xref: /netbsd-src/crypto/external/bsd/openssl/dist/util/find-doc-nits (revision 7330f729ccf0bd976a06f95fad452fe774fc7fd1)
1#! /usr/bin/env perl
2# Copyright 2002-2019 The OpenSSL Project Authors. All Rights Reserved.
3#
4# Licensed under the OpenSSL license (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;
13use Pod::Checker;
14use File::Find;
15use File::Basename;
16use File::Spec::Functions;
17use Getopt::Std;
18use lib catdir(dirname($0), "perl");
19use OpenSSL::Util::Pod;
20
21# Options.
22our($opt_d);
23our($opt_h);
24our($opt_l);
25our($opt_n);
26our($opt_p);
27our($opt_u);
28our($opt_c);
29
30sub help()
31{
32    print <<EOF;
33Find small errors (nits) in documentation.  Options:
34    -d Detailed list of undocumented (implies -u)
35    -l Print bogus links
36    -n Print nits in POD pages
37    -p Warn if non-public name documented (implies -n)
38    -u List undocumented functions
39    -h Print this help message
40    -c List undocumented commands and options
41EOF
42    exit;
43}
44
45my $temp = '/tmp/docnits.txt';
46my $OUT;
47my %public;
48
49my %mandatory_sections =
50    ( '*'    => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
51      1      => [ 'SYNOPSIS', 'OPTIONS' ],
52      3      => [ 'SYNOPSIS', 'RETURN VALUES' ],
53      5      => [ ],
54      7      => [ ] );
55
56# Cross-check functions in the NAME and SYNOPSIS section.
57sub name_synopsis()
58{
59    my $id = shift;
60    my $filename = shift;
61    my $contents = shift;
62
63    # Get NAME section and all words in it.
64    return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
65    my $tmp = $1;
66    $tmp =~ tr/\n/ /;
67    print "$id trailing comma before - in NAME\n" if $tmp =~ /, *-/;
68    $tmp =~ s/ -.*//g;
69    $tmp =~ s/  */ /g;
70    print "$id missing comma in NAME\n" if $tmp =~ /[^,] /;
71    $tmp =~ s/,//g;
72
73    my $dirname = dirname($filename);
74    my $simplename = basename($filename);
75    $simplename =~ s/.pod$//;
76    my $foundfilename = 0;
77    my %foundfilenames = ();
78    my %names;
79    foreach my $n ( split ' ', $tmp ) {
80        $names{$n} = 1;
81        $foundfilename++ if $n eq $simplename;
82        $foundfilenames{$n} = 1
83            if -f "$dirname/$n.pod" && $n ne $simplename;
84    }
85    print "$id the following exist as other .pod files:\n",
86        join(" ", sort keys %foundfilenames), "\n"
87        if %foundfilenames;
88    print "$id $simplename (filename) missing from NAME section\n"
89        unless $foundfilename;
90    foreach my $n ( keys %names ) {
91        print "$id $n is not public\n"
92            if $opt_p and !defined $public{$n};
93    }
94
95    # Find all functions in SYNOPSIS
96    return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
97    my $syn = $1;
98    foreach my $line ( split /\n+/, $syn ) {
99        my $sym;
100        $line =~ s/STACK_OF\([^)]+\)/int/g;
101        $line =~ s/__declspec\([^)]+\)//;
102        if ( $line =~ /env (\S*)=/ ) {
103            # environment variable env NAME=...
104            $sym = $1;
105        } elsif ( $line =~ /typedef.*\(\*(\S+)\)\(.*/ ) {
106            # a callback function pointer: typedef ... (*NAME)(...
107            $sym = $1;
108        } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
109            # a callback function signature: typedef ... NAME(...
110            $sym = $1;
111        } elsif ( $line =~ /typedef.* (\S+);/ ) {
112            # a simple typedef: typedef ... NAME;
113            $sym = $1;
114        } elsif ( $line =~ /enum (\S*) \{/ ) {
115            # an enumeration: enum ... {
116            $sym = $1;
117        } elsif ( $line =~ /#define ([A-Za-z0-9_]+)/ ) {
118            $sym = $1;
119        } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
120            $sym = $1;
121        }
122        else {
123            next;
124        }
125        print "$id $sym missing from NAME section\n"
126            unless defined $names{$sym};
127        $names{$sym} = 2;
128
129        # Do some sanity checks on the prototype.
