]> git.ipfire.org Git - thirdparty/openssl.git/blob - util/find-doc-nits
Exit non-zero if find-doc-nits finds nits
[thirdparty/openssl.git] / util / find-doc-nits
1 #! /usr/bin/env perl
2 # Copyright 2002-2019 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
10 require 5.10.0;
11 use warnings;
12 use strict;
13 use Pod::Checker;
14 use File::Find;
15 use File::Basename;
16 use File::Spec::Functions;
17 use Getopt::Std;
18 use lib catdir(dirname($0), "perl");
19 use OpenSSL::Util::Pod;
20
21 # Options.
22 our($opt_d);
23 our($opt_e);
24 our($opt_s);
25 our($opt_o);
26 our($opt_h);
27 our($opt_l);
28 our($opt_n);
29 our($opt_p);
30 our($opt_u);
31 our($opt_v);
32 our($opt_c);
33
34 sub help {
35 print <<EOF;
36 Find small errors (nits) in documentation. Options:
37 -d Detailed list of undocumented (implies -u)
38 -e Detailed list of new undocumented (implies -v)
39 -s Same as -e except no output is generated if nothing is undocumented
40 -o Causes -e/-v to count symbols added since 1.1.1 as new (implies -v)
41 -l Print bogus links
42 -n Print nits in POD pages
43 -p Warn if non-public name documented (implies -n)
44 -u Count undocumented functions
45 -v Count new undocumented functions
46 -h Print this help message
47 -c List undocumented commands and options
48 EOF
49 exit;
50 }
51
52 my $temp = '/tmp/docnits.txt';
53 my $OUT;
54 my %public;
55 my $status = 0;
56
57 my %mandatory_sections =
58 ( '*' => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
59 1 => [ 'SYNOPSIS', 'OPTIONS' ],
60 3 => [ 'SYNOPSIS', 'RETURN VALUES' ],
61 5 => [ ],
62 7 => [ ] );
63
64 # Print error message, set $status.
65 sub err {
66 print join(" ", @_), "\n";
67 $status = 1
68 }
69
70 # Cross-check functions in the NAME and SYNOPSIS section.
71 sub name_synopsis {
72 my $id = shift;
73 my $filename = shift;
74 my $contents = shift;
75
76 # Get NAME section and all words in it.
77 return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
78 my $tmp = $1;
79 $tmp =~ tr/\n/ /;
80 err($id, "trailing comma before - in NAME")
81 if $tmp =~ /, *-/;
82 $tmp =~ s/ -.*//g;
83 err($id, "POD markup among the names in NAME")
84 if $tmp =~ /[<>]/;
85 $tmp =~ s/ */ /g;
86 err($id, "missing comma in NAME")
87 if $tmp =~ /[^,] /;
88
89 my $dirname = dirname($filename);
90 my $simplename = basename(basename($filename, ".in"), ".pod");
91 my $foundfilename = 0;
92 my %foundfilenames = ();
93 my %names;
94 foreach my $n ( split ',', $tmp ) {
95 $n =~ s/^\s+//;
96 $n =~ s/\s+$//;
97 err($id, "the name '$n' contains white-space")
98 if $n =~ /\s/;
99 $names{$n} = 1;
100 $foundfilename++ if $n eq $simplename;
101 $foundfilenames{$n} = 1
102 if ((-f "$dirname/$n.pod.in" || -f "$dirname/$n.pod")
103 && $n ne $simplename);
104 }
105 err($id, "the following exist as other .pod or .pod.in files:",
106 sort keys %foundfilenames)
107 if %foundfilenames;
108 err($id, "$simplename (filename) missing from NAME section")
109 unless $foundfilename;
110 foreach my $n ( keys %names ) {
111 err($id, "$n is not public")
112 if $opt_p and !defined $public{$n};
113 }
114
115 # Find all functions in SYNOPSIS
116 return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
117 my $syn = $1;
118 foreach my $line ( split /\n+/, $syn ) {
119 next unless $line =~ /^\s/;
120 my $sym;
121 $line =~ s/STACK_OF\([^)]+\)/int/g;
122 $line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
123 $line =~ s/__declspec\([^)]+\)//;
124 if ( $line =~ /env (\S*)=/ ) {
125 # environment variable env NAME=...
126 $sym = $1;
127 } elsif ( $line =~ /typedef.*\(\*(\S+)\)\(.*/ ) {
128 # a callback function pointer: typedef ... (*NAME)(...
129 $sym = $1;
130 } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
131 # a callback function signature: typedef ... NAME(...
