util/find-doc-nits

   1 #! /usr/bin/env perl
   2 # Copyright 2002-2018 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($opt_h);
  24 our($opt_l);
  25 our($opt_n);
  26 our($opt_p);
  27 our($opt_u);
  28 our($opt_c);
  29
  30 sub help()
  31 {
  32     print <<EOF;
  33 Find 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 Count undocumented functions
  39     -h Print this help message
  40     -c List undocumented commands and options
  41 EOF
  42     exit;
  43 }
  44
  45 my $temp = '/tmp/docnits.txt';
  46 my $OUT;
  47 my %public;
  48
  49 my %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.
  57 sub 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
  72     my $dirname = dirname($filename);
  73     my $simplename = basename($filename);
  74     $simplename =~ s/.pod$//;
  75     my $foundfilename = 0;
  76     my %foundfilenames = ();
  77     my %names;
  78     foreach my $n ( split ',', $tmp ) {
  79         $n =~ s/^\s+//;
  80         $n =~ s/\s+$//;
  81         print "$id the name '$n' contains white-space\n"
  82             if $n =~ /\s/;
  83         $names{$n} = 1;
  84         $foundfilename++ if $n eq $simplename;
  85         $foundfilenames{$n} = 1
  86             if -f "$dirname/$n.pod" && $n ne $simplename;
  87     }
  88     print "$id the following exist as other .pod files:\n",
  89         join(" ", sort keys %foundfilenames), "\n"
  90         if %foundfilenames;
  91     print "$id $simplename (filename) missing from NAME section\n"
  92         unless $foundfilename;
  93     foreach my $n ( keys %names ) {
  94         print "$id $n is not public\n"
  95             if $opt_p and !defined $public{$n};
  96     }
  97
  98     # Find all functions in SYNOPSIS
  99     return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
 100     my $syn = $1;
 101     foreach my $line ( split /\n+/, $syn ) {
 102         next unless $line =~ /^\s/;
 103         my $sym;
 104         $line =~ s/STACK_OF\([^)]+\)/int/g;
 105         $line =~ s/__declspec\([^)]+\)//;
 106         if ( $line =~ /env (\S*)=/ ) {
 107             # environment variable env NAME=...
 108             $sym = $1;
 109         } elsif ( $line =~ /typedef.*\(\*(\S+)\)\(.*/ ) {
 110             # a callback function pointer: typedef ... (*NAME)(...
 111             $sym = $1;
 112         } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
 113             # a callback function signature: typedef ... NAME(...
 114             $sym = $1;
 115         } elsif ( $line =~ /typedef.* (\S+);/ ) {
 116             # a simple typedef: typedef ... NAME;
 117             $sym = $1;
 118         } elsif ( $line =~ /enum (\S*) \{/ ) {
 119             # an enumeration: enum ... {
 120             $sym = $1;
 121         } elsif ( $line =~ /#(?:define|undef) ([A-Za-z0-9_]+)/ ) {
 122             $sym = $1;
 123         } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
 124             $sym = $1;
 125         }
 126         else {
 127             next;
 128         }
 129         print "$id $sym missing from NAME section\n"
 130             unless defined $names{$sym};
 131         $names{$sym} = 2;
 132
 133         # Do some sanity checks on the prototype.
 134         print "$id prototype missing spaces around commas: $line\n"
 135             if ( $line =~ /[a-z0-9],[^ ]/ );
 136     }
 137
 138     foreach my $n ( keys %names ) {
 139         next if $names{$n} == 2;
 140         print "$id $n missing from SYNOPSIS\n";
 141     }
 142 }
 143
 144 sub check()
 145 {
 146     my $filename = shift;
 147     my $dirname = basename(dirname($filename));
 148
 149     my $contents = '';
 150     {
 151         local $/ = undef;
 152         open POD, $filename or die "Couldn't open $filename, $!";
 153         $contents = <POD>;
 154         close POD;
 155     }
 156
 157     my $id = "${filename}:1:";
 158
 159     &name_synopsis($id, $filename, $contents)
 160         unless $contents =~ /=for comment generic/
 161             or $filename =~ m@man[157]/@;
 162
 163     print "$id doesn't start with =pod\n"
 164         if $contents !~ /^=pod/;
 165     print "$id doesn't end with =cut\n"
 166         if $contents !~ /=cut\n$/;
 167     print "$id more than one cut line.\n"
 168         if $contents =~ /=cut.*=cut/ms;
 169     print "$id missing copyright\n"
 170         if $contents !~ /Copyright .* The OpenSSL Project Authors/;
 171     print "$id copyright not last\n"
 172         if $contents =~ /head1 COPYRIGHT.*=head/ms;
 173     print "$id head2 in All uppercase\n"
 174         if $contents =~ /head2\s+[A-Z ]+\n/;
 175     print "$id extra space after head\n"
 176         if $contents =~ /=head\d\s\s+/;
 177     print "$id period in NAME section\n"
 178         if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
 179     print "$id POD markup in NAME section\n"
 180         if $contents =~ /=head1 NAME.*[<>].*=head1 SYNOPSIS/ms;
 181     print "$id Duplicate $1 in L<>\n"
 182         if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
 183     print "$id Bad =over $1\n"
 184         if $contents =~ /=over([^ ][^24])/;
 185     print "$id Possible version style issue\n"
 186         if $contents =~ /OpenSSL version [019]/;
 187
 188     if ( $contents !~ /=for comment multiple includes/ ) {
 189         # Look for multiple consecutive openssl #include lines
 190         # (non-consecutive lines are okay; see man3/MD5.pod).