130        print "$id prototype missing spaces around commas: $line\n"
131            if ( $line =~ /[a-z0-9],[^ ]/ );
132    }
133
134    foreach my $n ( keys %names ) {
135        next if $names{$n} == 2;
136        print "$id $n missing from SYNOPSIS\n";
137    }
138}
139
140# Check if SECTION is located before BEFORE
141sub check_section_location()
142{
143    my $filename = shift;
144    my $contents = shift;
145    my $section = shift;
146    my $before = shift;
147
148    return unless $contents =~ /=head1 $section/
149        and $contents =~ /=head1 $before/;
150    print "$filename: $section should be placed before $before section\n"
151        if $contents =~ /=head1 $before.*=head1 $section/ms;
152}
153
154sub check()
155{
156    my $filename = shift;
157    my $dirname = basename(dirname($filename));
158
159    my $contents = '';
160    {
161        local $/ = undef;
162        open POD, $filename or die "Couldn't open $filename, $!";
163        $contents = <POD>;
164        close POD;
165    }
166
167    # Check if EXAMPLES is located after RETURN VALUES section.
168    &check_section_location($filename, $contents, "RETURN VALUES", "EXAMPLES") if $filename =~ m|man3/|;
169    # Check if HISTORY is located after SEE ALSO
170    &check_section_location($filename, $contents, "SEE ALSO", "HISTORY") if $filename =~ m|man3/|;
171    # Check if SEE ALSO is located after EXAMPLES
172    &check_section_location($filename, $contents, "EXAMPLES", "SEE ALSO") if $filename =~ m|man3/|;
173
174    my $id = "${filename}:1:";
175
176    &name_synopsis($id, $filename, $contents)
177        unless $contents =~ /=for comment generic/
178            or $filename =~ m@man[157]/@;
179
180    print "$id doesn't start with =pod\n"
181        if $contents !~ /^=pod/;
182    print "$id doesn't end with =cut\n"
183        if $contents !~ /=cut\n$/;
184    print "$id more than one cut line.\n"
185        if $contents =~ /=cut.*=cut/ms;
186    print "$id missing copyright\n"
187        if $contents !~ /Copyright .* The OpenSSL Project Authors/;
188    print "$id copyright not last\n"
189        if $contents =~ /head1 COPYRIGHT.*=head/ms;
190    print "$id head2 in All uppercase\n"
191        if $contents =~ /head2\s+[A-Z ]+\n/;
192    print "$id extra space after head\n"
193        if $contents =~ /=head\d\s\s+/;
194    print "$id period in NAME section\n"
195        if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
196    print "$id POD markup in NAME section\n"
197        if $contents =~ /=head1 NAME.*[<>].*=head1 SYNOPSIS/ms;
198    print "$id Duplicate $1 in L<>\n"
199        if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
200    print "$id Bad =over $1\n"
201        if $contents =~ /=over([^ ][^24])/;
202    print "$id Possible version style issue\n"
203        if $contents =~ /OpenSSL version [019]/;
204
205    if ( $contents !~ /=for comment multiple includes/ ) {
206        # Look for multiple consecutive openssl #include lines
207        # (non-consecutive lines are okay; see man3/MD5.pod).
208        if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
209            my $count = 0;
210            foreach my $line ( split /\n+/, $1 ) {
211                if ( $line =~ m@include <openssl/@ ) {
212                    print "$id has multiple includes\n" if ++$count == 2;
213                } else {
214                    $count = 0;
215                }
216            }
217        }
218    }
219
220    open my $OUT, '>', $temp
221        or die "Can't open $temp, $!";
222    podchecker($filename, $OUT);
223    close $OUT;
224    open $OUT, '<', $temp
225        or die "Can't read $temp, $!";
226    while ( <$OUT> ) {
227        next if /\(section\) in.*deprecated/;
228        print;
229    }
230    close $OUT;
231    unlink $temp || warn "Can't remove $temp, $!";
232
233    # Find what section this page is in; assume 3.
