xref: /openbsd-src/gnu/usr.bin/perl/utils/h2ph.PL (revision 09467b48e8bc8b4905716062da846024139afbf2)
1#!/usr/local/bin/perl
2
3use Config;
4use File::Basename qw(basename dirname);
5use Cwd;
6
7# List explicitly here the variables you want Configure to
8# generate.  Metaconfig only looks for shell variables, so you
9# have to mention them as if they were shell variables, not
10# %Config entries.  Thus you write
11#  $startperl
12# to ensure Configure will look for $Config{startperl}.
13# Wanted:  $archlibexp
14
15# This forces PL files to create target in same directory as PL file.
16# This is so that make depend always knows where to find PL derivatives.
17$origdir = cwd;
18chdir dirname($0);
19$file = basename($0, '.PL');
20$file .= '.com' if $^O eq 'VMS';
21
22open OUT, '>', $file or die "Can't create $file: $!";
23
24print "Extracting $file (with variable substitutions)\n";
25
26# In this section, perl variables will be expanded during extraction.
27# You can use $Config{...} to use Configure variables.
28
29print OUT <<"!GROK!THIS!";
30$Config{startperl}
31    eval 'exec $Config{perlpath} -S \$0 \${1+"\$@"}'
32	if \$running_under_some_shell;
33!GROK!THIS!
34
35# In the following, perl variables are not expanded during extraction.
36
37print OUT <<'!NO!SUBS!';
38
39BEGIN { pop @INC if $INC[-1] eq '.' }
40
41use strict;
42
43use Config;
44use File::Path qw(mkpath);
45use Getopt::Std;
46
47# Make sure read permissions for all are set:
48if (defined umask && (umask() & 0444)) {
49    umask (umask() & ~0444);
50}
51
52getopts('Dd:rlhaQe');
53use vars qw($opt_D $opt_d $opt_r $opt_l $opt_h $opt_a $opt_Q $opt_e);
54die "-r and -a options are mutually exclusive\n" if ($opt_r and $opt_a);
55my @inc_dirs = inc_dirs() if $opt_a;
56
57my $Exit = 0;
58
59my $Dest_dir = $opt_d || $Config{installsitearch};
60die "Destination directory $Dest_dir doesn't exist or isn't a directory\n"
61    unless -d $Dest_dir;
62
63my @isatype = qw(
64	char	uchar	u_char
65	short	ushort	u_short
66	int	uint	u_int
67	long	ulong	u_long
68	FILE	key_t	caddr_t
69	float	double	size_t
70);
71
72my %isatype;
73@isatype{@isatype} = (1) x @isatype;
74my $inif = 0;
75my %Is_converted;
76my %bad_file = ();
77
78@ARGV = ('-') unless @ARGV;
79
80build_preamble_if_necessary();
81
82sub reindent($) {
83    my($text) = shift;
84    $text =~ s/\n/\n    /g;
85    $text =~ s/        /\t/g;
86    $text;
87}
88
89my ($t, $tab, %curargs, $new, $eval_index, $dir, $name, $args, $outfile);
90my ($incl, $incl_type, $incl_quote, $next);
91while (defined (my $file = next_file())) {
92    if (-l $file and -d $file) {
93        link_if_possible($file) if ($opt_l);
94        next;
95    }
96
97    # Recover from header files with unbalanced cpp directives
98    $t = '';
99    $tab = 0;
100
101    # $eval_index goes into '#line' directives, to help locate syntax errors:
102    $eval_index = 1;
103
104    if ($file eq '-') {
105	open(IN, "-");
106	open(OUT, ">-");
107    } else {
108	($outfile = $file) =~ s/\.h$/.ph/ || next;
109	print "$file -> $outfile\n" unless $opt_Q;
110	if ($file =~ m|^(.*)/|) {
111	    $dir = $1;
112	    mkpath "$Dest_dir/$dir";
113	}
114
115	if ($opt_a) { # automagic mode:  locate header file in @inc_dirs
116	    foreach (@inc_dirs) {
117		chdir $_;
118		last if -f $file;
119	    }
120	}
121
122	open(IN, "<", "$file") || (($Exit = 1),(warn "Can't open $file: $!\n"),next);
123	open(OUT, ">", "$Dest_dir/$outfile") || die "Can't create $outfile: $!\n";
124    }
125
126    print OUT
127        "require '_h2ph_pre.ph';\n\n",
128        "no warnings qw(redefine misc);\n\n";
129
130    while (defined (local $_ = next_line($file))) {
131	if (s/^\s*\#\s*//) {
132	    if (s/^define\s+(\w+)//) {
133		$name = $1;
134		$new = '';
135		s/\s+$//;
136		s/\(\w+\s*\(\*\)\s*\(\w*\)\)\s*(-?\d+)/$1/; # (int (*)(foo_t))0
137		if (s/^\(([\w,\s]*)\)//) {
138		    $args = $1;
