Upgrade to I18N::LangTags 0.30.
[p5sagit/p5-mst-13.2.git] / utils / h2ph.PL
CommitLineData
4633a7c4 1#!/usr/local/bin/perl
2
3use Config;
ee580363 4use File::Basename qw(basename dirname);
8a5546a1 5use Cwd;
4633a7c4 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: $archlibexp
14
15# This forces PL files to create target in same directory as PL file.
16# This is so that make depend always knows where to find PL derivatives.
8a5546a1 17$origdir = cwd;
44a8e56a 18chdir dirname($0);
19$file = basename($0, '.PL');
774d564b 20$file .= '.com' if $^O eq 'VMS';
4633a7c4 21
22open OUT,">$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 $Config{...} to use Configure variables.
28
29print OUT <<"!GROK!THIS!";
5f05dabc 30$Config{startperl}
31 eval 'exec $Config{perlpath} -S \$0 \${1+"\$@"}'
32 if \$running_under_some_shell;
154e51a4 33!GROK!THIS!
34
4633a7c4 35# In the following, perl variables are not expanded during extraction.
36
37print OUT <<'!NO!SUBS!';
154e51a4 38
fc865b05 39use strict;
40
2c2acf7e 41use Config;
b306bf39 42use File::Path qw(mkpath);
50f6e060 43use Getopt::Std;
44
80d6dabb 45# Make sure read permissions for all are set:
46if (defined umask && (umask() & 0444)) {
47 umask (umask() & ~0444);
48}
49
917244ce 50getopts('Dd:rlhaQe');
51use vars qw($opt_D $opt_d $opt_r $opt_l $opt_h $opt_a $opt_Q $opt_e);
1d3434b8 52die "-r and -a options are mutually exclusive\n" if ($opt_r and $opt_a);
fc865b05 53my @inc_dirs = inc_dirs() if $opt_a;
2c2acf7e 54
b306bf39 55my $Exit = 0;
56
50f6e060 57my $Dest_dir = $opt_d || $Config{installsitearch};
b306bf39 58die "Destination directory $Dest_dir doesn't exist or isn't a directory\n"
59 unless -d $Dest_dir;
154e51a4 60
e77cf69f 61my @isatype = qw(
fe14fcc3 62 char uchar u_char
63 short ushort u_short
64 int uint u_int
65 long ulong u_long
fb73857a 66 FILE key_t caddr_t
e77cf69f 67 float double size_t
68);
fe14fcc3 69
fc865b05 70my %isatype;
55204971 71@isatype{@isatype} = (1) x @isatype;
fc865b05 72my $inif = 0;
73my %Is_converted;
917244ce 74my %bad_file = ();
fe14fcc3 75
76@ARGV = ('-') unless @ARGV;
154e51a4 77
7f04632d 78build_preamble_if_necessary();
79
917244ce 80sub reindent($) {
81 my($text) = shift;
82 $text =~ s/\n/\n /g;
83 $text =~ s/ /\t/g;
84 $text;
85}
86
fc865b05 87my ($t, $tab, %curargs, $new, $eval_index, $dir, $name, $args, $outfile);
917244ce 88my ($incl, $incl_type, $next);
fc865b05 89while (defined (my $file = next_file())) {
50f6e060 90 if (-l $file and -d $file) {
91 link_if_possible($file) if ($opt_l);
92 next;
93 }
94
5f05dabc 95 # Recover from header files with unbalanced cpp directives
96 $t = '';
97 $tab = 0;
98
50f6e060 99 # $eval_index goes into ``#line'' directives, to help locate syntax errors:
100 $eval_index = 1;
101
fe14fcc3 102 if ($file eq '-') {
103 open(IN, "-");
104 open(OUT, ">-");
ee580363 105 } else {
fe14fcc3 106 ($outfile = $file) =~ s/\.h$/.ph/ || next;
625ca0ef 107 print "$file -> $outfile\n" unless $opt_Q;
fe14fcc3 108 if ($file =~ m|^(.*)/|) {
109 $dir = $1;
b306bf39 110 mkpath "$Dest_dir/$dir";
154e51a4 111 }
1d3434b8 112
113 if ($opt_a) { # automagic mode: locate header file in @inc_dirs
114 foreach (@inc_dirs) {
115 chdir $_;
116 last if -f $file;
117 }
118 }
119
b306bf39 120 open(IN,"$file") || (($Exit = 1),(warn "Can't open $file: $!\n"),next);
121 open(OUT,">$Dest_dir/$outfile") || die "Can't create $outfile: $!\n";
154e51a4 122 }
7f04632d 123
ccfcdfed 124 print OUT
125 "require '_h2ph_pre.ph';\n\n",
126 "no warnings 'redefine';\n\n";
dccff43d 127
128 while (defined (local $_ = next_line($file))) {
1d2dff63 129 if (s/^\s*\#\s*//) {