 191         if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
 192             my $count = 0;
 193             foreach my $line ( split /\n+/, $1 ) {
 194                 if ( $line =~ m@include <openssl/@ ) {
 195                     print "$id has multiple includes\n" if ++$count == 2;
 196                 } else {
 197                     $count = 0;
 198                 }
 199             }
 200         }
 201     }
 202
 203     open my $OUT, '>', $temp
 204         or die "Can't open $temp, $!";
 205     podchecker($filename, $OUT);
 206     close $OUT;
 207     open $OUT, '<', $temp
 208         or die "Can't read $temp, $!";
 209     while ( <$OUT> ) {
 210         next if /\(section\) in.*deprecated/;
 211         print;
 212     }
 213     close $OUT;
 214     unlink $temp || warn "Can't remove $temp, $!";
 215
 216     # Find what section this page is in; assume 3.
 217     my $section = 3;
 218     $section = $1 if $dirname =~ /man([1-9])/;
 219
 220     foreach ((@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}})) {
 221         # Skip "return values" if not -s
 222         print "$id: missing $_ head1 section\n"
 223             if $contents !~ /^=head1\s+${_}\s*$/m;
 224     }
 225 }
 226
 227 my %dups;
 228
 229 sub parsenum()
 230 {
 231     my $file = shift;
 232     my @apis;
 233
 234     open my $IN, '<', $file
 235         or die "Can't open $file, $!, stopped";
 236
 237     while ( <$IN> ) {
 238         next if /^#/;
 239         next if /\bNOEXIST\b/;
 240         next if /\bEXPORT_VAR_AS_FUNC\b/;
 241         my @fields = split();
 242         die "Malformed line $_"
 243             if scalar @fields != 2 && scalar @fields != 4;
 244         push @apis, $fields[0];
 245     }
 246
 247     close $IN;
 248
 249     print "# Found ", scalar(@apis), " in $file\n" unless $opt_p;
 250     return sort @apis;
 251 }
 252
 253 sub getdocced
 254 {
 255     my $dir = shift;
 256     my %return;
 257
 258     foreach my $pod ( glob("$dir/*.pod") ) {
 259         my %podinfo = extract_pod_info($pod);
 260         foreach my $n ( @{$podinfo{names}} ) {
 261             $return{$n} = $pod;
 262             print "# Duplicate $n in $pod and $dups{$n}\n"
 263                 if defined $dups{$n} && $dups{$n} ne $pod;
 264             $dups{$n} = $pod;
 265         }
 266     }
 267
 268     return %return;
 269 }
 270
 271 my %docced;
 272
 273 sub checkmacros()
 274 {
 275     my $count = 0;
 276     my %seen;
 277
 278     print "# Checking macros (approximate)\n";
 279     foreach my $f ( glob('include/openssl/*.h') ) {
 280         # Skip some internals we don't want to document yet.