139		    my $proto = '() ';
140		    if ($args ne '') {
141			$proto = '';
142			foreach my $arg (split(/,\s*/,$args)) {
143			    $arg =~ s/^\s*([^\s].*[^\s])\s*$/$1/;
144			    $curargs{$arg} = 1;
145			}
146			$args =~ s/\b(\w)/\$$1/g;
147			$args = "my($args) = \@_;\n$t    ";
148		    }
149		    s/^\s+//;
150		    expr();
151		    $new =~ s/(["\\])/\\$1/g;       #"]);
152		    EMIT($proto);
153		} else {
154		    s/^\s+//;
155		    expr();
156
157		    $new = 1 if $new eq '';
158
159		    # Shunt around such directives as '#define FOO FOO':
160		    next if $new =~ /^\s*&\Q$name\E\s*\z/;
161
162		    $new = reindent($new);
163		    $args = reindent($args);
164		    $new =~ s/(['\\])/\\$1/g;        #']);
165
166	    	    print OUT $t, 'eval ';
167		    if ($opt_h) {
168			print OUT "\"\\n#line $eval_index $outfile\\n\" . ";
169			$eval_index++;
170		    }
171		    print OUT "'sub $name () {$new;}' unless defined(&$name);\n";
172		}
173	    } elsif (/^(include|import|include_next)\s*([<\"])(.*)[>\"]/) {
174                $incl_type = $1;
175                $incl_quote = $2;
176                $incl = $3;
177                if (($incl_type eq 'include_next') ||
178                    ($opt_e && exists($bad_file{$incl}))) {
179                    $incl =~ s/\.h$/.ph/;
180		print OUT ($t,
181			   "eval {\n");
182                $tab += 4;
183                $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
184                    print OUT ($t, "my(\@REM);\n");
185                    if ($incl_type eq 'include_next') {
186		print OUT ($t,
187			   "my(\%INCD) = map { \$INC{\$_} => 1 } ",
188			           "(grep { \$_ eq \"$incl\" } ",
189                                   "keys(\%INC));\n");
190		print OUT ($t,
191			           "\@REM = map { \"\$_/$incl\" } ",
192			   "(grep { not exists(\$INCD{\"\$_/$incl\"})",
193			           " and -f \"\$_/$incl\" } \@INC);\n");
194                    } else {
195                        print OUT ($t,
196                                   "\@REM = map { \"\$_/$incl\" } ",
197                                   "(grep {-r \"\$_/$incl\" } \@INC);\n");
198                    }
199		print OUT ($t,
200			   "require \"\$REM[0]\" if \@REM;\n");
201                $tab -= 4;
202                $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
203                print OUT ($t,
204			   "};\n");
205		print OUT ($t,
206			   "warn(\$\@) if \$\@;\n");
207                } else {
208                    $incl =~ s/\.h$/.ph/;
209                    # copy the prefix in the quote syntax (#include "x.h") case
210                    if ($incl !~ m|/| && $incl_quote eq q{"} && $file =~ m|^(.*)/|) {
211                        $incl = "$1/$incl";
212                    }
213		    print OUT $t,"require '$incl';\n";
214                }
215	    } elsif (/^ifdef\s+(\w+)/) {
216		print OUT $t,"if(defined(&$1)) {\n";
217		$tab += 4;
218		$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
219	    } elsif (/^ifndef\s+(\w+)/) {
220		print OUT $t,"unless(defined(&$1)) {\n";
221		$tab += 4;
222		$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
223	    } elsif (s/^if\s+//) {
224		$new = '';
225		$inif = 1;
226		expr();
227		$inif = 0;
228		print OUT $t,"if($new) {\n";
229		$tab += 4;
230		$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
231	    } elsif (s/^elif\s+//) {
232		$new = '';
233		$inif = 1;
234		expr();
235		$inif = 0;
236		$tab -= 4;
237		$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
238		print OUT $t,"}\n elsif($new) {\n";
239		$tab += 4;
240		$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
241	    } elsif (/^else/) {
242		$tab -= 4;
243		$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
244		print OUT $t,"} else {\n";
245		$tab += 4;
246		$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
247	    } elsif (/^endif/) {
248		$tab -= 4;