154e51a4 130 if (s/^define\s+(\w+)//) {
131 $name = $1;
132 $new = '';
133 s/\s+$//;
1be505aa 134 s/\(\w+\s*\(\*\)\s*\(\w*\)\)\s*(-?\d+)/$1/; # (int (*)(foo_t))0
154e51a4 135 if (s/^\(([\w,\s]*)\)//) {
136 $args = $1;
09f42789 137 my $proto = '() ';
154e51a4 138 if ($args ne '') {
09f42789 139 $proto = '';
fc865b05 140 foreach my $arg (split(/,\s*/,$args)) {
55204971 141 $arg =~ s/^\s*([^\s].*[^\s])\s*$/$1/;
154e51a4 142 $curargs{$arg} = 1;
143 }
144 $args =~ s/\b(\w)/\$$1/g;
145 $args = "local($args) = \@_;\n$t ";
146 }
147 s/^\s+//;
5f05dabc 148 expr();
ee580363 149 $new =~ s/(["\\])/\\$1/g; #"]);
0e885527 150 EMIT:
ee580363 151 $new = reindent($new);
152 $args = reindent($args);
154e51a4 153 if ($t ne '') {
ee580363 154 $new =~ s/(['\\])/\\$1/g; #']);
50f6e060 155 if ($opt_h) {
156 print OUT $t,
ee580363 157 "eval \"\\n#line $eval_index $outfile\\n\" . 'sub $name $proto\{\n$t ${args}eval q($new);\n$t}' unless defined(\&$name);\n";
50f6e060 158 $eval_index++;
159 } else {
160 print OUT $t,
ee580363 161 "eval 'sub $name $proto\{\n$t ${args}eval q($new);\n$t}' unless defined(\&$name);\n";
50f6e060 162 }
ee580363 163 } else {
164 print OUT "unless(defined(\&$name)) {\n sub $name $proto\{\n\t${args}eval q($new);\n }\n}\n";
154e51a4 165 }
166 %curargs = ();
ee580363 167 } else {
154e51a4 168 s/^\s+//;
5f05dabc 169 expr();
154e51a4 170 $new = 1 if $new eq '';
ee580363 171 $new = reindent($new);
172 $args = reindent($args);
154e51a4 173 if ($t ne '') {
ee580363 174 $new =~ s/(['\\])/\\$1/g; #']);
7f04632d 175
50f6e060 176 if ($opt_h) {
177 print OUT $t,"eval \"\\n#line $eval_index $outfile\\n\" . 'sub $name () {",$new,";}' unless defined(\&$name);\n";
178 $eval_index++;
179 } else {
180 print OUT $t,"eval 'sub $name () {",$new,";}' unless defined(\&$name);\n";
181 }
ee580363 182 } else {
7f04632d 183 # Shunt around such directives as `#define FOO FOO':
184 next if " \&$name" eq $new;
185
ee580363 186 print OUT $t,"unless(defined(\&$name)) {\n sub $name () {\t",$new,";}\n}\n";
154e51a4 187 }
188 }
917244ce 189 } elsif (/^(include|import|include_next)\s*[<\"](.*)[>\"]/) {
190 $incl_type = $1;
191 $incl = $2;
192 if (($incl_type eq 'include_next') ||
193 ($opt_e && exists($bad_file{$incl}))) {
194 $incl =~ s/\.h$/.ph/;
ee580363 195 print OUT ($t,
1d2dff63 196 "eval {\n");
197 $tab += 4;
198 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
917244ce 199 print OUT ($t, "my(\@REM);\n");
200 if ($incl_type eq 'include_next') {
1d2dff63 201 print OUT ($t,
202 "my(\%INCD) = map { \$INC{\$_} => 1 } ",
917244ce 203 "(grep { \$_ eq \"$incl\" } ",
204 "keys(\%INC));\n");
1d2dff63 205 print OUT ($t,
917244ce 206 "\@REM = map { \"\$_/$incl\" } ",
1d2dff63 207 "(grep { not exists(\$INCD{\"\$_/$incl\"})",
917244ce 208 " and -f \"\$_/$incl\" } \@INC);\n");
209 } else {
210 print OUT ($t,
211 "\@REM = map { \"\$_/$incl\" } ",
212 "(grep {-r \"\$_/$incl\" } \@INC);\n");
213 }
1d2dff63 214 print OUT ($t,
215 "require \"\$REM[0]\" if \@REM;\n");
216 $tab -= 4;
217 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
218 print OUT ($t,
219 "};\n");
220 print OUT ($t,
221 "warn(\$\@) if \$\@;\n");
917244ce 222 } else {
223 $incl =~ s/\.h$/.ph/;
224 print OUT $t,"require '$incl';\n";
225 }
ee580363 226 } elsif (/^ifdef\s+(\w+)/) {
227 print OUT $t,"if(defined(&$1)) {\n";
154e51a4 228 $tab += 4;
229 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
ee580363 230 } elsif (/^ifndef\s+(\w+)/) {
231 print OUT $t,"unless(defined(&$1)) {\n";
232 $tab += 4;
233 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
234 } elsif (s/^if\s+//) {
154e51a4 235 $new = '';
748a9306 236 $inif = 1;
