8 # Do not require test.pl, this file has its own framework.
13 use feature 'unicode_strings';
24 if ( $Config{usecrosscompile} ) {
25 print "1..0 # Not all files are available during cross-compilation\n";
28 if ($^O eq 'dec_osf') {
29 print "1..0 # $^O cannot handle this test\n";
32 require '../regen/regen_lib.pl';
41 podcheck.t - Look for possible problems in the Perl pods
46 ./perl -I../lib porting/podcheck.t [--show_all] [--cpan] [--deltas]
47 [--counts] [--pedantic] [FILE ...]
49 ./perl -I../lib porting/podcheck.t --add_link MODULE ...
51 ./perl -I../lib porting/podcheck.t --regen
55 podcheck.t is an extension of Pod::Checker. It looks for pod errors and
56 potential errors in the files given as arguments, or if none specified, in all
57 pods in the distribution workspace, except certain known special ones
58 (specified below). It does additional checking beyond that done by
59 Pod::Checker, and keeps a database of known potential problems, and will
60 fail a pod only if the number of such problems differs from that given in the
63 The additional checks it always makes are:
67 =item Cross-pod link checking
69 Pod::Checker verifies that links to an internal target in a pod are not
70 broken. podcheck.t extends that (when called without FILE arguments) to
71 external links. It does this by gathering up all the possible targets in the
72 workspace, and cross-checking them. It also checks that a non-broken link
73 points to just one target. (The destination pod could have two targets with
76 The way that the C<LE<lt>E<gt>> pod command works (for links outside the pod)
77 is to actually create a link to C<search.cpan.org> with an embedded query for
78 the desired pod or man page. That means that links outside the distribution
79 are valid. podcheck.t doesn't verify the validity of such links, but instead
80 keeps a database of those known to be valid. This means that if a link to a
81 target not on the list is created, the target needs to be added to the data
82 base. This is accomplished via the L<--add_link|/--add_link MODULE ...>
83 option to podcheck.t, described below.
85 =item An internal link that isn't so specified
87 If a link is broken, but there is an existing internal target of the same
88 name, it is likely that the internal target was meant, and the C<"/"> is
89 missing from the C<LE<lt>E<gt>> pod command.
91 =item Missing or duplicate NAME or missing NAME short description
93 A pod can't be linked to unless it has a unique name.
94 And a NAME should have a dash and short description after it.
96 If the C<PERL_POD_PEDANTIC> environment variable is set or the C<--pedantic>
97 command line argument is provided then a few more checks are made.
98 The pedantic checks are:
102 =item Verbatim paragraphs that wrap in an 80 (including 1 spare) column window
104 It's annoying to have lines wrap when displaying pod documentation in a
105 terminal window. This checks that all verbatim lines fit in a standard 80
106 column window, even when using a pager that reserves a column for its own use.
107 (Thus the check is for a net of 79 columns.)
108 For those lines that don't fit, it tells you how much needs to be cut in
111 Often, the easiest thing to do to gain space for these is to lower the indent
114 =item Items that perhaps should be links
116 There are mentions of apparent files in the pods that perhaps should be links
117 instead, using C<LE<lt>...E<gt>>
119 =item Items that perhaps should be C<FE<lt>...E<gt>>
121 What look like path names enclosed in C<CE<lt>...E<gt>> should perhaps have
122 C<FE<lt>...E<gt>> mark-up instead.
126 A number of issues raised by podcheck.t and by the base Pod::Checker are not
127 really problems, but merely potential problems, that is, false positives.
128 After inspecting them and
129 deciding that they aren't real problems, it is possible to shut up this program
130 about them, unlike base Pod::Checker. For a valid link to an outside module
131 or man page, call podcheck.t with the C<--add_link> option to add it to the
132 the database of known links; for other causes, call podcheck.t with the C<--regen>
133 option to regenerate the entire database. This tells it that all existing
134 issues are to not be mentioned again.
136 C<--regen> isn't fool-proof. The database merely keeps track of the number of these
137 potential problems of each type for each pod. If a new problem of a given
138 type is introduced into the pod, podcheck.t will spit out all of them. You
139 then have to figure out which is the new one, and should it be changed or not.
140 But doing it this way insulates the database from having to keep track of line
141 numbers of problems, which may change, or the exact wording of each problem
142 which might also change without affecting whether it is a problem or not.
144 Also, if the count of potential problems of a given type for a pod decreases,
145 the database must be regenerated so that it knows the new number. The program
146 gives instructions when this happens.
148 Some pods will have varying numbers of problems of a given type. This can
149 be handled by manually editing the database file (see L</FILES>), and setting
150 the number of those problems for that pod to a negative number. This will
151 cause the corresponding error to always be suppressed no matter how many there
154 Another problem is that there is currently no check that modules listed as
155 valid in the database
156 actually are. Thus any errors introduced there will remain there.
160 =head2 Specially handled pods
166 This pod is generated by pasting bits from other pods. Errors in those bits
167 will show up as errors here, as well as for those other pods. Therefore
168 errors here are suppressed, and the pod is checked only to verify that nodes
169 within it actually exist that are externally linked to.
173 The current perldelta pod is initialized from a template that contains
174 placeholder text. Some of this text is in the form of links that don't really
175 exist. Any such links that are listed in C<@perldelta_ignore_links> will not
176 generate messages. It is presumed that these links will be cleaned up when
177 the perldelta is cleaned up for release since they should be marked with
180 =item Porting/perldelta_template.pod
182 This is not a pod, but a template for C<perldelta>. Any errors introduced
183 here will show up when C<perldelta> is created from it.
185 =item cpan-upstream pods
187 See the L</--cpan> option documentation
191 See the L</--deltas> option documentation
199 =item --add_link MODULE ...
201 Use this option to teach podcheck.t that the C<MODULE>s or man pages actually
202 exist, and to silence any messages that links to them are broken.
204 podcheck.t checks that links within the Perl core distribution are valid, but
205 it doesn't check links to man pages or external modules. When it finds
206 a broken link, it checks its database of external modules and man pages,
207 and only if not found there does it raise a message. This option just adds
208 the list of modules and man page references that follow it on the command line
214 ./perl -I../lib porting/podcheck.t --add_link Unicode::Casing
216 causes the external module "Unicode::Casing" to be added to the database, so
217 C<LE<lt>Unicode::CasingE<gt>> will be considered valid.
221 Regenerate the database used by podcheck.t to include all the existing
222 potential problems. Future runs of the program will not then flag any of
223 these. Setting this option also sets C<--pedantic>.
227 Normally, all pods in the cpan directory are skipped, except to make sure that
228 any blead-upstream links to such pods are valid.
229 This option will cause cpan upstream pods to be fully checked.
233 Normally, all old perldelta pods are skipped, except to make sure that
234 any links to such pods are valid. This is because they are considered
235 stable, and perhaps trying to fix them will cause changes that will
236 misrepresent Perl's history. But, this option will cause them to be fully
241 Normally, if the number of potential problems of a given type found for a
242 pod matches the expected value in the database, they will not be displayed.
243 This option forces the database to be ignored during the run, so all potential
244 problems are displayed and will fail their respective pod test. Specifying
245 any particular FILES to operate on automatically selects this option.
249 Instead of testing, this just dumps the counts of the occurrences of the
250 various types of potential problems in the database.
254 There are three potential problems that are not checked for by default.
255 This options enables them. The environment variable C<PERL_POD_PEDANTIC>
256 can be set to 1 to enable this option also.
257 This option is set when C<--regen> is used.
263 The database is stored in F<t/porting/known_pod_issues.dat>
271 # VMS builds have a '.com' appended to utility and script names, and it adds a
272 # trailing dot for any other file name that doesn't have a dot in it. The db
273 # is stored without those things. This regex allows for these special file
274 # names to be dealt with. It needs to be interpolated into a larger regex
275 # that furnishes the closing boundary.
276 my $vms_re = qr/ \. (?: com )? /x;
278 # Some filenames in the MANIFEST match $vms_re, and so must not be handled the
279 # same way that that the special vms ones are. This hash lists those.
280 my %special_vms_files;
282 # This is to get this to work across multiple file systems, including those
283 # that are not case sensitive. The db is stored in lower case, Un*x style,
284 # and all file name comparisons are done that way.
285 sub canonicalize($) {
287 my ($volume, $directories, $file)
288 = File::Spec->splitpath(File::Spec->canonpath($input));
289 # Assumes $volume is constant for everything in this directory structure
290 $directories = "" if ! $directories;
291 $file = "" if ! $file;
292 $file = lc join '/', File::Spec->splitdir($directories), $file;
293 $file =~ s! / /+ !/!gx; # Multiple slashes => single slash
295 # The db is stored without the special suffixes that are there in VMS, so
296 # strip them off to get the comparable name. But some files on all
297 # platforms have these suffixes, so this shouldn't happen for them, as any
298 # of their db entries will have the suffixes in them. The hash has been
299 # populated with these files.