132 $sym = $1;
133 } elsif ( $line =~ /typedef.* (\S+);/ ) {
134 # a simple typedef: typedef ... NAME;
135 $sym = $1;
136 } elsif ( $line =~ /enum (\S*) \{/ ) {
137 # an enumeration: enum ... {
138 $sym = $1;
139 } elsif ( $line =~ /#(?:define|undef) ([A-Za-z0-9_]+)/ ) {
140 $sym = $1;
141 } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
142 $sym = $1;
143 }
144 else {
145 next;
146 }
147 err($id, "$sym missing from NAME section")
148 unless defined $names{$sym};
149 $names{$sym} = 2;
150
151 # Do some sanity checks on the prototype.
152 err($id, "prototype missing spaces around commas: $line")
153 if ( $line =~ /[a-z0-9],[^ ]/ );
154 }
155
156 foreach my $n ( keys %names ) {
157 next if $names{$n} == 2;
158 err($id, "$n missing from SYNOPSIS")
159 }
160 }
161
162 # Check if SECTION ($3) is located before BEFORE ($4)
163 sub check_section_location {
164 my $id = shift;
165 my $contents = shift;
166 my $section = shift;
167 my $before = shift;
168
169 return unless $contents =~ /=head1 $section/
170 and $contents =~ /=head1 $before/;
171 err($id, "$section should appear before $before section")
172 if $contents =~ /=head1 $before.*=head1 $section/ms;
173 }
174
175 # Check if a =head1 is duplicated, or a =headX is duplicated within a
176 # =head1. Treats =head2 =head3 as equivalent -- it doesn't reset the head3
177 # sets if it finds a =head2 -- but that is good enough for now. Also check
178 # for proper capitalization, trailing periods, etc.
179 sub check_head_style {
180 my $id = shift;
181 my $contents = shift;
182 my %head1;
183 my %subheads;
184
185 foreach my $line ( split /\n+/, $contents ) {
186 next unless $line =~ /^=head/;
187 if ( $line =~ /head1/ ) {
188 err($id, "duplicate section $line")
189 if defined $head1{$line};
190 $head1{$line} = 1;
191 %subheads = ();
192 } else {
193 err($id, "duplicate subsection $line")
194 if defined $subheads{$line};
195 $subheads{$line} = 1;
196 }
197 err($id, "period in =head")
198 if $line =~ /\.[^\w]/ or $line =~ /\.$/;
199 err($id, "not all uppercase in =head1")
200 if $line =~ /head1.*[a-z]/;
201 err($id, "all uppercase in subhead")
202 if $line =~ /head[234][ A-Z0-9]+$/;
203 }
204 }
205
206 sub check {
207 my $filename = shift;
208 my $dirname = basename(dirname($filename));
209
210 my $contents = '';
211 {
212 local $/ = undef;
213 open POD, $filename or die "Couldn't open $filename, $!";
214 $contents = <POD>;
215 close POD;
216 }
217
218 my $id = "${filename}:1:";
219 check_head_style($id, $contents);
220
221 # Check ordering of some sections in man3
222 if ( $filename =~ m|man3/| ) {
223 check_section_location($id, $contents, "RETURN VALUES", "EXAMPLES");
224 check_section_location($id, $contents, "SEE ALSO", "HISTORY");
225 check_section_location($id, $contents, "EXAMPLES", "SEE ALSO");
226 }
227
228 name_synopsis($id, $filename, $contents)
229 unless $contents =~ /=for comment generic/
230 or $filename =~ m@man[157]/@;
231
232 err($id, "doesn't start with =pod")
233 if $contents !~ /^=pod/;
234 err($id, "doesn't end with =cut")
235 if $contents !~ /=cut\n$/;
236 err($id, "more than one cut line.")
237 if $contents =~ /=cut.*=cut/ms;
238 err($id, "EXAMPLE not EXAMPLES section.")
239 if $contents =~ /=head1 EXAMPLE[^S]/;
240 err($id, "WARNING not WARNINGS section.")
241 if $contents =~ /=head1 WARNING[^S]/;
242 err($id, "missing copyright")
243 if $contents !~ /Copyright .* The OpenSSL Project Authors/;
244 err($id, "copyright not last")
245 if $contents =~ /head1 COPYRIGHT.*=head/ms;
246 err($id, "head2 in All uppercase")
247 if $contents =~ /head2\s+[A-Z ]+\n/;
248 err($id, "extra space after head")
249 if $contents =~ /=head\d\s\s+/;
250 err($id, "period in NAME section")
251 if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
252 err($id, "Duplicate $1 in L<>")
253 if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
254 err($id, "Bad =over $1")
255 if $contents =~ /=over([^ ][^24])/;
256 err($id, "Possible version style issue")
257 if $contents =~ /OpenSSL version [019]/;
258
259 if ( $contents !~ /=for comment multiple includes/ ) {
260 # Look for multiple consecutive openssl #include lines
261 # (non-consecutive lines are okay; see man3/MD5.pod).