5f05dabc 237 expr();
748a9306 238 $inif = 0;
ee580363 239 print OUT $t,"if($new) {\n";
154e51a4 240 $tab += 4;
241 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
ee580363 242 } elsif (s/^elif\s+//) {
154e51a4 243 $new = '';
748a9306 244 $inif = 1;
5f05dabc 245 expr();
748a9306 246 $inif = 0;
154e51a4 247 $tab -= 4;
248 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
ee580363 249 print OUT $t,"}\n elsif($new) {\n";
154e51a4 250 $tab += 4;
251 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
ee580363 252 } elsif (/^else/) {
154e51a4 253 $tab -= 4;
254 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
ee580363 255 print OUT $t,"} else {\n";
154e51a4 256 $tab += 4;
257 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
ee580363 258 } elsif (/^endif/) {
154e51a4 259 $tab -= 4;
260 $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
261 print OUT $t,"}\n";
ee580363 262 } elsif(/^undef\s+(\w+)/) {
263 print OUT $t, "undef(&$1) if defined(&$1);\n";
d3e00f1c 264 } elsif(/^error\s+(".*")/) {
265 print OUT $t, "die($1);\n";
ee580363 266 } elsif(/^error\s+(.*)/) {
5d42aa7b 267 print OUT $t, "die(\"", quotemeta($1), "\");\n";
ee580363 268 } elsif(/^warning\s+(.*)/) {
5d42aa7b 269 print OUT $t, "warn(\"", quotemeta($1), "\");\n";
ee580363 270 } elsif(/^ident\s+(.*)/) {
271 print OUT $t, "# $1\n";
154e51a4 272 }
09f42789 273 } elsif (/^\s*(typedef\s*)?enum\s*(\s+[a-zA-Z_]\w*\s*)?/) { # { for vi
e7cba2ba 274 until(/\{[^}]*\}.*;/ || /;/) {
dccff43d 275 last unless defined ($next = next_line($file));
e7cba2ba 276 chomp $next;
277 # drop "#define FOO FOO" in enums
278 $next =~ s/^\s*#\s*define\s+(\w+)\s+\1\s*$//;
1d2dff63 279 $_ .= $next;
280 print OUT "# $next\n" if $opt_D;
281 }
e7cba2ba 282 s/#\s*if.*?#\s*endif//g; # drop #ifdefs
1d2dff63 283 s@/\*.*?\*/@@g;
284 s/\s+/ /g;
e7cba2ba 285 next unless /^\s?(typedef\s?)?enum\s?([a-zA-Z_]\w*)?\s?\{(.*)\}\s?([a-zA-Z_]\w*)?\s?;/;
fc865b05 286 (my $enum_subs = $3) =~ s/\s//g;
287 my @enum_subs = split(/,/, $enum_subs);
288 my $enum_val = -1;
289 foreach my $enum (@enum_subs) {
290 my ($enum_name, $enum_value) = $enum =~ /^([a-zA-Z_]\w*)(=.+)?$/;
1d2dff63 291 $enum_value =~ s/^=//;
292 $enum_val = (length($enum_value) ? $enum_value : $enum_val + 1);
293 if ($opt_h) {
294 print OUT ($t,
295 "eval(\"\\n#line $eval_index $outfile\\n",
296 "sub $enum_name () \{ $enum_val; \}\") ",
297 "unless defined(\&$enum_name);\n");
298 ++ $eval_index;
299 } else {
300 print OUT ($t,
301 "eval(\"sub $enum_name () \{ $enum_val; \}\") ",
302 "unless defined(\&$enum_name);\n");
303 }
304 }
5fff27bd 305 } elsif (/^(?:__extension__\s+)?(?:extern|static)\s+(?:__)?inline(?:__)?\s+/
306 and !/;\s*$/ and !/{\s*}\s*$/)
307 { # { for vi
09f42789 308 # This is a hack to parse the inline functions in the glibc headers.
5fff27bd 309 # Warning: massive kludge ahead. We suppose inline functions
310 # are mainly constructed like macros.
09f42789 311 while (1) {
312 last unless defined ($next = next_line($file));
313 chomp $next;
5fff27bd 314 undef $_, last if $next =~ /__THROW\s*;/
315 or $next =~ /^(__extension__|extern|static)\b/;
09f42789 316 $_ .= " $next";
317 print OUT "# $next\n" if $opt_D;
318 last if $next =~ /^}|^{.*}\s*$/;
319 }
320 next if not defined; # because it's only a prototype
0e885527 321 s/\b(__extension__|extern|static|(?:__)?inline(?:__)?)\b//g;
322 # violently drop #ifdefs
323 s/#\s*if.*?#\s*endif//g
324 and print OUT "# some #ifdef were dropped here -- fill in the blanks\n";
09f42789 325 if (s/^(?:\w|\s|\*)*\s(\w+)\s*//) {
326 $name = $1;
327 } else {
328 warn "name not found"; next; # shouldn't occur...