301 && $file =~ / ( $vms_re ) $ /x
302 && ! exists $special_vms_files{$file})
304 $file =~ s/ $1 $ //x;
309 #####################################################
310 # HOW IT WORKS (in general)
312 # If not called with specific files to check, the directory structure is
313 # examined for files that have pods in them. Files that might not have to be
314 # fully parsed (e.g. in cpan) are parsed enough at this time to find their
315 # pod's NAME, and to get a checksum.
317 # Those kinds of files are sorted last, but otherwise the pods are parsed with
318 # the package coded here, My::Pod::Checker, which is an extension to
319 # Pod::Checker that adds some tests and suppresses others that aren't
320 # appropriate. The latter module has no provision for capturing diagnostics,
321 # so a package, Tie_Array_to_FH, is used to force them to be placed into an
322 # array instead of printed.
324 # Parsing the files builds up a list of links. The files are gone through
325 # again, doing cross-link checking and outputting all saved-up problems with
328 # Sorting the files last that potentially don't need to be fully parsed allows
329 # us to not parse them unless there is a link to an internal anchor in them
330 # from something that we have already parsed. Keeping checksums allows us to
331 # not parse copies of other pods.
333 #####################################################
335 # 1 => Exclude low priority messages that aren't likely to be problems, and
336 # has many false positives; higher numbers give more messages.
337 my $Warnings_Level = 200;
339 # perldelta during construction may have place holder links. N.B. This
340 # variable is referred to by name in release_managers_guide.pod
341 our @perldelta_ignore_links = ( "XXX", "perl5YYYdelta", "perldiag/message" );
343 # To see if two pods with the same NAME are actually copies of the same pod,
344 # which is not an error, it uses a checksum to save work.
345 my $digest_type = "SHA-1";
347 my $original_dir = File::Spec->rel2abs(File::Spec->curdir);
348 my $data_dir = File::Spec->catdir($original_dir, 'porting');
349 my $known_issues = File::Spec->catfile($data_dir, 'known_pod_issues.dat');
350 my $MANIFEST = File::Spec->catfile(File::Spec->updir($original_dir), 'MANIFEST');
353 my $MAX_LINE_LENGTH = 79; # 79 columns
354 my $INDENT = 7; # default nroff indent
356 # Our warning messages. Better not have [('"] in them, as those are used as
357 # delimiters for variable parts of the messages by poderror.
358 my $broken_link = "Apparent broken link";
359 my $broken_internal_link = "Apparent internal link is missing its forward slash";
360 my $multiple_targets = "There is more than one target";
361 my $duplicate_name = "Pod NAME already used";
362 my $no_name = "There is no NAME";
363 my $missing_name_description = "The NAME should have a dash and short description after it";
364 # the pedantic warnings messages
365 my $line_length = "Verbatim line length including indents exceeds $MAX_LINE_LENGTH by";
366 my $C_not_linked = "? Should you be using L<...> instead of";
367 my $C_with_slash = "? Should you be using F<...> or maybe L<...> instead of";
369 # objects, tests, etc can't be pods, so don't look for them. Also skip
370 # files output by the patch program. Could also ignore most of .gitignore
371 # files, but not all, so don't.
373 my $obj_ext = $Config{'obj_ext'}; $obj_ext =~ tr/.//d; # dot will be added back
374 my $lib_ext = $Config{'lib_ext'}; $lib_ext =~ tr/.//d;
375 my $lib_so = $Config{'so'}; $lib_so =~ tr/.//d;
376 my $dl_ext = $Config{'dlext'}; $dl_ext =~ tr/.//d;
378 # Not really pods, but can look like them.
379 my %excluded_files = (
380 canonicalize("lib/unicore/mktables") => 1,
381 canonicalize("Porting/make-rmg-checklist") => 1,
382 canonicalize("Porting/perldelta_template.pod") => 1,
383 canonicalize("regen/feature.pl") => 1,
384 canonicalize("regen/warnings.pl") => 1,
385 canonicalize("autodoc.pl") => 1,
386 canonicalize("configpm") => 1,
387 canonicalize("miniperl") => 1,
388 canonicalize("perl") => 1,
389 canonicalize('cpan/Pod-Perldoc/corpus/no-head.pod') => 1,
390 canonicalize('cpan/Pod-Perldoc/corpus/perlfunc.pod') => 1,
391 canonicalize('cpan/Pod-Perldoc/corpus/utf8.pod') => 1,
392 canonicalize("lib/unicore/mktables") => 1,
395 # This list should not include anything for which case sensitivity is
396 # important, as it won't work on VMS, and won't show up until tested on VMS.
397 # All or almost all such files should be listed in the MANIFEST, so that can
398 # be examined for them, and each such file explicitly excluded, as is done for
399 # .PL files in the loop just below this. For files not catchable this way,
400 # is_pod_file() can be used to exclude these at a finer grained level.
401 my $non_pods = qr/ (?: \.
402 (?: [achot] | zip | gz | bz2 | jar | tar | tgz
403 | orig | rej | patch # Patch program output
404 | sw[op] | \#.* # Editor droppings
405 | old # buildtoc output
406 | xs # pod should be in the .pm file
407 | al # autosplit files
408 | bs # bootstrap files
409 | (?i:sh) # shell scripts, hints, templates
410 | lst # assorted listing files
411 | bat # Windows,Netware,OS2 batch files
412 | cmd # Windows,Netware,OS2 command files
413 | lis # VMS compiler listings
414 | map # VMS linker maps
415 | opt # VMS linker options files
416 | mms # MM(K|S) description files
417 | ts # timestamp files generated during build
418 | $obj_ext # object files
419 | exe # $Config{'exe_ext'} might be empty string
420 | $lib_ext # object libraries
421 | $lib_so # shared libraries
422 | $dl_ext # dynamic libraries
423 | gif # GIF images (example files from CGI.pm)
424 | eg # examples from libnet
428 ) | ~$ | \ \(Autosaved\)\.txt$ # Other editor droppings
429 | ^cxx\$demangler_db\.$ # VMS name mangler database
430 | ^typemap\.?$ # typemap files
431 | ^(?i:Makefile\.PL)$
434 # Matches something that looks like a file name, but is enclosed in C<...>
435 my $C_path_re = qr{ ^
436 # exclude various things that have slashes
437 # in them but aren't paths
439 (?: (?: s | qr | m | tr | y ) / ) # regexes
440 | \d+/\d+ \b # probable fractions
448 /? # Optional initial slash
449 \w+ # First component of path, doesn't begin with
451 (?: / [-\w]+ )+ # Subsequent path components
452 (?: \. \w+ )? # Optional trailing dot and suffix
453 >* # Any enclosed L< F< have matching closing >
457 # '.PL' files should be excluded, as they aren't final pods, but often contain
458 # material used in generating pods, and so can look like a pod. We can't use
459 # the regexp above because case sensisitivity is important for these, as some
460 # '.pl' files should be examined for pods. Instead look through the MANIFEST
461 # for .PL files and get their full path names, so we can exclude each such
462 # file explicitly. This works because other porting tests prohibit having two
463 # files with the same names except for case.
464 open my $manifest_fh, '<:bytes', $MANIFEST or die "Can't open $MANIFEST";
465 while (<$manifest_fh>) {
467 # While we have MANIFEST open, on VMS platforms, look for files that match
468 # the magic VMS file names that have to be handled specially. Add these
469 # to the list of them.
470 if ($^O eq 'VMS' && / ^ ( [^\t]* $vms_re ) \t /x) {
471 $special_vms_files{$1} = 1;
473 if (/ ^ ( [^\t]* \. PL ) \t /x) {
474 $excluded_files{canonicalize($1)} = 1;
477 close $manifest_fh, or die "Can't close $MANIFEST";
480 # Pod::Checker messages to suppress
481 my @suppressed_messages = (
482 # We catch independently the ones that are real problems.
483 qr/multiple occurrences \(\d+\) of link target/,
485 "unescaped <>", # Not every '<' or '>' need be escaped
486 qr/No items in =over/, # i.e., a blockquote, which we consider legal
490 # Returns bool as to if input message is one that is to be suppressed
494 return grep { $message =~ /^$_/i } @suppressed_messages;
497 { # Closure to contain a simple subset of test.pl. This is to get rid of the
498 # unnecessary 'failed at' messages that would otherwise be output pointing
499 # to a particular line in this file.
