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($opt_h);
  24 our($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/SPARSE_ARRAY_OF\([^)]+\)/int/g;
 106         $line =~ s/__declspec\([^)]+\)//;
 107         if ( $line =~ /env (\S*)=/ ) {
 108             # environment variable env NAME=...
 109             $sym = $1;
 110         } elsif ( $line =~ /typedef.*\(\*(\S+)\)\(.*/ ) {
 111             # a callback function pointer: typedef ... (*NAME)(...
 112             $sym = $1;
 113         } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
 114             # a callback function signature: typedef ... NAME(...
 115             $sym = $1;
 116         } elsif ( $line =~ /typedef.* (\S+);/ ) {
 117             # a simple typedef: typedef ... NAME;
 118             $sym = $1;
 119         } elsif ( $line =~ /enum (\S*) \{/ ) {
 120             # an enumeration: enum ... {
 121             $sym = $1;
 122         } elsif ( $line =~ /#(?:define|undef) ([A-Za-z0-9_]+)/ ) {
 123             $sym = $1;
 124         } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
 125             $sym = $1;
 126         }
 127         else {
 128             next;
 129         }
 130         print "$id $sym missing from NAME section\n"
 131             unless defined $names{$sym};
 132         $names{$sym} = 2;
 133
 134         # Do some sanity checks on the prototype.
 135         print "$id prototype missing spaces around commas: $line\n"
 136             if ( $line =~ /[a-z0-9],[^ ]/ );
 137     }
 138
 139     foreach my $n ( keys %names ) {
 140         next if $names{$n} == 2;
 141         print "$id $n missing from SYNOPSIS\n";
 142     }
 143 }
 144
 145 # Check if SECTION is located before BEFORE
 146 sub check_section_location()
 147 {
 148     my $filename = shift;
 149     my $contents = shift;
 150     my $section = shift;
 151     my $before = shift;
 152
 153     return unless $contents =~ /=head1 $section/
 154         and $contents =~ /=head1 $before/;
 155     print "$filename: $section should be placed before $before section\n"
 156         if $contents =~ /=head1 $before.*=head1 $section/ms;
 157 }
 158
 159 sub check()
 160 {
 161     my $filename = shift;
 162     my $dirname = basename(dirname($filename));
 163
 164     my $contents = '';
 165     {
 166         local $/ = undef;
 167         open POD, $filename or die "Couldn't open $filename, $!";
 168         $contents = <POD>;
 169         close POD;
 170     }
 171
 172     # Check if EXAMPLES is located after RETURN VALUES section.
 173     &check_section_location($filename, $contents, "RETURN VALUES", "EXAMPLES") if $filename =~ m|man3/|;
 174     # Check if HISTORY is located after SEE ALSO
 175     &check_section_location($filename, $contents, "SEE ALSO", "HISTORY") if $filename =~ m|man3/|;
 176     # Check if SEE ALSO is located after EXAMPLES
 177     &check_section_location($filename, $contents, "EXAMPLES", "SEE ALSO") if $filename =~ m|man3/|;
 178
 179     my $id = "${filename}:1:";
 180
 181     &name_synopsis($id, $filename, $contents)
 182         unless $contents =~ /=for comment generic/
 183             or $filename =~ m@man[157]/@;
 184
 185     print "$id doesn't start with =pod\n"
 186         if $contents !~ /^=pod/;
 187     print "$id doesn't end with =cut\n"
 188         if $contents !~ /=cut\n$/;
 189     print "$id more than one cut line.\n"
 190         if $contents =~ /=cut.*=cut/ms;
 191     print "$id missing copyright\n"
 192         if $contents !~ /Copyright .* The OpenSSL Project Authors/;
 193     print "$id copyright not last\n"
 194         if $contents =~ /head1 COPYRIGHT.*=head/ms;
 195     print "$id head2 in All uppercase\n"
 196         if $contents =~ /head2\s+[A-Z ]+\n/;
 197     print "$id extra space after head\n"
 198         if $contents =~ /=head\d\s\s+/;
 199     print "$id period in NAME section\n"
 200         if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
 201     print "$id POD markup in NAME section\n"
 202         if $contents =~ /=head1 NAME.*[<>].*=head1 SYNOPSIS/ms;
 203     print "$id Duplicate $1 in L<>\n"
 204         if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
 205     print "$id Bad =over $1\n"
 206         if $contents =~ /=over([^ ][^24])/;
 207     print "$id Possible version style issue\n"
 208         if $contents =~ /OpenSSL version [019]/;
 209
 210     if ( $contents !~ /=for comment multiple includes/ ) {
 211         # Look for multiple consecutive openssl #include lines
 212         # (non-consecutive lines are okay; see man3/MD5.pod).
