Commit | Line | Data |
5ac756c6 |
1 | # -*- mode: cperl; tab-width: 8; indent-tabs-mode: nil; basic-offset: 2 -*- |
2 | # vim:ts=8:sw=2:et:sta:sts=2 |
3 | package Module::Metadata; |
4 | |
cd41f0db |
5 | # Adapted from Perl-licensed code originally distributed with |
6 | # Module-Build by Ken Williams |
5ac756c6 |
7 | |
8 | # This module provides routines to gather information about |
9 | # perl modules (assuming this may be expanded in the distant |
10 | # parrot future to look at other types of modules). |
11 | |
12 | use strict; |
eed8b6fa |
13 | use warnings; |
14 | |
21db36b1 |
15 | our $VERSION = '1.000016'; |
5ac756c6 |
16 | $VERSION = eval $VERSION; |
17 | |
6f3c7f28 |
18 | use Carp qw/croak/; |
5ac756c6 |
19 | use File::Spec; |
20 | use IO::File; |
4850170c |
21 | use version 0.87; |
3db27017 |
22 | BEGIN { |
23 | if ($INC{'Log/Contextual.pm'}) { |
24 | Log::Contextual->import('log_info'); |
25 | } else { |
e6ddd765 |
26 | *log_info = sub (&) { warn $_[0]->() }; |
3db27017 |
27 | } |
28 | } |
5ac756c6 |
29 | use File::Find qw(find); |
30 | |
31 | my $V_NUM_REGEXP = qr{v?[0-9._]+}; # crudely, a v-string or decimal |
32 | |
8e4bef01 |
33 | my $PKG_FIRST_WORD_REGEXP = qr{ # the FIRST word in a package name |
34 | [a-zA-Z_] # the first word CANNOT start with a digit |
35 | (?: |
36 | [\w']? # can contain letters, digits, _, or ticks |
37 | \w # But, NO multi-ticks or trailing ticks |
38 | )* |
39 | }x; |
40 | |
41 | my $PKG_ADDL_WORD_REGEXP = qr{ # the 2nd+ word in a package name |
42 | \w # the 2nd+ word CAN start with digits |
43 | (?: |
44 | [\w']? # and can contain letters or ticks |
45 | \w # But, NO multi-ticks or trailing ticks |
46 | )* |
47 | }x; |
48 | |
49 | my $PKG_NAME_REGEXP = qr{ # match a package name |
50 | (?: :: )? # a pkg name can start with aristotle |
51 | $PKG_FIRST_WORD_REGEXP # a package word |
52 | (?: |
53 | (?: :: )+ ### aristotle (allow one or many times) |
54 | $PKG_ADDL_WORD_REGEXP ### a package word |
55 | )* # ^ zero, one or many times |
56 | (?: |
57 | :: # allow trailing aristotle |
58 | )? |
59 | }x; |
60 | |
5ac756c6 |
61 | my $PKG_REGEXP = qr{ # match a package declaration |
62 | ^[\s\{;]* # intro chars on a line |
63 | package # the word 'package' |
64 | \s+ # whitespace |
8e4bef01 |
65 | ($PKG_NAME_REGEXP) # a package name |
5ac756c6 |
66 | \s* # optional whitespace |
67 | ($V_NUM_REGEXP)? # optional version number |
68 | \s* # optional whitesapce |
710f253f |
69 | [;\{] # semicolon line terminator or block start (since 5.16) |
5ac756c6 |
70 | }x; |
71 | |
72 | my $VARNAME_REGEXP = qr{ # match fully-qualified VERSION name |
73 | ([\$*]) # sigil - $ or * |
74 | ( |
75 | ( # optional leading package name |
76 | (?:::|\')? # possibly starting like just :: (Ì la $::VERSION) |
77 | (?:\w+(?:::|\'))* # Foo::Bar:: ... |
78 | )? |
79 | VERSION |
80 | )\b |
81 | }x; |
82 | |
83 | my $VERS_REGEXP = qr{ # match a VERSION definition |
84 | (?: |
85 | \(\s*$VARNAME_REGEXP\s*\) # with parens |
86 | | |
87 | $VARNAME_REGEXP # without parens |
88 | ) |
89 | \s* |
90 | =[^=~] # = but not ==, nor =~ |
91 | }x; |
92 | |
5ac756c6 |
93 | sub new_from_file { |
94 | my $class = shift; |
95 | my $filename = File::Spec->rel2abs( shift ); |
96 | |
97 | return undef unless defined( $filename ) && -f $filename; |
98 | return $class->_init(undef, $filename, @_); |
99 | } |
100 | |
f33c0a6c |
101 | sub new_from_handle { |
102 | my $class = shift; |
103 | my $handle = shift; |
104 | my $filename = shift; |
105 | return undef unless defined($handle) && defined($filename); |
106 | $filename = File::Spec->rel2abs( $filename ); |
107 | |
108 | return $class->_init(undef, $filename, @_, handle => $handle); |
109 | |
110 | } |
111 | |
112 | |
5ac756c6 |
113 | sub new_from_module { |
114 | my $class = shift; |
115 | my $module = shift; |
116 | my %props = @_; |
117 | |
118 | $props{inc} ||= \@INC; |
119 | my $filename = $class->find_module_by_name( $module, $props{inc} ); |
120 | return undef unless defined( $filename ) && -f $filename; |
121 | return $class->_init($module, $filename, %props); |
122 | } |
123 | |
124 | { |
dd5a4b10 |
125 | |
5ac756c6 |
126 | my $compare_versions = sub { |
127 | my ($v1, $op, $v2) = @_; |
4850170c |
128 | $v1 = version->new($v1) |
129 | unless UNIVERSAL::isa($v1,'version'); |
dd5a4b10 |
130 | |
5ac756c6 |
131 | my $eval_str = "\$v1 $op \$v2"; |
132 | my $result = eval $eval_str; |
133 | log_info { "error comparing versions: '$eval_str' $@" } if $@; |