234    my $section = 3;
235    $section = $1 if $dirname =~ /man([1-9])/;
236
237    foreach ((@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}})) {
238        # Skip "return values" if not -s
239        print "$id: missing $_ head1 section\n"
240            if $contents !~ /^=head1\s+${_}\s*$/m;
241    }
242}
243
244my %dups;
245
246sub parsenum()
247{
248    my $file = shift;
249    my @apis;
250
251    open my $IN, '<', $file
252        or die "Can't open $file, $!, stopped";
253
254    while ( <$IN> ) {
255        next if /^#/;
256        next if /\bNOEXIST\b/;
257        next if /\bEXPORT_VAR_AS_FUNC\b/;
258        my @fields = split();
259        die "Malformed line $_"
260            if scalar @fields != 2 && scalar @fields != 4;
261        push @apis, $fields[0];
262    }
263
264    close $IN;
265
266    print "# Found ", scalar(@apis), " in $file\n" unless $opt_p;
267    return sort @apis;
268}
269
270sub getdocced()
271{
272    my $dir = shift;
273    my %return;
274
275    foreach my $pod ( glob("$dir/*.pod") ) {
276        my %podinfo = extract_pod_info($pod);
277        foreach my $n ( @{$podinfo{names}} ) {
278            $return{$n} = $pod;
279            print "# Duplicate $n in $pod and $dups{$n}\n"
280                if defined $dups{$n} && $dups{$n} ne $pod;
281            $dups{$n} = $pod;
282        }
283    }
284
285    return %return;
286}
287
288my %docced;
289
290sub checkmacros()
291{
292    my $count = 0;
293
294    print "# Checking macros (approximate)\n";
295    foreach my $f ( glob('include/openssl/*.h') ) {
296        # Skip some internals we don't want to document yet.
297        next if $f eq 'include/openssl/asn1.h';
298        next if $f eq 'include/openssl/asn1t.h';
299        next if $f eq 'include/openssl/err.h';
300        open(IN, $f) || die "Can't open $f, $!";
301        while ( <IN> ) {
302            next unless /^#\s*define\s*(\S+)\(/;
303            my $macro = $1;
304            next if $docced{$macro};
305            next if $macro =~ /i2d_/
306                || $macro =~ /d2i_/
307                || $macro =~ /DEPRECATEDIN/
308                || $macro =~ /IMPLEMENT_/
309                || $macro =~ /DECLARE_/;
310            print "$f:$macro\n" if $opt_d;
311            $count++;
312        }
313        close(IN);
314    }
315    print "# Found $count macros missing (not all should be documented)\n"
316}
317
318sub printem()
319{
320    my $libname = shift;
321    my $numfile = shift;
322    my $count = 0;
323
324    foreach my $func ( &parsenum($numfile) ) {
325        next if $docced{$func};
326
327        # Skip ASN1 utilities
328        next if $func =~ /^ASN1_/;
329
330        print "$libname:$func\n" if $opt_d;
331        $count++;
332    }
333    print "# Found $count missing from $numfile\n\n";
334}
335
336
337# Collection of links in each POD file.
338# filename => [ "foo(1)", "bar(3)", ... ]
339my %link_collection = ();
340# Collection of names in each POD file.
341# "name(s)" => filename
342my %name_collection = ();
343
344sub collectnames {
345    my $filename = shift;
346    $filename =~ m|man(\d)/|;
347    my $section = $1;
348    my $simplename = basename($filename, ".pod");
349    my $id = "${filename}:1:";
350
351    my $contents = '';
352    {
353        local $/ = undef;
354        open POD, $filename or die "Couldn't open $filename, $!";
355        $contents = <POD>;
356        close POD;
357    }
358
359    $contents =~ /=head1 NAME([^=]*)=head1 /ms;
360    my $tmp = $1;
361    unless (defined $tmp) {
362        print "$id weird name section\n";
363        return;
364    }
365    $tmp =~ tr/\n/ /;
366    $tmp =~ s/-.*//g;
367
368    my @names = map { s/\s+//g; $_ } split(/,/, $tmp);
369    unless (grep { $simplename eq $_ } @names) {
370        print "$id missing $simplename\n";
371        push @names, $simplename;
372    }
373    foreach my $name (@names) {
374        next if $name eq "";
375        my $name_sec = "$name($section)";
376        if (! exists $name_collection{$name_sec}) {
377            $name_collection{$name_sec} = $filename;
378        } else { #elsif ($filename ne $name_collection{$name_sec}) {
379            print "$id $name_sec also in $name_collection{$name_sec}\n";
380        }
381    }
382
383    my @foreign_names =
384        map { map { s/\s+//g; $_ } split(/,/, $_) }
385        $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
386    foreach (@foreign_names) {
387        $name_collection{$_} = undef; # It still exists!
388    }
389
390    my @links = $contents =~ /L<
391                              # if the link is of the form L<something|name(s)>,
392                              # then remove 'something'.  Note that 'something'
393                              # may contain POD codes as well...
394                              (?:(?:[^\|]|<[^>]*>)*\|)?