 281         next if $f eq 'include/openssl/asn1.h';
 282         next if $f eq 'include/openssl/asn1t.h';
 283         next if $f eq 'include/openssl/err.h';
 284         open(IN, $f) || die "Can't open $f, $!";
 285         while ( <IN> ) {
 286             next unless /^#\s*define\s*(\S+)\(/;
 287             my $macro = $1;
 288             next if $docced{$macro} || defined $seen{$macro};
 289             next if $macro =~ /i2d_/
 290                 || $macro =~ /d2i_/
 291                 || $macro =~ /DEPRECATEDIN/
 292                 || $macro =~ /IMPLEMENT_/
 293                 || $macro =~ /DECLARE_/;
 294             print "$f:$macro\n" if $opt_d;
 295             $count++;
 296             $seen{$macro} = 1;
 297         }
 298         close(IN);
 299     }
 300     print "# Found $count macros missing (not all should be documented)\n"
 301 }
 302
 303 sub printem()
 304 {
 305     my $libname = shift;
 306     my $numfile = shift;
 307     my $count = 0;
 308     my %seen;
 309
 310     foreach my $func ( &parsenum($numfile) ) {
 311         next if $docced{$func} || defined $seen{$func};
 312
 313         # Skip ASN1 utilities
 314         next if $func =~ /^ASN1_/;
 315
 316         print "$libname:$func\n" if $opt_d;
 317         $count++;
 318         $seen{$func} = 1;
 319     }
 320     print "# Found $count missing from $numfile\n\n";
 321 }
 322
 323
 324 # Collection of links in each POD file.
 325 # filename => [ "foo(1)", "bar(3)", ... ]
 326 my %link_collection = ();
 327 # Collection of names in each POD file.
 328 # "name(s)" => filename
 329 my %name_collection = ();
 330
 331 sub collectnames {
 332     my $filename = shift;
 333     $filename =~ m|man(\d)/|;
 334     my $section = $1;
 335     my $simplename = basename($filename, ".pod");
 336     my $id = "${filename}:1:";
 337
 338     my $contents = '';
 339     {
 340         local $/ = undef;
 341         open POD, $filename or die "Couldn't open $filename, $!";
 342         $contents = <POD>;
 343         close POD;
 344     }
 345
 346     $contents =~ /=head1 NAME([^=]*)=head1 /ms;
 347     my $tmp = $1;
 348     unless (defined $tmp) {
 349         print "$id weird name section\n";
 350         return;
 351     }
 352     $tmp =~ tr/\n/ /;
 353     $tmp =~ s/-.*//g;
 354
 355     my @names = map { s/^\s+//g; s/\s+$//g; $_ } split(/,/, $tmp);
 356     unless (grep { $simplename eq $_ } @names) {
 357         print "$id missing $simplename\n";
 358         push @names, $simplename;
 359     }
 360     foreach my $name (@names) {
 361         next if $name eq "";
 362         if ($name =~ /\s/) {
 363             print "$id '$name' contains white space\n";
 364         }
 365         my $name_sec = "$name($section)";
 366         if (! exists $name_collection{$name_sec}) {
 367             $name_collection{$name_sec} = $filename;
 368         } else { #elsif ($filename ne $name_collection{$name_sec}) {
 369             print "$id $name_sec also in $name_collection{$name_sec}\n";
 370         }
 371     }
 372
 373     my @foreign_names =
 374         map { map { s/\s+//g; $_ } split(/,/, $_) }
 375         $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
 376     foreach (@foreign_names) {
 377         $name_collection{$_} = undef; # It still exists!
 378     }
 379
 380     my @links = $contents =~ /L<
 381                               # if the link is of the form L<something|name(s)>,
 382                               # then remove 'something'.  Note that 'something'
 383                               # may contain POD codes as well...
 384                               (?:(?:[^\|]|<[^>]*>)*\|)?
 385                               # we're only interested in references that have
 386                               # a one digit section number
 387                               ([^\/>\(]+\(\d\))
 388                              /gx;
 389     $link_collection{$filename} = [ @links ];
 390 }
 391
 392 sub checklinks {
 393     foreach my $filename (sort keys %link_collection) {
 394         foreach my $link (@{$link_collection{$filename}}) {
 395             print "${filename}:1: reference to non-existing $link\n"
 396                 unless exists $name_collection{$link};
 397         }
 398     }
 399 }
 400
 401 sub publicize() {
 402     foreach my $name ( &parsenum('util/libcrypto.num') ) {
 403         $public{$name} = 1;
 404     }
 405     foreach my $name ( &parsenum('util/libssl.num') ) {
 406         $public{$name} = 1;
 407     }
 408     foreach my $name ( &parsenum('util/private.num') ) {
 409         $public{$name} = 1;
 410     }
 411 }
 412
 413 my %skips = (
 414     'aes128' => 1,
 415     'aes192' => 1,
 416     'aes256' => 1,
 417     'aria128' => 1,
 418     'aria192' => 1,
 419     'aria256' => 1,
 420     'camellia128' => 1,
 421     'camellia192' => 1,
 422     'camellia256' => 1,
 423     'des' => 1,
 424     'des3' => 1,
 425     'idea' => 1,
 426     '[cipher]' => 1,
 427     '[digest]' => 1,
 428 );
 429
 430 sub checkflags() {
 431     my $cmd = shift;
 432     my %cmdopts;
 433     my %docopts;
 434     my $ok = 1;
 435
 436     # Get the list of options in the command.