dd5a4b10 |
134 | |
5ac756c6 |
135 | return $result; |
136 | }; |
137 | |
138 | my $normalize_version = sub { |
139 | my ($version) = @_; |
140 | if ( $version =~ /[=<>!,]/ ) { # logic, not just version |
141 | # take as is without modification |
142 | } |
4850170c |
143 | elsif ( ref $version eq 'version' ) { # version objects |
5ac756c6 |
144 | $version = $version->is_qv ? $version->normal : $version->stringify; |
145 | } |
146 | elsif ( $version =~ /^[^v][^.]*\.[^.]+\./ ) { # no leading v, multiple dots |
147 | # normalize string tuples without "v": "1.2.3" -> "v1.2.3" |
148 | $version = "v$version"; |
149 | } |
150 | else { |
151 | # leave alone |
152 | } |
153 | return $version; |
154 | }; |
155 | |
156 | # separate out some of the conflict resolution logic |
157 | |
158 | my $resolve_module_versions = sub { |
159 | my $packages = shift; |
dd5a4b10 |
160 | |
5ac756c6 |
161 | my( $file, $version ); |
162 | my $err = ''; |
163 | foreach my $p ( @$packages ) { |
164 | if ( defined( $p->{version} ) ) { |
165 | if ( defined( $version ) ) { |
166 | if ( $compare_versions->( $version, '!=', $p->{version} ) ) { |
167 | $err .= " $p->{file} ($p->{version})\n"; |
168 | } else { |
169 | # same version declared multiple times, ignore |
170 | } |
171 | } else { |
172 | $file = $p->{file}; |
173 | $version = $p->{version}; |
174 | } |
175 | } |
176 | $file ||= $p->{file} if defined( $p->{file} ); |
177 | } |
dd5a4b10 |
178 | |
5ac756c6 |
179 | if ( $err ) { |
180 | $err = " $file ($version)\n" . $err; |
181 | } |
dd5a4b10 |
182 | |
5ac756c6 |
183 | my %result = ( |
184 | file => $file, |
185 | version => $version, |
186 | err => $err |
187 | ); |
dd5a4b10 |
188 | |
5ac756c6 |
189 | return \%result; |
190 | }; |
191 | |
ca33f3bd |
192 | sub provides { |
193 | my $class = shift; |
194 | |
6f3c7f28 |
195 | croak "provides() requires key/value pairs \n" if @_ % 2; |
ca33f3bd |
196 | my %args = @_; |
197 | |
6f3c7f28 |
198 | croak "provides() takes only one of 'dir' or 'files'\n" |
ca33f3bd |
199 | if $args{dir} && $args{files}; |
200 | |
6f3c7f28 |
201 | croak "provides() requires a 'version' argument" |
c06d0187 |
202 | unless defined $args{version}; |
203 | |
6f3c7f28 |
204 | croak "provides() does not support version '$args{version}' metadata" |
c06d0187 |
205 | unless grep { $args{version} eq $_ } qw/1.4 2/; |
206 | |
ca33f3bd |
207 | $args{prefix} = 'lib' unless defined $args{prefix}; |
208 | |
209 | my $p; |
210 | if ( $args{dir} ) { |
211 | $p = $class->package_versions_from_directory($args{dir}); |
212 | } |
213 | else { |
6f3c7f28 |
214 | croak "provides() requires 'files' to be an array reference\n" |
ca33f3bd |
215 | unless ref $args{files} eq 'ARRAY'; |
216 | $p = $class->package_versions_from_directory($args{files}); |
217 | } |
218 | |
219 | # Now, fix up files with prefix |
220 | if ( length $args{prefix} ) { # check in case disabled with q{} |
221 | $args{prefix} =~ s{/$}{}; |
222 | for my $v ( values %$p ) { |
223 | $v->{file} = "$args{prefix}/$v->{file}"; |
224 | } |
225 | } |
226 | |
227 | return $p |
228 | } |
229 | |
5ac756c6 |
230 | sub package_versions_from_directory { |
231 | my ( $class, $dir, $files ) = @_; |
232 | |
233 | my @files; |
234 | |
235 | if ( $files ) { |
236 | @files = @$files; |
237 | } else { |
238 | find( { |
239 | wanted => sub { |
240 | push @files, $_ if -f $_ && /\.pm$/; |
241 | }, |
242 | no_chdir => 1, |
243 | }, $dir ); |
244 | } |
245 | |
246 | # First, we enumerate all packages & versions, |
247 | # separating into primary & alternative candidates |
248 | my( %prime, %alt ); |
249 | foreach my $file (@files) { |
713aab0e |
250 | my $mapped_filename = File::Spec::Unix->abs2rel( $file, $dir ); |
5ac756c6 |
251 | my @path = split( /\//, $mapped_filename ); |
252 | (my $prime_package = join( '::', @path )) =~ s/\.pm$//; |
dd5a4b10 |
253 | |
5ac756c6 |
254 | my $pm_info = $class->new_from_file( $file ); |
dd5a4b10 |
255 | |
5ac756c6 |
256 | foreach my $package ( $pm_info->packages_inside ) { |
257 | next if $package eq 'main'; # main can appear numerous times, ignore |
258 | next if $package eq 'DB'; # special debugging package, ignore |
259 | next if grep /^_/, split( /::/, $package ); # private package, ignore |
dd5a4b10 |
260 | |
5ac756c6 |
261 | my $version = $pm_info->version( $package ); |
dd5a4b10 |
262 | |
713aab0e |
263 | $prime_package = $package if lc($prime_package) eq lc($package); |
5ac756c6 |
264 | if ( $package eq $prime_package ) { |
