xref: /openssl/util/find-doc-nits (revision da1c088f)
1#! /usr/bin/env perl
2# Copyright 2002-2023 The OpenSSL Project Authors. All Rights Reserved.
3#
4# Licensed under the Apache License 2.0 (the "License").  You may not use
5# this file except in compliance with the License.  You can obtain a copy
6# in the file LICENSE in the source distribution or at
7# https://www.openssl.org/source/license.html
8
9
10require 5.10.0;
11use warnings;
12use strict;
13
14use Carp qw(:DEFAULT cluck);
15use Pod::Checker;
16use File::Find;
17use File::Basename;
18use File::Spec::Functions;
19use Getopt::Std;
20use FindBin;
21use lib "$FindBin::Bin/perl";
22
23use OpenSSL::Util::Pod;
24
25use lib '.';
26use configdata;
27
28# Set to 1 for debug output
29my $debug = 0;
30
31# Options.
32our($opt_d);
33our($opt_e);
34our($opt_s);
35our($opt_o);
36our($opt_h);
37our($opt_l);
38our($opt_m);
39our($opt_n);
40our($opt_p);
41our($opt_u);
42our($opt_v);
43our($opt_c);
44
45# Print usage message and exit.
46sub help {
47    print <<EOF;
48Find small errors (nits) in documentation.  Options:
49    -c List undocumented commands, undocumented options and unimplemented options.
50    -d Detailed list of undocumented (implies -u)
51    -e Detailed list of new undocumented (implies -v)
52    -h Print this help message
53    -l Print bogus links
54    -m Name(s) of manuals to focus on. Default: man1,man3,man5,man7
55    -n Print nits in POD pages
56    -o Causes -e/-v to count symbols added since 1.1.1 as new (implies -v)
57    -u Count undocumented functions
58    -v Count new undocumented functions
59EOF
60    exit;
61}
62
63getopts('cdehlm:nouv');
64
65help() if $opt_h;
66$opt_u = 1 if $opt_d;
67$opt_v = 1 if $opt_o || $opt_e;
68die "Cannot use both -u and -v"
69    if $opt_u && $opt_v;
70die "Cannot use both -d and -e"
71    if $opt_d && $opt_e;
72
73# We only need to check c, l, n, u and v.
74# Options d, e, o imply one of the above.
75die "Need one of -[cdehlnouv] flags.\n"
76    unless $opt_c or $opt_l or $opt_n or $opt_u or $opt_v;
77
78
79my $temp = '/tmp/docnits.txt';
80my $OUT;
81my $status = 0;
82
83$opt_m = "man1,man3,man5,man7" unless $opt_m;
84die "Argument of -m option may contain only man1, man3, man5, and/or man7"
85    unless $opt_m =~ /^(man[1357][, ]?)*$/;
86my @sections = ( split /[, ]/, $opt_m );
87
88my %mandatory_sections = (
89    '*' => [ 'NAME', 'COPYRIGHT' ],
90    1   => [ 'DESCRIPTION', 'SYNOPSIS', 'OPTIONS' ],
91    3   => [ 'DESCRIPTION', 'SYNOPSIS', 'RETURN VALUES' ],
92    5   => [ 'DESCRIPTION' ],
93    7   => [ ]
94                         );
95
96# Symbols that we ignored.
97# They are reserved macros that we currently don't document
98my $ignored = qr/(?| ^i2d_
99                 |   ^d2i_
100                 |   ^DEPRECATEDIN
101                 |   ^OSSL_DEPRECATED
102                 |   \Q_fnsig(3)\E$
103                 |   ^IMPLEMENT_
104                 |   ^_?DECLARE_
105                 |   ^sk_
106                 |   ^SKM_DEFINE_STACK_OF_INTERNAL
107                 |   ^lh_
108                 |   ^DEFINE_LHASH_OF_(INTERNAL|DEPRECATED)
109                 )/x;
110
111# A common regexp for C symbol names
112my $C_symbol = qr/\b[[:alpha:]][_[:alnum:]]*\b/;
113
114# Collect all POD files, both internal and public, and regardless of location
115# We collect them in a hash table with each file being a key, so we can attach
116# tags to them.  For example, internal docs will have the word "internal"
117# attached to them.
118my %files = ();
119# We collect files names on the fly, on known tag basis
120my %collected_tags = ();
121# We cache results based on tags
122my %collected_results = ();
123
124# files OPTIONS
125#
126# Example:
127#
128#       files(TAGS => 'manual');
129#       files(TAGS => [ 'manual', 'man1' ]);
130#
131# This function returns an array of files corresponding to a set of tags
132# given with the options "TAGS".  The value of this option can be a single
133# word, or an array of several words, which work as inclusive or exclusive
134# selectors.  Inclusive selectors are used to add one more set of files to
135# the returned array, while exclusive selectors limit the set of files added
136# to the array.  The recognised tag values are:
137#
138# 'public_manual'       - inclusive selector, adds public manuals to the
139#                         returned array of files.
140# 'internal_manual'     - inclusive selector, adds internal manuals to the
141#                         returned array of files.
142# 'manual'              - inclusive selector, adds any manual to the returned
143#                         array of files.  This is really a shorthand for
144#                         'public_manual' and 'internal_manual' combined.
145# 'public_header'       - inclusive selector, adds public headers to the
146#                         returned array of files.
147# 'header'              - inclusive selector, adds any header file to the
148#                         returned array of files.  Since we currently only
149#                         care about public headers, this is exactly
150#                         equivalent to 'public_header', but is present for
151#                         consistency.
152#
153# 'man1', 'man3', 'man5', 'man7'
154#                       - exclusive selectors, only applicable together with
155#                         any of the manual selectors.  If any of these are
156#                         present, only the manuals from the given sections
157#                         will be included.  If none of these are present,
158#                         the manuals from all sections will be returned.
159#
160# All returned manual files come from configdata.pm.
