xref: /openbsd-src/gnu/usr.bin/perl/utils/h2ph.PL (revision d13be5d47e4149db2549a9828e244d59dbc43f15)
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 $Config{perlpath} -S \$0 \${1+"\$@"}'
32	if \$running_under_some_shell;
33!GROK!THIS!
34
35# In the following, perl variables are not expanded during extraction.
36
37print OUT <<'!NO!SUBS!';
38
39use strict;
40
41use Config;
42use File::Path qw(mkpath);
43use Getopt::Std;
44
45# Make sure read permissions for all are set:
46if (defined umask && (umask() & 0444)) {
47    umask (umask() & ~0444);
48}
49
50getopts('Dd:rlhaQe');
51use vars qw($opt_D $opt_d $opt_r $opt_l $opt_h $opt_a $opt_Q $opt_e);
52die "-r and -a options are mutually exclusive\n" if ($opt_r and $opt_a);
53my @inc_dirs = inc_dirs() if $opt_a;
54
55my $Exit = 0;
56
57my $Dest_dir = $opt_d || $Config{installsitearch};
58die "Destination directory $Dest_dir doesn't exist or isn't a directory\n"
59    unless -d $Dest_dir;
60
61my @isatype = qw(
62	char	uchar	u_char
63	short	ushort	u_short
64	int	uint	u_int
65	long	ulong	u_long
66	FILE	key_t	caddr_t
67	float	double	size_t
68);
69
70my %isatype;
71@isatype{@isatype} = (1) x @isatype;
72my $inif = 0;
73my %Is_converted;
74my %bad_file = ();
75
76@ARGV = ('-') unless @ARGV;
77
78build_preamble_if_necessary();
79
80sub reindent($) {
81    my($text) = shift;
82    $text =~ s/\n/\n    /g;
83    $text =~ s/        /\t/g;
84    $text;
85}
86
87my ($t, $tab, %curargs, $new, $eval_index, $dir, $name, $args, $outfile);
88my ($incl, $incl_type, $incl_quote, $next);
89while (defined (my $file = next_file())) {
90    if (-l $file and -d $file) {
91        link_if_possible($file) if ($opt_l);
92        next;
93    }
94
95    # Recover from header files with unbalanced cpp directives
96    $t = '';
97    $tab = 0;
98
99    # $eval_index goes into ``#line'' directives, to help locate syntax errors:
100    $eval_index = 1;
101
102    if ($file eq '-') {
103	open(IN, "-");
104	open(OUT, ">-");
105    } else {
106	($outfile = $file) =~ s/\.h$/.ph/ || next;
107	print "$file -> $outfile\n" unless $opt_Q;
108	if ($file =~ m|^(.*)/|) {
109	    $dir = $1;
110	    mkpath "$Dest_dir/$dir";
111	}
112
113	if ($opt_a) { # automagic mode:  locate header file in @inc_dirs
114	    foreach (@inc_dirs) {
115		chdir $_;
116		last if -f $file;
117	    }
118	}
119
120	open(IN,"$file") || (($Exit = 1),(warn "Can't open $file: $!\n"),next);
121	open(OUT,">$Dest_dir/$outfile") || die "Can't create $outfile: $!\n";
122    }
123
124    print OUT
125        "require '_h2ph_pre.ph';\n\n",
126        "no warnings qw(redefine misc);\n\n";
127
128    while (defined (local $_ = next_line($file))) {
129	if (s/^\s*\#\s*//) {
130	    if (s/^define\s+(\w+)//) {
131		$name = $1;
132		$new = '';
133		s/\s+$//;
134		s/\(\w+\s*\(\*\)\s*\(\w*\)\)\s*(-?\d+)/$1/; # (int (*)(foo_t))0
135		if (s/^\(([\w,\s]*)\)//) {
136		    $args = $1;
137		    my $proto = '() ';
138		    if ($args ne '') {
139			$proto = '';
140			foreach my $arg (split(/,\s*/,$args)) {
141			    $arg =~ s/^\s*([^\s].*[^\s])\s*$/$1/;
142			    $curargs{$arg} = 1;
143			}
144			$args =~ s/\b(\w)/\$$1/g;
145			$args = "my($args) = \@_;\n$t    ";
146		    }
147		    s/^\s+//;
148		    expr();
149		    $new =~ s/(["\\])/\\$1/g;       #"]);
150		    EMIT($proto);
151		} else {
152		    s/^\s+//;
153		    expr();
154		    $new = 1 if $new eq '';
155		    $new = reindent($new);
156		    $args = reindent($args);
157		    if ($t ne '') {
158			$new =~ s/(['\\])/\\$1/g;        #']);
159
160			if ($opt_h) {
161			    print OUT $t,"eval \"\\n#line $eval_index $outfile\\n\" . 'sub $name () {",$new,";}' unless defined(\&$name);\n";
162			    $eval_index++;
163			} else {
164			    print OUT $t,"eval 'sub $name () {",$new,";}' unless defined(\&$name);\n";
165			}
166		    } else {
167		    	# Shunt around such directives as `#define FOO FOO':
168		    	next if " \&$name" eq $new;
169
170                      print OUT $t,"unless(defined(\&$name)) {\n    sub $name () {\t",$new,";}\n}\n";
171		    }
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    $new = '"(assembly code)"' and return if /\b__asm__\b/; # freak out.