249		$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
250		print OUT $t,"}\n";
251	    } elsif(/^undef\s+(\w+)/) {
252		print OUT $t, "undef(&$1) if defined(&$1);\n";
253	    } elsif(/^error\s+(".*")/) {
254		print OUT $t, "die($1);\n";
255	    } elsif(/^error\s+(.*)/) {
256		print OUT $t, "die(\"", quotemeta($1), "\");\n";
257	    } elsif(/^warning\s+(.*)/) {
258		print OUT $t, "warn(\"", quotemeta($1), "\");\n";
259	    } elsif(/^ident\s+(.*)/) {
260		print OUT $t, "# $1\n";
261	    }
262	} elsif (/^\s*(typedef\s*)?enum\s*(\s+[a-zA-Z_]\w*\s*)?/) { # { for vi
263	    until(/\{[^}]*\}.*;/ || /;/) {
264		last unless defined ($next = next_line($file));
265		chomp $next;
266		# drop "#define FOO FOO" in enums
267		$next =~ s/^\s*#\s*define\s+(\w+)\s+\1\s*$//;
268		# #defines in enums (aliases)
269		$next =~ s/^\s*#\s*define\s+(\w+)\s+(\w+)\s*$/$1 = $2,/;
270		$_ .= $next;
271		print OUT "# $next\n" if $opt_D;
272	    }
273	    s/#\s*if.*?#\s*endif//g; # drop #ifdefs
274	    s@/\*.*?\*/@@g;
275	    s/\s+/ /g;
276	    next unless /^\s?(typedef\s?)?enum\s?([a-zA-Z_]\w*)?\s?\{(.*)\}\s?([a-zA-Z_]\w*)?\s?;/;
277	    (my $enum_subs = $3) =~ s/\s//g;
278	    my @enum_subs = split(/,/, $enum_subs);
279	    my $enum_val = -1;
280	    foreach my $enum (@enum_subs) {
281		my ($enum_name, $enum_value) = $enum =~ /^([a-zA-Z_]\w*)(=.+)?$/;
282		$enum_name or next;
283		$enum_value =~ s/^=//;
284		$enum_val = (length($enum_value) ? $enum_value : $enum_val + 1);
285		if ($opt_h) {
286		    print OUT ($t,
287			       "eval(\"\\n#line $eval_index $outfile\\n",
288			       "sub $enum_name () \{ $enum_val; \}\") ",
289			       "unless defined(\&$enum_name);\n");
290		    ++ $eval_index;
291		} else {
292		    print OUT ($t,
293			       "eval(\"sub $enum_name () \{ $enum_val; \}\") ",
294			       "unless defined(\&$enum_name);\n");
295		}
296	    }
297	} elsif (/^(?:__extension__\s+)?(?:extern|static)\s+(?:__)?inline(?:__)?\s+/
298	    and !/;\s*$/ and !/{\s*}\s*$/)
299	{ # { for vi
300	    # This is a hack to parse the inline functions in the glibc headers.
301	    # Warning: massive kludge ahead. We suppose inline functions
302	    # are mainly constructed like macros.
303	    while (1) {
304		last unless defined ($next = next_line($file));
305		chomp $next;
306		undef $_, last if $next =~ /__THROW\s*;/
307			       or $next =~ /^(__extension__|extern|static)\b/;
308		$_ .= " $next";
309		print OUT "# $next\n" if $opt_D;
310		last if $next =~ /^}|^{.*}\s*$/;
311	    }
312	    next if not defined; # because it's only a prototype
313	    s/\b(__extension__|extern|static|(?:__)?inline(?:__)?)\b//g;
314	    # violently drop #ifdefs
315	    s/#\s*if.*?#\s*endif//g
316		and print OUT "# some #ifdef were dropped here -- fill in the blanks\n";
317	    if (s/^(?:\w|\s|\*)*\s(\w+)\s*//) {
318		$name = $1;
319	    } else {
320		warn "name not found"; next; # shouldn't occur...
321	    }
322	    my @args;
323	    if (s/^\(([^()]*)\)\s*(\w+\s*)*//) {
324		for my $arg (split /,/, $1) {
325		    if ($arg =~ /(\w+)\s*$/) {
326			$curargs{$1} = 1;
327			push @args, $1;
328		    }
329		}
330	    }
331	    $args = (
332		@args
333		? "my(" . (join ',', map "\$$_", @args) . ") = \@_;\n$t    "
334		: ""
335	    );
336	    my $proto = @args ? '' : '() ';
337	    $new = '';
338	    s/\breturn\b//g; # "return" doesn't occur in macros usually...
339	    expr();
340	    # try to find and perlify local C variables
341	    our @local_variables = (); # needs to be a our(): (?{...}) bug workaround
342	    {
343		use re "eval";
344		my $typelist = join '|', keys %isatype;
345		$new =~ s['
346		  (?:(?:__)?const(?:__)?\s+)?
347		  (?:(?:un)?signed\s+)?
348		  (?:long\s+)?
349		  (?:$typelist)\s+
350		  (\w+)
351		  (?{ push @local_variables, $1 })
352		  ']
353		 [my \$$1]gx;
354		$new =~ s['
355		  (?:(?:__)?const(?:__)?\s+)?
356		  (?:(?:un)?signed\s+)?
357		  (?:long\s+)?