501 my $current_test = 0;
506 $planned = $plan{tests} + 1; # +1 for final test that files haven't
508 print "1..$planned\n";
519 print "not " unless $success;
520 print "ok $current_test - $message\n";
526 my $n = @_ ? shift : 1;
529 print "ok $current_test # skip $why\n";
531 no warnings 'exiting';
536 my ($andle, $message) = @_;
540 print $andle $message =~ s/^/# /mgr;
545 sub note { unshift @_, \*STDOUT; goto &_note }
547 sub diag { unshift @_, \*STDERR; goto &_note }
550 if ($planned && $planned != $current_test) {
552 "# Looks like you planned $planned tests but ran $current_test.\n";
557 # List of known potential problems by pod and type.
560 # Pods given by the keys contain an interior node that is referred to from
562 my %has_referred_to_node;
570 my $do_upstream_cpan = 0; # Assume that are to skip anything in /cpan
571 my $do_deltas = 0; # And stable perldeltas
573 while (@ARGV && substr($ARGV[0], 0, 1) eq '-') {
574 my $arg = shift @ARGV;
576 $arg =~ s/^--/-/; # Treat '--' the same as a single '-'
577 if ($arg eq '-regen') {
581 elsif ($arg eq '-add_link') {
584 elsif ($arg eq '-cpan') {
585 $do_upstream_cpan = 1;
587 elsif ($arg eq '-deltas') {
590 elsif ($arg eq '-show_all') {
593 elsif ($arg eq '-counts') {
596 elsif ($arg eq '-pedantic') {
601 Unknown option '$arg'
603 Usage: $0 [ --regen | --cpan | --show_all | FILE ... | --add_link MODULE ... ]\n"
604 --add_link -> Add the MODULE and man page references to the database
605 --regen -> Regenerate the data file for $0
606 --cpan -> Include files in the cpan subdirectory.
607 --deltas -> Include stable perldeltas
608 --show_all -> Show all known potential problems
609 --counts -> Don't test, but give summary counts of the currently
611 --pedantic -> Check for overly long lines in verbatim blocks
616 $pedantic = 1 if exists $ENV{PERL_POD_PEDANTIC} and $ENV{PERL_POD_PEDANTIC};
619 my $cpan_or_deltas = $do_upstream_cpan || $do_deltas;
620 if (($regen + $show_all + $show_counts + $add_link + $cpan_or_deltas ) > 1) {
621 croak "--regen, --show_all, --counts, and --add_link are mutually exclusive\n and none can be run with --cpan nor --deltas";
624 my $has_input_files = @files;
628 if (! $has_input_files) {
629 croak "--add_link requires at least one module or man page reference";
632 elsif ($has_input_files) {
633 if ($regen || $show_counts || $do_upstream_cpan || $do_deltas) {
634 croak "--regen, --counts, --deltas, and --cpan can't be used since using specific files";
636 foreach my $file (@files) {
637 croak "Can't read file '$file'" if ! -r $file;
641 our %problems; # potential problems found in this run
643 package My::Pod::Checker { # Extend Pod::Checker
644 use parent 'Pod::Checker';
646 # Uses inside out hash to protect from typos
647 # For new fields, remember to add to destructor DESTROY()
648 my %CFL_text; # The text comprising the current C<>, F<>, or L<>
649 my %C_text; # If defined, are in a C<> section, and includes
650 # the accumulated text from that
651 my %current_indent; # Current line's indent
652 my %filename; # The pod is store in this file
653 my %in_CFL; # count of stacked C<>, F<>, L<> directives
654 my %indents; # Stack of indents from =over's in effect for
656 my %in_for; # true if in a =for or =begin
657 my %in_NAME; # true if within NAME section
658 my %in_begin; # true if within =begin section
659 my %in_X; # true if in a X<>
660 my %linkable_item; # Bool: if the latest =item is linkable. It isn't
661 # for bullet and number lists
662 my %linkable_nodes; # Pod::Checker adds all =items to its node list,
663 # but not all =items are linkable to
664 my %running_CFL_text; # The current text that is being accumulated until
665 # an end_FOO is found, and this includes any C<>,
666 # F<>, or L<> directives.
667 my %running_simple_text; # The currentt text that is being accumulated
668 # until an end_FOO is found, and all directives
669 # have been expanded into plain text
670 my %command_count; # Number of commands seen
671 my %seen_pod_cmd; # true if have =pod earlier
672 my %skip; # is SKIP set for this pod
673 my %start_line; # the first input line number in the the thing
674 # currently being worked on
677 my $addr = Scalar::Util::refaddr $_[0];
678 delete $CFL_text{$addr};
679 delete $C_text{$addr};
680 delete $command_count{$addr};
681 delete $current_indent{$addr};
682 delete $filename{$addr};
683 delete $in_begin{$addr};
684 delete $in_CFL{$addr};
685 delete $indents{$addr};
686 delete $in_for{$addr};
687 delete $in_NAME{$addr};
689 delete $linkable_item{$addr};
690 delete $linkable_nodes{$addr};
691 delete $running_CFL_text{$addr};
692 delete $running_simple_text{$addr};
693 delete $seen_pod_cmd{$addr};
695 delete $start_line{$addr};
701 my $filename = shift;
703 my $self = $class->SUPER::new(-quiet => 1,
704 -warnings => $Warnings_Level);
705 my $addr = Scalar::Util::refaddr $self;
706 $command_count{$addr} = 0;
707 $current_indent{$addr} = 0;
708 $filename{$addr} = $filename;
709 $in_begin{$addr} = 0;
713 $linkable_item{$addr} = 0;
714 $seen_pod_cmd{$addr} = 0;
718 # re's for messages that Pod::Checker outputs
719 my $location = qr/ \b (?:in|at|on|near) \s+ /xi;
720 my $optional_location = qr/ (?: $location )? /xi;
721 my $line_reference = qr/ [('"]? $optional_location \b line \s+
722 (?: \d+ | EOF | \Q???\E | - )
725 sub poderror { # Called to register a potential problem
727 # This adds an extra field to the parent hash, 'parameter'. It is
728 # used to extract the variable parts of a message leaving just the
729 # constant skeleton. This in turn allows the message to be
730 # categorized better, so that it shows up as a single type in our
731 # database, with the specifics of each occurrence not being stored with
737 my $addr = Scalar::Util::refaddr $self;
738 return if $skip{$addr};
740 # Input can be a string or hash. If a string, parse it to separate
741 # out the line number and convert to a hash for easier further
744 if (ref $opts ne 'HASH') {
745 $message = join "", $opts, @_;
747 if ($message =~ s/\s*($line_reference)//) {
748 ($line_number = $1) =~ s/\s*$optional_location//;
751 $line_number = '???';
753 $opts = { -msg => $message, -line => $line_number };
755 $message = $opts->{'-msg'};
759 $message =~ s/^\d+\s+//;
760 return if main::suppressed($message);
762 $self->SUPER::poderror($opts, @_);
764 $opts->{parameter} = "" unless $opts->{parameter};
766 # The variable parts of the message tend to be enclosed in '...',
767 # "....", or (...). Extract them and put them in an extra field,
768 # 'parameter'. This is trickier because the matching delimiter to a
769 # '(' is its mirror, and not itself. Text::Balanced could be used
771 while ($message =~ m/ \s* $optional_location ( [('"] )/xg) {
774 $delimiter = ')' if $delimiter eq '(';
776 # If there is no ending delimiter, don't consider it to be a
777 # variable part. Most likely it is a contraction like "Don't"
778 last unless $message =~ m/\G .+? \Q$delimiter/xg;
780 my $length = $+[0] - $start;
782 # Get the part up through the closing delimiter
783 my $special = substr($message, $start, $length);
784 $special =~ s/^\s+//; # No leading whitespace
786 # And add that variable part to the parameter, while removing it
787 # from the message. This isn't a foolproof way of finding the
788 # variable part. For example '(s)' can occur in e.g.,
790 if ($special ne '(s)') {
791 substr($message, $start, $length) = "";
792 pos $message = $start;
793 $opts->{-msg} = $message;
794 $opts->{parameter} .= " " if $opts->{parameter};
795 $opts->{parameter} .= $special;
799 # Extract any additional line number given. This is often the
800 # beginning location of something whereas the main line number gives
802 if ($message =~ /( $line_reference )/xi) {
804 while ($message =~ s/\s*\Q$line_ref//) {
805 $opts->{-msg} = $message;
806 $opts->{parameter} .= " " if $opts->{parameter};
807 $opts->{parameter} .= $line_ref;
811 Carp::carp("Couldn't extract line number from '$message'") if $message =~ /line \d+/;
812 push @{$problems{$filename{$addr}}{$message}}, $opts;
813 #push @{$problems{$self->get_filename}{$message}}, $opts;
816 # In the next subroutines, we keep track of the text of the current
817 # innermost thing, like F<fooC<bar>baz>. The things we care about raising
818 # messages about in this program all come from a single sequence of
819 # characters uninterrupted by other pod commands. Therefore we don't have
820 # to worry about recursion, and we can just set the string we care about
821 # to empty on entrance to each command.