395                              # we're only interested in references that have
396                              # a one digit section number
397                              ([^\/>\(]+\(\d\))
398                             /gx;
399    $link_collection{$filename} = [ @links ];
400}
401
402sub checklinks {
403    foreach my $filename (sort keys %link_collection) {
404        foreach my $link (@{$link_collection{$filename}}) {
405            print "${filename}:1: reference to non-existing $link\n"
406                unless exists $name_collection{$link};
407        }
408    }
409}
410
411sub publicize() {
412    foreach my $name ( &parsenum('util/libcrypto.num') ) {
413        $public{$name} = 1;
414    }
415    foreach my $name ( &parsenum('util/libssl.num') ) {
416        $public{$name} = 1;
417    }
418    foreach my $name ( &parsenum('util/private.num') ) {
419        $public{$name} = 1;
420    }
421}
422
423my %skips = (
424    'aes128' => 1,
425    'aes192' => 1,
426    'aes256' => 1,
427    'aria128' => 1,
428    'aria192' => 1,
429    'aria256' => 1,
430    'camellia128' => 1,
431    'camellia192' => 1,
432    'camellia256' => 1,
433    'des' => 1,
434    'des3' => 1,
435    'idea' => 1,
436    '[cipher]' => 1,
437    '[digest]' => 1,
438);
439
440sub checkflags() {
441    my $cmd = shift;
442    my %cmdopts;
443    my %docopts;
444    my $ok = 1;
445
446    # Get the list of options in the command.
447    open CFH, "./apps/openssl list --options $cmd|"
448        || die "Can list options for $cmd, $!";
449    while ( <CFH> ) {
450        chop;
451        s/ .$//;
452        $cmdopts{$_} = 1;
453    }
454    close CFH;
455
456    # Get the list of flags from the synopsis
457    open CFH, "<doc/man1/$cmd.pod"
458        || die "Can't open $cmd.pod, $!";
459    while ( <CFH> ) {
460        chop;
461        last if /DESCRIPTION/;
462        next unless /\[B<-([^ >]+)/;
463        $docopts{$1} = 1;
464    }
465    close CFH;
466
467    # See what's in the command not the manpage.
468    my @undocced = ();
469    foreach my $k ( keys %cmdopts ) {
470        push @undocced, $k unless $docopts{$k};
471    }
472    if ( scalar @undocced > 0 ) {
473        $ok = 0;
474        foreach ( @undocced ) {
475            print "doc/man1/$cmd.pod: Missing -$_\n";
476        }
477    }
478
479    # See what's in the command not the manpage.
480    my @unimpl = ();
481    foreach my $k ( keys %docopts ) {
482        push @unimpl, $k unless $cmdopts{$k};
483    }
484    if ( scalar @unimpl > 0 ) {
485        $ok = 0;
486        foreach ( @unimpl ) {
487            next if defined $skips{$_};
488            print "doc/man1/$cmd.pod: Not implemented -$_\n";
489        }
490    }
491
492    return $ok;
493}
494
495getopts('cdlnphu');
496
497&help() if $opt_h;
498$opt_n = 1 if $opt_p;
499$opt_u = 1 if $opt_d;
500
501die "Need one of -[cdlnpu] flags.\n"
502    unless $opt_c or $opt_l or $opt_n or $opt_u;
503
504if ( $opt_c ) {
505    my $ok = 1;
506    my @commands = ();
507
508    # Get list of commands.
509    open FH, "./apps/openssl list -1 -commands|"
510        || die "Can't list commands, $!";
511    while ( <FH> ) {
512        chop;
513        push @commands, $_;
514    }
515    close FH;
516
517    # See if each has a manpage.
518    foreach ( @commands ) {
519        next if $_ eq 'help' || $_ eq 'exit';
520        if ( ! -f "doc/man1/$_.pod" ) {
521            print "doc/man1/$_.pod does not exist\n";
522            $ok = 0;
523        } else {
524            $ok = 0 if not &checkflags($_);
525        }
526    }
527
528    # See what help is missing.
529    open FH, "./apps/openssl list --missing-help |"
530        || die "Can't list missing help, $!";
531    while ( <FH> ) {
532        chop;
533        my ($cmd, $flag) = split;
534        print "$cmd has no help for -$flag\n";
535        $ok = 0;
536    }
537    close FH;
538
539    exit 1 if not $ok;
540}
541
542if ( $opt_l ) {
543    foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
544        collectnames($_);
545    }
546    checklinks();
547}
548
549if ( $opt_n ) {
550    &publicize() if $opt_p;
551    foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
552        &check($_);
553    }
554}
555
556if ( $opt_u ) {
557    my %temp = &getdocced('doc/man3');
558    foreach ( keys %temp ) {
559        $docced{$_} = $temp{$_};
560    }
561    &printem('crypto', 'util/libcrypto.num');
562    &printem('ssl', 'util/libssl.num');
563    &checkmacros();
564}
565
566exit;
567