358		  (?:$typelist)\s+
359		  ' \s+ &(\w+) \s* ;
360		  (?{ push @local_variables, $1 })
361		  ]
362		 [my \$$1;]gx;
363	     }
364	    $new =~ s/&$_\b/\$$_/g for @local_variables;
365	    $new =~ s/(["\\])/\\$1/g;       #"]);
366	    # now that's almost like a macro (we hope)
367	    EMIT($proto);
368	}
369    }
370    $Is_converted{$file} = 1;
371    if ($opt_e && exists($bad_file{$file})) {
372        unlink($Dest_dir . '/' . $outfile);
373        $next = '';
374    } else {
375        print OUT "1;\n";
376	queue_includes_from($file) if $opt_a;
377    }
378}
379
380if ($opt_e && (scalar(keys %bad_file) > 0)) {
381    warn "Was unable to convert the following files:\n";
382    warn "\t" . join("\n\t",sort(keys %bad_file)) . "\n";
383}
384
385exit $Exit;
386
387sub EMIT {
388    my $proto = shift;
389
390    $new = reindent($new);
391    $args = reindent($args);
392    if ($t ne '') {
393    $new =~ s/(['\\])/\\$1/g;   #']);
394    if ($opt_h) {
395        print OUT $t,
396                    "eval \"\\n#line $eval_index $outfile\\n\" . 'sub $name $proto\{\n$t    ${args}eval q($new);\n$t}' unless defined(\&$name);\n";
397                    $eval_index++;
398    } else {
399        print OUT $t,
400                    "eval 'sub $name $proto\{\n$t    ${args}eval q($new);\n$t}' unless defined(\&$name);\n";
401    }
402    } else {
403              print OUT "unless(defined(\&$name)) {\n    sub $name $proto\{\n\t${args}eval q($new);\n    }\n}\n";
404    }
405    %curargs = ();
406    return;
407}
408
409sub expr {
410    if (/\b__asm__\b/) {	# freak out
411	$new = '"(assembly code)"';
412	return
413    }
414    my $joined_args;
415    if(keys(%curargs)) {
416	$joined_args = join('|', keys(%curargs));
417    }
418    while ($_ ne '') {
419	s/^\&\&// && do { $new .= " &&"; next;}; # handle && operator
420	s/^\&([\(a-z\)]+)/$1/i;	# hack for things that take the address of
421	s/^(\s+)//		&& do {$new .= ' '; next;};
422	s/^0X([0-9A-F]+)[UL]*//i
423	    && do {my $hex = $1;
424		   $hex =~ s/^0+//;
425		   if (length $hex > 8 && !$Config{use64bitint}) {
426		       # Croak if nv_preserves_uv_bits < 64 ?
427		       $new .=         hex(substr($hex, -8)) +
428			       2**32 * hex(substr($hex,  0, -8));
429		       # The above will produce "erroneous" code
430		       # if the hex constant was e.g. inside UINT64_C
431		       # macro, but then again, h2ph is an approximation.
432		   } else {
433		       $new .= lc("0x$hex");
434		   }
435		   next;};
436	s/^(-?\d+\.\d+E[-+]?\d+)[FL]?//i	&& do {$new .= $1; next;};
437	s/^(\d+)\s*[LU]*//i	&& do {$new .= $1; next;};
438	s/^("(\\"|[^"])*")//	&& do {$new .= $1; next;};
439	s/^'((\\"|[^"])*)'//	&& do {
440	    if ($curargs{$1}) {
441		$new .= "ord('\$$1')";
442	    } else {
443		$new .= "ord('$1')";
444	    }
445	    next;
446	};
447        # replace "sizeof(foo)" with "{foo}"
448        # also, remove * (C dereference operator) to avoid perl syntax
449        # problems.  Where the %sizeof array comes from is anyone's
450        # guess (c2ph?), but this at least avoids fatal syntax errors.
451        # Behavior is undefined if sizeof() delimiters are unbalanced.
452        # This code was modified to able to handle constructs like this:
453        #   sizeof(*(p)), which appear in the HP-UX 10.01 header files.
454        s/^sizeof\s*\(// && do {
455            $new .= '$sizeof';
456            my $lvl = 1;  # already saw one open paren
457            # tack { on the front, and skip it in the loop
458            $_ = "{" . "$_";
459            my $index = 1;
460            # find balanced closing paren
461            while ($index <= length($_) && $lvl > 0) {
462                $lvl++ if substr($_, $index, 1) eq "(";
463                $lvl-- if substr($_, $index, 1) eq ")";
464                $index++;
465            }
466            # tack } on the end, replacing )
467            substr($_, $index - 1, 1) = "}";
468            # remove pesky * operators within the sizeof argument
469            substr($_, 0, $index - 1) =~ s/\*//g;
470            next;
471        };
472	# Eliminate typedefs
473	/\(([\w\s]+)[\*\s]*\)\s*[\w\(]/ && do {
474	    my $doit = 1;
475	    foreach (split /\s+/, $1) {  # Make sure all the words are types,
476	        unless($isatype{$_} or $_ eq 'struct' or $_ eq 'union'){
477		    $doit = 0;
478		    last;
479		}
480	    }
481	    if( $doit ){
482		s/\([\w\s]+[\*\s]*\)// && next;      # then eliminate them.
