xref: /netbsd-src/usr.bin/make/for.c (revision b1c86f5f087524e68db12794ee9c3e3da1ab17a0)
1 /*	$NetBSD: for.c,v 1.47 2010/02/06 20:37:13 dholland Exp $	*/
2 
3 /*
4  * Copyright (c) 1992, The Regents of the University of California.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the University nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 #ifndef MAKE_NATIVE
33 static char rcsid[] = "$NetBSD: for.c,v 1.47 2010/02/06 20:37:13 dholland Exp $";
34 #else
35 #include <sys/cdefs.h>
36 #ifndef lint
37 #if 0
38 static char sccsid[] = "@(#)for.c	8.1 (Berkeley) 6/6/93";
39 #else
40 __RCSID("$NetBSD: for.c,v 1.47 2010/02/06 20:37:13 dholland Exp $");
41 #endif
42 #endif /* not lint */
43 #endif
44 
45 /*-
46  * for.c --
47  *	Functions to handle loops in a makefile.
48  *
49  * Interface:
50  *	For_Eval 	Evaluate the loop in the passed line.
51  *	For_Run		Run accumulated loop
52  *
53  */
54 
55 #include    <assert.h>
56 #include    <ctype.h>
57 
58 #include    "make.h"
59 #include    "hash.h"
60 #include    "dir.h"
61 #include    "buf.h"
62 #include    "strlist.h"
63 
64 #define FOR_SUB_ESCAPE_CHAR  1
65 #define FOR_SUB_ESCAPE_BRACE 2
66 #define FOR_SUB_ESCAPE_PAREN 4
67 
68 /*
69  * For statements are of the form:
70  *
71  * .for <variable> in <varlist>
72  * ...
73  * .endfor
74  *
75  * The trick is to look for the matching end inside for for loop
76  * To do that, we count the current nesting level of the for loops.
77  * and the .endfor statements, accumulating all the statements between
78  * the initial .for loop and the matching .endfor;
79  * then we evaluate the for loop for each variable in the varlist.
80  *
81  * Note that any nested fors are just passed through; they get handled
82  * recursively in For_Eval when we're expanding the enclosing for in
83  * For_Run.
84  */
85 
86 static int  	  forLevel = 0;  	/* Nesting level	*/
87 
88 /*
89  * State of a for loop.
90  */
91 typedef struct _For {
92     Buffer	  buf;			/* Body of loop		*/
93     strlist_t     vars;			/* Iteration variables	*/
94     strlist_t     items;		/* Substitution items */
95     char          *parse_buf;
96     int           short_var;
97     int           sub_next;
98 } For;
99 
100 static For        *accumFor;            /* Loop being accumulated */
101 
102 
103 
104 static char *
105 make_str(const char *ptr, int len)
106 {
107 	char *new_ptr;
108 
109 	new_ptr = bmake_malloc(len + 1);
110 	memcpy(new_ptr, ptr, len);
111 	new_ptr[len] = 0;
112 	return new_ptr;
113 }
114 
115 static void
116 For_Free(For *arg)
117 {
118     Buf_Destroy(&arg->buf, TRUE);
119     strlist_clean(&arg->vars);
120     strlist_clean(&arg->items);
121     free(arg->parse_buf);
122 
123     free(arg);
124 }
125 
126 /*-
127  *-----------------------------------------------------------------------
128  * For_Eval --
129  *	Evaluate the for loop in the passed line. The line
130  *	looks like this:
131  *	    .for <variable> in <varlist>
132  *
133  * Input:
134  *	line		Line to parse
135  *
136  * Results:
137  *      0: Not a .for statement, parse the line
138  *	1: We found a for loop
139  *     -1: A .for statement with a bad syntax error, discard.
140  *
141  * Side Effects:
142  *	None.
143  *
144  *-----------------------------------------------------------------------
145  */
146 int
147 For_Eval(char *line)
148 {
149     For *new_for;
150     char *ptr = line, *sub;
151     int len;
152     int escapes;
153     unsigned char ch;
154 
155     /* Skip the '.' and any following whitespace */
156     for (ptr++; *ptr && isspace((unsigned char) *ptr); ptr++)
157 	continue;
158 
159     /*
160      * If we are not in a for loop quickly determine if the statement is
161      * a for.
162      */
163     if (ptr[0] != 'f' || ptr[1] != 'o' || ptr[2] != 'r' ||
164 	    !isspace((unsigned char) ptr[3])) {
165 	if (ptr[0] == 'e' && strncmp(ptr+1, "ndfor", 5) == 0) {
166 	    Parse_Error(PARSE_FATAL, "for-less endfor");
167 	    return -1;
168 	}
169 	return 0;
170     }
171     ptr += 3;
172 
173     /*
174      * we found a for loop, and now we are going to parse it.