265 | if ( exists( $prime{$package} ) ) { |
6f3c7f28 |
266 | croak "Unexpected conflict in '$package'; multiple versions found.\n"; |
5ac756c6 |
267 | } else { |
713aab0e |
268 | $mapped_filename = "$package.pm" if lc("$package.pm") eq lc($mapped_filename); |
5ac756c6 |
269 | $prime{$package}{file} = $mapped_filename; |
270 | $prime{$package}{version} = $version if defined( $version ); |
271 | } |
272 | } else { |
273 | push( @{$alt{$package}}, { |
274 | file => $mapped_filename, |
275 | version => $version, |
276 | } ); |
277 | } |
278 | } |
279 | } |
dd5a4b10 |
280 | |
5ac756c6 |
281 | # Then we iterate over all the packages found above, identifying conflicts |
282 | # and selecting the "best" candidate for recording the file & version |
283 | # for each package. |
284 | foreach my $package ( keys( %alt ) ) { |
285 | my $result = $resolve_module_versions->( $alt{$package} ); |
dd5a4b10 |
286 | |
5ac756c6 |
287 | if ( exists( $prime{$package} ) ) { # primary package selected |
dd5a4b10 |
288 | |
5ac756c6 |
289 | if ( $result->{err} ) { |
290 | # Use the selected primary package, but there are conflicting |
291 | # errors among multiple alternative packages that need to be |
292 | # reported |
293 | log_info { |
294 | "Found conflicting versions for package '$package'\n" . |
295 | " $prime{$package}{file} ($prime{$package}{version})\n" . |
296 | $result->{err} |
297 | }; |
dd5a4b10 |
298 | |
5ac756c6 |
299 | } elsif ( defined( $result->{version} ) ) { |
300 | # There is a primary package selected, and exactly one |
301 | # alternative package |
dd5a4b10 |
302 | |
5ac756c6 |
303 | if ( exists( $prime{$package}{version} ) && |
304 | defined( $prime{$package}{version} ) ) { |
305 | # Unless the version of the primary package agrees with the |
306 | # version of the alternative package, report a conflict |
307 | if ( $compare_versions->( |
308 | $prime{$package}{version}, '!=', $result->{version} |
309 | ) |
310 | ) { |
311 | |
312 | log_info { |
313 | "Found conflicting versions for package '$package'\n" . |
314 | " $prime{$package}{file} ($prime{$package}{version})\n" . |
315 | " $result->{file} ($result->{version})\n" |
316 | }; |
317 | } |
dd5a4b10 |
318 | |
5ac756c6 |
319 | } else { |
320 | # The prime package selected has no version so, we choose to |
321 | # use any alternative package that does have a version |
322 | $prime{$package}{file} = $result->{file}; |
323 | $prime{$package}{version} = $result->{version}; |
324 | } |
dd5a4b10 |
325 | |
5ac756c6 |
326 | } else { |
327 | # no alt package found with a version, but we have a prime |
328 | # package so we use it whether it has a version or not |
329 | } |
dd5a4b10 |
330 | |
5ac756c6 |
331 | } else { # No primary package was selected, use the best alternative |
dd5a4b10 |
332 | |
5ac756c6 |
333 | if ( $result->{err} ) { |
334 | log_info { |
335 | "Found conflicting versions for package '$package'\n" . |
336 | $result->{err} |
337 | }; |
338 | } |
dd5a4b10 |
339 | |
5ac756c6 |
340 | # Despite possible conflicting versions, we choose to record |
341 | # something rather than nothing |
342 | $prime{$package}{file} = $result->{file}; |
343 | $prime{$package}{version} = $result->{version} |
344 | if defined( $result->{version} ); |
345 | } |
346 | } |
dd5a4b10 |
347 | |
5ac756c6 |
348 | # Normalize versions. Can't use exists() here because of bug in YAML::Node. |
868feb8a |
349 | # XXX "bug in YAML::Node" comment seems irrelevant -- dagolden, 2009-05-18 |
5ac756c6 |
350 | for (grep defined $_->{version}, values %prime) { |
351 | $_->{version} = $normalize_version->( $_->{version} ); |
352 | } |
dd5a4b10 |
353 | |
5ac756c6 |
354 | return \%prime; |
355 | } |
dd5a4b10 |
356 | } |
357 | |
5ac756c6 |
358 | |
359 | sub _init { |
360 | my $class = shift; |
361 | my $module = shift; |
362 | my $filename = shift; |
363 | my %props = @_; |
364 | |
f33c0a6c |
365 | my $handle = delete $props{handle}; |
5ac756c6 |
366 | my( %valid_props, @valid_props ); |
367 | @valid_props = qw( collect_pod inc ); |
368 | @valid_props{@valid_props} = delete( @props{@valid_props} ); |
369 | warn "Unknown properties: @{[keys %props]}\n" if scalar( %props ); |
370 | |
371 | my %data = ( |
372 | module => $module, |
373 | filename => $filename, |
374 | version => undef, |
375 | packages => [], |
376 | versions => {}, |
377 | pod => {}, |
378 | pod_headings => [], |
379 | collect_pod => 0, |
380 | |
381 | %valid_props, |
382 | ); |
383 | |
384 | my $self = bless(\%data, $class); |