483	    }
484	};
485	# struct/union member, including arrays:
486	s/^([_A-Z]\w*(\[[^\]]+\])?((\.|->)[_A-Z]\w*(\[[^\]]+\])?)+)//i && do {
487	    my $id = $1;
488	    $id =~ s/(\.|(->))([^\.\-]*)/->\{$3\}/g;
489	    $id =~ s/\b([^\$])($joined_args)/$1\$$2/g if length($joined_args);
490	    while($id =~ /\[\s*([^\$\&\d\]]+)\]/) {
491		my($index) = $1;
492		$index =~ s/\s//g;
493		if(exists($curargs{$index})) {
494		    $index = "\$$index";
495		} else {
496		    $index = "&$index";
497		}
498		$id =~ s/\[\s*([^\$\&\d\]]+)\]/[$index]/;
499	    }
500	    $new .= " (\$$id)";
501	};
502	s/^([_a-zA-Z]\w*)//	&& do {
503	    my $id = $1;
504	    if ($id eq 'struct' || $id eq 'union') {
505		s/^\s+(\w+)//;
506		$id .= ' ' . $1;
507		$isatype{$id} = 1;
508	    } elsif ($id =~ /^((un)?signed)|(long)|(short)$/) {
509		while (s/^\s+(\w+)//) { $id .= ' ' . $1; }
510		$isatype{$id} = 1;
511	    }
512	    if ($curargs{$id}) {
513		$new .= "\$$id";
514		$new .= '->' if /^[\[\{]/;
515	    } elsif ($id eq 'defined') {
516		$new .= 'defined';
517	    } elsif (/^\s*\(/) {
518		s/^\s*\((\w),/("$1",/ if $id =~ /^_IO[WR]*$/i;	# cheat
519		$new .= " &$id";
520	    } elsif ($isatype{$id}) {
521		if ($new =~ /\{\s*$/) {
522		    $new .= "'$id'";
523		} elsif ($new =~ /\(\s*$/ && /^[\s*]*\)/) {
524		    $new =~ s/\(\s*$//;
525		    s/^[\s*]*\)//;
526		} else {
527		    $new .= q(').$id.q(');
528		}
529	    } else {
530		if ($inif) {
531		    if ($new =~ /defined\s*$/) {
532			$new .= '(&' . $id . ')';
533		    } elsif ($new =~ /defined\s*\($/) {
534			$new .= '&' . $id;
535		    } else {
536			$new .= '(defined(&' . $id . ') ? &' . $id . ' : undef)';
537		    }
538		} elsif (/^\[/) {
539		    $new .= " \$$id";
540		} else {
541		    $new .= ' &' . $id;
542		}
543	    }
544	    next;
545	};
546	s/^(.)// && do { if ($1 ne '#') { $new .= $1; } next;};
547    }
548}
549
550
551sub next_line
552{
553    my $file = shift;
554    my ($in, $out);
555    my $pre_sub_tri_graphs = 1;
556
557    READ: while (not eof IN) {
558        $in  .= <IN>;
559        chomp $in;
560        next unless length $in;
561
562        while (length $in) {
563            if ($pre_sub_tri_graphs) {
564                # Preprocess all tri-graphs
565                # including things stuck in quoted string constants.
566                $in =~ s/\?\?=/#/g;                         # | ??=|  #|
567                $in =~ s/\?\?\!/|/g;                        # | ??!|  ||
568                $in =~ s/\?\?'/^/g;                         # | ??'|  ^|
569                $in =~ s/\?\?\(/[/g;                        # | ??(|  [|
570                $in =~ s/\?\?\)/]/g;                        # | ??)|  ]|
571                $in =~ s/\?\?\-/~/g;                        # | ??-|  ~|
572                $in =~ s/\?\?\//\\/g;                       # | ??/|  \|
573                $in =~ s/\?\?</{/g;                         # | ??<|  {|
574                $in =~ s/\?\?>/}/g;                         # | ??>|  }|
575            }
576	    if ($in =~ s/^\#ifdef __LANGUAGE_PASCAL__//) {
577		# Tru64 disassembler.h evilness: mixed C and Pascal.
578		while (<IN>) {
579		    last if /^\#endif/;
580		}
581		$in = "";
582		next READ;
583	    }
584	    # Skip inlined functions in headers
585	    if ($in =~ s/^(extern|static) (__inline__|inline) .*[^;]\s*$//) {
586		while (<IN>) {
587		    last if /^}/;
588		}
589		$in = "";
590		next READ;
591	    }
592            if ($in =~ s/\\$//) {                           # \-newline
593                $out    .= ' ';
594                next READ;
595            } elsif ($in =~ s/^([^"'\\\/]+)//) {            # Passthrough
596                $out    .= $1;
597            } elsif ($in =~ s/^(\\.)//) {                   # \...