262 if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
263 my $count = 0;
264 foreach my $line ( split /\n+/, $1 ) {
265 if ( $line =~ m@include <openssl/@ ) {
266 err($id, "has multiple includes")
267 if ++$count == 2;
268 } else {
269 $count = 0;
270 }
271 }
272 }
273 }
274
275 open my $OUT, '>', $temp
276 or die "Can't open $temp, $!";
277 podchecker($filename, $OUT);
278 close $OUT;
279 open $OUT, '<', $temp
280 or die "Can't read $temp, $!";
281 while ( <$OUT> ) {
282 next if /\(section\) in.*deprecated/;
283 print;
284 }
285 close $OUT;
286 unlink $temp || warn "Can't remove $temp, $!";
287
288 # Find what section this page is in; assume 3.
289 my $section = 3;
290 $section = $1 if $dirname =~ /man([1-9])/;
291
292 foreach ((@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}})) {
293 # Skip "return values" if not -s
294 err($id, "missing $_ head1 section")
295 if $contents !~ /^=head1\s+${_}\s*$/m;
296 }
297 }
298
299 my %dups;
300
301 sub parsenum {
302 my $file = shift;
303 my @apis;
304
305 open my $IN, '<', $file
306 or die "Can't open $file, $!, stopped";
307
308 while ( <$IN> ) {
309 next if /^#/;
310 next if /\bNOEXIST\b/;
311 my @fields = split();
312 die "Malformed line $_"
313 if scalar @fields != 2 && scalar @fields != 4;
314 push @apis, $fields[0];
315 }
316
317 close $IN;
318
319 print "# Found ", scalar(@apis), " in $file\n" unless $opt_p;
320 return sort @apis;
321 }
322
323 sub getdocced
324 {
325 my $dir = shift;
326 my %return;
327
328 foreach my $pod ( glob("$dir/*.pod"), glob("$dir/*.pod.in") ) {
329 my %podinfo = extract_pod_info($pod);
330 foreach my $n ( @{$podinfo{names}} ) {
331 $return{$n} = $pod;
332 print "# Duplicate $n in $pod and $dups{$n}\n"
333 if defined $dups{$n} && $dups{$n} ne $pod;
334 $dups{$n} = $pod;
335 }
336 }
337
338 return %return;
339 }
340
341 my %docced;
342
343 sub loadmissing($)
344 {
345 my $missingfile = shift;
346 my @missing;
347
348 open FH, $missingfile
349 || die "Can't open $missingfile";
350 while ( <FH> ) {
351 chomp;
352 next if /^#/;
353 push @missing, $_;
354 }
355 close FH;
356
357 return @missing;
358 }
359
360 sub checkmacros {
361 my $count = 0;
362 my %seen;
363 my @missing;
364
365 if ($opt_o) {
366 @missing = loadmissing('util/missingmacro111.txt');
367 } elsif ($opt_v) {
368 @missing = loadmissing('util/missingmacro.txt');
369 }
370
371 print "# Checking macros (approximate)\n"
372 if !$opt_s;
373 foreach my $f ( glob('include/openssl/*.h') ) {
374 # Skip some internals we don't want to document yet.
375 next if $f eq 'include/openssl/asn1.h';
376 next if $f eq 'include/openssl/asn1t.h';
377 next if $f eq 'include/openssl/err.h';
378 open(IN, $f) || die "Can't open $f, $!";
379 while ( <IN> ) {
380 next unless /^#\s*define\s*(\S+)\(/;
381 my $macro = $1;
382 next if $docced{$macro} || defined $seen{$macro};
383 next if $macro =~ /i2d_/
384 || $macro =~ /d2i_/
385 || $macro =~ /DEPRECATEDIN/
386 || $macro =~ /IMPLEMENT_/
387 || $macro =~ /DECLARE_/;
388
389 # Skip macros known to be missing
390 next if $opt_v && grep( /^$macro$/, @missing);
391
392 print "$f:$macro\n"
393 if $opt_d || $opt_e;
394 $count++;
395 $seen{$macro} = 1;
396 }
397 close(IN);
398 }
399 print "# Found $count macros missing\n"
400 if !$opt_s || $count > 0;
401 }
402
403 sub printem {
404 my $libname = shift;
405 my $numfile = shift;
406 my $missingfile = shift;
407 my $count = 0;
408 my %seen;
409
410 my @missing = loadmissing($missingfile) if ($opt_v);
411
412 foreach my $func ( parsenum($numfile) ) {
413 next if $docced{$func} || defined $seen{$func};
414
415 # Skip ASN1 utilities
416 next if $func =~ /^ASN1_/;
417
418 # Skip functions known to be missing
419 next if $opt_v && grep( /^$func$/, @missing);
420
421 print "$libname:$func\n"
422 if $opt_d || $opt_e;
423 $count++;
424 $seen{$func} = 1;
425 }
426 print "# Found $count missing from $numfile\n\n"
427 if !$opt_s || $count > 0;
428 }
429
430
431 # Collection of links in each POD file.