161# All returned header files come from looking inside
162# "$config{sourcedir}/include/openssl"
163#
164sub files {
165    my %opts = ( @_ );          # Make a copy of the arguments
166
167    $opts{TAGS} = [ $opts{TAGS} ] if ref($opts{TAGS}) eq '';
168
169    croak "No tags given, or not an array"
170        unless exists $opts{TAGS} && ref($opts{TAGS}) eq 'ARRAY';
171
172    my %tags = map { $_ => 1 } @{$opts{TAGS}};
173    $tags{public_manual} = 1
174        if $tags{manual} && ($tags{public} // !$tags{internal});
175    $tags{internal_manual} = 1
176        if $tags{manual} && ($tags{internal} // !$tags{public});
177    $tags{public_header} = 1
178        if $tags{header} && ($tags{public} // !$tags{internal});
179    delete $tags{manual};
180    delete $tags{header};
181    delete $tags{public};
182    delete $tags{internal};
183
184    my $tags_as_key = join(':', sort keys %tags);
185
186    cluck  "DEBUG[files]: This is how we got here!" if $debug;
187    print STDERR "DEBUG[files]: tags: $tags_as_key\n" if $debug;
188
189    my %tags_to_collect = ( map { $_ => 1 }
190                            grep { !exists $collected_tags{$_} }
191                            keys %tags );
192
193    if ($tags_to_collect{public_manual}) {
194        print STDERR "DEBUG[files]: collecting public manuals\n"
195            if $debug;
196
197        # The structure in configdata.pm is that $unified_info{mandocs}
198        # contains lists of man files, and in turn, $unified_info{depends}
199        # contains hash tables showing which POD file each of those man
200        # files depend on.  We use that information to find the POD files,
201        # and to attach the man section they belong to as tags
202        foreach my $mansect ( @sections ) {
203            foreach ( map { @{$unified_info{depends}->{$_}} }
204                      @{$unified_info{mandocs}->{$mansect}} ) {
205                $files{$_} = { $mansect => 1, public_manual => 1 };
206            }
207        }
208        $collected_tags{public_manual} = 1;
209    }
210
211    if ($tags_to_collect{internal_manual}) {
212        print STDERR "DEBUG[files]: collecting internal manuals\n"
213            if $debug;
214
215        # We don't have the internal docs in configdata.pm.  However, they
216        # are all in the source tree, so they're easy to find.
217        foreach my $mansect ( @sections ) {
218            foreach ( glob(catfile($config{sourcedir},
219                                   'doc', 'internal', $mansect, '*.pod')) ) {
220                $files{$_} = { $mansect => 1, internal_manual => 1 };
221            }
222        }
223        $collected_tags{internal_manual} = 1;
224    }
225
226    if ($tags_to_collect{public_header}) {
227        print STDERR "DEBUG[files]: collecting public headers\n"
228            if $debug;
229
230        foreach ( glob(catfile($config{sourcedir},
231                               'include', 'openssl', '*.h')) ) {
232            $files{$_} = { public_header => 1 };
233        }
234    }
235
236    my @result = @{$collected_results{$tags_as_key} // []};
237
238    if (!@result) {
239        # Produce a result based on caller tags
240        foreach my $type ( ( 'public_manual', 'internal_manual' ) ) {
241            next unless $tags{$type};
242
243            # If caller asked for specific sections, we care about sections.
244            # Otherwise, we give back all of them.
245            my @selected_sections =
246                grep { $tags{$_} } @sections;
247            @selected_sections = @sections unless @selected_sections;
248
249            foreach my $section ( ( @selected_sections ) ) {
250                push @result,
251                    ( sort { basename($a) cmp basename($b) }
252                      grep { $files{$_}->{$type} && $files{$_}->{$section} }
253                      keys %files );
254            }
255        }
256        if ($tags{public_header}) {
257            push @result,
258                ( sort { basename($a) cmp basename($b) }
259                  grep { $files{$_}->{public_header} }
260                  keys %files );
261        }
262
263        if ($debug) {
264            print STDERR "DEBUG[files]: result:\n";
265            print STDERR "DEBUG[files]:     $_\n" foreach @result;
266        }
267        $collected_results{$tags_as_key} = [ @result ];
268    }
269
270    return @result;
271}
272
273# Print error message, set $status.
274sub err {
275    print join(" ", @_), "\n";
276    $status = 1
277}
278
279# Cross-check functions in the NAME and SYNOPSIS section.
280sub name_synopsis {
281    my $id = shift;
282    my $filename = shift;
283    my $contents = shift;
284
285    # Get NAME section and all words in it.
286    return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
287    my $tmp = $1;
288    $tmp =~ tr/\n/ /;
289    err($id, "Trailing comma before - in NAME")
290        if $tmp =~ /, *-/;
291    $tmp =~ s/ -.*//g;
292    err($id, "POD markup among the names in NAME")
293        if $tmp =~ /[<>]/;
294    $tmp =~ s/  */ /g;
295    err($id, "Missing comma in NAME")
296        if $tmp =~ /[^,] /;
297
298    my $dirname = dirname($filename);
299    my $section = basename($dirname);
300    my $simplename = basename($filename, ".pod");
301    my $foundfilename = 0;
302    my %foundfilenames = ();
303    my %names;
304    foreach my $n ( split ',', $tmp ) {
305        $n =~ s/^\s+//;
306        $n =~ s/\s+$//;
307        err($id, "The name '$n' contains white-space")
308            if $n =~ /\s/;
309        $names{$n} = 1;
310        $foundfilename++ if $n eq $simplename;
311        $foundfilenames{$n} = 1
312            if ( ( grep { basename($_) eq "$n.pod" }
313                   files(TAGS => [ 'manual', $section ]) )
314                 && $n ne $simplename );
315    }
316    err($id, "The following exist as other .pod files:",
317         sort keys %foundfilenames)
318        if %foundfilenames;
319    err($id, "$simplename (filename) missing from NAME section")
320        unless $foundfilename;
321
322    # Find all functions in SYNOPSIS
323    return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
324    my $syn = $1;
325    my $ignore_until = undef;   # If defined, this is a regexp
326    # Remove all non-code lines
327    $syn =~ s/^(?:\s*?|\S.*?)$//msg;
328    # Remove all comments
329    $syn =~ s/\/\*.*?\*\///msg;
330    while ( $syn ) {
331        # "env" lines end at a newline.