598                $out    .= $1;
599            } elsif ($in =~ /^'/) {                         # '...
600                if ($in =~ s/^('(\\.|[^'\\])*')//) {
601                    $out    .= $1;
602                } else {
603                    next READ;
604                }
605            } elsif ($in =~ /^"/) {                         # "...
606                if ($in =~ s/^("(\\.|[^"\\])*")//) {
607                    $out    .= $1;
608                } else {
609                    next READ;
610                }
611            } elsif ($in =~ s/^\/\/.*//) {                  # //...
612                # fall through
613            } elsif ($in =~ m/^\/\*/) {                     # /*...
614                # C comment removal adapted from perlfaq6:
615                if ($in =~ s/^\/\*[^*]*\*+([^\/*][^*]*\*+)*\///) {
616                    $out    .= ' ';
617                } else {                                    # Incomplete /* */
618                    next READ;
619                }
620            } elsif ($in =~ s/^(\/)//) {                    # /...
621                $out    .= $1;
622            } elsif ($in =~ s/^([^\'\"\\\/]+)//) {
623                $out    .= $1;
624            } elsif ($^O eq 'linux' &&
625                     $file =~ m!(?:^|/)linux/byteorder/pdp_endian\.h$! &&
626                     $in   =~ s!\'T KNOW!!) {
627                $out    =~ s!I DON$!I_DO_NOT_KNOW!;
628            } else {
629                if ($opt_e) {
630                    warn "Cannot parse $file:\n$in\n";
631                    $bad_file{$file} = 1;
632                    $in = '';
633                    $out = undef;
634                    last READ;
635                } else {
636		die "Cannot parse:\n$in\n";
637                }
638            }
639        }
640
641        last READ if $out =~ /\S/;
642    }
643
644    return $out;
645}
646
647
648# Handle recursive subdirectories without getting a grotesquely big stack.
649# Could this be implemented using File::Find?
650sub next_file
651{
652    my $file;
653
654    while (@ARGV) {
655        $file = shift @ARGV;
656
657        if ($file eq '-' or -f $file or -l $file) {
658            return $file;
659        } elsif (-d $file) {
660            if ($opt_r) {
661                expand_glob($file);
662            } else {
663                print STDERR "Skipping directory '$file'\n";
664            }
665        } elsif ($opt_a) {
666            return $file;
667        } else {
668            print STDERR "Skipping '$file':  not a file or directory\n";
669        }
670    }
671
672    return undef;
673}
674
675
676# Put all the files in $directory into @ARGV for processing.
677sub expand_glob
678{
679    my ($directory)  = @_;
680
681    $directory =~ s:/$::;
682
683    opendir DIR, $directory;
684        foreach (readdir DIR) {
685            next if ($_ eq '.' or $_ eq '..');
686
687            # expand_glob() is going to be called until $ARGV[0] isn't a
688            # directory; so push directories, and unshift everything else.
689            if (-d "$directory/$_") { push    @ARGV, "$directory/$_" }
690            else                    { unshift @ARGV, "$directory/$_" }
691        }
692    closedir DIR;
693}
694
695
696# Given $file, a symbolic link to a directory in the C include directory,
697# make an equivalent symbolic link in $Dest_dir, if we can figure out how.
698# Otherwise, just duplicate the file or directory.
699sub link_if_possible
700{
701    my ($dirlink)  = @_;
702    my $target  = eval 'readlink($dirlink)';
703
704    if ($target =~ m:^\.\./: or $target =~ m:^/:) {
705        # The target of a parent or absolute link could leave the $Dest_dir
706        # hierarchy, so let's put all of the contents of $dirlink (actually,
707        # the contents of $target) into @ARGV; as a side effect down the
708        # line, $dirlink will get created as an _actual_ directory.
709        expand_glob($dirlink);
710    } else {
711        if (-l "$Dest_dir/$dirlink") {
712            unlink "$Dest_dir/$dirlink" or
713                print STDERR "Could not remove link $Dest_dir/$dirlink:  $!\n";
714        }
715
716        if (eval 'symlink($target, "$Dest_dir/$dirlink")') {
717            print "Linking $target -> $Dest_dir/$dirlink\n";
718
719            # Make sure that the link _links_ to something:
720            if (! -e "$Dest_dir/$target") {
721                mkpath("$Dest_dir/$target", 0755) or
722                    print STDERR "Could not create $Dest_dir/$target/\n";
723            }
724        } else {
725            print STDERR "Could not symlink $target -> $Dest_dir/$dirlink:  $!\n";
726        }
727    }
728}
729
730
731# Push all #included files in $file onto our stack, except for STDIN
732# and files we've already processed.