329 }
330 my @args;
331 if (s/^\(([^()]*)\)\s*(\w+\s*)*//) {
332 for my $arg (split /,/, $1) {
333 if ($arg =~ /(\w+)\s*$/) {
334 $curargs{$1} = 1;
335 push @args, $1;
336 }
337 }
338 }
339 $args = (
340 @args
341 ? "local(" . (join ',', map "\$$_", @args) . ") = \@_;\n$t "
342 : ""
343 );
344 my $proto = @args ? '' : '() ';
345 $new = '';
346 s/\breturn\b//g; # "return" doesn't occur in macros usually...
347 expr();
0e885527 348 # try to find and perlify local C variables
349 our @local_variables = (); # needs to be a our(): (?{...}) bug workaround
e77cf69f 350 {
351 use re "eval";
352 my $typelist = join '|', keys %isatype;
353 $new =~ s['
354 (?:(?:un)?signed\s+)?
355 (?:long\s+)?
356 (?:$typelist)\s+
357 (\w+)
358 (?{ push @local_variables, $1 })
359 ']
360 [my \$$1]gx;
361 $new =~ s['
362 (?:(?:un)?signed\s+)?
363 (?:long\s+)?
364 (?:$typelist)\s+
365 ' \s+ &(\w+) \s* ;
366 (?{ push @local_variables, $1 })
367 ]
368 [my \$$1;]gx;
369 }
0e885527 370 $new =~ s/&$_\b/\$$_/g for @local_variables;
09f42789 371 $new =~ s/(["\\])/\\$1/g; #"]);
0e885527 372 # now that's almost like a macro (we hope)
373 goto EMIT;
154e51a4 374 }
375 }
fc865b05 376 $Is_converted{$file} = 1;
917244ce 377 if ($opt_e && exists($bad_file{$file})) {
378 unlink($Dest_dir . '/' . $outfile);
379 $next = '';
380 } else {
381 print OUT "1;\n";
09f42789 382 queue_includes_from($file) if $opt_a;
917244ce 383 }
154e51a4 384}
385
917244ce 386if ($opt_e && (scalar(keys %bad_file) > 0)) {
387 warn "Was unable to convert the following files:\n";
388 warn "\t" . join("\n\t",sort(keys %bad_file)) . "\n";
ee580363 389}
390
917244ce 391exit $Exit;
fc865b05 392
154e51a4 393sub expr {
fc865b05 394 my $joined_args;
ee580363 395 if(keys(%curargs)) {
fc865b05 396 $joined_args = join('|', keys(%curargs));
ee580363 397 }
154e51a4 398 while ($_ ne '') {
ee580363 399 s/^\&\&// && do { $new .= " &&"; next;}; # handle && operator
400 s/^\&([\(a-z\)]+)/$1/i; # hack for things that take the address of
154e51a4 401 s/^(\s+)// && do {$new .= ' '; next;};
25146a1a 402 s/^0X([0-9A-F]+)[UL]*//i
403 && do {my $hex = $1;
404 $hex =~ s/^0+//;
405 if (length $hex > 8 && !$Config{use64bitint}) {
406 # Croak if nv_preserves_uv_bits < 64 ?
407 $new .= hex(substr($hex, -8)) +
408 2**32 * hex(substr($hex, 0, -8));
409 # The above will produce "errorneus" code
410 # if the hex constant was e.g. inside UINT64_C
411 # macro, but then again, h2ph is an approximation.
412 } else {
413 $new .= lc("0x$hex");
414 }
415 next;};
fd3f0aff 416 s/^(-?\d+\.\d+E[-+]?\d+)[FL]?//i && do {$new .= $1; next;};
50f6e060 417 s/^(\d+)\s*[LU]*//i && do {$new .= $1; next;};
154e51a4 418 s/^("(\\"|[^"])*")// && do {$new .= $1; next;};
419 s/^'((\\"|[^"])*)'// && do {
420 if ($curargs{$1}) {
421 $new .= "ord('\$$1')";
ee580363 422 } else {
154e51a4 423 $new .= "ord('$1')";
424 }
425 next;
426 };
5f05dabc 427 # replace "sizeof(foo)" with "{foo}"
428 # also, remove * (C dereference operator) to avoid perl syntax
429 # problems. Where the %sizeof array comes from is anyone's
430 # guess (c2ph?), but this at least avoids fatal syntax errors.
431 # Behavior is undefined if sizeof() delimiters are unbalanced.