332        # Preprocessor lines start with a # and end at a newline.
333        # Other lines end with a semicolon, and may cover more than
334        # one physical line.
335        if ( $syn !~ /^ \s*(env .*?|#.*?|.*?;)\s*$/ms ) {
336            err($id, "Can't parse rest of synopsis:\n$syn\n(declarations not ending with a semicolon (;)?)");
337            last;
338        }
339        my $line = $1;
340        $syn = $';
341
342        print STDERR "DEBUG[name_synopsis] \$line = '$line'\n" if $debug;
343
344        # Special code to skip over documented structures
345        if ( defined $ignore_until) {
346            next if $line !~ /$ignore_until/;
347            $ignore_until = undef;
348            next;
349        }
350        if ( $line =~ /^\s*(?:typedef\s+)?struct(?:\s+\S+)\s*\{/ ) {
351            $ignore_until = qr/\}.*?;/;
352            next;
353        }
354
355        my $sym;
356        my $is_prototype = 1;
357        $line =~ s/LHASH_OF\([^)]+\)/int/g;
358        $line =~ s/STACK_OF\([^)]+\)/int/g;
359        $line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
360        $line =~ s/__declspec\([^)]+\)//;
361
362        ## We don't prohibit that space, to allow typedefs looking like
363        ## this:
364        ##
365        ## typedef int (fantastically_long_name_breaks_80char_limit)
366        ##     (fantastically_long_name_breaks_80char_limit *something);
367        ##
368        #if ( $line =~ /typedef.*\(\*?\S+\)\s+\(/ ) {
369        #    # a callback function with whitespace before the argument list:
370        #    # typedef ... (*NAME) (...
371        #    # typedef ... (NAME) (...
372        #    err($id, "Function typedef has space before arg list: $line");
373        #}
374
375        if ( $line =~ /env (\S*)=/ ) {
376            # environment variable env NAME=...
377            $sym = $1;
378        } elsif ( $line =~ /typedef.*\(\*?($C_symbol)\)\s*\(/ ) {
379            # a callback function pointer: typedef ... (*NAME)(...
380            # a callback function signature: typedef ... (NAME)(...
381            $sym = $1;
382        } elsif ( $line =~ /typedef.*($C_symbol)\s*\(/ ) {
383            # a callback function signature: typedef ... NAME(...
384            $sym = $1;
385        } elsif ( $line =~ /typedef.*($C_symbol);/ ) {
386            # a simple typedef: typedef ... NAME;
387            $is_prototype = 0;
388            $sym = $1;
389        } elsif ( $line =~ /enum ($C_symbol) \{/ ) {
390            # an enumeration: enum ... {
391            $sym = $1;
392        } elsif ( $line =~ /#\s*(?:define|undef) ($C_symbol)/ ) {
393            $is_prototype = 0;
394            $sym = $1;
395        } elsif ( $line =~ /^[^\(]*?\(\*($C_symbol)\s*\(/ ) {
396            # a function returning a function pointer: TYPE (*NAME(args))(args)
397            $sym = $1;
398        } elsif ( $line =~ /^[^\(]*?($C_symbol)\s*\(/ ) {
399            # a simple function declaration
400            $sym = $1;
401        }
402        else {
403            next;
404        }
405
406        print STDERR "DEBUG[name_synopsis] \$sym = '$sym'\n" if $debug;
407
408        err($id, "$sym missing from NAME section")
409            unless defined $names{$sym};
410        $names{$sym} = 2;
411
412        # Do some sanity checks on the prototype.
413        err($id, "Prototype missing spaces around commas: $line")
414            if $is_prototype && $line =~ /[a-z0-9],[^\s]/;
415    }
416
417    foreach my $n ( keys %names ) {
418        next if $names{$n} == 2;
419        err($id, "$n missing from SYNOPSIS")
420    }
421}
422
423# Check if SECTION ($3) is located before BEFORE ($4)
424sub check_section_location {
425    my $id = shift;
426    my $contents = shift;
427    my $section = shift;
428    my $before = shift;
429
430    return unless $contents =~ /=head1 $section/
431        and $contents =~ /=head1 $before/;
432    err($id, "$section should appear before $before section")
433        if $contents =~ /=head1 $before.*=head1 $section/ms;
434}
435
436# Check if a =head1 is duplicated, or a =headX is duplicated within a
437# =head1.  Treats =head2 =head3 as equivalent -- it doesn't reset the head3
438# sets if it finds a =head2 -- but that is good enough for now. Also check
439# for proper capitalization, trailing periods, etc.
440sub check_head_style {
441    my $id = shift;
442    my $contents = shift;
443    my %head1;
444    my %subheads;
445
446    foreach my $line ( split /\n+/, $contents ) {
447        next unless $line =~ /^=head/;
448        if ( $line =~ /head1/ ) {
449            err($id, "Duplicate section $line")
450                if defined $head1{$line};
451            $head1{$line} = 1;
452            %subheads = ();
453        } else {
454            err($id, "Duplicate subsection $line")
455                if defined $subheads{$line};
456            $subheads{$line} = 1;
457        }
458        err($id, "Period in =head")
459            if $line =~ /\.[^\w]/ or $line =~ /\.$/;
460        err($id, "not all uppercase in =head1")
461            if $line =~ /head1.*[a-z]/;
462        err($id, "All uppercase in subhead")
463            if $line =~ /head[234][ A-Z0-9]+$/;
464    }
465}
466
467# Because we have options and symbols with extra markup, we need
468# to take that into account, so we need a regexp that extracts
469# markup chunks, including recursive markup.
470# please read up on /(?R)/ in perlre(1)
471# (note: order is important, (?R) needs to come before .)
472# (note: non-greedy is important, or something like 'B<foo> and B<bar>'
473# will be captured as one item)
474my $markup_re =
475    qr/(                        # Capture group
476           [BIL]<               # The start of what we recurse on
477           (?:(?-1)|.)*?        # recurse the whole regexp (referring to
478                                # the last opened capture group, i.e. the
479                                # start of this regexp), or pick next
480                                # character.  Do NOT be greedy!