 213         if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
 214             my $count = 0;
 215             foreach my $line ( split /\n+/, $1 ) {
 216                 if ( $line =~ m@include <openssl/@ ) {
 217                     print "$id has multiple includes\n" if ++$count == 2;
 218                 } else {
 219                     $count = 0;
 220                 }
 221             }
 222         }
 223     }
 224
 225     open my $OUT, '>', $temp
 226         or die "Can't open $temp, $!";
 227     podchecker($filename, $OUT);
 228     close $OUT;
 229     open $OUT, '<', $temp
 230         or die "Can't read $temp, $!";
 231     while ( <$OUT> ) {
 232         next if /\(section\) in.*deprecated/;
 233         print;
 234     }
 235     close $OUT;
 236     unlink $temp || warn "Can't remove $temp, $!";
 237
 238     # Find what section this page is in; assume 3.
 239     my $section = 3;
 240     $section = $1 if $dirname =~ /man([1-9])/;
 241
 242     foreach ((@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}})) {
 243         # Skip "return values" if not -s
 244         print "$id: missing $_ head1 section\n"
 245             if $contents !~ /^=head1\s+${_}\s*$/m;
 246     }
 247 }
 248
 249 my %dups;
 250
 251 sub parsenum()
 252 {
 253     my $file = shift;
 254     my @apis;
 255
 256     open my $IN, '<', $file
 257         or die "Can't open $file, $!, stopped";
 258
 259     while ( <$IN> ) {
 260         next if /^#/;
 261         next if /\bNOEXIST\b/;
 262         next if /\bEXPORT_VAR_AS_FUNC\b/;
 263         my @fields = split();
 264         die "Malformed line $_"
 265             if scalar @fields != 2 && scalar @fields != 4;
 266         push @apis, $fields[0];
 267     }
 268
 269     close $IN;
 270
 271     print "# Found ", scalar(@apis), " in $file\n" unless $opt_p;
 272     return sort @apis;
 273 }
 274
 275 sub getdocced
 276 {
 277     my $dir = shift;
 278     my %return;
 279
 280     foreach my $pod ( glob("$dir/*.pod") ) {
 281         my %podinfo = extract_pod_info($pod);
 282         foreach my $n ( @{$podinfo{names}} ) {
 283             $return{$n} = $pod;
 284             print "# Duplicate $n in $pod and $dups{$n}\n"
 285                 if defined $dups{$n} && $dups{$n} ne $pod;
 286             $dups{$n} = $pod;
 287         }
 288     }
 289
 290     return %return;
 291 }
 292
 293 my %docced;
 294
 295 sub checkmacros()
 296 {
 297     my $count = 0;
 298     my %seen;
 299
 300     print "# Checking macros (approximate)\n";
 301     foreach my $f ( glob('include/openssl/*.h') ) {
 302         # Skip some internals we don't want to document yet.