411    my $joined_args;
412    if(keys(%curargs)) {
413	$joined_args = join('|', keys(%curargs));
414    }
415    while ($_ ne '') {
416	s/^\&\&// && do { $new .= " &&"; next;}; # handle && operator
417	s/^\&([\(a-z\)]+)/$1/i;	# hack for things that take the address of
418	s/^(\s+)//		&& do {$new .= ' '; next;};
419	s/^0X([0-9A-F]+)[UL]*//i
420	    && do {my $hex = $1;
421		   $hex =~ s/^0+//;
422		   if (length $hex > 8 && !$Config{use64bitint}) {
423		       # Croak if nv_preserves_uv_bits < 64 ?
424		       $new .=         hex(substr($hex, -8)) +
425			       2**32 * hex(substr($hex,  0, -8));
426		       # The above will produce "errorneus" code
427		       # if the hex constant was e.g. inside UINT64_C
428		       # macro, but then again, h2ph is an approximation.
429		   } else {
430		       $new .= lc("0x$hex");
431		   }
432		   next;};
433	s/^(-?\d+\.\d+E[-+]?\d+)[FL]?//i	&& do {$new .= $1; next;};
434	s/^(\d+)\s*[LU]*//i	&& do {$new .= $1; next;};
435	s/^("(\\"|[^"])*")//	&& do {$new .= $1; next;};
436	s/^'((\\"|[^"])*)'//	&& do {
437	    if ($curargs{$1}) {
438		$new .= "ord('\$$1')";
439	    } else {
440		$new .= "ord('$1')";
441	    }
442	    next;
443	};
444        # replace "sizeof(foo)" with "{foo}"
445        # also, remove * (C dereference operator) to avoid perl syntax
446        # problems.  Where the %sizeof array comes from is anyone's
447        # guess (c2ph?), but this at least avoids fatal syntax errors.
448        # Behavior is undefined if sizeof() delimiters are unbalanced.
449        # This code was modified to able to handle constructs like this:
450        #   sizeof(*(p)), which appear in the HP-UX 10.01 header files.
451        s/^sizeof\s*\(// && do {
452            $new .= '$sizeof';
453            my $lvl = 1;  # already saw one open paren
454            # tack { on the front, and skip it in the loop
455            $_ = "{" . "$_";
456            my $index = 1;
457            # find balanced closing paren
458            while ($index <= length($_) && $lvl > 0) {
459                $lvl++ if substr($_, $index, 1) eq "(";
460                $lvl-- if substr($_, $index, 1) eq ")";
461                $index++;
462            }
463            # tack } on the end, replacing )
464            substr($_, $index - 1, 1) = "}";
465            # remove pesky * operators within the sizeof argument
466            substr($_, 0, $index - 1) =~ s/\*//g;
467            next;
468        };
469	# Eliminate typedefs
470	/\(([\w\s]+)[\*\s]*\)\s*[\w\(]/ && do {
471	    my $doit = 1;
472	    foreach (split /\s+/, $1) {  # Make sure all the words are types,
473	        unless($isatype{$_} or $_ eq 'struct' or $_ eq 'union'){
474		    $doit = 0;
475		    last;
476		}
477	    }
478	    if( $doit ){
479		s/\([\w\s]+[\*\s]*\)// && next;      # then eliminate them.