481           >                    # The end of what we recurse on
482       )/x;                     # (the x allows this sort of split up regexp)
483
484# Options must start with a dash, followed by a letter, possibly
485# followed by letters, digits, dashes and underscores, and the last
486# character must be a letter or a digit.
487# We do also accept the single -? or -n, where n is a digit
488my $option_re =
489    qr/(?:
490            \?                  # Single question mark
491            |
492            \d                  # Single digit
493            |
494            -                   # Single dash (--)
495            |
496            [[:alpha:]](?:[-_[:alnum:]]*?[[:alnum:]])?
497       )/x;
498
499# Helper function to check if a given $thing is properly marked up
500# option.  It returns one of these values:
501#     undef         if it's not an option
502#     ""            if it's a malformed option
503#     $unwrapped    the option with the outermost B<> wrapping removed.
504sub normalise_option {
505    my $id = shift;
506    my $filename = shift;
507    my $thing = shift;
508
509    my $unwrapped = $thing;
510    my $unmarked = $thing;
511
512    # $unwrapped is the option with the outer B<> markup removed
513    $unwrapped =~ s/^B<//;
514    $unwrapped =~ s/>$//;
515    # $unmarked is the option with *all* markup removed
516    $unmarked =~ s/[BIL]<|>//msg;
517
518
519    # If we found an option, check it, collect it
520    if ( $unwrapped =~ /^\s*-/ ) {
521        return $unwrapped       # return option with outer B<> removed
522            if $unmarked =~ /^-${option_re}$/;
523        return "";              # Malformed option
524    }
525    return undef;               # Something else
526}
527
528# Checks of command option (man1) formatting.  The man1 checks are
529# restricted to the SYNOPSIS and OPTIONS sections, the rest is too
530# free form, we simply cannot be too strict there.
531
532sub option_check {
533    my $id = shift;
534    my $filename = shift;
535    my $contents = shift;
536
537    my $synopsis = ($contents =~ /=head1\s+SYNOPSIS(.*?)=head1/s, $1);
538
539    # Some pages have more than one OPTIONS section, let's make sure
540    # to get them all
541    my $options = '';
542    while ( $contents =~ /=head1\s+[A-Z ]*?OPTIONS$(.*?)(?==head1)/msg ) {
543        $options .= $1;
544    }
545
546    # Look for options with no or incorrect markup
547    while ( $synopsis =~
548            /(?<![-<[:alnum:]])-(?:$markup_re|.)*(?![->[:alnum:]])/msg ) {
549        err($id, "Malformed option [1] in SYNOPSIS: $&");
550    }
551
552    my @synopsis;
553    while ( $synopsis =~ /$markup_re/msg ) {
554        my $found = $&;
555        push @synopsis, $found if $found =~ /^B<-/;
556        print STDERR "$id:DEBUG[option_check] SYNOPSIS: found $found\n"
557            if $debug;
558        my $option_uw = normalise_option($id, $filename, $found);
559        err($id, "Malformed option [2] in SYNOPSIS: $found")
560            if defined $option_uw && $option_uw eq '';
561    }
562
563    # In OPTIONS, we look for =item paragraphs.
564    # (?=^\s*$) detects an empty line.
565    my @options;
566    while ( $options =~ /=item\s+(.*?)(?=^\s*$)/msg ) {
567        my $item = $&;
568
569        while ( $item =~ /(\[\s*)?($markup_re)/msg ) {
570            my $found = $2;
571            print STDERR "$id:DEBUG[option_check] OPTIONS: found $&\n"
572                if $debug;
573            err($id, "Unexpected bracket in OPTIONS =item: $item")
574                if ($1 // '') ne '' && $found =~ /^B<\s*-/;
575
576            my $option_uw = normalise_option($id, $filename, $found);
577            err($id, "Malformed option in OPTIONS: $found")
578                if defined $option_uw && $option_uw eq '';
579            if ($found =~ /^B<-/) {
580                push @options, $found;
581                err($id, "OPTIONS entry $found missing from SYNOPSIS")
582                    unless (grep /^\Q$found\E$/, @synopsis)
583                         || $id =~ /(openssl|-options)\.pod:1:$/;
584            }
585        }
586    }
587    foreach (@synopsis) {
588        my $option = $_;
589        err($id, "SYNOPSIS entry $option missing from OPTIONS")
590            unless (grep /^\Q$option\E$/, @options);
591    }
592}
593
594# Normal symbol form
595my $symbol_re = qr/[[:alpha:]_][_[:alnum:]]*?/;
596
597# Checks of function name (man3) formatting.  The man3 checks are
598# easier than the man1 checks, we only check the names followed by (),
599# and only the names that have POD markup.