175      */
176 
177     new_for = bmake_malloc(sizeof *new_for);
178     memset(new_for, 0, sizeof *new_for);
179 
180     /* Grab the variables. Terminate on "in". */
181     for (;; ptr += len) {
182 	while (*ptr && isspace((unsigned char) *ptr))
183 	    ptr++;
184 	if (*ptr == '\0') {
185 	    Parse_Error(PARSE_FATAL, "missing `in' in for");
186 	    For_Free(new_for);
187 	    return -1;
188 	}
189 	for (len = 1; ptr[len] && !isspace((unsigned char)ptr[len]); len++)
190 	    continue;
191 	if (len == 2 && ptr[0] == 'i' && ptr[1] == 'n') {
192 	    ptr += 2;
193 	    break;
194 	}
195 	if (len == 1)
196 	    new_for->short_var = 1;
197 	strlist_add_str(&new_for->vars, make_str(ptr, len), len);
198     }
199 
200     if (strlist_num(&new_for->vars) == 0) {
201 	Parse_Error(PARSE_FATAL, "no iteration variables in for");
202 	For_Free(new_for);
203 	return -1;
204     }
205 
206     while (*ptr && isspace((unsigned char) *ptr))
207 	ptr++;
208 
209     /*
210      * Make a list with the remaining words
211      * The values are substituted as ${:U<value>...} so we must \ escape
212      * characters that break that syntax.
213      * Variables are fully expanded - so it is safe for escape $.
214      * We can't do the escapes here - because we don't know whether
215      * we are substuting into ${...} or $(...).
216      */
217     sub = Var_Subst(NULL, ptr, VAR_GLOBAL, FALSE);
218 
219     for (ptr = sub;; ptr += len) {
220 	while (*ptr && isspace((unsigned char)*ptr))
221 	    ptr++;
222 	if (*ptr == 0)
223 	    break;
224 	escapes = 0;
225 	for (len = 0; (ch = ptr[len]) != 0 && !isspace(ch); len++) {
226 	    if (ch == ':' || ch == '$' || ch == '\\')
227 		escapes |= FOR_SUB_ESCAPE_CHAR;
228 	    else if (ch == ')')
229 		escapes |= FOR_SUB_ESCAPE_PAREN;
230 	    else if (ch == /*{*/ '}')
231 		escapes |= FOR_SUB_ESCAPE_BRACE;
232 	}
233 	strlist_add_str(&new_for->items, make_str(ptr, len), escapes);
234     }
235 
236     free(sub);
237 
238     if (strlist_num(&new_for->items) % strlist_num(&new_for->vars)) {
239 	Parse_Error(PARSE_FATAL,
240 		"Wrong number of words (%d) in .for substitution list"
241 		" with %d vars",
242 		strlist_num(&new_for->items), strlist_num(&new_for->vars));
243 	/*
244 	 * Return 'success' so that the body of the .for loop is accumulated.
245 	 * Remove all items so that the loop doesn't iterate.
246 	 */
247 	strlist_clean(&new_for->items);
248     }
249 
250     Buf_Init(&new_for->buf, 0);
251     accumFor = new_for;
252     forLevel = 1;
253     return 1;
254 }
255 
256 /*
257  * Add another line to a .for loop.
258  * Returns 0 when the matching .endfor is reached.
259  */
260 
261 int
262 For_Accum(char *line)
263 {
264     char *ptr = line;
265 
266     if (*ptr == '.') {
267 
268 	for (ptr++; *ptr && isspace((unsigned char) *ptr); ptr++)
269 	    continue;
270 
271 	if (strncmp(ptr, "endfor", 6) == 0 &&
272 		(isspace((unsigned char) ptr[6]) || !ptr[6])) {
273 	    if (DEBUG(FOR))
274 		(void)fprintf(debug_file, "For: end for %d\n", forLevel);
275 	    if (--forLevel <= 0)
276 		return 0;
277 	} else if (strncmp(ptr, "for", 3) == 0 &&
278 		 isspace((unsigned char) ptr[3])) {
279 	    forLevel++;
280 	    if (DEBUG(FOR))
281 		(void)fprintf(debug_file, "For: new loop %d\n", forLevel);
282 	}
283     }
284 
285     Buf_AddBytes(&accumFor->buf, strlen(line), line);
286     Buf_AddByte(&accumFor->buf, '\n');
287     return 1;
288 }
289 
290 
291 /*-
292  *-----------------------------------------------------------------------
293  * For_Run --
294  *	Run the for loop, imitating the actions of an include file
295  *
296  * Results:
297  *	None.
298  *
299  * Side Effects:
300  *	None.