385 | |
f33c0a6c |
386 | if ( $handle ) { |
387 | $self->_parse_fh($handle); |
388 | } |
389 | else { |
390 | $self->_parse_file(); |
391 | } |
5ac756c6 |
392 | |
393 | unless($self->{module} and length($self->{module})) { |
394 | my ($v, $d, $f) = File::Spec->splitpath($self->{filename}); |
395 | if($f =~ /\.pm$/) { |
396 | $f =~ s/\..+$//; |
397 | my @candidates = grep /$f$/, @{$self->{packages}}; |
398 | $self->{module} = shift(@candidates); # punt |
399 | } |
400 | else { |
401 | if(grep /main/, @{$self->{packages}}) { |
402 | $self->{module} = 'main'; |
403 | } |
404 | else { |
405 | $self->{module} = $self->{packages}[0] || ''; |
406 | } |
407 | } |
408 | } |
409 | |
410 | $self->{version} = $self->{versions}{$self->{module}} |
411 | if defined( $self->{module} ); |
412 | |
413 | return $self; |
414 | } |
415 | |
416 | # class method |
417 | sub _do_find_module { |
418 | my $class = shift; |
6f3c7f28 |
419 | my $module = shift || croak 'find_module_by_name() requires a package name'; |
5ac756c6 |
420 | my $dirs = shift || \@INC; |
421 | |
422 | my $file = File::Spec->catfile(split( /::/, $module)); |
423 | foreach my $dir ( @$dirs ) { |
424 | my $testfile = File::Spec->catfile($dir, $file); |
425 | return [ File::Spec->rel2abs( $testfile ), $dir ] |
426 | if -e $testfile and !-d _; # For stuff like ExtUtils::xsubpp |
427 | return [ File::Spec->rel2abs( "$testfile.pm" ), $dir ] |
428 | if -e "$testfile.pm"; |
429 | } |
430 | return; |
431 | } |
432 | |
433 | # class method |
434 | sub find_module_by_name { |
435 | my $found = shift()->_do_find_module(@_) or return; |
436 | return $found->[0]; |
437 | } |
438 | |
439 | # class method |
440 | sub find_module_dir_by_name { |
441 | my $found = shift()->_do_find_module(@_) or return; |
442 | return $found->[1]; |
443 | } |
444 | |
445 | |
446 | # given a line of perl code, attempt to parse it if it looks like a |
447 | # $VERSION assignment, returning sigil, full name, & package name |
448 | sub _parse_version_expression { |
449 | my $self = shift; |
450 | my $line = shift; |
451 | |
452 | my( $sig, $var, $pkg ); |
69859aa0 |
453 | if ( $line =~ /$VERS_REGEXP/o ) { |
5ac756c6 |
454 | ( $sig, $var, $pkg ) = $2 ? ( $1, $2, $3 ) : ( $4, $5, $6 ); |
455 | if ( $pkg ) { |
456 | $pkg = ($pkg eq '::') ? 'main' : $pkg; |
457 | $pkg =~ s/::$//; |
458 | } |
459 | } |
460 | |
461 | return ( $sig, $var, $pkg ); |
462 | } |
463 | |
464 | sub _parse_file { |
465 | my $self = shift; |
466 | |
467 | my $filename = $self->{filename}; |
468 | my $fh = IO::File->new( $filename ) |
6f3c7f28 |
469 | or croak( "Can't open '$filename': $!" ); |
5ac756c6 |
470 | |
f77c3f08 |
471 | $self->_handle_bom($fh, $filename); |
472 | |
5ac756c6 |
473 | $self->_parse_fh($fh); |
474 | } |
475 | |
f77c3f08 |
476 | # Look for a UTF-8/UTF-16BE/UTF-16LE BOM at the beginning of the stream. |
477 | # If there's one, then skip it and set the :encoding layer appropriately. |
478 | sub _handle_bom { |
479 | my ($self, $fh, $filename) = @_; |
480 | |
481 | my $pos = $fh->getpos; |
482 | return unless defined $pos; |
483 | |
484 | my $buf = ' ' x 2; |
485 | my $count = $fh->read( $buf, length $buf ); |
486 | return unless defined $count and $count >= 2; |
487 | |
488 | my $encoding; |
489 | if ( $buf eq "\x{FE}\x{FF}" ) { |
490 | $encoding = 'UTF-16BE'; |
491 | } elsif ( $buf eq "\x{FF}\x{FE}" ) { |
492 | $encoding = 'UTF-16LE'; |
493 | } elsif ( $buf eq "\x{EF}\x{BB}" ) { |
494 | $buf = ' '; |
495 | $count = $fh->read( $buf, length $buf ); |
496 | if ( defined $count and $count >= 1 and $buf eq "\x{BF}" ) { |
497 | $encoding = 'UTF-8'; |
498 | } |
499 | } |
500 | |
501 | if ( defined $encoding ) { |
502 | if ( "$]" >= 5.008 ) { |
503 | # $fh->binmode requires perl 5.10 |
504 | binmode( $fh, ":encoding($encoding)" ); |
505 | } |
506 | } else { |
507 | $fh->setpos($pos) |
508 | or croak( sprintf "Can't reset position to the top of '$filename'" ); |
509 | } |
510 | |
511 | return $encoding; |
512 | } |
513 | |
5ac756c6 |
514 | sub _parse_fh { |
515 | my ($self, $fh) = @_; |
516 | |
517 | my( $in_pod, $seen_end, $need_vers ) = ( 0, 0, 0 ); |
518 | my( @pkgs, %vers, %pod, @pod ); |
519 | my $pkg = 'main'; |
520 | my $pod_sect = ''; |
521 | my $pod_data = ''; |
30890a12 |
522 | my $in_end = 0; |
5ac756c6 |
523 | |
524 | while (defined( my $line = <$fh> )) { |
525 | my $line_num = $.; |
526 | |
527 | chomp( $line ); |
5ac756c6 |
528 | |
cfedad89 |