600sub functionname_check {
601    my $id = shift;
602    my $filename = shift;
603    my $contents = shift;
604
605    while ( $contents =~ /($markup_re)\(\)/msg ) {
606        print STDERR "$id:DEBUG[functionname_check] SYNOPSIS: found $&\n"
607            if $debug;
608
609        my $symbol = $1;
610        my $unmarked = $symbol;
611        $unmarked =~ s/[BIL]<|>//msg;
612
613        err($id, "Malformed symbol: $symbol")
614            unless $symbol =~ /^B<.*?>$/ && $unmarked =~ /^${symbol_re}$/
615    }
616
617    # We can't do the kind of collecting coolness that option_check()
618    # does, because there are too many things that can't be found in
619    # name repositories like the NAME sections, such as symbol names
620    # with a variable part (typically marked up as B<foo_I<TYPE>_bar>
621}
622
623# This is from http://man7.org/linux/man-pages/man7/man-pages.7.html
624my %preferred_words = (
625    '16bit'         => '16-bit',
626    'a.k.a.'        => 'aka',
627    'bitmask'       => 'bit mask',
628    'builtin'       => 'built-in',
629   #'epoch'         => 'Epoch', # handled specially, below
630    'fall-back'     => 'fallback',
631    'file name'     => 'filename',
632    'file system'   => 'filesystem',
633    'host name'     => 'hostname',
634    'i-node'        => 'inode',
635    'lower case'    => 'lowercase',
636    'lower-case'    => 'lowercase',
637    'manpage'       => 'man page',
638    'non-blocking'  => 'nonblocking',
639    'non-default'   => 'nondefault',
640    'non-empty'     => 'nonempty',
641    'non-negative'  => 'nonnegative',
642    'non-zero'      => 'nonzero',
643    'path name'     => 'pathname',
644    'pre-allocated' => 'preallocated',
645    'pseudo-terminal' => 'pseudoterminal',
646    'real time'     => 'real-time',
647    'realtime'      => 'real-time',
648    'reserved port' => 'privileged port',
649    'runtime'       => 'run time',
650    'saved group ID'=> 'saved set-group-ID',
651    'saved set-GID' => 'saved set-group-ID',
652    'saved set-UID' => 'saved set-user-ID',
653    'saved user ID' => 'saved set-user-ID',
654    'set-GID'       => 'set-group-ID',
655    'set-UID'       => 'set-user-ID',
656    'setgid'        => 'set-group-ID',
657    'setuid'        => 'set-user-ID',
658    'sub-system'    => 'subsystem',
659    'super block'   => 'superblock',
660    'super-block'   => 'superblock',
661    'super user'    => 'superuser',
662    'super-user'    => 'superuser',
663    'system port'   => 'privileged port',
664    'time stamp'    => 'timestamp',
665    'time zone'     => 'timezone',
666    'upper case'    => 'uppercase',
667    'upper-case'    => 'uppercase',
668    'useable'       => 'usable',
669    'user name'     => 'username',
670    'userspace'     => 'user space',
671    'zeroes'        => 'zeros'
672);
673
674# Search manpage for words that have a different preferred use.
675sub wording {
676    my $id = shift;
677    my $contents = shift;
678
679    foreach my $k ( keys %preferred_words ) {
680        # Sigh, trademark
681        next if $k eq 'file system'
682            and $contents =~ /Microsoft Encrypted File System/;
683        err($id, "Found '$k' should use '$preferred_words{$k}'")
684            if $contents =~ /\b\Q$k\E\b/i;
685    }
686    err($id, "Found 'epoch' should use 'Epoch'")
687        if $contents =~ /\bepoch\b/;
688    if ( $id =~ m@man1/@ ) {
689        err($id, "found 'tool' in NAME, should use 'command'")
690            if $contents =~ /=head1 NAME.*\btool\b.*=head1 SYNOPSIS/s;
691        err($id, "found 'utility' in NAME, should use 'command'")
692            if $contents =~ /NAME.*\butility\b.*=head1 SYNOPSIS/s;
693
694    }
695}
696
697# Perform all sorts of nit/error checks on a manpage
698sub check {
699    my %podinfo = @_;
700    my $filename = $podinfo{filename};
701    my $dirname = basename(dirname($filename));
702    my $contents = $podinfo{contents};
703
704    # Find what section this page is in; presume 3.
705    my $mansect = 3;
706    $mansect = $1 if $filename =~ /man([1-9])/;
707
708    my $id = "${filename}:1:";
709    check_head_style($id, $contents);
710
711    # Check ordering of some sections in man3
712    if ( $mansect == 3 ) {
713        check_section_location($id, $contents, "RETURN VALUES", "EXAMPLES");
714        check_section_location($id, $contents, "SEE ALSO", "HISTORY");
715        check_section_location($id, $contents, "EXAMPLES", "SEE ALSO");
716    }
717
718    # Make sure every link has a man section number.
719    while ( $contents =~ /$markup_re/msg ) {
720        my $target = $1;
721        next unless $target =~ /^L<(.*)>$/;     # Skip if not L<...>
722        $target = $1;                           # Peal away L< and >
723        $target =~ s/\/[^\/]*$//;               # Peal away possible anchor
724        $target =~ s/.*\|//g;                   # Peal away possible link text
725        next if $target eq '';                  # Skip if links within page, or
726        next if $target =~ /::/;                #   links to a Perl module, or
727        next if $target =~ /^https?:/;          #   is a URL link, or
728        next if $target =~ /\([1357]\)$/;       #   it has a section
729        err($id, "Missing man section number (likely, $mansect) in L<$target>")
730    }
731    # Check for proper links to commands.
732    while ( $contents =~ /L<([^>]*)\(1\)(?:\/.*)?>/g ) {
733        my $target = $1;
734        next if $target =~ /openssl-?/;
735        next if ( grep { basename($_) eq "$target.pod" }
736                  files(TAGS => [ 'manual', 'man1' ]) );
737        next if $target =~ /ps|apropos|sha1sum|procmail|perl/;
738        err($id, "Bad command link L<$target(1)>") if grep /man1/, @sections;
739    }
740    # Check for proper in-man-3 API links.
741    while ( $contents =~ /L<([^>]*)\(3\)(?:\/.*)?>/g ) {
742        my $target = $1;
743        err($id, "Bad L<$target>")
744            unless $target =~ /^[_[:alpha:]][_[:alnum:]]*$/
745    }
746
747    unless ( $contents =~ /^=for openssl generic/ms ) {
748        if ( $mansect == 3 ) {
749            name_synopsis($id, $filename, $contents);
750            functionname_check($id, $filename, $contents);
751        } elsif ( $mansect == 1 ) {
752            option_check($id, $filename, $contents)
753        }
754    }
755
756    wording($id, $contents);
757
758    err($id, "Doesn't start with =pod")
759        if $contents !~ /^=pod/;
760    err($id, "Doesn't end with =cut")
761        if $contents !~ /=cut\n$/;
762    err($id, "More than one cut line.")
763        if $contents =~ /=cut.*=cut/ms;
764    err($id, "EXAMPLE not EXAMPLES section.")
765        if $contents =~ /=head1 EXAMPLE[^S]/;
766    err($id, "WARNING not WARNINGS section.")
