_stmtseq(pTHX_ U32 flags)
12142 {
12143     OP *stmtseqop;
12144     I32 c;
12145     if (flags)
12146 	Perl_croak(aTHX_ "Parsing code internal error (%s)", "parse_stmtseq");
12147     stmtseqop = parse_recdescent_for_op(GRAMSTMTSEQ, LEX_FAKEEOF_CLOSING);
12148     c = lex_peek_unichar(0);
12149     if (c != -1 && c != /*{*/'}')
12150 	qerror(Perl_mess(aTHX_ "Parse error"));
12151     return stmtseqop;
12152 }
12153 
12154 /*
12155  * Local variables:
12156  * c-indentation-style: bsd
12157  * c-basic-offset: 4
12158  * indent-tabs-mode: nil
12159  * End:
12160  *
12161  * ex: set ts=8 sts=4 sw=4 et:
12162  */
12163