432 # This code was modified to able to handle constructs like this:
433 # sizeof(*(p)), which appear in the HP-UX 10.01 header files.
434 s/^sizeof\s*\(// && do {
435 $new .= '$sizeof';
436 my $lvl = 1; # already saw one open paren
437 # tack { on the front, and skip it in the loop
438 $_ = "{" . "$_";
439 my $index = 1;
440 # find balanced closing paren
441 while ($index <= length($_) && $lvl > 0) {
442 $lvl++ if substr($_, $index, 1) eq "(";
443 $lvl-- if substr($_, $index, 1) eq ")";
444 $index++;
445 }
446 # tack } on the end, replacing )
447 substr($_, $index - 1, 1) = "}";
448 # remove pesky * operators within the sizeof argument
449 substr($_, 0, $index - 1) =~ s/\*//g;
450 next;
451 };
50f6e060 452 # Eliminate typedefs
453 /\(([\w\s]+)[\*\s]*\)\s*[\w\(]/ && do {
55c8af0d 454 my $doit = 1;
50f6e060 455 foreach (split /\s+/, $1) { # Make sure all the words are types,
55c8af0d 456 unless($isatype{$_} or $_ eq 'struct' or $_ eq 'union'){
457 $doit = 0;
458 last;
459 }
460 }
461 if( $doit ){
462 s/\([\w\s]+[\*\s]*\)// && next; # then eliminate them.
50f6e060 463 }
50f6e060 464 };
ee580363 465 # struct/union member, including arrays:
466 s/^([_A-Z]\w*(\[[^\]]+\])?((\.|->)[_A-Z]\w*(\[[^\]]+\])?)+)//i && do {
fc865b05 467 my $id = $1;
ee580363 468 $id =~ s/(\.|(->))([^\.\-]*)/->\{$3\}/g;
469 $id =~ s/\b([^\$])($joined_args)/$1\$$2/g if length($joined_args);
470 while($id =~ /\[\s*([^\$\&\d\]]+)\]/) {
471 my($index) = $1;
472 $index =~ s/\s//g;
473 if(exists($curargs{$index})) {
474 $index = "\$$index";
475 } else {
476 $index = "&$index";
477 }
478 $id =~ s/\[\s*([^\$\&\d\]]+)\]/[$index]/;
479 }
480 $new .= " (\$$id)";
50f6e060 481 };
154e51a4 482 s/^([_a-zA-Z]\w*)// && do {
fc865b05 483 my $id = $1;
99ed927b 484 if ($id eq 'struct' || $id eq 'union') {
fe14fcc3 485 s/^\s+(\w+)//;
486 $id .= ' ' . $1;
487 $isatype{$id} = 1;
ee580363 488 } elsif ($id =~ /^((un)?signed)|(long)|(short)$/) {
50f6e060 489 while (s/^\s+(\w+)//) { $id .= ' ' . $1; }
fe14fcc3 490 $isatype{$id} = 1;
491 }
154e51a4 492 if ($curargs{$id}) {
ee580363 493 $new .= "\$$id";
494 $new .= '->' if /^[\[\{]/;
495 } elsif ($id eq 'defined') {
154e51a4 496 $new .= 'defined';
cd4e1efa 497 } elsif (/^\s*\(/) {
498 s/^\s*\((\w),/("$1",/ if $id =~ /^_IO[WR]*$/i; # cheat
154e51a4 499 $new .= " &$id";
ee580363 500 } elsif ($isatype{$id}) {
fe14fcc3 501 if ($new =~ /{\s*$/) {
502 $new .= "'$id'";
ee580363 503 } elsif ($new =~ /\(\s*$/ && /^[\s*]*\)/) {
fe14fcc3 504 $new =~ s/\(\s*$//;
505 s/^[\s*]*\)//;
ee580363 506 } else {
b276c83d 507 $new .= q(').$id.q(');
fe14fcc3 508 }
ee580363 509 } else {
c07a80fd 510 if ($inif && $new !~ /defined\s*\($/) {
748a9306 511 $new .= '(defined(&' . $id . ') ? &' . $id . ' : 0)';
ee580363 512 } elsif (/^\[/) {
513 $new .= " \$$id";
514 } else {
748a9306 515 $new .= ' &' . $id;
516 }
154e51a4 517 }
518 next;
519 };
fb21d8eb 520 s/^(.)// && do { if ($1 ne '#') { $new .= $1; } next;};
154e51a4 521 }
522}
50f6e060 523
524
79c1b905 525sub next_line
526{
dccff43d 527 my $file = shift;
79c1b905 528 my ($in, $out);
b7bcf494 529 my $pre_sub_tri_graphs = 1;
79c1b905 530
531 READ: while (not eof IN) {
532 $in .= <IN>;
533 chomp $in;
534 next unless length $in;
535
536 while (length $in) {
b7bcf494 537 if ($pre_sub_tri_graphs) {
538 # Preprocess all tri-graphs
539 # including things stuck in quoted string constants.