767        if $contents =~ /=head1 WARNING[^S]/;
768    err($id, "Missing copyright")
769        if $contents !~ /Copyright .* The OpenSSL Project Authors/;
770    err($id, "Copyright not last")
771        if $contents =~ /head1 COPYRIGHT.*=head/ms;
772    err($id, "head2 in All uppercase")
773        if $contents =~ /head2\s+[A-Z ]+\n/;
774    err($id, "Extra space after head")
775        if $contents =~ /=head\d\s\s+/;
776    err($id, "Period in NAME section")
777        if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
778    err($id, "Duplicate $1 in L<>")
779        if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
780    err($id, "Bad =over $1")
781        if $contents =~ /=over([^ ][^24])/;
782    err($id, "Possible version style issue")
783        if $contents =~ /OpenSSL version [019]/;
784
785    if ( $contents !~ /=for openssl multiple includes/ ) {
786        # Look for multiple consecutive openssl #include lines
787        # (non-consecutive lines are okay; see man3/MD5.pod).
788        if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
789            my $count = 0;
790            foreach my $line ( split /\n+/, $1 ) {
791                if ( $line =~ m@include <openssl/@ ) {
792                    err($id, "Has multiple includes")
793                        if ++$count == 2;
794                } else {
795                    $count = 0;
796                }
797            }
798        }
799    }
800
801    open my $OUT, '>', $temp
802        or die "Can't open $temp, $!";
803    err($id, "POD errors")
804        if podchecker($filename, $OUT) != 0;
805    close $OUT;
806    open $OUT, '<', $temp
807        or die "Can't read $temp, $!";
808    while ( <$OUT> ) {
809        next if /\(section\) in.*deprecated/;
810        print;
811    }
812    close $OUT;
813    unlink $temp || warn "Can't remove $temp, $!";
814
815    # Find what section this page is in; presume 3.
816    my $section = 3;
817    $section = $1 if $dirname =~ /man([1-9])/;
818
819    foreach ( (@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}}) ) {
820        err($id, "Missing $_ head1 section")
821            if $contents !~ /^=head1\s+${_}\s*$/m;
822    }
823}
824
825# Information database ###############################################
826
827# Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
828my %link_map = ();
829# Map of names in each POD file or from "missing" files; possible values are:
830# If found in a POD files, "name(s)" => filename
831# If found in a "missing" file or external, "name(s)" => ''
832my %name_map = ();
833
834# State of man-page names.
835# %state is affected by loading util/*.num and util/*.syms
836# Values may be one of:
837# 'crypto' : belongs in libcrypto (loaded from libcrypto.num)
838# 'ssl' : belongs in libssl (loaded from libssl.num)
839# 'other' : belongs in libcrypto or libssl (loaded from other.syms)
840# 'internal' : Internal
841# 'public' : Public (generic name or external documentation)
842# Any of these values except 'public' may be prefixed with 'missing_'
843# to indicate that they are known to be missing.
844my %state;
845# %missing is affected by loading util/missing*.txt.  Values may be one of:
846# 'crypto' : belongs in libcrypto (loaded from libcrypto.num)
847# 'ssl' : belongs in libssl (loaded from libssl.num)
848# 'other' : belongs in libcrypto or libssl (loaded from other.syms)
849# 'internal' : Internal
850my %missing;
851
852# Parse libcrypto.num, etc., and return sorted list of what's there.
853sub loadnum ($;$) {
854    my $file = shift;
855    my $type = shift;
856    my @symbols;
857
858    open my $IN, '<', catfile($config{sourcedir}, $file)
859        or die "Can't open $file, $!, stopped";
860
861    while ( <$IN> ) {
862        next if /^#/;
863        next if /\bNOEXIST\b/;
864        my @fields = split();
865        die "Malformed line $. in $file: $_"
866            if scalar @fields != 2 && scalar @fields != 4;
867        $state{$fields[0].'(3)'} = $type // 'internal';
868    }
869    close $IN;
870}
871
872# Load file of symbol names that we know aren't documented.
873sub loadmissing($;$)
874{
875    my $missingfile = shift;
876    my $type = shift;
877
878    open FH, catfile($config{sourcedir}, $missingfile)
879        or die "Can't open $missingfile";
880    while ( <FH> ) {
881        chomp;
882        next if /^#/;
883        $missing{$_} = $type // 'internal';
884    }
885    close FH;
886}
887
888# Check that we have consistent public / internal documentation and declaration
889sub checkstate () {
890    # Collect all known names, no matter where they come from
891    my %names = map { $_ => 1 } (keys %name_map, keys %state, keys %missing);
892
893    # Check section 3, i.e. functions and macros
894    foreach ( grep { $_ =~ /\(3\)$/ } sort keys %names ) {
895        next if ( $name_map{$_} // '') eq '' || $_ =~ /$ignored/;
896
897        # If a man-page isn't recorded public or if it's recorded missing
898        # and internal, it's declared to be internal.
899        my $declared_internal =
900            ($state{$_} // 'internal') eq 'internal'
901            || ($missing{$_} // '') eq 'internal';
902        # If a man-page isn't recorded internal or if it's recorded missing
903        # and not internal, it's declared to be public
904        my $declared_public =
905            ($state{$_} // 'internal') ne 'internal'
906            || ($missing{$_} // 'internal') ne 'internal';
907
908        err("$_ is supposedly public but is documented as internal")
909            if ( $declared_public && $name_map{$_} =~ /\/internal\// );
910        err("$_ is supposedly internal (maybe missing from other.syms) but is documented as public")
911            if ( $declared_internal && $name_map{$_} !~ /\/internal\// );
912    }
913}
914
915# Check for undocumented macros; ignore those in the "missing" file
916# and do simple check for #define in our header files.
917sub checkmacros {
918    my $count = 0;
919    my %seen;
920
921    foreach my $f ( files(TAGS => 'public_header') ) {
922        # Skip some internals we don't want to document yet.