824 # This is called by the parent class to deal with any straight text.
825 # We mostly just append this to the running current value which will
826 # be dealt with upon the end of the current construct, like a
827 # paragraph. But certain things don't contribute to checking the pod
828 # and are ignored. We also have set flags to indicate this text is
829 # going towards constructing certain constructs, and handle those
833 my $addr = Scalar::Util::refaddr $self;
835 my $return = $self->SUPER::handle_text(@_);
837 if ($in_X{$addr} || $in_for{$addr}) { # ignore
841 my $text = join "\n", @_;
842 $running_simple_text{$addr} .= $text;
844 # Keep separate tabs on C<>, F<>, and L<> directives, and one
845 # especially for C<> ones.
846 if ($in_CFL{$addr}) {
847 $CFL_text{$addr} .= $text;
848 $C_text{$addr} .= $text if defined $C_text{$addr};
851 # This variable is updated instead in the corresponding C, F, or L
853 $running_CFL_text{$addr} .= $text;
859 # The start_FOO routines check that somehow a C<> construct hasn't escaped
860 # without being checked, and initialize things, and call the parent
861 # class's equivalent routine.
863 # The end_FOO routines close things off, and check the text that has been
864 # accumulated for FOO, then call the parent's corresponding routine.
868 check_see_but_not_link($self);
870 my $addr = Scalar::Util::refaddr $self;
871 $start_line{$addr} = $_[0]->{start_line};
872 $running_CFL_text{$addr} = "";
873 $running_simple_text{$addr} = "";
874 return $self->SUPER::start_Para(@_);
877 sub start_item_text {
879 check_see_but_not_link($self);
881 my $addr = Scalar::Util::refaddr $self;
882 $start_line{$addr} = $_[0]->{start_line};
883 $running_CFL_text{$addr} = "";
884 $running_simple_text{$addr} = "";
886 # This is the only =item that is linkable
887 $linkable_item{$addr} = 1;
889 return $self->SUPER::start_item_text(@_);
892 sub start_item_number {
894 check_see_but_not_link($self);
896 my $addr = Scalar::Util::refaddr $self;
897 $start_line{$addr} = $_[0]->{start_line};
898 $running_CFL_text{$addr} = "";
899 $running_simple_text{$addr} = "";
901 return $self->SUPER::start_item_number(@_);
904 sub start_item_bullet {
906 check_see_but_not_link($self);
908 my $addr = Scalar::Util::refaddr $self;
909 $start_line{$addr} = $_[0]->{start_line};
910 $running_CFL_text{$addr} = "";
911 $running_simple_text{$addr} = "";
913 return $self->SUPER::start_item_bullet(@_);
916 sub end_item { # No difference in =item types endings
918 check_see_but_not_link($self);
919 return $self->SUPER::end_item(@_);
924 check_see_but_not_link($self);
926 my $addr = Scalar::Util::refaddr $self;
927 $start_line{$addr} = $_[0]->{start_line};
928 $running_CFL_text{$addr} = "";
929 $running_simple_text{$addr} = "";
931 # Save this indent on a stack, and keep track of total indent
932 my $indent = $_[0]{'indent'};
933 push @{$indents{$addr}}, $indent;
934 $current_indent{$addr} += $indent;
936 return $self->SUPER::start_over(@_);
939 sub end_over_bullet { shift->end_over(@_) }
940 sub end_over_number { shift->end_over(@_) }
941 sub end_over_text { shift->end_over(@_) }
942 sub end_over_block { shift->end_over(@_) }
943 sub end_over_empty { shift->end_over(@_) }
946 check_see_but_not_link($self);
948 my $addr = Scalar::Util::refaddr $self;
951 if (@{$indents{$addr}}) {
952 $current_indent{$addr} -= pop @{$indents{$addr}};
955 # =back without corresponding =over, but should have
957 $current_indent{$addr} = 0;
961 sub check_see_but_not_link {
963 # Looks through accumulated text for current element that includes the
964 # C<>, F<>, and L<> directives for ones that look like they are
965 # C<link> instead of L<link>.
968 my $addr = Scalar::Util::refaddr $self;
970 return unless defined $running_CFL_text{$addr};
972 while ($running_CFL_text{$addr} =~ m{
973 ( (?: \w+ \s+ )* ) # The phrase before, if any
977 ( [^<]*? ) # The not < excludes nested C<L<...
980 ( \s+ (?: under | in ) \s+ L< )?
983 my $prefix = $1 // "";
984 my $construct = $2; # The whole thing, like C<...>
987 my $trailing = $5; # After the whole thing ending in "L<"
989 # If the full phrase is something like, "you might see C<", or
990 # similar, it really isn't a reference to a link. The ones I saw
991 # all had the word "you" in them; and the "you" wasn't the
992 # beginning of a sentence.
993 if ($prefix !~ / \b you \b /x) {
995 # Now, find what the module or man page name within the
996 # construct would be if it actually has L<> syntax. If it
997 # doesn't have that syntax, will set the module to the entire
999 if (! defined $trailing # not referring to something in another
1001 && $interior !~ /$non_pods/
1003 # There can't be spaces (I think) in module names or man
1005 && $interior !~ / \s /x
1007 # F<> that end in eg \.pl are almost certainly ok, as are
1008 # those that look like a path with multiple "/" chars
1011 && $interior !~ /\.\w+$/
1012 && $interior !~ /\/.+\//)
1015 # TODO: move the checking of $pedantic higher up
1016 $self->poderror({ -line => $start_line{$addr},
1017 -msg => $C_not_linked,
1018 parameter => $construct
1024 undef $running_CFL_text{$addr};
1029 check_see_but_not_link($self);
1031 my $addr = Scalar::Util::refaddr $self;
1032 if ($in_NAME{$addr}) {
1033 if ($running_simple_text{$addr} =~ /^\s*(\S+?)\s*$/) {
1034 $self->poderror({ -line => $start_line{$addr},
1035 -msg => $missing_name_description,
1038 $in_NAME{$addr} = 0;
1040 $self->SUPER::end_Para(@_);
1045 check_see_but_not_link($self);
1047 my $addr = Scalar::Util::refaddr $self;
1048 $start_line{$addr} = $_[0]->{start_line};
1049 $running_CFL_text{$addr} = "";
1050 $running_simple_text{$addr} = "";
1052 return $self->SUPER::start_head1(@_);
1055 sub end_head1 { # This is called at the end of the =head line.
1057 check_see_but_not_link($self);
1059 my $addr = Scalar::Util::refaddr $self;
1061 $in_NAME{$addr} = 1 if $running_simple_text{$addr} eq 'NAME';
1062 return $self->SUPER::end_head(@_);
1065 sub start_Verbatim {
1067 check_see_but_not_link($self);
1069 my $addr = Scalar::Util::refaddr $self;
1070 $running_simple_text{$addr} = "";
1071 $start_line{$addr} = $_[0]->{start_line};
1072 return $self->SUPER::start_Verbatim(@_);
1077 my $addr = Scalar::Util::refaddr $self;
1079 # Pick up the name if it looks like one, since the parent class
1080 # doesn't handle verbatim NAMEs
1082 && $running_simple_text{$addr} =~ /^\s*(\S+?)\s*[,-]/)
1087 my $indent = $self->get_current_indent;
1089 # Look at each line to verify it is short enough
1090 my @lines = split /^/, $running_simple_text{$addr};
1091 for my $i (0 .. @lines - 1) {
1092 $lines[$i] =~ s/\s+$//;
1093 my $exceeds = length(Text::Tabs::expand($lines[$i]))
1094 + $indent - $MAX_LINE_LENGTH;
1095 next unless $exceeds > 0;
1097 $self->poderror({ -line => $start_line{$addr} + $i,
1098 -msg => $line_length,
1099 parameter => "+$exceeds (including " . ($indent - $INDENT) . " from =over's)",
1103 undef $running_simple_text{$addr};
1105 # Parent class didn't bother to define this
1106 #return $self->SUPER::SUPER::end_Verbatim(@_);
1111 my $addr = Scalar::Util::refaddr $self;
1113 $C_text{$addr} = "";
1115 # If not in a stacked set of C<>, F<> and L<>, initialize the text for
1117 $CFL_text{$addr} = "" if ! $in_CFL{$addr};
1120 return $self->SUPER::start_C(@_);
1125 my $addr = Scalar::Util::refaddr $self;
1127 $CFL_text{$addr} = "" if ! $in_CFL{$addr};
1129 return $self->SUPER::start_F(@_);
1134 my $addr = Scalar::Util::refaddr $self;
1136 $CFL_text{$addr} = "" if ! $in_CFL{$addr};
1138 return $self->SUPER::start_L(@_);
1143 my $addr = Scalar::Util::refaddr $self;
1145 # Warn if looks like a file or link enclosed instead by this C<>
1146 if ($C_text{$addr} =~ qr/^ $C_path_re $/x) {
1147 # Here it does look like it could be be a file path or a link.