540 $in =~ s/\?\?=/#/g; # | ??=| #|
541 $in =~ s/\?\?\!/|/g; # | ??!| ||
542 $in =~ s/\?\?'/^/g; # | ??'| ^|
543 $in =~ s/\?\?\(/[/g; # | ??(| [|
544 $in =~ s/\?\?\)/]/g; # | ??)| ]|
545 $in =~ s/\?\?\-/~/g; # | ??-| ~|
546 $in =~ s/\?\?\//\\/g; # | ??/| \|
547 $in =~ s/\?\?</{/g; # | ??<| {|
548 $in =~ s/\?\?>/}/g; # | ??>| }|
549 }
9efe82d3 550 if ($in =~ /^\#ifdef __LANGUAGE_PASCAL__/) {
551 # Tru64 disassembler.h evilness: mixed C and Pascal.
552 while (<IN>) {
553 last if /^\#endif/;
554 }
555 next READ;
556 }
37723803 557 if ($in =~ /^extern inline / && # Inlined assembler.
dccff43d 558 $^O eq 'linux' && $file =~ m!(?:^|/)asm/[^/]+\.h$!) {
559 while (<IN>) {
560 last if /^}/;
561 }
562 next READ;
563 }
79c1b905 564 if ($in =~ s/\\$//) { # \-newline
565 $out .= ' ';
566 next READ;
567 } elsif ($in =~ s/^([^"'\\\/]+)//) { # Passthrough
568 $out .= $1;
569 } elsif ($in =~ s/^(\\.)//) { # \...
570 $out .= $1;
ab5fe4d6 571 } elsif ($in =~ /^'/) { # '...
572 if ($in =~ s/^('(\\.|[^'\\])*')//) {
573 $out .= $1;
574 } else {
575 next READ;
576 }
577 } elsif ($in =~ /^"/) { # "...
578 if ($in =~ s/^("(\\.|[^"\\])*")//) {
579 $out .= $1;
580 } else {
581 next READ;
582 }
79c1b905 583 } elsif ($in =~ s/^\/\/.*//) { # //...
edf6e4ec 584 # fall through
79c1b905 585 } elsif ($in =~ m/^\/\*/) { # /*...
586 # C comment removal adapted from perlfaq6:
587 if ($in =~ s/^\/\*[^*]*\*+([^\/*][^*]*\*+)*\///) {
588 $out .= ' ';
589 } else { # Incomplete /* */
590 next READ;
591 }
592 } elsif ($in =~ s/^(\/)//) { # /...
593 $out .= $1;
594 } elsif ($in =~ s/^([^\'\"\\\/]+)//) {
595 $out .= $1;
889e303a 596 } elsif ($^O eq 'linux' &&
597 $file =~ m!(?:^|/)linux/byteorder/pdp_endian\.h$! &&
598 $in =~ s!\'T KNOW!!) {
599 $out =~ s!I DON$!I_DO_NOT_KNOW!;
79c1b905 600 } else {
917244ce 601 if ($opt_e) {
602 warn "Cannot parse $file:\n$in\n";
603 $bad_file{$file} = 1;
604 $in = '';
605 $out = undef;
606 last READ;
607 } else {
dccff43d 608 die "Cannot parse:\n$in\n";
917244ce 609 }
79c1b905 610 }
611 }
612
edf6e4ec 613 last READ if $out =~ /\S/;
79c1b905 614 }
615
616 return $out;
617}
618
619
50f6e060 620# Handle recursive subdirectories without getting a grotesquely big stack.
621# Could this be implemented using File::Find?
622sub next_file
623{
624 my $file;
625
626 while (@ARGV) {
627 $file = shift @ARGV;
628
629 if ($file eq '-' or -f $file or -l $file) {
630 return $file;
631 } elsif (-d $file) {
632 if ($opt_r) {
633 expand_glob($file);
634 } else {
635 print STDERR "Skipping directory `$file'\n";
636 }
1d3434b8 637 } elsif ($opt_a) {
638 return $file;
639 } else {
50f6e060 640 print STDERR "Skipping `$file': not a file or directory\n";
641 }
642 }
643
644 return undef;
645}
646
647
648# Put all the files in $directory into @ARGV for processing.
649sub expand_glob
650{
651 my ($directory) = @_;
652
653 $directory =~ s:/$::;
654
655 opendir DIR, $directory;
656 foreach (readdir DIR) {
657 next if ($_ eq '.' or $_ eq '..');
658
659 # expand_glob() is going to be called until $ARGV[0] isn't a
660 # directory; so push directories, and unshift everything else.
1d3434b8 661 if (-d "$directory/$_") { push @ARGV, "$directory/$_" }
662 else { unshift @ARGV, "$directory/$_" }
50f6e060 663 }
664 closedir DIR;
665}
666
667
668# Given $file, a symbolic link to a directory in the C include directory,
669# make an equivalent symbolic link in $Dest_dir, if we can figure out how.
