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