432 # filename => [ "foo(1)", "bar(3)", ... ]
433 my %link_collection = ();
434 # Collection of names in each POD file.
435 # "name(s)" => filename
436 my %name_collection = ();
437
438 sub collectnames {
439 my $filename = shift;
440 $filename =~ m|man(\d)/|;
441 my $section = $1;
442 my $simplename = basename(basename($filename, ".in"), ".pod");
443 my $id = "${filename}:1:";
444
445 my $contents = '';
446 {
447 local $/ = undef;
448 open POD, $filename or die "Couldn't open $filename, $!";
449 $contents = <POD>;
450 close POD;
451 }
452
453 $contents =~ /=head1 NAME([^=]*)=head1 /ms;
454 my $tmp = $1;
455 unless (defined $tmp) {
456 err($id, "weird name section");
457 return;
458 }
459 $tmp =~ tr/\n/ /;
460 $tmp =~ s/ -.*//g;
461
462 my @names =
463 map { s|/|-|g; $_ } # Treat slash as dash
464 map { s/^\s+//g; s/\s+$//g; $_ } # Trim prefix and suffix blanks
465 split(/,/, $tmp);
466 unless (grep { $simplename eq $_ } @names) {
467 err($id, "missing $simplename");
468 push @names, $simplename;
469 }
470 foreach my $name (@names) {
471 next if $name eq "";
472 if ($name =~ /\s/) {
473 err($id, "'$name' contains white space")
474 }
475 my $name_sec = "$name($section)";
476 if (! exists $name_collection{$name_sec}) {
477 $name_collection{$name_sec} = $filename;
478 } elsif ($filename eq $name_collection{$name_sec}) {
479 err($id, "$name_sec repeated in NAME section of",
480 $name_collection{$name_sec});
481 } else {
482 err($id, "$name_sec also in NAME section of",
483 $name_collection{$name_sec});
484 }
485 }
486
487 my @foreign_names =
488 map { map { s/\s+//g; $_ } split(/,/, $_) }
489 $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
490 foreach (@foreign_names) {
491 $name_collection{$_} = undef; # It still exists!
492 }
493
494 my @links = $contents =~ /L<
495 # if the link is of the form L<something|name(s)>,
496 # then remove 'something'. Note that 'something'
497 # may contain POD codes as well...
498 (?:(?:[^\|]|<[^>]*>)*\|)?
499 # we're only interested in references that have
500 # a one digit section number
501 ([^\/>\(]+\(\d\))
502 /gx;
503 $link_collection{$filename} = [ @links ];
504 }
505
506 sub checklinks {
507 foreach my $filename (sort keys %link_collection) {
508 foreach my $link (@{$link_collection{$filename}}) {
509 err("${filename}:1:", "reference to non-existing $link")
510 unless exists $name_collection{$link};
511 }
512 }
513 }
514
515 sub publicize {
516 foreach my $name ( parsenum('util/libcrypto.num') ) {
517 $public{$name} = 1;
518 }
519 foreach my $name ( parsenum('util/libssl.num') ) {
520 $public{$name} = 1;
521 }
522 foreach my $name ( parsenum('util/private.num') ) {
523 $public{$name} = 1;
524 }
525 }
526
527 my %skips = (
528 'aes128' => 1,
529 'aes192' => 1,
530 'aes256' => 1,
531 'aria128' => 1,
532 'aria192' => 1,
533 'aria256' => 1,
534 'camellia128' => 1,
535 'camellia192' => 1,
536 'camellia256' => 1,
537 'des' => 1,
538 'des3' => 1,
539 'idea' => 1,
540 '[cipher]' => 1,
541 '[digest]' => 1,
542 );
543
544 sub checkflags {
545 my $cmd = shift;
546 my $doc = shift;
547 my %cmdopts;
548 my %docopts;
549
550 # Get the list of options in the command.