670# Otherwise, just duplicate the file or directory.
671sub link_if_possible
672{
673 my ($dirlink) = @_;
674 my $target = eval 'readlink($dirlink)';
675
676 if ($target =~ m:^\.\./: or $target =~ m:^/:) {
677 # The target of a parent or absolute link could leave the $Dest_dir
678 # hierarchy, so let's put all of the contents of $dirlink (actually,
679 # the contents of $target) into @ARGV; as a side effect down the
680 # line, $dirlink will get created as an _actual_ directory.
681 expand_glob($dirlink);
682 } else {
683 if (-l "$Dest_dir/$dirlink") {
684 unlink "$Dest_dir/$dirlink" or
685 print STDERR "Could not remove link $Dest_dir/$dirlink: $!\n";
686 }
1d3434b8 687
50f6e060 688 if (eval 'symlink($target, "$Dest_dir/$dirlink")') {
689 print "Linking $target -> $Dest_dir/$dirlink\n";
690
691 # Make sure that the link _links_ to something:
692 if (! -e "$Dest_dir/$target") {
1d3434b8 693 mkpath("$Dest_dir/$target", 0755) or
50f6e060 694 print STDERR "Could not create $Dest_dir/$target/\n";
695 }
696 } else {
697 print STDERR "Could not symlink $target -> $Dest_dir/$dirlink: $!\n";
698 }
699 }
700}
701
702
1d3434b8 703# Push all #included files in $file onto our stack, except for STDIN
704# and files we've already processed.
705sub queue_includes_from
706{
707 my ($file) = @_;
708 my $line;
709
710 return if ($file eq "-");
711
712 open HEADER, $file or return;
713 while (defined($line = <HEADER>)) {
714 while (/\\$/) { # Handle continuation lines
715 chop $line;
716 $line .= <HEADER>;
717 }
718
719 if ($line =~ /^#\s*include\s+<(.*?)>/) {
fc865b05 720 push(@ARGV, $1) unless $Is_converted{$1};
1d3434b8 721 }
722 }
723 close HEADER;
724}
725
726
727# Determine include directories; $Config{usrinc} should be enough for (all
728# non-GCC?) C compilers, but gcc uses an additional include directory.
729sub inc_dirs
730{
731 my $from_gcc = `$Config{cc} -v 2>&1`;
732 $from_gcc =~ s:^Reading specs from (.*?)/specs\b.*:$1/include:s;
733
734 length($from_gcc) ? ($from_gcc, $Config{usrinc}) : ($Config{usrinc});
735}
736
737
7f04632d 738# Create "_h2ph_pre.ph", if it doesn't exist or was built by a different
739# version of h2ph.
740sub build_preamble_if_necessary
741{
742 # Increment $VERSION every time this function is modified:
00f0ad8c 743 my $VERSION = 2;
7f04632d 744 my $preamble = "$Dest_dir/_h2ph_pre.ph";
745
746 # Can we skip building the preamble file?
747 if (-r $preamble) {
748 # Extract version number from first line of preamble:
749 open PREAMBLE, $preamble or die "Cannot open $preamble: $!";
750 my $line = <PREAMBLE>;
751 $line =~ /(\b\d+\b)/;
752 close PREAMBLE or die "Cannot close $preamble: $!";
753
754 # Don't build preamble if a compatible preamble exists:
755 return if $1 == $VERSION;
756 }
757
758 my (%define) = _extract_cc_defines();
759
760 open PREAMBLE, ">$preamble" or die "Cannot open $preamble: $!";
761 print PREAMBLE "# This file was created by h2ph version $VERSION\n";
762
763 foreach (sort keys %define) {
764 if ($opt_D) {
765 print PREAMBLE "# $_=$define{$_}\n";
766 }
767
4322f456 768 if ($define{$_} =~ /^(\d+)U?L{0,2}$/i) {
7f04632d 769 print PREAMBLE
4322f456 770 "unless (defined &$_) { sub $_() { $1 } }\n\n";
00f0ad8c 771 } elsif ($define{$_} =~ /^\w+$/) {
772 print PREAMBLE
773 "unless (defined &$_) { sub $_() { &$define{$_} } }\n\n";
7f04632d 774 } else {
775 print PREAMBLE
776 "unless (defined &$_) { sub $_() { \"",
777 quotemeta($define{$_}), "\" } }\n\n";
778 }
779 }
780 close PREAMBLE or die "Cannot close $preamble: $!";
781}
782
783
784# %Config contains information on macros that are pre-defined by the
785# system's compiler. We need this information to make the .ph files
786# function with perl as the .h files do with cc.