480	    }
481	};
482	# struct/union member, including arrays:
483	s/^([_A-Z]\w*(\[[^\]]+\])?((\.|->)[_A-Z]\w*(\[[^\]]+\])?)+)//i && do {
484	    my $id = $1;
485	    $id =~ s/(\.|(->))([^\.\-]*)/->\{$3\}/g;
486	    $id =~ s/\b([^\$])($joined_args)/$1\$$2/g if length($joined_args);
487	    while($id =~ /\[\s*([^\$\&\d\]]+)\]/) {
488		my($index) = $1;
489		$index =~ s/\s//g;
490		if(exists($curargs{$index})) {
491		    $index = "\$$index";
492		} else {
493		    $index = "&$index";
494		}
495		$id =~ s/\[\s*([^\$\&\d\]]+)\]/[$index]/;
496	    }
497	    $new .= " (\$$id)";
498	};
499	s/^([_a-zA-Z]\w*)//	&& do {
500	    my $id = $1;
501	    if ($id eq 'struct' || $id eq 'union') {
502		s/^\s+(\w+)//;
503		$id .= ' ' . $1;
504		$isatype{$id} = 1;
505	    } elsif ($id =~ /^((un)?signed)|(long)|(short)$/) {
506		while (s/^\s+(\w+)//) { $id .= ' ' . $1; }
507		$isatype{$id} = 1;
508	    }
509	    if ($curargs{$id}) {
510		$new .= "\$$id";
511		$new .= '->' if /^[\[\{]/;
512	    } elsif ($id eq 'defined') {
513		$new .= 'defined';
514	    } elsif (/^\s*\(/) {
515		s/^\s*\((\w),/("$1",/ if $id =~ /^_IO[WR]*$/i;	# cheat
516		$new .= " &$id";
517	    } elsif ($isatype{$id}) {
518		if ($new =~ /\{\s*$/) {
519		    $new .= "'$id'";
520		} elsif ($new =~ /\(\s*$/ && /^[\s*]*\)/) {
521		    $new =~ s/\(\s*$//;
522		    s/^[\s*]*\)//;
523		} else {
524		    $new .= q(').$id.q(');
525		}
526	    } else {
527		if ($inif) {
528		    if ($new =~ /defined\s*$/) {
529			$new .= '(&' . $id . ')';
530		    } elsif ($new =~ /defined\s*\($/) {
531			$new .= '&' . $id;
532		    } else {
533			$new .= '(defined(&' . $id . ') ? &' . $id . ' : undef)';
534		    }
535		} elsif (/^\[/) {
536		    $new .= " \$$id";
537		} else {
538		    $new .= ' &' . $id;
539		}
540	    }
541	    next;
542	};
543	s/^(.)// && do { if ($1 ne '#') { $new .= $1; } next;};
544    }
545}
546
547
548sub next_line
549{
550    my $file = shift;
551    my ($in, $out);
552    my $pre_sub_tri_graphs = 1;
553
554    READ: while (not eof IN) {
555        $in  .= <IN>;
556        chomp $in;
557        next unless length $in;
558
559        while (length $in) {
560            if ($pre_sub_tri_graphs) {
561                # Preprocess all tri-graphs
562                # including things stuck in quoted string constants.
563                $in =~ s/\?\?=/#/g;                         # | ??=|  #|
564                $in =~ s/\?\?\!/|/g;                        # | ??!|  ||
565                $in =~ s/\?\?'/^/g;                         # | ??'|  ^|
566                $in =~ s/\?\?\(/[/g;                        # | ??(|  [|
567                $in =~ s/\?\?\)/]/g;                        # | ??)|  ]|
568                $in =~ s/\?\?\-/~/g;                        # | ??-|  ~|
569                $in =~ s/\?\?\//\\/g;                       # | ??/|  \|
570                $in =~ s/\?\?</{/g;                         # | ??<|  {|
571                $in =~ s/\?\?>/}/g;                         # | ??>|  }|
572            }
573	    if ($in =~ s/^\#ifdef __LANGUAGE_PASCAL__//) {
574		# Tru64 disassembler.h evilness: mixed C and Pascal.
575		while (<IN>) {
576		    last if /^\#endif/;
577		}
578		$in = "";
579		next READ;
580	    }
581	    # Skip inlined functions in headers
582	    if ($in =~ s/^(extern|static) (__inline__|inline) .*[^;]\s*$//) {
583		while (<IN>) {
584		    last if /^}/;
585		}
586		$in = "";
587		next READ;
588	    }
589            if ($in =~ s/\\$//) {                           # \-newline
590                $out    .= ' ';
591                next READ;
592            } elsif ($in =~ s/^([^"'\\\/]+)//) {            # Passthrough
593                $out    .= $1;
594            } elsif ($in =~ s/^(\\.)//) {                   # \...