529 | # From toke.c : any line that begins by "=X", where X is an alphabetic |
530 | # character, introduces a POD segment. |
921abefc |
531 | my $is_cut; |
cfedad89 |
532 | if ( $line =~ /^=([a-zA-Z].*)/ ) { |
533 | my $cmd = $1; |
534 | # Then it goes back to Perl code for "=cutX" where X is a non-alphabetic |
535 | # character (which includes the newline, but here we chomped it away). |
536 | $is_cut = $cmd =~ /^cut(?:[^a-zA-Z]|$)/; |
921abefc |
537 | $in_pod = !$is_cut; |
538 | } |
5ac756c6 |
539 | |
617f8754 |
540 | if ( $in_pod ) { |
5ac756c6 |
541 | |
a4aafbc2 |
542 | if ( $line =~ /^=head[1-4]\s+(.+)\s*$/ ) { |
5ac756c6 |
543 | push( @pod, $1 ); |
544 | if ( $self->{collect_pod} && length( $pod_data ) ) { |
545 | $pod{$pod_sect} = $pod_data; |
546 | $pod_data = ''; |
547 | } |
548 | $pod_sect = $1; |
549 | |
5ac756c6 |
550 | } elsif ( $self->{collect_pod} ) { |
551 | $pod_data .= "$line\n"; |
552 | |
553 | } |
554 | |
617f8754 |
555 | } elsif ( $is_cut ) { |
5ac756c6 |
556 | |
617f8754 |
557 | if ( $self->{collect_pod} && length( $pod_data ) ) { |
558 | $pod{$pod_sect} = $pod_data; |
559 | $pod_data = ''; |
560 | } |
5ac756c6 |
561 | $pod_sect = ''; |
617f8754 |
562 | |
563 | } else { |
5ac756c6 |
564 | |
30890a12 |
565 | # Skip after __END__ |
566 | next if $in_end; |
567 | |
81ce8c82 |
568 | # Skip comments in code |
569 | next if $line =~ /^\s*#/; |
570 | |
b48c592a |
571 | # Would be nice if we could also check $in_string or something too |
30890a12 |
572 | if ($line eq '__END__') { |
573 | $in_end++; |
574 | next; |
575 | } |
576 | last if $line eq '__DATA__'; |
b48c592a |
577 | |
5ac756c6 |
578 | # parse $line to see if it's a $VERSION declaration |
579 | my( $vers_sig, $vers_fullname, $vers_pkg ) = |
9922478c |
580 | ($line =~ /VERSION/) |
581 | ? $self->_parse_version_expression( $line ) |
582 | : (); |
5ac756c6 |
583 | |
69859aa0 |
584 | if ( $line =~ /$PKG_REGEXP/o ) { |
5ac756c6 |
585 | $pkg = $1; |
586 | push( @pkgs, $pkg ) unless grep( $pkg eq $_, @pkgs ); |
a88210e4 |
587 | $vers{$pkg} = $2 unless exists( $vers{$pkg} ); |
5ac756c6 |
588 | $need_vers = defined $2 ? 0 : 1; |
589 | |
590 | # VERSION defined with full package spec, i.e. $Module::VERSION |
591 | } elsif ( $vers_fullname && $vers_pkg ) { |
592 | push( @pkgs, $vers_pkg ) unless grep( $vers_pkg eq $_, @pkgs ); |
593 | $need_vers = 0 if $vers_pkg eq $pkg; |
594 | |
595 | unless ( defined $vers{$vers_pkg} && length $vers{$vers_pkg} ) { |
596 | $vers{$vers_pkg} = |
597 | $self->_evaluate_version_line( $vers_sig, $vers_fullname, $line ); |
5ac756c6 |
598 | } |
599 | |
600 | # first non-comment line in undeclared package main is VERSION |
601 | } elsif ( !exists($vers{main}) && $pkg eq 'main' && $vers_fullname ) { |
602 | $need_vers = 0; |
603 | my $v = |
604 | $self->_evaluate_version_line( $vers_sig, $vers_fullname, $line ); |
605 | $vers{$pkg} = $v; |
606 | push( @pkgs, 'main' ); |
607 | |
608 | # first non-comment line in undeclared package defines package main |
609 | } elsif ( !exists($vers{main}) && $pkg eq 'main' && $line =~ /\w+/ ) { |
610 | $need_vers = 1; |
611 | $vers{main} = ''; |
612 | push( @pkgs, 'main' ); |
613 | |
614 | # only keep if this is the first $VERSION seen |
615 | } elsif ( $vers_fullname && $need_vers ) { |
616 | $need_vers = 0; |
617 | my $v = |
618 | $self->_evaluate_version_line( $vers_sig, $vers_fullname, $line ); |
619 | |
620 | |
621 | unless ( defined $vers{$pkg} && length $vers{$pkg} ) { |
622 | $vers{$pkg} = $v; |
dd5a4b10 |
623 | } |
5ac756c6 |
624 | |
625 | } |
626 | |
627 | } |
628 | |
629 | } |
630 | |
631 | if ( $self->{collect_pod} && length($pod_data) ) { |
632 | $pod{$pod_sect} = $pod_data; |
633 | } |
634 | |
635 | $self->{versions} = \%vers; |
636 | $self->{packages} = \@pkgs; |
637 | $self->{pod} = \%pod; |
638 | $self->{pod_headings} = \@pod; |
639 | } |
640 | |
641 | { |
642 | my $pn = 0; |
643 | sub _evaluate_version_line { |
644 | my $self = shift; |
645 | my( $sigil, $var, $line ) = @_; |
646 | |
647 | # Some of this code came from the ExtUtils:: hierarchy. |
648 | |
649 | # We compile into $vsub because 'use version' would cause |
650 | # compiletime/runtime issues with local() |
651 | my $vsub; |
652 | $pn++; # everybody gets their own package |
a8a9baef |
653 | my $eval = qq{BEGIN { my \$dummy = q# Hide from _packages_inside() |
5ac756c6 |
654 | #; package Module::Metadata::_version::p$pn; |
4850170c |
655 | use version; |
5ac756c6 |
656 | no strict; |
657 | |
5ac756c6 |