1148 # But some varieties of regex patterns could also fit with what we
1149 # have so far. Weed those out as best we can. '/foo/' is almost
1150 # certainly meant to be a pattern, as is '/foo/g'.
1152 if ($C_text{$addr} !~ qr| ^ / [^/]* / ( [msixpodualngcr]* ) $ |x) {
1157 # Here, it looks like a pattern potentially followed by some
1158 # modifiers. To make doubly sure, don't count as patterns
1159 # those constructs which have more occurrences (generally 1)
1160 # of a modifier than is legal.
1162 map { $counts{$_}++ } split "", $1;
1163 foreach my $modifier (keys %counts) {
1164 if ($counts{$modifier} > (($modifier eq 'a')
1172 $is_pattern = 1 unless defined $is_pattern;
1175 unless ($is_pattern) {
1176 $self->poderror({ -line => $start_line{$addr},
1177 -msg => $C_with_slash,
1178 parameter => "C<$C_text{$addr}>"
1182 undef $C_text{$addr};
1184 # Add the current text to the running total. This was not done in
1185 # handle_text(), because it just sees the plain text of the innermost
1186 # stacked directive. We want to keep all the directive names
1187 # enclosing the text. Otherwise the fact that C<L<foobar>> is to a
1188 # link would be lost, as the L<> would be gone.
1189 $CFL_text{$addr} = "C<$CFL_text{$addr}>";
1191 # Add this text to the the whole running total only if popping this
1192 # directive off the stack leaves it empty. As long as something is on
1193 # the stack, it gets added to $CFL_text (just above). It is only
1194 # entirely constructed when the stack is empty.
1196 $running_CFL_text{$addr} .= $CFL_text{$addr} if ! $in_CFL{$addr};
1198 return $self->SUPER::end_C(@_);
1203 my $addr = Scalar::Util::refaddr $self;
1205 $CFL_text{$addr} = "F<$CFL_text{$addr}>";
1207 $running_CFL_text{$addr} .= $CFL_text{$addr} if ! $in_CFL{$addr};
1208 return $self->SUPER::end_F(@_);
1213 my $addr = Scalar::Util::refaddr $self;
1215 $CFL_text{$addr} = "L<$CFL_text{$addr}>";
1217 $running_CFL_text{$addr} .= $CFL_text{$addr} if ! $in_CFL{$addr};
1218 return $self->SUPER::end_L(@_);
1223 my $addr = Scalar::Util::refaddr $self;
1226 return $self->SUPER::start_X(@_);
1231 my $addr = Scalar::Util::refaddr $self;
1234 return $self->SUPER::end_X(@_);
1239 my $addr = Scalar::Util::refaddr $self;
1242 return $self->SUPER::start_for(@_);
1247 my $addr = Scalar::Util::refaddr $self;
1250 return $self->SUPER::end_for(@_);
1254 my ($self, $link) = @_;
1256 if ($link && $link->type eq 'pod') {
1257 my $page = $link->page;
1258 my $node = $link->node;
1260 # If the hyperlink is to an interior node of another page, save it
1261 # so that we can see if we need to parse normally skipped files.
1262 $has_referred_to_node{$page} = 1 if $node;
1264 # Ignore certain placeholder links in perldelta. Check if the
1265 # link is page-level, and also check if to a node within the page
1266 if ( $self->name && $self->name eq "perldelta"
1267 && (( grep { $page eq $_ } @perldelta_ignore_links)
1269 && (grep { "$page/$node" eq $_ } @perldelta_ignore_links)
1275 return $self->SUPER::hyperlink($link);
1282 $text =~ s/\s+$//s; # strip trailing whitespace
1283 $text =~ s/\s+/ /gs; # collapse whitespace
1284 my $addr = Scalar::Util::refaddr $self;
1285 push(@{$linkable_nodes{$addr}}, $text) if
1286 ! $current_indent{$addr}
1287 || $linkable_item{$addr};
1289 return $self->SUPER::node($_[0]);
1292 sub get_current_indent {
1293 return $INDENT + $current_indent{Scalar::Util::refaddr $_[0]};
1297 return $filename{Scalar::Util::refaddr $_[0]};
1300 sub linkable_nodes {
1301 my $linkables = $linkable_nodes{Scalar::Util::refaddr $_[0]};
1302 return undef unless $linkables;
1307 return $skip{Scalar::Util::refaddr $_[0]} // 0;
1312 $skip{Scalar::Util::refaddr $self} = shift;
1314 # If skipping, no need to keep the problems for it
1315 delete $problems{$self->get_filename};
1319 sub parse_from_file {
1320 # This overrides the super class method so that if an open fails on a
1321 # transitory file, it doesn't croak. It returns 1 if it did find the
1322 # file, 0 if it didn't
1325 my $filename = shift;
1326 # ignores 2nd param, which is output file. Always uses undef
1328 if (open my $in_fh, '<:bytes', $filename) {
1329 $self->SUPER::parse_from_file($in_fh, undef);
1334 # If couldn't open file, perhaps it was transitory, and hence not an error
1335 return 0 unless -e $filename;
1337 die "Can't open '$filename': $!\n";
1341 package Tie_Array_to_FH { # So printing actually goes to an array
1347 my $array_ref = shift;
1349 my $self = bless \do{ my $anonymous_scalar }, $class;
1350 $array{Scalar::Util::refaddr $self} = $array_ref;
1357 push @{$array{Scalar::Util::refaddr $self}}, @_;
1363 my %filename_to_checker; # Map a filename to its pod checker object
1364 my %id_to_checker; # Map a checksum to its pod checker object
1365 my %nodes; # key is filename, values are nodes in that file.
1366 my %nodes_first_word; # same, but value is first word of each node
1367 my %valid_modules; # List of modules known to exist outside us.
1368 my %digests; # checksums of files, whose names are the keys
1369 my %filename_to_pod; # Map a filename to its pod NAME
1370 my %files_with_unknown_issues;
1371 my %files_with_fixes;
1374 open $data_fh, '<:bytes', $known_issues or die "Can't open $known_issues";
1376 my %counts; # For --counts param, count of each issue type
1377 my %suppressed_files; # Files with at least one issue type to suppress
1379 # This file is the data file for $0.
1380 # There are three types of lines.
1381 # Comment lines are white-space only or begin with a '#', like this one. Any
1382 # changes you make to the comment lines will be lost when the file is
1384 # Lines without tab characters are simply NAMES of pods that the program knows
1385 # will have links to them and the program does not check if those links are
1387 # All other lines should have three fields, each separated by a tab. The
1388 # first field is the name of a pod; the second field is an error message
1389 # generated by this program; and the third field is a count of how many
1390 # known instances of that message there are in the pod. -1 means that the
1391 # program can expect any number of this type of message.
1394 my @existing_issues;
1397 while (<$data_fh>) { # Read the database
1399 next if /^\s*(?:#|$)/; # Skip comment and empty lines
1402 if ($add_link) { # The issues are saved and later output unchanged
1403 push @existing_issues, $_;
1407 # Keep track of counts of each issue type for each file
1408 my ($filename, $message, $count) = split /\t/;
1409 $known_problems{$filename}{$message} = $count;
1412 if ($count < 0) { # -1 means to suppress this issue type
1413 $suppressed_files{$filename} = $filename;
1416 $counts{$message} += $count;
1420 else { # Lines without a tab are modules known to be valid
1421 $valid_modules{$_} = 1
1427 $copy_fh = open_new($known_issues);
1429 # Check for basic sanity, and add each command line argument
1430 foreach my $module (@files) {
1431 die "\"$module\" does not look like a module or man page"
1432 # Must look like (A or A::B or A::B::C ..., or foo(3C)
1433 if $module !~ /^ (?: \w+ (?: :: \w+ )* | \w+ \( \d \w* \) ) $/x;
1434 $valid_modules{$module} = 1
1436 my_safer_print($copy_fh, $HEADER);
1437 foreach (sort { lc $a cmp lc $b } keys %valid_modules) {
1438 my_safer_print($copy_fh, $_, "\n");
1441 # The rest of the db file is output unchanged.