733sub queue_includes_from
734{
735    my ($file)    = @_;
736    my $line;
737
738    return if ($file eq "-");
739
740    open HEADER, "<", $file or return;
741        while (defined($line = <HEADER>)) {
742            while (/\\$/) { # Handle continuation lines
743                chop $line;
744                $line .= <HEADER>;
745            }
746
747            if ($line =~ /^#\s*include\s+([<"])(.*?)[>"]/) {
748                my ($delimiter, $new_file) = ($1, $2);
749                # copy the prefix in the quote syntax (#include "x.h") case
750                if ($delimiter eq q{"} && $file =~ m|^(.*)/|) {
751                    $new_file = "$1/$new_file";
752                }
753                push(@ARGV, $new_file) unless $Is_converted{$new_file};
754            }
755        }
756    close HEADER;
757}
758
759
760# Determine include directories; $Config{usrinc} should be enough for (all
761# non-GCC?) C compilers, but gcc uses additional include directories.
762sub inc_dirs
763{
764    my $from_gcc   = `LC_ALL=C $Config{cc} -v -E - < /dev/null 2>&1 | awk '/^#include/, /^End of search list/' | grep '^ '`;
765    length($from_gcc) ? (split(' ', $from_gcc), $Config{usrinc}) : ($Config{usrinc});
766}
767
768
769# Create "_h2ph_pre.ph", if it doesn't exist or was built by a different
770# version of h2ph.
771sub build_preamble_if_necessary
772{
773    # Increment $VERSION every time this function is modified:
774    my $VERSION     = 4;
775    my $preamble    = "$Dest_dir/_h2ph_pre.ph";
776
777    # Can we skip building the preamble file?
778    if (-r $preamble) {
779        # Extract version number from first line of preamble:
780        open  PREAMBLE, "<", $preamble or die "Cannot open $preamble:  $!";
781            my $line = <PREAMBLE>;
782            $line =~ /(\b\d+\b)/;
783        close PREAMBLE            or die "Cannot close $preamble:  $!";
784
785        # Don't build preamble if a compatible preamble exists:
786        return if $1 == $VERSION;
787    }
788
789    my (%define) = _extract_cc_defines();
790
791    open  PREAMBLE, ">", $preamble or die "Cannot open $preamble:  $!";
792	print PREAMBLE "# This file was created by h2ph version $VERSION\n";
793        # Prevent non-portable hex constants from warning.
794        #
795        # We still produce an overflow warning if we can't represent
796        # a hex constant as an integer.
797        print PREAMBLE "no warnings qw(portable);\n";
798
799	foreach (sort keys %define) {
800	    if ($opt_D) {
801		print PREAMBLE "# $_=$define{$_}\n";
802	    }
803	    if ($define{$_} =~ /^\((.*)\)$/) {
804		# parenthesized value:  d=(v)
805		$define{$_} = $1;
806	    }
807	    if (/^(\w+)\((\w)\)$/) {
808		my($macro, $arg) = ($1, $2);
809		my $def = $define{$_};
810		$def =~ s/$arg/\$\{$arg\}/g;
811		print PREAMBLE <<DEFINE;
812unless (defined &$macro) { sub $macro(\$) { my (\$$arg) = \@_; \"$def\" } }
813
814DEFINE
815	    } elsif
816		($define{$_} =~ /^([+-]?(\d+)?\.\d+([eE][+-]?\d+)?)[FL]?$/) {
817		# float:
818		print PREAMBLE
819		    "unless (defined &$_) { sub $_() { $1 } }\n\n";
820	    } elsif ($define{$_} =~ /^([+-]?\d+)U?L{0,2}$/i) {
821		# integer:
822		print PREAMBLE
823		    "unless (defined &$_) { sub $_() { $1 } }\n\n";
824            } elsif ($define{$_} =~ /^([+-]?0x[\da-f]+)U?L{0,2}$/i) {
825                # hex integer
826                # Special cased, since perl warns on hex integers
827                # that can't be represented in a UV.
828                #
829                # This way we get the warning at time of use, so the user
830                # only gets the warning if they happen to use this
831                # platform-specific definition.
832                my $code = $1;
833                $code = "hex('$code')" if length $code > 10;
834                print PREAMBLE
835                    "unless (defined &$_) { sub $_() { $code } }\n\n";
836	    } elsif ($define{$_} =~ /^\w+$/) {
837		my $def = $define{$_};
838		if ($isatype{$def}) {
839		  print PREAMBLE
840		    "unless (defined &$_) { sub $_() { \"$def\" } }\n\n";
841		} else {
842		  print PREAMBLE
843		    "unless (defined &$_) { sub $_() { &$def } }\n\n";
844	        }
845	    } else {
846		print PREAMBLE
847		    "unless (defined &$_) { sub $_() { \"",
848		    quotemeta($define{$_}), "\" } }\n\n";
849	    }
850	}
851	print PREAMBLE "\n1;\n";  # avoid 'did not return a true value' when empty
852    close PREAMBLE               or die "Cannot close $preamble:  $!";
853}
854
855
856# %Config contains information on macros that are pre-defined by the
857# system's compiler.  We need this information to make the .ph files
858# function with perl as the .h files do with cc.