658 | \$vsub = sub { |
398fe5a2 |
659 | local $sigil$var; |
660 | \$$var=undef; |
5ac756c6 |
661 | $line; |
662 | \$$var |
663 | }; |
664 | }}; |
665 | |
666 | local $^W; |
667 | # Try to get the $VERSION |
668 | eval $eval; |
669 | # some modules say $VERSION = $Foo::Bar::VERSION, but Foo::Bar isn't |
670 | # installed, so we need to hunt in ./lib for it |
671 | if ( $@ =~ /Can't locate/ && -d 'lib' ) { |
672 | local @INC = ('lib',@INC); |
673 | eval $eval; |
674 | } |
675 | warn "Error evaling version line '$eval' in $self->{filename}: $@\n" |
676 | if $@; |
677 | (ref($vsub) eq 'CODE') or |
6f3c7f28 |
678 | croak "failed to build version sub for $self->{filename}"; |
5ac756c6 |
679 | my $result = eval { $vsub->() }; |
6f3c7f28 |
680 | croak "Could not get version from $self->{filename} by executing:\n$eval\n\nThe fatal error was: $@\n" |
5ac756c6 |
681 | if $@; |
682 | |
d880ef1f |
683 | # Upgrade it into a version object |
92ad06ed |
684 | my $version = eval { _dwim_version($result) }; |
685 | |
6f3c7f28 |
686 | croak "Version '$result' from $self->{filename} does not appear to be valid:\n$eval\n\nThe fatal error was: $@\n" |
92ad06ed |
687 | unless defined $version; # "0" is OK! |
5ac756c6 |
688 | |
92ad06ed |
689 | return $version; |
5ac756c6 |
690 | } |
691 | } |
692 | |
92ad06ed |
693 | # Try to DWIM when things fail the lax version test in obvious ways |
694 | { |
695 | my @version_prep = ( |
696 | # Best case, it just works |
697 | sub { return shift }, |
698 | |
699 | # If we still don't have a version, try stripping any |
700 | # trailing junk that is prohibited by lax rules |
701 | sub { |
702 | my $v = shift; |
703 | $v =~ s{([0-9])[a-z-].*$}{$1}i; # 1.23-alpha or 1.23b |
704 | return $v; |
705 | }, |
706 | |
707 | # Activestate apparently creates custom versions like '1.23_45_01', which |
708 | # cause version.pm to think it's an invalid alpha. So check for that |
709 | # and strip them |
710 | sub { |
711 | my $v = shift; |
712 | my $num_dots = () = $v =~ m{(\.)}g; |
713 | my $num_unders = () = $v =~ m{(_)}g; |
714 | my $leading_v = substr($v,0,1) eq 'v'; |
715 | if ( ! $leading_v && $num_dots < 2 && $num_unders > 1 ) { |
716 | $v =~ s{_}{}g; |
717 | $num_unders = () = $v =~ m{(_)}g; |
718 | } |
719 | return $v; |
720 | }, |
721 | |
722 | # Worst case, try numifying it like we would have before version objects |
723 | sub { |
724 | my $v = shift; |
725 | no warnings 'numeric'; |
726 | return 0 + $v; |
727 | }, |
728 | |
729 | ); |
730 | |
731 | sub _dwim_version { |
732 | my ($result) = shift; |
733 | |
734 | return $result if ref($result) eq 'version'; |
735 | |
736 | my ($version, $error); |
737 | for my $f (@version_prep) { |
738 | $result = $f->($result); |
739 | $version = eval { version->new($result) }; |
740 | $error ||= $@ if $@; # capture first failure |
741 | last if defined $version; |
742 | } |
743 | |
6f3c7f28 |
744 | croak $error unless defined $version; |
92ad06ed |
745 | |
746 | return $version; |
747 | } |
748 | } |
5ac756c6 |
749 | |
750 | ############################################################ |
751 | |
752 | # accessors |
b50a9801 |
753 | sub name { $_[0]->{module} } |
5ac756c6 |
754 | |
b50a9801 |
755 | sub filename { $_[0]->{filename} } |
756 | sub packages_inside { @{$_[0]->{packages}} } |
757 | sub pod_inside { @{$_[0]->{pod_headings}} } |
758 | sub contains_pod { 0+@{$_[0]->{pod_headings}} } |
5ac756c6 |
759 | |
760 | sub version { |
761 | my $self = shift; |
762 | my $mod = shift || $self->{module}; |
763 | my $vers; |
764 | if ( defined( $mod ) && length( $mod ) && |
765 | exists( $self->{versions}{$mod} ) ) { |
766 | return $self->{versions}{$mod}; |
767 | } else { |
768 | return undef; |
769 | } |
770 | } |
771 | |
772 | sub pod { |
773 | my $self = shift; |
774 | my $sect = shift; |
775 | if ( defined( $sect ) && length( $sect ) && |
776 | exists( $self->{pod}{$sect} ) ) { |
777 | return $self->{pod}{$sect}; |
778 | } else { |
779 | return undef; |
780 | } |
781 | } |
782 | |
783 | 1; |
784 | |
5ac756c6 |
785 | =head1 NAME |
786 | |
2c11e51d |
787 | Module::Metadata - Gather package and POD information from perl module files |
5ac756c6 |
788 | |
6290f67c |
789 | =head1 SYNOPSIS |
790 | |
791 | use Module::Metadata; |
792 | |
793 | # information about a .pm file |
794 | my $info = Module::Metadata->new_from_file( $file ); |
795 | my $version = $info->version; |
796 | |
ca33f3bd |
797 | # CPAN META 'provides' field for .pm files in a directory |