1442 my_safer_print($copy_fh, join "\n", @existing_issues, "");
1444 close_and_rename($copy_fh);
1450 foreach my $message (sort keys %counts) {
1451 $total += $counts{$message};
1452 note(Text::Tabs::expand("$counts{$message}\t$message"));
1454 note("-----\n" . Text::Tabs::expand("$total\tknown potential issues"));
1455 if (%suppressed_files) {
1456 note("\nFiles that have all messages of at least one type suppressed:");
1457 note(join ",", keys %suppressed_files);
1462 # re to match files that are to be parsed only if there is an internal link
1463 # to them. It does not include cpan, as whether those are parsed depends
1464 # on a switch. Currently, only perltoc and the stable perldelta.pod's
1465 # are included. The latter all have characters between 'perl' and
1466 # 'delta'. (Actually the currently developed one matches as well, but
1467 # is a duplicate of perldelta.pod, so can be skipped, so fine for it to
1469 my $only_for_interior_links_re = qr/ ^ pod\/perltoc.pod $
1471 unless ($do_deltas) {
1472 $only_for_interior_links_re = qr/$only_for_interior_links_re |
1473 \b perl \d+ delta \. pod \b
1480 sub output_thanks ($$$$) { # Called when an issue has been fixed
1481 my $filename = shift;
1482 my $original_count = shift;
1483 my $current_count = shift;
1484 my $message = shift;
1486 $files_with_fixes{$filename} = 1;
1488 my $fixed_count = $original_count - $current_count;
1489 my $a_problem = ($fixed_count == 1) ? "a problem" : "multiple problems";
1490 my $another_problem = ($fixed_count == 1) ? "another problem" : "another set of problems";
1494 There were $original_count occurrences (now $current_count) in this pod of type
1499 There are no longer any problems found in this pod!
1506 Thanks for fixing $a_problem!
1508 Now you must teach $0 that this was fixed.
1513 Thanks for fixing $another_problem.
1522 sub my_safer_print { # print, with error checking for outputting to db
1523 my ($fh, @lines) = @_;
1525 if (! print $fh @lines) {
1526 my $save_error = $!;
1528 die "Write failure: $save_error";
1532 sub extract_pod { # Extracts just the pod from a file; returns undef if file
1534 my $filename = shift;
1539 # Arrange for the output of Pod::Parser to be collected in an array we can
1540 # look at instead of being printed
1541 tie *ALREADY_FH, 'Tie_Array_to_FH', \@pod;
1542 if (open my $in_fh, '<:bytes', $filename) {
1543 my $parser = Pod::Parser->new();
1544 $parser->parse_from_filehandle($in_fh, *ALREADY_FH);
1547 return join "", @pod
1550 # The file should already have been opened once to get here, so if that
1551 # fails, something is wrong. It's possible that a transitory file
1552 # containing a pod would get here, so if the file no longer exists just
1554 return unless -e $filename;
1555 die "Can't open '$filename': $!\n";
1558 my $digest = Digest->new($digest_type);
1560 # This is used as a callback from File::Find::find(), which always constructs
1561 # pathnames using Unix separators
1563 # If $_ is a pod file, add it to the lists and do other prep work.
1566 # Don't look at files in directories that are for tests, nor those
1567 # beginning with a dot
1568 if (m!/t\z! || m!/\.!) {
1569 $File::Find::prune = 1;
1574 return unless -r && -s; # Can't check it if can't read it; no need to
1576 return unless -f || -l; # Weird file types won't be pods
1578 my ($leaf) = m!([^/]+)\z!;
1579 if (m!/\.! # No hidden Unix files
1580 || $leaf =~ $non_pods) {
1581 note("Not considering $_") if DEBUG;
1585 my $filename = $File::Find::name;
1587 # $filename is relative, like './path'. Strip that initial part away.
1588 $filename =~ s!^\./!! or die 'Unexpected pathname "$filename"';
1590 return if $excluded_files{canonicalize($filename)};
1595 if (! open $candidate, '<:bytes', $_) {
1597 # If a transitory file was found earlier, the open could fail
1598 # legitimately and we just skip the file; also skip it if it is a
1599 # broken symbolic link, as it is probably just a build problem;
1600 # certainly not a file that we would want to check the pod of.
1601 # Otherwise fail it here and no reason to process it further.
1602 # (But the test count will be off too)
1603 ok(0, "Can't open '$filename': $!")
1604 if -r $filename && ! -l $filename;
1610 # If the file is a .pm or .pod, having any initial '=' on a line is
1611 # grounds for testing it. Otherwise, require a head1 NAME line to
1612 # consider it as a potential pod
1613 if ($filename =~ /\.(?:pm|pod)/) {
1614 return unless $contents =~ /^=/m;
1616 return unless $contents =~ /^=head1 +NAME/m;
1619 # Here, we know that the file is a pod. Add it to the list of files
1620 # to check and create a checker object for it.
1622 push @files, $filename;
1623 my $checker = My::Pod::Checker->new($filename);
1624 $filename_to_checker{$filename} = $checker;
1626 # In order to detect duplicate pods and only analyze them once, we
1627 # compute checksums for the file, so don't have to do an exact
1628 # compare. Note that if the pod is just part of the file, the
1629 # checksums can differ for the same pod. That special case is handled
1630 # later, since if the checksums of the whole file are the same, that
1631 # case won't even come up. We don't need the checksums for files that
1632 # we parse only if there is a link to its interior, but we do need its
1633 # NAME, which is also retrieved in the code below.
1635 if ($filename =~ / (?: ^(cpan|lib|ext|dist)\/ )
1636 | $only_for_interior_links_re
1638 $digest->add($contents);
1639 $digests{$filename} = $digest->digest;
1641 # lib files aren't analyzed if they are duplicates of files copied
1642 # there from some other directory. But to determine this, we need
1643 # to know their NAMEs. We might as well find the NAME now while
1644 # the file is open. Similarly, cpan files aren't analyzed unless
1645 # we're analyzing all of them, or this particular file is linked
1646 # to by a file we are analyzing, and thus we will want to verify
1647 # that the target exists in it. We need to know at least the NAME
1648 # to see if it's worth analyzing, or so we can determine if a lib
1649 # file is a copy of a cpan one.
1650 if ($filename =~ m{ (?: ^ (?: cpan | lib ) / )
1651 | $only_for_interior_links_re
1653 if ($contents =~ /^=head1 +NAME.*/mg) {
1654 # The NAME is the first non-spaces on the line up to a
1655 # comma, dash or end of line. Otherwise, it's invalid and
1656 # this pod doesn't have a legal name that we're smart
1657 # enough to find currently. But the parser will later
1658 # find it if it thinks there is a legal name, and set the
1660 if ($contents =~ /\G # continue from the line after =head1
1661 \s* # ignore any empty lines
1663 # ignore =for paragraphs followed by empty
1665 (?: ^ =for .*? \n (?: [^\s]*? \n )* \s* )*
1667 ^ \s* ( \S+?) \s* (?: [,-] | $ )/mx) {
1669 $checker->name($name);
1670 $id_to_checker{$name} = $checker
1671 if $filename =~ m{^cpan/};
1674 elsif ($filename =~ m{^cpan/}) {
1675 $id_to_checker{$digests{$filename}} = $checker;
1681 } # End of is_pod_file()
1683 # Start of real code that isn't processing the command line (except the
1684 # db is read in above, as is processing of the --add_link option).
1685 # Here, @files contains list of files on the command line. If have any of
1686 # these, unconditionally test them, and show all the errors, even the known
1687 # ones, and, since not testing other pods, don't do cross-pod link tests.
1688 # (Could add extra code to do cross-pod tests for the ones in the list.)
1690 if ($has_input_files) {
1691 undef %known_problems;
1692 $do_upstream_cpan = $do_deltas = 1; # In case one of the inputs is one
1695 else { # No input files -- go find all the possibilities.
1697 $copy_fh = open_new($known_issues);
1698 note("Regenerating $known_issues, please be patient...");
1699 print $copy_fh $HEADER;
1702 # Move to the directory above us, but have to adjust @INC to account for
1704 s{^\.\./lib$}{lib} for @INC;
1705 chdir File::Spec->updir;
1707 # And look in this directory and all its subdirectories
1708 find( {wanted => \&is_pod_file, no_chdir => 1}, '.');
1710 # Add ourselves to the test
1711 push @files, "t/porting/podcheck.t";
1714 # Now we know how many tests there will be.