923        my $b = basename($f);
924        next if $b eq 'asn1.h';
925        next if $b eq 'asn1t.h';
926        next if $b eq 'err.h';
927        open(IN, $f)
928            or die "Can't open $f, $!";
929        while ( <IN> ) {
930            next unless /^#\s*define\s*(\S+)\(/;
931            my $macro = "$1(3)"; # We know they're all in section 3
932            next if defined $name_map{$macro}
933                || defined $missing{$macro}
934                || defined $seen{$macro}
935                || $macro =~ /$ignored/;
936
937            err("$f:", "macro $macro undocumented")
938                if $opt_d || $opt_e;
939            $count++;
940            $seen{$macro} = 1;
941        }
942        close(IN);
943    }
944    err("# $count macros undocumented (count is approximate)")
945        if $count > 0;
946}
947
948# Find out what is undocumented (filtering out the known missing ones)
949# and display them.
950sub printem ($) {
951    my $type = shift;
952    my $count = 0;
953
954    foreach my $func ( grep { $state{$_} eq $type } sort keys %state ) {
955        next if defined $name_map{$func}
956            || defined $missing{$func};
957
958        err("$type:", "function $func undocumented")
959            if $opt_d || $opt_e;
960        $count++;
961    }
962    err("# $count lib$type names are not documented")
963        if $count > 0;
964}
965
966# Collect all the names in a manpage.
967sub collectnames {
968    my %podinfo = @_;
969    my $filename = $podinfo{filename};
970    $filename =~ m|man(\d)/|;
971    my $section = $1;
972    my $simplename = basename($filename, ".pod");
973    my $id = "${filename}:1:";
974    my $is_generic = $podinfo{contents} =~ /^=for openssl generic/ms;
975
976    unless ( grep { $simplename eq $_ } @{$podinfo{names}} ) {
977        err($id, "$simplename not in NAME section");
978        push @{$podinfo{names}}, $simplename;
979    }
980    foreach my $name ( @{$podinfo{names}} ) {
981        next if $name eq "";
982        err($id, "'$name' contains whitespace")
983            if $name =~ /\s/;
984        my $name_sec = "$name($section)";
985        if ( !defined $name_map{$name_sec} ) {
986            $name_map{$name_sec} = $filename;
987            $state{$name_sec} //=
988                ( $filename =~ /\/internal\// ? 'internal' : 'public' )
989                if $is_generic;
990        } elsif ( $filename eq $name_map{$name_sec} ) {
991            err($id, "$name_sec duplicated in NAME section of",
992                 $name_map{$name_sec});
993        } elsif ( $name_map{$name_sec} ne '' ) {
994            err($id, "$name_sec also in NAME section of",
995                 $name_map{$name_sec});
996        }
997    }
998
999    if ( $podinfo{contents} =~ /=for openssl foreign manual (.*)\n/ ) {
1000        foreach my $f ( split / /, $1 ) {
1001            $name_map{$f} = ''; # It still exists!
1002            $state{$f} = 'public'; # We assume!
1003        }
1004    }
1005
1006    my @links = ();
1007    # Don't use this regexp directly on $podinfo{contents}, as it causes
1008    # a regexp recursion, which fails on really big PODs.  Instead, use
1009    # $markup_re to pick up general markup, and use this regexp to check
1010    # that the markup that was found is indeed a link.
1011    my $linkre = qr/L<
1012                    # if the link is of the form L<something|name(s)>,
1013                    # then remove 'something'.  Note that 'something'
1014                    # may contain POD codes as well...
1015                    (?:(?:[^\|]|<[^>]*>)*\|)?
1016                    # we're only interested in references that have
1017                    # a one digit section number
1018                    ([^\/>\(]+\(\d\))
1019                   /x;
1020    while ( $podinfo{contents} =~ /$markup_re/msg ) {
1021        my $x = $1;
1022
1023        if ($x =~ $linkre) {
1024            push @links, $1;
1025        }
1026    }
1027    $link_map{$filename} = [ @links ];
1028}
1029
1030# Look for L<> ("link") references that point to files that do not exist.
1031sub checklinks {
1032    foreach my $filename ( sort keys %link_map ) {
1033        foreach my $link ( @{$link_map{$filename}} ) {
1034            err("${filename}:1:", "reference to non-existing $link")
1035                unless defined $name_map{$link} || defined $missing{$link};
1036            err("${filename}:1:", "reference of internal $link in public documentation $filename")
1037                if ( ( ($state{$link} // '') eq 'internal'
1038                       || ($missing{$link} // '') eq 'internal' )
1039                     && $filename !~ /\/internal\// );
1040        }
1041    }
1042}
1043
1044# Cipher/digests to skip if they show up as "not implemented"
1045# because they are, via the "-*" construct.
1046my %skips = (
1047    'aes128' => 1,
1048    'aes192' => 1,
1049    'aes256' => 1,
1050    'aria128' => 1,
1051    'aria192' => 1,
1052    'aria256' => 1,
1053    'camellia128' => 1,
1054    'camellia192' => 1,
1055    'camellia256' => 1,
1056    'des' => 1,
1057    'des3' => 1,
1058    'idea' => 1,
1059    'cipher' => 1,
1060    'digest' => 1,
1061);
1062
1063my %genopts; # generic options parsed from apps/include/opt.h
1064
1065# Check the flags of a command and see if everything is in the manpage
1066sub checkflags {
1067    my $cmd = shift;
1068    my $doc = shift;
1069    my @cmdopts;
1070    my %docopts;
1071
1072    # Get the list of options in the command source file.