595                $out    .= $1;
596            } elsif ($in =~ /^'/) {                         # '...
597                if ($in =~ s/^('(\\.|[^'\\])*')//) {
598                    $out    .= $1;
599                } else {
600                    next READ;
601                }
602            } elsif ($in =~ /^"/) {                         # "...
603                if ($in =~ s/^("(\\.|[^"\\])*")//) {
604                    $out    .= $1;
605                } else {
606                    next READ;
607                }
608            } elsif ($in =~ s/^\/\/.*//) {                  # //...
609                # fall through
610            } elsif ($in =~ m/^\/\*/) {                     # /*...
611                # C comment removal adapted from perlfaq6:
612                if ($in =~ s/^\/\*[^*]*\*+([^\/*][^*]*\*+)*\///) {
613                    $out    .= ' ';
614                } else {                                    # Incomplete /* */
615                    next READ;
616                }
617            } elsif ($in =~ s/^(\/)//) {                    # /...
618                $out    .= $1;
619            } elsif ($in =~ s/^([^\'\"\\\/]+)//) {
620                $out    .= $1;
621            } elsif ($^O eq 'linux' &&
622                     $file =~ m!(?:^|/)linux/byteorder/pdp_endian\.h$! &&
623                     $in   =~ s!\'T KNOW!!) {
624                $out    =~ s!I DON$!I_DO_NOT_KNOW!;
625            } else {
626                if ($opt_e) {
627                    warn "Cannot parse $file:\n$in\n";
628                    $bad_file{$file} = 1;
629                    $in = '';
630                    $out = undef;
631                    last READ;
632                } else {
633		die "Cannot parse:\n$in\n";
634                }
635            }
636        }
637
638        last READ if $out =~ /\S/;
639    }
640
641    return $out;
642}
643
644
645# Handle recursive subdirectories without getting a grotesquely big stack.
646# Could this be implemented using File::Find?
647sub next_file
648{
649    my $file;
650
651    while (@ARGV) {
652        $file = shift @ARGV;
653
654        if ($file eq '-' or -f $file or -l $file) {
655            return $file;
656        } elsif (-d $file) {
657            if ($opt_r) {
658                expand_glob($file);
659            } else {
660                print STDERR "Skipping directory `$file'\n";
661            }
662        } elsif ($opt_a) {
663            return $file;
664        } else {
665            print STDERR "Skipping `$file':  not a file or directory\n";
666        }
667    }
668
669    return undef;
670}
671
672
673# Put all the files in $directory into @ARGV for processing.
674sub expand_glob
675{
676    my ($directory)  = @_;
677
678    $directory =~ s:/$::;
679
680    opendir DIR, $directory;
681        foreach (readdir DIR) {
682            next if ($_ eq '.' or $_ eq '..');
683
684            # expand_glob() is going to be called until $ARGV[0] isn't a
685            # directory; so push directories, and unshift everything else.
686            if (-d "$directory/$_") { push    @ARGV, "$directory/$_" }
687            else                    { unshift @ARGV, "$directory/$_" }
688        }
689    closedir DIR;
690}
691
692
693# Given $file, a symbolic link to a directory in the C include directory,
694# make an equivalent symbolic link in $Dest_dir, if we can figure out how.
695# Otherwise, just duplicate the file or directory.
696sub link_if_possible
697{
698    my ($dirlink)  = @_;
699    my $target  = eval 'readlink($dirlink)';
700
701    if ($target =~ m:^\.\./: or $target =~ m:^/:) {
702        # The target of a parent or absolute link could leave the $Dest_dir
703        # hierarchy, so let's put all of the contents of $dirlink (actually,
704        # the contents of $target) into @ARGV; as a side effect down the
705        # line, $dirlink will get created as an _actual_ directory.
706        expand_glob($dirlink);
707    } else {
708        if (-l "$Dest_dir/$dirlink") {
709            unlink "$Dest_dir/$dirlink" or
710                print STDERR "Could not remove link $Dest_dir/$dirlink:  $!\n";
711        }
712
713        if (eval 'symlink($target, "$Dest_dir/$dirlink")') {
714            print "Linking $target -> $Dest_dir/$dirlink\n";
715
716            # Make sure that the link _links_ to something:
717            if (! -e "$Dest_dir/$target") {
718                mkpath("$Dest_dir/$target", 0755) or
719                    print STDERR "Could not create $Dest_dir/$target/\n";
720            }
721        } else {
722            print STDERR "Could not symlink $target -> $Dest_dir/$dirlink:  $!\n";
723        }
724    }
725}
726
727
728# Push all #included files in $file onto our stack, except for STDIN
729# and files we've already processed.