1715 plan (tests => scalar @files) if ! $regen;
1718 # Sort file names so we get consistent results, and to put cpan last,
1719 # preceded by the ones that we don't generally parse. This is because both
1720 # these classes are generally parsed only if there is a link to the interior
1721 # of them, and we have to parse all others first to guarantee that they don't
1722 # have such a link. 'lib' files come just before these, as some of these are
1723 # duplicates of others. We already have figured this out when gathering the
1724 # data as a special case for all such files, but this, while unnecessary,
1725 # puts the derived file last in the output. 'readme' files come before those,
1726 # as those also could be duplicates of others, which are considered the
1727 # primary ones. These currently aren't figured out when gathering data, so
1729 @files = sort { if ($a =~ /^cpan/) {
1730 return 1 if $b !~ /^cpan/;
1731 return lc $a cmp lc $b;
1733 elsif ($b =~ /^cpan/) {
1736 elsif ($a =~ /$only_for_interior_links_re/) {
1737 return 1 if $b !~ /$only_for_interior_links_re/;
1738 return lc $a cmp lc $b;
1740 elsif ($b =~ /$only_for_interior_links_re/) {
1743 elsif ($a =~ /^lib/) {
1744 return 1 if $b !~ /^lib/;
1745 return lc $a cmp lc $b;
1747 elsif ($b =~ /^lib/) {
1749 } elsif ($a =~ /\breadme\b/i) {
1750 return 1 if $b !~ /\breadme\b/i;
1751 return lc $a cmp lc $b;
1753 elsif ($b =~ /\breadme\b/i) {
1757 return lc $a cmp lc $b;
1762 # Now go through all the files and parse them
1764 foreach my $filename (@files) {
1766 note("parsing $filename") if DEBUG;
1768 # We may have already figured out some things in the process of generating
1769 # the file list. If so, we have a $checker object already. But if not,
1771 my $checker = $filename_to_checker{$filename};
1773 $checker = My::Pod::Checker->new($filename);
1774 $filename_to_checker{$filename} = $checker;
1777 # We have set the name in the checker object if there is a possibility
1778 # that no further parsing is necessary, but otherwise do the parsing now.
1779 if (! $checker->name) {
1780 if (! $checker->parse_from_file($filename, undef)) {
1781 $checker->set_skip("$filename is transitory");
1788 if ($checker->num_errors() < 0) { # Returns negative if not a pod
1789 $checker->set_skip("$filename is not a pod");
1793 # Here, is a pod. See if it is one that has already been tested,
1794 # or should be tested under another directory. Use either its NAME
1795 # if it has one, or a checksum if not.
1796 my $name = $checker->name;
1803 my $digest = Digest->new($digest_type);
1804 my $contents = extract_pod($filename);
1806 # If the return is undef, it means that $filename was a transitory
1808 next FILE unless defined $contents;
1809 $digest->add($contents);
1810 $id = $digest->digest;
1813 # If there is a match for this pod with something that we've already
1814 # processed, don't process it, and output why.
1816 if (defined ($prior_checker = $id_to_checker{$id})
1817 && $prior_checker != $checker) # Could have defined the checker
1818 # earlier without pursuing it
1821 # If the pods are identical, then it's just a copy, and isn't an
1822 # error. First use the checksums we have already computed to see
1823 # if the entire files are identical, which means that the pods are
1825 my $prior_filename = $prior_checker->get_filename;
1827 || ($digests{$prior_filename}
1828 && $digests{$filename}
1829 && $digests{$prior_filename} eq $digests{$filename}));
1831 # If they differ, it could be that the files differ for some
1832 # reason, but the pods they contain are identical. Extract the
1833 # pods and do the comparisons on just those.
1834 if (! $same && $name) {
1835 my $contents = extract_pod($filename);
1837 # If return is <undef>, it means that $filename no longer
1838 # exists. This means it was a transitory file, and should not
1840 next FILE unless defined $contents;
1842 my $prior_contents = extract_pod($prior_filename);
1844 # If return is <undef>, it means that $prior_filename no
1845 # longer exists. This means it was a transitory file, and
1846 # should not have been tested, but we already did process it.
1847 # What we should do now is to back-out its records, and
1848 # process $filename in its stead. But backing out is not so
1849 # simple, and so I'm (khw) skipping that unless and until
1850 # experience shows that it is needed. We do go process
1851 # $filename, and there are potential false positive conflicts
1852 # with the transitory $prior_contents, and rerunning the test
1853 # should cause it to succeed.
1854 goto process_this_pod unless defined $prior_contents;
1856 $same = $prior_contents eq $contents;
1859 use File::Basename 'basename';
1861 $checker->set_skip("The pod of $filename is a duplicate of "
1862 . "the pod for $prior_filename");
1863 } elsif ($prior_filename =~ /\breadme\b/i) {
1864 $checker->set_skip("$prior_filename is a README apparently for $filename");
1865 } elsif ($filename =~ /\breadme\b/i) {
1866 $checker->set_skip("$filename is a README apparently for $prior_filename");
1867 } elsif (! $do_upstream_cpan
1868 && $filename =~ /^cpan/
1869 && $prior_filename =~ /^cpan/)
1871 $checker->set_skip("CPAN is upstream for $filename");
1872 } elsif ( $filename =~ /^utils/ or $prior_filename =~ /^utils/ ) {
1873 $checker->set_skip("$filename copy is in utils/");
1874 } elsif ($prior_filename =~ /^(?:cpan|ext|dist)/
1875 && $filename !~ /^(?:cpan|ext|dist)/
1876 && basename($prior_filename) eq basename($filename))
1878 $checker->set_skip("$filename: Need to run make?");
1879 } else { # Here have two pods with identical names that differ
1880 $prior_checker->poderror(
1881 { -msg => $duplicate_name,
1883 parameter => "'$filename' also has NAME '$name'"
1886 { -msg => $duplicate_name,
1888 parameter => "'$prior_filename' also has NAME '$name'"
1891 # Changing the names helps later.
1892 $prior_checker->name("$name version arbitrarily numbered 1");
1893 $checker->name("$name version arbitrarily numbered 2");
1896 # In any event, don't process this pod that has the same name as
1904 $id_to_checker{$id} = $checker;
1906 my $parsed_for_links = ", but parsed for its interior links";
1907 if ((! $do_upstream_cpan && $filename =~ /^cpan/)
1908 || $filename =~ $only_for_interior_links_re)
1910 if ($filename =~ /^cpan/) {
1911 $checker->set_skip("CPAN is upstream for $filename");
1913 elsif ($filename =~ /perl\d+delta/) {
1915 $checker->set_skip("$filename is a stable perldelta");
1918 elsif ($filename =~ /perltoc/) {
1919 $checker->set_skip("$filename dependent on component pods");
1922 croak("Unexpected file '$filename' encountered that has parsing for interior-linking only");
1925 if ($name && $has_referred_to_node{$name}) {
1926 $checker->set_skip($checker->get_skip() . $parsed_for_links);
1930 # Need a name in order to process it, because not meaningful
1931 # otherwise, and also can't test links to this without a name.
1932 if (!defined $name) {
1933 $checker->poderror( { -msg => $no_name,
1939 # For skipped files, just get its NAME
1941 if (($skip = $checker->get_skip()) && $skip !~ /$parsed_for_links/)
1943 $checker->node($name) if $name;
1946 if (! $checker->parse_from_file($filename, undef)) {
1947 $checker->set_skip("$filename is transitory");
1952 # Go through everything in the file that could be an anchor that
1953 # could be a link target. Count how many there are of the same name.
1954 foreach my $node ($checker->linkable_nodes) {
1955 next FILE if ! $node; # Can be empty is like '=item *'
1956 $nodes{$name}{$node}++;
1958 # Experiments have shown that cpan search can figure out the
1959 # target of a link even if the exact wording is incorrect, as long
1960 # as the first word is. This happens frequently in perlfunc.pod,
1961 # where the link will be just to the function, but the target
1962 # entry also includes parameters to the function.
1963 my $first_word = $node;
1964 if ($first_word =~ s/^(\S+)\s+\S.*/$1/) {
1965 $nodes_first_word{$name}{$first_word} = $node;
1968 $filename_to_pod{$filename} = $name;
1972 # Here, all files have been parsed, and all links and link targets are stored.
1973 # Now go through the files again and see which don't have matches.