551 open CFH, "./apps/openssl list --options $cmd|"
552 || die "Can list options for $cmd, $!";
553 while ( <CFH> ) {
554 chop;
555 s/ .$//;
556 $cmdopts{$_} = 1;
557 }
558 close CFH;
559
560 # Get the list of flags from the synopsis
561 open CFH, "<$doc"
562 || die "Can't open $doc, $!";
563 while ( <CFH> ) {
564 chop;
565 last if /DESCRIPTION/;
566 next unless /\[B<-([^ >]+)/;
567 $docopts{$1} = 1;
568 }
569 close CFH;
570
571 # See what's in the command not the manpage.
572 my @undocced = ();
573 foreach my $k ( keys %cmdopts ) {
574 push @undocced, $k unless $docopts{$k};
575 }
576 if ( scalar @undocced > 0 ) {
577 foreach ( @undocced ) {
578 err("doc/man1/$cmd.pod: Missing -$_");
579 }
580 }
581
582 # See what's in the command not the manpage.
583 my @unimpl = ();
584 foreach my $k ( keys %docopts ) {
585 push @unimpl, $k unless $cmdopts{$k};
586 }
587 if ( scalar @unimpl > 0 ) {
588 foreach ( @unimpl ) {
589 next if defined $skips{$_};
590 err("doc/man1/$cmd.pod: Not implemented -$_");
591 }
592 }
593 }
594
595 getopts('cdesolnphuv');
596
597 help() if $opt_h;
598
599 $opt_n = 1 if $opt_p;
600 $opt_u = 1 if $opt_d;
601 $opt_e = 1 if $opt_s;
602 $opt_v = 1 if $opt_o || $opt_e;
603
604 die "Cannot use both -u and -v"
605 if $opt_u && $opt_v;
606 die "Cannot use both -d and -e"
607 if $opt_d && $opt_e;
608
609 # We only need to check c, l, n, u and v.
610 # Options d, e, s, o and p imply one of the above.
611 die "Need one of -[cdesolnpuv] flags.\n"
612 unless $opt_c or $opt_l or $opt_n or $opt_u or $opt_v;
613
614 if ( $opt_c ) {
615 my @commands = ();
616
617 # Get list of commands.
618 open FH, "./apps/openssl list -1 -commands|"
619 || die "Can't list commands, $!";
620 while ( <FH> ) {
621 chop;
622 push @commands, $_;
623 }
624 close FH;
625
626 # See if each has a manpage.
627 foreach my $cmd ( @commands ) {
628 next if $cmd eq 'help' || $cmd eq 'exit';
629 my $doc = "doc/man1/$cmd.pod";
630 $doc = "doc/man1/openssl-$cmd.pod" if -f "doc/man1/openssl-$cmd.pod";
631 if ( ! -f "$doc" ) {
632 err("$doc does not exist");
633 } else {
634 checkflags($cmd, $doc);
635 }
636 }
637
638 # See what help is missing.
639 open FH, "./apps/openssl list --missing-help |"
640 || die "Can't list missing help, $!";
641 while ( <FH> ) {
642 chop;
643 my ($cmd, $flag) = split;
644 err("$cmd has no help for -$flag");
645 }
646 close FH;
647
648 exit $status;
649 }
650
651 if ( $opt_l ) {
652 foreach (@ARGV ? @ARGV : (glob('doc/*/*.pod'), glob('doc/*/*.pod.in'),
653 glob('doc/internal/*/*.pod'))) {
654 collectnames($_);
655 }
656 checklinks();
657 }
658
659 if ( $opt_n ) {
660 publicize() if $opt_p;
661 foreach (@ARGV ? @ARGV : (glob('doc/*/*.pod'), glob('doc/*/*.pod.in'))) {
662 check($_);
663 }
664 {
665 local $opt_p = undef;
666 foreach (@ARGV ? @ARGV : glob('doc/internal/*/*.pod')) {
667 check($_);
668 }
669 }
670 }
671
672 if ( $opt_u || $opt_v) {
673 my %temp = getdocced('doc/man3');
674 foreach ( keys %temp ) {
675 $docced{$_} = $temp{$_};
676 }
677 if ($opt_o) {
678 printem('crypto', 'util/libcrypto.num', 'util/missingcrypto111.txt');
679 printem('ssl', 'util/libssl.num', 'util/missingssl111.txt');
680 } else {
681 printem('crypto', 'util/libcrypto.num', 'util/missingcrypto.txt');
682 printem('ssl', 'util/libssl.num', 'util/missingssl.txt');
683 }
684 checkmacros();
685 }
686
687 exit $status;