730sub queue_includes_from
731{
732    my ($file)    = @_;
733    my $line;
734
735    return if ($file eq "-");
736
737    open HEADER, $file or return;
738        while (defined($line = <HEADER>)) {
739            while (/\\$/) { # Handle continuation lines
740                chop $line;
741                $line .= <HEADER>;
742            }
743
744            if ($line =~ /^#\s*include\s+([<"])(.*?)[>"]/) {
745                my ($delimiter, $new_file) = ($1, $2);
746                # copy the prefix in the quote syntax (#include "x.h") case
747                if ($delimiter eq q{"} && $file =~ m|^(.*)/|) {
748                    $new_file = "$1/$new_file";
749                }
750                push(@ARGV, $new_file) unless $Is_converted{$new_file};
751            }
752        }
753    close HEADER;
754}
755
756
757# Determine include directories; $Config{usrinc} should be enough for (all
758# non-GCC?) C compilers, but gcc uses additional include directories.
759sub inc_dirs
760{
761    my $from_gcc    = `LC_ALL=C $Config{cc} -v 2>&1`;
762    if( !( $from_gcc =~ s:^Reading specs from (.*?)/specs\b.*:$1/include:s ) )
763    { # gcc-4+ :
764       $from_gcc   = `LC_ALL=C $Config{cc} -print-search-dirs 2>&1`;
765       if ( !($from_gcc =~ s/^install:\s*([^\s]+[^\s\/])([\s\/]*).*$/$1\/include/s) )
766       {
767           $from_gcc = '';
768       };
769    };
770    length($from_gcc) ? ($from_gcc, $from_gcc . "-fixed", $Config{usrinc}) : ($Config{usrinc});
771}
772
773
774# Create "_h2ph_pre.ph", if it doesn't exist or was built by a different
775# version of h2ph.
776sub build_preamble_if_necessary
777{
778    # Increment $VERSION every time this function is modified:
779    my $VERSION     = 2;
780    my $preamble    = "$Dest_dir/_h2ph_pre.ph";
781
782    # Can we skip building the preamble file?
783    if (-r $preamble) {
784        # Extract version number from first line of preamble:
785        open  PREAMBLE, $preamble or die "Cannot open $preamble:  $!";
786            my $line = <PREAMBLE>;
787            $line =~ /(\b\d+\b)/;
788        close PREAMBLE            or die "Cannot close $preamble:  $!";
789
790        # Don't build preamble if a compatible preamble exists:
791        return if $1 == $VERSION;
792    }
793
794    my (%define) = _extract_cc_defines();
795
796    open  PREAMBLE, ">$preamble" or die "Cannot open $preamble:  $!";
797	print PREAMBLE "# This file was created by h2ph version $VERSION\n";
798
799	foreach (sort keys %define) {
800	    if ($opt_D) {
801		print PREAMBLE "# $_=$define{$_}\n";
802	    }
803	    if ($define{$_} =~ /^\((.*)\)$/) {
804		# parenthesized value:  d=(v)
805		$define{$_} = $1;
806	    }
807	    if ($define{$_} =~ /^([+-]?(\d+)?\.\d+([eE][+-]?\d+)?)[FL]?$/) {
808		# float:
809		print PREAMBLE
810		    "unless (defined &$_) { sub $_() { $1 } }\n\n";
811	    } elsif ($define{$_} =~ /^([+-]?\d+)U?L{0,2}$/i) {
812		# integer:
813		print PREAMBLE
814		    "unless (defined &$_) { sub $_() { $1 } }\n\n";
815	    } elsif ($define{$_} =~ /^\w+$/) {
816		print PREAMBLE
817		    "unless (defined &$_) { sub $_() { &$define{$_} } }\n\n";
818	    } else {
819		print PREAMBLE
820		    "unless (defined &$_) { sub $_() { \"",
821		    quotemeta($define{$_}), "\" } }\n\n";
822	    }
823	}
824	print PREAMBLE "\n1;\n";  # avoid 'did not return a true value' when empty
825    close PREAMBLE               or die "Cannot close $preamble:  $!";
826}
827
828
829# %Config contains information on macros that are pre-defined by the
830# system's compiler.  We need this information to make the .ph files
831# function with perl as the .h files do with cc.