c06d0187 |
798 | my $provides = Module::Metadata->provides( |
799 | dir => 'lib', version => 2 |
800 | ); |
6290f67c |
801 | |
5ac756c6 |
802 | =head1 DESCRIPTION |
803 | |
c0278e58 |
804 | This module provides a standard way to gather metadata about a .pm file through |
805 | (mostly) static analysis and (some) code execution. When determining the |
806 | version of a module, the C<$VERSION> assignment is C<eval>ed, as is traditional |
807 | in the CPAN toolchain. |
6290f67c |
808 | |
809 | =head1 USAGE |
810 | |
811 | =head2 Class methods |
812 | |
5ac756c6 |
813 | =over 4 |
814 | |
6290f67c |
815 | =item C<< new_from_file($filename, collect_pod => 1) >> |
5ac756c6 |
816 | |
4c401e3f |
817 | Constructs a C<Module::Metadata> object given the path to a file. Returns |
818 | undef if the filename does not exist. |
819 | |
820 | C<collect_pod> is a optional boolean argument that determines whether POD |
821 | data is collected and stored for reference. POD data is not collected by |
822 | default. POD headings are always collected. |
823 | |
824 | If the file begins by an UTF-8, UTF-16BE or UTF-16LE byte-order mark, then |
825 | it is skipped before processing, and the content of the file is also decoded |
826 | appropriately starting from perl 5.8. |
5ac756c6 |
827 | |
6290f67c |
828 | =item C<< new_from_handle($handle, $filename, collect_pod => 1) >> |
f33c0a6c |
829 | |
830 | This works just like C<new_from_file>, except that a handle can be provided |
4c401e3f |
831 | as the first argument. |
832 | |
833 | Note that there is no validation to confirm that the handle is a handle or |
834 | something that can act like one. Passing something that isn't a handle will |
835 | cause a exception when trying to read from it. The C<filename> argument is |
836 | mandatory or undef will be returned. |
837 | |
838 | You are responsible for setting the decoding layers on C<$handle> if |
839 | required. |
f33c0a6c |
840 | |
6290f67c |
841 | =item C<< new_from_module($module, collect_pod => 1, inc => \@dirs) >> |
5ac756c6 |
842 | |
4c401e3f |
843 | Constructs a C<Module::Metadata> object given a module or package name. |
844 | Returns undef if the module cannot be found. |
845 | |
846 | In addition to accepting the C<collect_pod> argument as described above, |
847 | this method accepts a C<inc> argument which is a reference to an array of |
848 | directories to search for the module. If none are given, the default is |
849 | @INC. |
850 | |
851 | If the file that contains the module begins by an UTF-8, UTF-16BE or |
852 | UTF-16LE byte-order mark, then it is skipped before processing, and the |
853 | content of the file is also decoded appropriately starting from perl 5.8. |
5ac756c6 |
854 | |
6290f67c |
855 | =item C<< find_module_by_name($module, \@dirs) >> |
5ac756c6 |
856 | |
857 | Returns the path to a module given the module or package name. A list |
858 | of directories can be passed in as an optional parameter, otherwise |
859 | @INC is searched. |
860 | |
861 | Can be called as either an object or a class method. |
862 | |
6290f67c |
863 | =item C<< find_module_dir_by_name($module, \@dirs) >> |
5ac756c6 |
864 | |
865 | Returns the entry in C<@dirs> (or C<@INC> by default) that contains |
866 | the module C<$module>. A list of directories can be passed in as an |
867 | optional parameter, otherwise @INC is searched. |
868 | |
869 | Can be called as either an object or a class method. |
870 | |
ca33f3bd |
871 | =item C<< provides( %options ) >> |
872 | |
873 | This is a convenience wrapper around C<package_versions_from_directory> |
874 | to generate a CPAN META C<provides> data structure. It takes key/value |
875 | pairs. Valid option keys include: |
876 | |
877 | =over |
878 | |
c06d0187 |
879 | =item version B<(required)> |
880 | |
881 | Specifies which version of the L<CPAN::Meta::Spec> should be used as |
882 | the format of the C<provides> output. Currently only '1.4' and '2' |
883 | are supported (and their format is identical). This may change in |
884 | the future as the definition of C<provides> changes. |
885 | |
886 | The C<version> option is required. If it is omitted or if |
887 | an unsupported version is given, then C<provides> will throw an error. |
888 | |
ca33f3bd |
889 | =item dir |
890 | |
891 | Directory to search recursively for F<.pm> files. May not be specified with |
892 | C<files>. |
893 | |
894 | =item files |
895 | |
896 | Array reference of files to examine. May not be specified with C<dir>. |