787sub _extract_cc_defines
788{
789 my %define;
fc865b05 790 my $allsymbols = join " ",
791 @Config{'ccsymbols', 'cppsymbols', 'cppccsymbols'};
7f04632d 792
793 # Split compiler pre-definitions into `key=value' pairs:
794 foreach (split /\s+/, $allsymbols) {
00f0ad8c 795 /(.+?)=(.+)/ and $define{$1} = $2;
7f04632d 796
797 if ($opt_D) {
798 print STDERR "$_: $1 -> $2\n";
799 }
800 }
801
802 return %define;
803}
804
805
50f6e060 8061;
807
154e51a4 808##############################################################################
1fef88e7 809__END__
810
811=head1 NAME
812
813h2ph - convert .h C header files to .ph Perl header files
814
815=head1 SYNOPSIS
816
1d3434b8 817B<h2ph [-d destination directory] [-r | -a] [-l] [headerfiles]>
1fef88e7 818
819=head1 DESCRIPTION
154e51a4 820
1fef88e7 821I<h2ph>
154e51a4 822converts any C header files specified to the corresponding Perl header file
823format.
824It is most easily run while in /usr/include:
154e51a4 825
826 cd /usr/include; h2ph * sys/*
827
50f6e060 828or
829
ef0ae776 830 cd /usr/include; h2ph * sys/* arpa/* netinet/*
831
832or
833
50f6e060 834 cd /usr/include; h2ph -r -l .
835
b306bf39 836The output files are placed in the hierarchy rooted at Perl's
837architecture dependent library directory. You can specify a different
838hierarchy with a B<-d> switch.
839
fe14fcc3 840If run with no arguments, filters standard input to standard output.
1fef88e7 841
50f6e060 842=head1 OPTIONS
843
844=over 4
845
846=item -d destination_dir
847
848Put the resulting B<.ph> files beneath B<destination_dir>, instead of
849beneath the default Perl library location (C<$Config{'installsitsearch'}>).
850
851=item -r
852
853Run recursively; if any of B<headerfiles> are directories, then run I<h2ph>
1d3434b8 854on all files in those directories (and their subdirectories, etc.). B<-r>
855and B<-a> are mutually exclusive.
856
857=item -a
858
859Run automagically; convert B<headerfiles>, as well as any B<.h> files
860which they include. This option will search for B<.h> files in all
861directories which your C compiler ordinarily uses. B<-a> and B<-r> are
862mutually exclusive.
50f6e060 863
864=item -l
865
866Symbolic links will be replicated in the destination directory. If B<-l>
867is not specified, then links are skipped over.
868
869=item -h
870
871Put ``hints'' in the .ph files which will help in locating problems with
872I<h2ph>. In those cases when you B<require> a B<.ph> file containing syntax
873errors, instead of the cryptic
874
875 [ some error condition ] at (eval mmm) line nnn
876
877you will see the slightly more helpful
878
879 [ some error condition ] at filename.ph line nnn
880
881However, the B<.ph> files almost double in size when built using B<-h>.
882
1d3434b8 883=item -D
884
885Include the code from the B<.h> file as a comment in the B<.ph> file.
886This is primarily used for debugging I<h2ph>.
887
7f04632d 888=item -Q
889
890``Quiet'' mode; don't print out the names of the files being converted.
891
50f6e060 892=back
893
1fef88e7 894=head1 ENVIRONMENT
895
154e51a4 896No environment variables are used.
1fef88e7 897
898=head1 FILES
899
900 /usr/include/*.h
901 /usr/include/sys/*.h
902
154e51a4 903etc.
1fef88e7 904
905=head1 AUTHOR
906
154e51a4 907Larry Wall
1fef88e7 908
909=head1 SEE ALSO
910
154e51a4 911perl(1)
1fef88e7 912
913=head1 DIAGNOSTICS
914
154e51a4 915The usual warnings if it can't read or write the files involved.
1fef88e7 916
917=head1 BUGS
918
154e51a4 919Doesn't construct the %sizeof array for you.
1fef88e7 920
154e51a4 921It doesn't handle all C constructs, but it does attempt to isolate
922definitions inside evals so that you can get at the definitions
923that it can translate.
1fef88e7 924
154e51a4 925It's only intended as a rough tool.
926You may need to dicker with the files produced.
1fef88e7 927
7f04632d 928You have to run this program by hand; it's not run as part of the Perl
929installation.
930
931Doesn't handle complicated expressions built piecemeal, a la:
932
933 enum {
934 FIRST_VALUE,
935 SECOND_VALUE,
936 #ifdef ABC
937 THIRD_VALUE
938 #endif
939 };
940
941Doesn't necessarily locate all of your C compiler's internally-defined
942symbols.
943
1fef88e7 944=cut
945
154e51a4 946!NO!SUBS!
4633a7c4 947
948close OUT or die "Can't close $file: $!";
949chmod 0755, $file or die "Can't reset permissions for $file: $!\n";
950exec("$Config{'eunicefix'} $file") if $Config{'eunicefix'} ne ':';
8a5546a1 951chdir $origdir;