832sub _extract_cc_defines
833{
834    my %define;
835    my $allsymbols  = join " ",
836	@Config{'ccsymbols', 'cppsymbols', 'cppccsymbols'};
837
838    # Split compiler pre-definitions into `key=value' pairs:
839    while ($allsymbols =~ /([^\s]+)=((\\\s|[^\s])+)/g) {
840	$define{$1} = $2;
841	if ($opt_D) {
842	    print STDERR "$_:  $1 -> $2\n";
843	}
844    }
845
846    return %define;
847}
848
849
8501;
851
852##############################################################################
853__END__
854
855=head1 NAME
856
857h2ph - convert .h C header files to .ph Perl header files
858
859=head1 SYNOPSIS
860
861B<h2ph [-d destination directory] [-r | -a] [-l] [headerfiles]>
862
863=head1 DESCRIPTION
864
865I<h2ph>
866converts any C header files specified to the corresponding Perl header file
867format.
868It is most easily run while in /usr/include:
869
870	cd /usr/include; h2ph * sys/*
871
872or
873
874	cd /usr/include; h2ph * sys/* arpa/* netinet/*
875
876or
877
878	cd /usr/include; h2ph -r -l .
879
880The output files are placed in the hierarchy rooted at Perl's
881architecture dependent library directory.  You can specify a different
882hierarchy with a B<-d> switch.
883
884If run with no arguments, filters standard input to standard output.
885
886=head1 OPTIONS
887
888=over 4
889
890=item -d destination_dir
891
892Put the resulting B<.ph> files beneath B<destination_dir>, instead of
893beneath the default Perl library location (C<$Config{'installsitearch'}>).
894
895=item -r
896
897Run recursively; if any of B<headerfiles> are directories, then run I<h2ph>
898on all files in those directories (and their subdirectories, etc.).  B<-r>
899and B<-a> are mutually exclusive.
900
901=item -a
902
903Run automagically; convert B<headerfiles>, as well as any B<.h> files
904which they include.  This option will search for B<.h> files in all
905directories which your C compiler ordinarily uses.  B<-a> and B<-r> are
906mutually exclusive.
907
908=item -l
909
910Symbolic links will be replicated in the destination directory.  If B<-l>
911is not specified, then links are skipped over.
912
913=item -h
914
915Put ``hints'' in the .ph files which will help in locating problems with
916I<h2ph>.  In those cases when you B<require> a B<.ph> file containing syntax
917errors, instead of the cryptic
918
919	[ some error condition ] at (eval mmm) line nnn
920
921you will see the slightly more helpful
922
923	[ some error condition ] at filename.ph line nnn
924
925However, the B<.ph> files almost double in size when built using B<-h>.
926
927=item -D
928
929Include the code from the B<.h> file as a comment in the B<.ph> file.
930This is primarily used for debugging I<h2ph>.
931
932=item -Q
933
934``Quiet'' mode; don't print out the names of the files being converted.
935
936=back
937
938=head1 ENVIRONMENT
939
940No environment variables are used.
941
942=head1 FILES
943
944 /usr/include/*.h
945 /usr/include/sys/*.h
946
947etc.
948
949=head1 AUTHOR
950
951Larry Wall
952
953=head1 SEE ALSO
954
955perl(1)
956
957=head1 DIAGNOSTICS
958
959The usual warnings if it can't read or write the files involved.
960
961=head1 BUGS
962
963Doesn't construct the %sizeof array for you.
964
965It doesn't handle all C constructs, but it does attempt to isolate
966definitions inside evals so that you can get at the definitions
967that it can translate.
968
969It's only intended as a rough tool.
970You may need to dicker with the files produced.
971
972You have to run this program by hand; it's not run as part of the Perl
973installation.
974
975Doesn't handle complicated expressions built piecemeal, a la:
976
977    enum {
978	FIRST_VALUE,
979	SECOND_VALUE,
980    #ifdef ABC
981	THIRD_VALUE
982    #endif
983    };
984
985Doesn't necessarily locate all of your C compiler's internally-defined
986symbols.
987
988=cut
989
990!NO!SUBS!
991
992close OUT or die "Can't close $file: $!";
993chmod 0755, $file or die "Can't reset permissions for $file: $!\n";
994exec("$Config{'eunicefix'} $file") if $Config{'eunicefix'} ne ':';
995chdir $origdir;
996