859sub _extract_cc_defines
860{
861    my %define;
862    my $allsymbols  = join " ",
863	@Config{'ccsymbols', 'cppsymbols', 'cppccsymbols'};
864
865    # Split compiler pre-definitions into 'key=value' pairs:
866    while ($allsymbols =~ /([^\s]+)=((\\\s|[^\s])+)/g) {
867	$define{$1} = $2;
868	if ($opt_D) {
869	    print STDERR "$_:  $1 -> $2\n";
870	}
871    }
872
873    return %define;
874}
875
876
8771;
878
879##############################################################################
880__END__
881
882=head1 NAME
883
884h2ph - convert .h C header files to .ph Perl header files
885
886=head1 SYNOPSIS
887
888B<h2ph [-d destination directory] [-r | -a] [-l] [-h] [-e] [-D] [-Q]
889[headerfiles]>
890
891=head1 DESCRIPTION
892
893I<h2ph>
894converts any C header files specified to the corresponding Perl header file
895format.
896It is most easily run while in /usr/include:
897
898	cd /usr/include; h2ph * sys/*
899
900or
901
902	cd /usr/include; h2ph * sys/* arpa/* netinet/*
903
904or
905
906	cd /usr/include; h2ph -r -l .
907
908The output files are placed in the hierarchy rooted at Perl's
909architecture dependent library directory.  You can specify a different
910hierarchy with a B<-d> switch.
911
912If run with no arguments, filters standard input to standard output.
913
914=head1 OPTIONS
915
916=over 4
917
918=item -d destination_dir
919
920Put the resulting B<.ph> files beneath B<destination_dir>, instead of
921beneath the default Perl library location (C<$Config{'installsitearch'}>).
922
923=item -r
924
925Run recursively; if any of B<headerfiles> are directories, then run I<h2ph>
926on all files in those directories (and their subdirectories, etc.).  B<-r>
927and B<-a> are mutually exclusive.
928
929=item -a
930
931Run automagically; convert B<headerfiles>, as well as any B<.h> files
932which they include.  This option will search for B<.h> files in all
933directories which your C compiler ordinarily uses.  B<-a> and B<-r> are
934mutually exclusive.
935
936=item -l
937
938Symbolic links will be replicated in the destination directory.  If B<-l>
939is not specified, then links are skipped over.
940
941=item -h
942
943Put 'hints' in the .ph files which will help in locating problems with
944I<h2ph>.  In those cases when you B<require> a B<.ph> file containing syntax
945errors, instead of the cryptic
946
947	[ some error condition ] at (eval mmm) line nnn
948
949you will see the slightly more helpful
950
951	[ some error condition ] at filename.ph line nnn
952
953However, the B<.ph> files almost double in size when built using B<-h>.
954
955=item -e
956
957If an error is encountered during conversion, output file will be removed and
958a warning emitted instead of terminating the conversion immediately.
959
960=item -D
961
962Include the code from the B<.h> file as a comment in the B<.ph> file.
963This is primarily used for debugging I<h2ph>.
964
965=item -Q
966
967'Quiet' mode; don't print out the names of the files being converted.
968
969=back
970
971=head1 ENVIRONMENT
972
973No environment variables are used.
974
975=head1 FILES
976
977 /usr/include/*.h
978 /usr/include/sys/*.h
979
980etc.
981
982=head1 AUTHOR
983
984Larry Wall
985
986=head1 SEE ALSO
987
988perl(1)
989
990=head1 DIAGNOSTICS
991
992The usual warnings if it can't read or write the files involved.
993
994=head1 BUGS
995
996Doesn't construct the %sizeof array for you.
997
998It doesn't handle all C constructs, but it does attempt to isolate
999definitions inside evals so that you can get at the definitions
1000that it can translate.
1001
1002It's only intended as a rough tool.
1003You may need to dicker with the files produced.
1004
1005You have to run this program by hand; it's not run as part of the Perl
1006installation.
1007
1008Doesn't handle complicated expressions built piecemeal, a la:
1009
1010    enum {
1011	FIRST_VALUE,
1012	SECOND_VALUE,
1013    #ifdef ABC
1014	THIRD_VALUE
1015    #endif
1016    };
1017
1018Doesn't necessarily locate all of your C compiler's internally-defined
1019symbols.
1020
1021=cut
1022
1023!NO!SUBS!
1024
1025close OUT or die "Can't close $file: $!";
1026chmod 0755, $file or die "Can't reset permissions for $file: $!\n";
1027exec("$Config{'eunicefix'} $file") if $Config{'eunicefix'} ne ':';
1028chdir $origdir;
1029