 437     open CFH, "./apps/openssl list --options $cmd|"
 438         || die "Can list options for $cmd, $!";
 439     while ( <CFH> ) {
 440         chop;
 441         s/ .$//;
 442         $cmdopts{$_} = 1;
 443     }
 444     close CFH;
 445
 446     # Get the list of flags from the synopsis
 447     open CFH, "<doc/man1/$cmd.pod"
 448         || die "Can't open $cmd.pod, $!";
 449     while ( <CFH> ) {
 450         chop;
 451         last if /DESCRIPTION/;
 452         next unless /\[B<-([^ >]+)/;
 453         $docopts{$1} = 1;
 454     }
 455     close CFH;
 456
 457     # See what's in the command not the manpage.
 458     my @undocced = ();
 459     foreach my $k ( keys %cmdopts ) {
 460         push @undocced, $k unless $docopts{$k};
 461     }
 462     if ( scalar @undocced > 0 ) {
 463         $ok = 0;
 464         foreach ( @undocced ) {
 465             print "doc/man1/$cmd.pod: Missing -$_\n";
 466         }
 467     }
 468
 469     # See what's in the command not the manpage.
 470     my @unimpl = ();
 471     foreach my $k ( keys %docopts ) {
 472         push @unimpl, $k unless $cmdopts{$k};
 473     }
 474     if ( scalar @unimpl > 0 ) {
 475         $ok = 0;
 476         foreach ( @unimpl ) {
 477             next if defined $skips{$_};
 478             print "doc/man1/$cmd.pod: Not implemented -$_\n";
 479         }
 480     }
 481
 482     return $ok;
 483 }
 484
 485 getopts('cdlnphu');
 486
 487 &help() if $opt_h;
 488 $opt_n = 1 if $opt_p;
 489 $opt_u = 1 if $opt_d;
 490
 491 die "Need one of -[cdlnpu] flags.\n"
 492     unless $opt_c or $opt_l or $opt_n or $opt_u;
 493
 494 if ( $opt_c ) {
 495     my $ok = 1;
 496     my @commands = ();
 497
 498     # Get list of commands.
 499     open FH, "./apps/openssl list -1 -commands|"
 500         || die "Can't list commands, $!";
 501     while ( <FH> ) {
 502         chop;
 503         push @commands, $_;
 504     }
 505     close FH;
 506
 507     # See if each has a manpage.
 508     foreach ( @commands ) {
 509         next if $_ eq 'help' || $_ eq 'exit';
 510         if ( ! -f "doc/man1/$_.pod" ) {
 511             print "doc/man1/$_.pod does not exist\n";
 512             $ok = 0;
 513         } else {
 514             $ok = 0 if not &checkflags($_);
 515         }
 516     }
 517
 518     # See what help is missing.
 519     open FH, "./apps/openssl list --missing-help |"
 520         || die "Can't list missing help, $!";
 521     while ( <FH> ) {
 522         chop;
 523         my ($cmd, $flag) = split;
 524         print "$cmd has no help for -$flag\n";
 525         $ok = 0;
 526     }
 527     close FH;
 528
 529     exit 1 if not $ok;
 530 }
 531
 532 if ( $opt_l ) {
 533     foreach (@ARGV ? @ARGV : (glob('doc/*/*.pod'),
 534                               glob('doc/internal/*/*.pod'))) {
 535         collectnames($_);
 536     }
 537     checklinks();
 538 }
 539
 540 if ( $opt_n ) {
 541     &publicize() if $opt_p;
 542     foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
 543         &check($_);
 544     }
 545     {
 546         local $opt_p = undef;
 547         foreach (@ARGV ? @ARGV : glob('doc/internal/*/*.pod')) {
 548             &check($_);
 549         }
 550     }
 551 }
 552
 553 if ( $opt_u ) {
 554     my %temp = getdocced('doc/man3');
 555     foreach ( keys %temp ) {
 556         $docced{$_} = $temp{$_};
 557     }
 558     &printem('crypto', 'util/libcrypto.num');
 559     &printem('ssl', 'util/libssl.num');
 560     &checkmacros();
 561 }
 562
 563 exit;