897 | |
898 | =item prefix |
899 | |
900 | String to prepend to the C<file> field of the resulting output. This defaults |
901 | to F<lib>, which is the common case for most CPAN distributions with their |
902 | F<.pm> files in F<lib>. This option ensures the META information has the |
903 | correct relative path even when the C<dir> or C<files> arguments are |
904 | absolute or have relative paths from a location other than the distribution |
905 | root. |
906 | |
907 | =back |
908 | |
909 | For example, given C<dir> of 'lib' and C<prefix> of 'lib', the return value |
910 | is a hashref of the form: |
911 | |
912 | { |
913 | 'Package::Name' => { |
914 | version => '0.123', |
915 | file => 'lib/Package/Name.pm' |
916 | }, |
917 | 'OtherPackage::Name' => ... |
918 | } |
919 | |
6290f67c |
920 | =item C<< package_versions_from_directory($dir, \@files?) >> |
2c11e51d |
921 | |
922 | Scans C<$dir> for .pm files (unless C<@files> is given, in which case looks |
923 | for those files in C<$dir> - and reads each file for packages and versions, |
924 | returning a hashref of the form: |
5ac756c6 |
925 | |
2c11e51d |
926 | { |
927 | 'Package::Name' => { |
928 | version => '0.123', |
929 | file => 'Package/Name.pm' |
930 | }, |
931 | 'OtherPackage::Name' => ... |
932 | } |
933 | |
ca33f3bd |
934 | The C<DB> and C<main> packages are always omitted, as are any "private" |
935 | packages that have leading underscores in the namespace (e.g. |
936 | C<Foo::_private>) |
937 | |
938 | Note that the file path is relative to C<$dir> if that is specified. |
939 | This B<must not> be used directly for CPAN META C<provides>. See |
940 | the C<provides> method instead. |
941 | |
6290f67c |
942 | =item C<< log_info (internal) >> |
2c11e51d |
943 | |
944 | Used internally to perform logging; imported from Log::Contextual if |
945 | Log::Contextual has already been loaded, otherwise simply calls warn. |
946 | |
947 | =back |
5ac756c6 |
948 | |
6290f67c |
949 | =head2 Object methods |
950 | |
951 | =over 4 |
952 | |
953 | =item C<< name() >> |
954 | |
955 | Returns the name of the package represented by this module. If there |
956 | are more than one packages, it makes a best guess based on the |
957 | filename. If it's a script (i.e. not a *.pm) the package name is |
958 | 'main'. |
959 | |
960 | =item C<< version($package) >> |
961 | |
962 | Returns the version as defined by the $VERSION variable for the |
963 | package as returned by the C<name> method if no arguments are |
964 | given. If given the name of a package it will attempt to return the |
965 | version of that package if it is specified in the file. |
966 | |
967 | =item C<< filename() >> |
968 | |
969 | Returns the absolute path to the file. |
970 | |
971 | =item C<< packages_inside() >> |
972 | |
1f67ffd7 |
973 | Returns a list of packages. Note: this is a raw list of packages |
974 | discovered (or assumed, in the case of C<main>). It is not |
975 | filtered for C<DB>, C<main> or private packages the way the |
8e4bef01 |
976 | C<provides> method does. Invalid package names are not returned, |
977 | for example "Foo:Bar". Strange but valid package names are |
978 | returned, for example "Foo::Bar::", and are left up to the caller |
979 | on how to handle. |
6290f67c |
980 | |
981 | =item C<< pod_inside() >> |
982 | |
983 | Returns a list of POD sections. |
984 | |
985 | =item C<< contains_pod() >> |
986 | |
987 | Returns true if there is any POD in the file. |
988 | |
989 | =item C<< pod($section) >> |
990 | |
991 | Returns the POD data in the given section. |
992 | |
993 | =back |
994 | |
5ac756c6 |
995 | =head1 AUTHOR |
996 | |
6290f67c |
997 | Original code from Module::Build::ModuleInfo by Ken Williams |
998 | <kwilliams@cpan.org>, Randy W. Sims <RandyS@ThePierianSpring.org> |
5ac756c6 |
999 | |
2c11e51d |
1000 | Released as Module::Metadata by Matt S Trout (mst) <mst@shadowcat.co.uk> with |
6290f67c |
1001 | assistance from David Golden (xdg) <dagolden@cpan.org>. |
5ac756c6 |
1002 | |
4c401e3f |
1003 | =head1 COPYRIGHT & LICENSE |
5ac756c6 |
1004 | |
6290f67c |
1005 | Original code Copyright (c) 2001-2011 Ken Williams. |
1006 | Additional code Copyright (c) 2010-2011 Matt Trout and David Golden. |
1007 | All rights reserved. |
5ac756c6 |
1008 | |
1009 | This library is free software; you can redistribute it and/or |
1010 | modify it under the same terms as Perl itself. |
1011 | |
5ac756c6 |
1012 | =cut |
1013 | |