1974 if (! $has_input_files) {
1975 foreach my $filename (@files) {
1976 next if $filename_to_checker{$filename}->get_skip;
1978 my $checker = $filename_to_checker{$filename};
1979 foreach my $link ($checker->hyperlinks()) {
1980 my $linked_to_page = $link->page;
1981 next unless $linked_to_page; # intra-file checks are handled by std
1983 # Currently, we assume all external links are valid
1984 next if $link->type eq 'url';
1986 # Initialize the potential message.
1987 my %problem = ( -msg => $broken_link,
1988 -line => $link->line,
1989 parameter => "to \"$linked_to_page\"",
1992 # See if we have found the linked-to_file in our parse
1993 if (exists $nodes{$linked_to_page}) {
1994 my $node = $link->node;
1996 # If link is only to the page-level, already have it
1999 # If link is to a node that exists in the file, is ok
2000 if ($nodes{$linked_to_page}{$node}) {
2002 # But if the page has multiple targets with the same name,
2003 # it's ambiguous which one this should be to.
2004 if ($nodes{$linked_to_page}{$node} > 1) {
2005 $problem{-msg} = $multiple_targets;
2006 $problem{parameter} = "in $linked_to_page that $node could be pointing to";
2007 $checker->poderror(\%problem);
2009 } elsif (! $nodes_first_word{$linked_to_page}{$node}) {
2011 # Here the link target was not found, either exactly or to
2012 # the first word. Is an error.
2013 $problem{parameter} =~ s,"$,/$node",;
2014 $checker->poderror(\%problem);
2017 } # Linked-to-file not in parse; maybe is in exception list
2018 elsif (! exists $valid_modules{$link->page}) {
2020 # Here, is a link to a target that we can't find. Check if
2021 # there is an internal link on the page with the target name.
2022 # If so, it could be that they just forgot the initial '/'
2023 # But perldelta is handled specially: only do this if the
2024 # broken link isn't one of the known bad ones (that are
2025 # placemarkers and should be removed for the final)
2026 my $NAME = $filename_to_pod{$filename};
2027 if (! defined $NAME) {
2028 $checker->poderror(\%problem);
2031 if ($nodes{$NAME}{$linked_to_page}) {
2032 $problem{-msg} = $broken_internal_link;
2034 $checker->poderror(\%problem);
2041 # If regenerating the data file, start with the modules for which we don't
2042 # check targets. If you change the sort order, you need to run --regen before
2043 # committing so that future commits that do run regen don't show irrelevant
2046 foreach (sort { lc $a cmp lc $b } keys %valid_modules) {
2047 my_safer_print($copy_fh, $_, "\n");
2051 # Now ready to output the messages.
2052 foreach my $filename (@files) {
2053 my $canonical = canonicalize($filename);
2055 my $skip = $filename_to_checker{$filename}->get_skip // "";
2058 foreach my $message ( sort keys %{$problems{$filename}}) {
2061 # Preserve a negative setting.
2062 if ($known_problems{$canonical}{$message}
2063 && $known_problems{$canonical}{$message} < 0)
2065 $count = $known_problems{$canonical}{$message};
2068 $count = @{$problems{$filename}{$message}};
2070 my_safer_print($copy_fh, $canonical . "\t$message\t$count\n");
2075 skip($skip, 1) if $skip;
2077 my $thankful_diagnostics = 0;
2080 my $total_known = 0;
2081 foreach my $message ( sort keys %{$problems{$filename}}) {
2082 $known_problems{$canonical}{$message} = 0
2083 if ! $known_problems{$canonical}{$message};
2084 my $diagnostic = "";
2085 my $problem_count = scalar @{$problems{$filename}{$message}};
2086 $total_known += $problem_count;
2087 next if $known_problems{$canonical}{$message} < 0;
2088 if ($problem_count > $known_problems{$canonical}{$message}) {
2090 # Here we are about to output all the messages for this type,
2091 # subtract back this number we previously added in.
2092 $total_known -= $problem_count;
2094 $diagnostic .= $indent . qq{"$message"};
2095 if ($problem_count > 2) {
2096 $diagnostic .= " ($problem_count occurrences,"
2097 . " expected $known_problems{$canonical}{$message})";
2099 foreach my $problem (@{$problems{$filename}{$message}}) {
2100 $diagnostic .= " " if $problem_count == 1;
2101 $diagnostic .= "\n$indent$indent";
2102 $diagnostic .= "$problem->{parameter}" if $problem->{parameter};
2103 $diagnostic .= " near line $problem->{-line} of "
2105 $diagnostic .= " $problem->{comment}" if $problem->{comment};
2107 $diagnostic .= "\n";
2108 $files_with_unknown_issues{$filename} = 1;
2109 } elsif ($problem_count < $known_problems{$canonical}{$message}) {
2110 $diagnostic = output_thanks($filename, $known_problems{$canonical}{$message}, $problem_count, $message);
2111 $thankful_diagnostics++;
2113 push @diagnostics, $diagnostic if $diagnostic;
2116 # The above loop has output messages where there are current potential
2117 # issues. But it misses where there were some that have been entirely
2118 # fixed. For those, we need to look through the old issues
2119 foreach my $message ( sort keys %{$known_problems{$canonical}}) {
2120 next if $problems{$filename}{$message};
2121 next if ! $known_problems{$canonical}{$message};
2122 next if $known_problems{$canonical}{$message} < 0; # Preserve negs
2124 next if !$pedantic and $message =~
2125 /^(?:\Q$line_length\E|\Q$C_not_linked\E|\Q$C_with_slash\E)/;
2127 my $diagnostic = output_thanks($filename, $known_problems{$canonical}{$message}, 0, $message);
2128 push @diagnostics, $diagnostic if $diagnostic;
2129 $thankful_diagnostics++ if $diagnostic;
2132 my $output = "POD of $filename";
2133 $output .= ", excluding $total_known not shown known potential problems"
2135 if (@diagnostics && @diagnostics == $thankful_diagnostics) {
2136 # Output fixed issues as passing to-do tests, so they do not
2137 # cause failures, but t/harness still flags them.
2138 $output .= " # TODO"
2140 ok(@diagnostics == $thankful_diagnostics, $output);
2142 diag(join "", @diagnostics,
2143 "See end of this test output for your options on silencing this");
2146 delete $known_problems{$canonical};
2151 && ! ok (keys %known_problems == 0, "The known problems database ($data_dir/known_pod_issues.dat) includes no references to non-existent files"))
2153 note("The following files were not found: "
2154 . join ", ", keys %known_problems);
2155 note("They will automatically be removed from the db the next time");
2156 note(" cd t; ./perl -I../lib porting/podcheck.t --regen");
2161 run this test script by hand, using the following formula (on
2162 Un*x-like machines):
2164 ./perl -I../lib porting/podcheck.t --regen
2167 if (%files_with_unknown_issues) {
2168 my $were_count_files = scalar keys %files_with_unknown_issues;
2169 $were_count_files = ($were_count_files == 1)
2170 ? "was $were_count_files file"
2171 : "were $were_count_files files";
2172 my $message = <<EOF;
2174 HOW TO GET ${\__FILE__} TO PASS
2176 There $were_count_files that had new potential problems identified.
2177 Some of them may be real, and some of them may be false positives because
2178 this program isn't as smart as it likes to think it is. You can teach this
2179 program to ignore the issues it has identified, and hence pass, by doing the
2182 1) If a problem is about a link to an unknown module or man page that
2183 you know exists, re-run the command something like:
2184 ./perl -I../lib porting/podcheck.t --add_link MODULE man_page ...
2185 (MODULEs should look like Foo::Bar, and man_pages should look like
2186 bar(3c); don't do this for a module or man page that you aren't sure
2187 about; instead treat as another type of issue and follow the
2188 instructions below.)
2190 2) For other issues, decide if each should be fixed now or not. Fix the
2191 ones you decided to, and rerun this test to verify that the fixes
2194 3) If there remain false positive or problems that you don't plan to fix right
2197 That should cause all current potential problems to be accepted by
2198 the program, so that the next time it runs, they won't be flagged.
2200 if (%files_with_fixes) {
2201 $message .= " This step will also take care of the files that have fixes in them\n";
2205 For a few files, such as perltoc, certain issues will always be
2206 expected, and more of the same will be added over time. For those,
2207 before you do the regen, you can edit
2209 and find the entry for the module's file and specific error message,
2210 and change the count of known potential problems to -1.
2214 } elsif (%files_with_fixes) {
2216 To teach this test script that the potential problems have been fixed,
2223 chdir $original_dir || die "Can't change directories to $original_dir";
2224 close_and_rename($copy_fh);