1073    my $active = 0;
1074    my $expect_helpstr = "";
1075    open CFH, "apps/$cmd.c"
1076        or die "Can't open apps/$cmd.c to list options for $cmd, $!";
1077    while ( <CFH> ) {
1078        chop;
1079        if ($active) {
1080            last if m/^\s*};/;
1081            if ($expect_helpstr ne "") {
1082                next if m/^\s*#\s*if/;
1083                err("$cmd does not implement help for -$expect_helpstr") unless m/^\s*"/;
1084                $expect_helpstr = "";
1085            }
1086            if (m/\{\s*"([^"]+)"\s*,\s*OPT_[A-Z0-9_]+\s*,\s*('[-\/:<>cAEfFlMnNpsuU]'|0)(.*)$/
1087                    && !($cmd eq "s_client" && $1 eq "wdebug")) {
1088                push @cmdopts, $1;
1089                $expect_helpstr = $1;
1090                $expect_helpstr = "" if $3 =~ m/^\s*,\s*"/;
1091            } elsif (m/[\s,](OPT_[A-Z]+_OPTIONS?)\s*(,|$)/) {
1092                push @cmdopts, @{ $genopts{$1} };
1093            }
1094        } elsif (m/^const\s+OPTIONS\s*/) {
1095            $active = 1;
1096        }
1097    }
1098    close CFH;
1099
1100    # Get the list of flags from the synopsis
1101    open CFH, "<$doc"
1102        or die "Can't open $doc, $!";
1103    while ( <CFH> ) {
1104        chop;
1105        last if /DESCRIPTION/;
1106        my $opt;
1107        if ( /\[B<-([^ >]+)/ ) {
1108            $opt = $1;
1109        } elsif ( /^B<-([^ >]+)/ ) {
1110            $opt = $1;
1111        } else {
1112            next;
1113        }
1114        $opt = $1 if $opt =~ /I<(.*)/;
1115        $docopts{$1} = 1;
1116    }
1117    close CFH;
1118
1119    # See what's in the command not the manpage.
1120    my @undocced = sort grep { !defined $docopts{$_} } @cmdopts;
1121    foreach ( @undocced ) {
1122        err("$doc: undocumented $cmd option -$_");
1123    }
1124
1125    # See what's in the manpage not the command.
1126    my @unimpl = sort grep { my $e = $_; !(grep /^\Q$e\E$/, @cmdopts) } keys %docopts;
1127    foreach ( @unimpl ) {
1128        next if $_ eq "-"; # Skip the -- end-of-flags marker
1129        next if defined $skips{$_};
1130        err("$doc: $cmd does not implement -$_");
1131    }
1132}
1133
1134##
1135##  MAIN()
1136##  Do the work requested by the various getopt flags.
1137##  The flags are parsed in alphabetical order, just because we have
1138##  to have *some way* of listing them.
1139##
1140
1141if ( $opt_c ) {
1142    my @commands = ();
1143
1144    # Get the lists of generic options.
1145    my $active = "";
1146    open OFH, catdir($config{sourcedir}, "apps/include/opt.h")
1147        or die "Can't open apps/include/opt.h to list generic options, $!";
1148    while ( <OFH> ) {
1149        chop;
1150        push @{ $genopts{$active} }, $1 if $active ne "" && m/^\s+\{\s*"([^"]+)"\s*,\s*OPT_/;
1151        $active = $1 if m/^\s*#\s*define\s+(OPT_[A-Z]+_OPTIONS?)\s*\\\s*$/;
1152        $active = "" if m/^\s*$/;
1153    }
1154    close OFH;
1155
1156    # Get list of commands.
1157    opendir(DIR, "apps");
1158    @commands = grep(/\.c$/, readdir(DIR));
1159    closedir(DIR);
1160
1161    # See if each has a manpage.
1162    foreach my $cmd ( @commands ) {
1163        $cmd =~ s/\.c$//;
1164        next if $cmd eq 'progs' || $cmd eq 'vms_decc_init';
1165        my @doc = ( grep { basename($_) eq "openssl-$cmd.pod"
1166                           # For "tsget" and "CA.pl" pod pages
1167                           || basename($_) eq "$cmd.pod" }
1168                    files(TAGS => [ 'manual', 'man1' ]) );
1169        my $num = scalar @doc;
1170        if ($num > 1) {
1171            err("$num manuals for 'openssl $cmd': ".join(", ", @doc));
1172        } elsif ($num < 1) {
1173            err("no manual for 'openssl $cmd'");
1174        } else {
1175            checkflags($cmd, @doc);
1176        }
1177    }
1178}
1179
1180# Populate %state
1181loadnum('util/libcrypto.num', 'crypto');
1182loadnum('util/libssl.num', 'ssl');
1183loadnum('util/other.syms', 'other');
1184loadnum('util/other-internal.syms');
1185if ( $opt_o ) {
1186    loadmissing('util/missingmacro111.txt', 'crypto');
1187    loadmissing('util/missingcrypto111.txt', 'crypto');
1188    loadmissing('util/missingssl111.txt', 'ssl');
1189} elsif ( !$opt_u ) {
1190    loadmissing('util/missingmacro.txt', 'crypto');
1191    loadmissing('util/missingcrypto.txt', 'crypto');
1192    loadmissing('util/missingssl.txt', 'ssl');
1193    loadmissing('util/missingcrypto-internal.txt');
1194    loadmissing('util/missingssl-internal.txt');
1195}
1196
1197if ( $opt_n || $opt_l || $opt_u || $opt_v ) {
1198    my @files_to_read = ( $opt_n && @ARGV ) ? @ARGV : files(TAGS => 'manual');
1199
1200    foreach (@files_to_read) {
1201        my %podinfo = extract_pod_info($_, { debug => $debug });
1202
1203        collectnames(%podinfo)
1204            if ( $opt_l || $opt_u || $opt_v );
1205
1206        check(%podinfo)
1207            if ( $opt_n );
1208    }
1209}
1210
1211if ( $opt_l ) {
1212    checklinks();
1213}
1214
1215if ( $opt_n ) {
1216    # If not given args, check that all man1 commands are named properly.
1217    if ( scalar @ARGV == 0 && grep /man1/, @sections ) {
1218        foreach ( files(TAGS => [ 'public_manual', 'man1' ]) ) {
1219            next if /openssl\.pod/
1220                || /CA\.pl/ || /tsget\.pod/; # these commands are special cases
1221            err("$_ doesn't start with openssl-") unless /openssl-/;
1222        }
1223    }
1224}
1225
1226checkstate();
1227
1228if ( $opt_u || $opt_v) {
1229    printem('crypto');
1230    printem('ssl');
1231    checkmacros();
1232}
1233
1234exit $status;
1235