301  *
302  *-----------------------------------------------------------------------
303  */
304 
305 static int
306 for_var_len(const char *var)
307 {
308     char ch, var_start, var_end;
309     int depth;
310     int len;
311 
312     var_start = *var;
313     if (var_start == 0)
314 	/* just escape the $ */
315 	return 0;
316 
317     if (var_start == '(')
318 	var_end = ')';
319     else if (var_start == '{')
320 	var_end = '}';
321     else
322 	/* Single char variable */
323 	return 1;
324 
325     depth = 1;
326     for (len = 1; (ch = var[len++]) != 0;) {
327 	if (ch == var_start)
328 	    depth++;
329 	else if (ch == var_end && --depth == 0)
330 	    return len;
331     }
332 
333     /* Variable end not found, escape the $ */
334     return 0;
335 }
336 
337 static void
338 for_substitute(Buffer *cmds, strlist_t *items, unsigned int item_no, char ech)
339 {
340     const char *item = strlist_str(items, item_no);
341     int len;
342     char ch;
343 
344     /* If there were no escapes, or the only escape is the other variable
345      * terminator, then just substitute the full string */
346     if (!(strlist_info(items, item_no) &
347 	    (ech == ')' ? ~FOR_SUB_ESCAPE_BRACE : ~FOR_SUB_ESCAPE_PAREN))) {
348 	Buf_AddBytes(cmds, strlen(item), item);
349 	return;
350     }
351 
352     /* Escape ':', '$', '\\' and 'ech' - removed by :U processing */
353     while ((ch = *item++) != 0) {
354 	if (ch == '$') {
355 	    len = for_var_len(item);
356 	    if (len != 0) {
357 		Buf_AddBytes(cmds, len + 1, item - 1);
358 		item += len;
359 		continue;
360 	    }
361 	    Buf_AddByte(cmds, '\\');
362 	} else if (ch == ':' || ch == '\\' || ch == ech)
363 	    Buf_AddByte(cmds, '\\');
364 	Buf_AddByte(cmds, ch);
365     }
366 }
367 
368 static char *
369 For_Iterate(void *v_arg)
370 {
371     For *arg = v_arg;
372     int i, len;
373     char *var;
374     char *cp;
375     char *cmd_cp;
376     char *body_end;
377     char ch;
378     Buffer cmds;
379 
380     if (arg->sub_next + strlist_num(&arg->vars) > strlist_num(&arg->items)) {
381 	/* No more iterations */
382 	For_Free(arg);
383 	return NULL;
384     }
385 
386     free(arg->parse_buf);
387     arg->parse_buf = NULL;
388 
389     /*
390      * Scan the for loop body and replace references to the loop variables
391      * with variable references that expand to the required text.
392      * Using variable expansions ensures that the .for loop can't generate
393      * syntax, and that the later parsing will still see a variable.
394      * We assume that the null variable will never be defined.
395      *
396      * The detection of substitions of the loop control variable is naive.
397      * Many of the modifiers use \ to escape $ (not $) so it is possible
398      * to contrive a makefile where an unwanted substitution happens.
399      */
400 
401     cmd_cp = Buf_GetAll(&arg->buf, &len);
402     body_end = cmd_cp + len;
403     Buf_Init(&cmds, len + 256);
404     for (cp = cmd_cp; (cp = strchr(cp, '$')) != NULL;) {
405 	char ech;
406 	ch = *++cp;
407 	if ((ch == '(' && (ech = ')')) || (ch == '{' && (ech = '}'))) {
408 	    cp++;
409 	    /* Check variable name against the .for loop variables */
410 	    STRLIST_FOREACH(var, &arg->vars, i) {
411 		len = strlist_info(&arg->vars, i);
412 		if (memcmp(cp, var, len) != 0)
413 		    continue;
414 		if (cp[len] != ':' && cp[len] != ech && cp[len] != '\\')
415 		    continue;
416 		/* Found a variable match. Replace with :U<value> */
417 		Buf_AddBytes(&cmds, cp - cmd_cp, cmd_cp);
418 		Buf_AddBytes(&cmds, 2, ":U");
419 		cp += len;
420 		cmd_cp = cp;
421 		for_substitute(&cmds, &arg->items, arg->sub_next + i, ech);
422 		break;
423 	    }
424 	    continue;
425 	}
426 	if (ch == 0)
427 	    break;
428 	/* Probably a single character name, ignore $$ and stupid ones. {*/
429 	if (!arg->short_var || strchr("}):$", ch) != NULL) {
430 	    cp++;
431 	    continue;
432 	}
433 	STRLIST_FOREACH(var, &arg->vars, i) {
434 	    if (var[0] != ch || var[1] != 0)
435 		continue;
436 	    /* Found a variable match. Replace with ${:U<value>} */
437 	    Buf_AddBytes(&cmds, cp - cmd_cp, cmd_cp);
438 	    Buf_AddBytes(&cmds, 3, "{:U");
439 	    cmd_cp = ++cp;
440 	    for_substitute(&cmds, &arg->items, arg->sub_next + i, /*{*/ '}');
441 	    Buf_AddBytes(&cmds, 1, "}");
442 	    break;
443 	}
444     }
445     Buf_AddBytes(&cmds, body_end - cmd_cp, cmd_cp);
446 
447     cp = Buf_Destroy(&cmds, FALSE);
448     if (DEBUG(FOR))
449 	(void)fprintf(debug_file, "For: loop body:\n%s", cp);
450 
451     arg->sub_next += strlist_num(&arg->vars);
452 
453     arg->parse_buf = cp;
454     return cp;
455 }
456 
457 void
458 For_Run(int lineno)
459 {
460     For *arg;
461 
462     arg = accumFor;
463     accumFor = NULL;
464 
465     if (strlist_num(&arg->items) == 0) {
466         /* Nothing to expand - possibly due to an earlier syntax error. */
467         For_Free(arg);
468         return;
469     }
470 
471     Parse_SetInput(NULL, lineno, -1, For_Iterate, arg);
472 }
473