 303         next if $f eq 'include/openssl/asn1.h';
 304         next if $f eq 'include/openssl/asn1t.h';
 305         next if $f eq 'include/openssl/err.h';
 306         open(IN, $f) || die "Can't open $f, $!";
 307         while ( <IN> ) {
 308             next unless /^#\s*define\s*(\S+)\(/;
 309             my $macro = $1;
 310             next if $docced{$macro} || defined $seen{$macro};
 311             next if $macro =~ /i2d_/
 312                 || $macro =~ /d2i_/
 313                 || $macro =~ /DEPRECATEDIN/
 314                 || $macro =~ /IMPLEMENT_/
 315                 || $macro =~ /DECLARE_/;
 316             print "$f:$macro\n" if $opt_d;
 317             $count++;
 318             $seen{$macro} = 1;
 319         }
 320         close(IN);
 321     }
 322     print "# Found $count macros missing (not all should be documented)\n"
 323 }
 324
 325 sub printem()
 326 {
 327     my $libname = shift;
 328     my $numfile = shift;
 329     my $count = 0;
 330     my %seen;
 331
 332     foreach my $func ( &parsenum($numfile) ) {
 333         next if $docced{$func} || defined $seen{$func};
 334
 335         # Skip ASN1 utilities
 336         next if $func =~ /^ASN1_/;
 337
 338         print "$libname:$func\n" if $opt_d;
 339         $count++;
 340         $seen{$func} = 1;
 341     }
 342     print "# Found $count missing from $numfile\n\n";
 343 }
 344
 345
 346 # Collection of links in each POD file.
 347 # filename => [ "foo(1)", "bar(3)", ... ]
 348 my %link_collection = ();
 349 # Collection of names in each POD file.
 350 # "name(s)" => filename
 351 my %name_collection = ();
 352
 353 sub collectnames {
 354     my $filename = shift;
 355     $filename =~ m|man(\d)/|;
 356     my $section = $1;
 357     my $simplename = basename($filename, ".pod");
 358     my $id = "${filename}:1:";
 359
 360     my $contents = '';
 361     {
 362         local $/ = undef;
 363         open POD, $filename or die "Couldn't open $filename, $!";
 364         $contents = <POD>;
 365         close POD;
 366     }
 367
 368     $contents =~ /=head1 NAME([^=]*)=head1 /ms;
 369     my $tmp = $1;
 370     unless (defined $tmp) {
 371         print "$id weird name section\n";
 372         return;
 373     }
 374     $tmp =~ tr/\n/ /;
 375     $tmp =~ s/-.*//g;
 376
 377     my @names = map { s/^\s+//g; s/\s+$//g; $_ } split(/,/, $tmp);
 378     unless (grep { $simplename eq $_ } @names) {
 379         print "$id missing $simplename\n";
 380         push @names, $simplename;
 381     }
 382     foreach my $name (@names) {
 383         next if $name eq "";
 384         if ($name =~ /\s/) {
 385             print "$id '$name' contains white space\n";
 386         }
 387         my $name_sec = "$name($section)";
 388         if (! exists $name_collection{$name_sec}) {
 389             $name_collection{$name_sec} = $filename;
 390         } else { #elsif ($filename ne $name_collection{$name_sec}) {
 391             print "$id $name_sec also in $name_collection{$name_sec}\n";
 392         }
 393     }
 394
 395     my @foreign_names =
 396         map { map { s/\s+//g; $_ } split(/,/, $_) }
 397         $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
 398     foreach (@foreign_names) {
 399         $name_collection{$_} = undef; # It still exists!
 400     }
 401
 402     my @links = $contents =~ /L<
 403                               # if the link is of the form L<something|name(s)>,
 404                               # then remove 'something'.  Note that 'something'
 405                               # may contain POD codes as well...
 406                               (?:(?:[^\|]|<[^>]*>)*\|)?
 407                               # we're only interested in references that have
 408                               # a one digit section number
 409                               ([^\/>\(]+\(\d\))
 410                              /gx;
 411     $link_collection{$filename} = [ @links ];
 412 }
 413
 414 sub checklinks {
 415     foreach my $filename (sort keys %link_collection) {
 416         foreach my $link (@{$link_collection{$filename}}) {
 417             print "${filename}:1: reference to non-existing $link\n"
 418                 unless exists $name_collection{$link};
 419         }
 420     }
 421 }
 422
 423 sub publicize() {
 424     foreach my $name ( &parsenum('util/libcrypto.num') ) {
 425         $public{$name} = 1;
 426     }
 427     foreach my $name ( &parsenum('util/libssl.num') ) {
 428         $public{$name} = 1;
 429     }
 430     foreach my $name ( &parsenum('util/private.num') ) {
 431         $public{$name} = 1;
 432     }
 433 }
 434
 435 my %skips = (
 436     'aes128' => 1,
 437     'aes192' => 1,
 438     'aes256' => 1,
 439     'aria128' => 1,
 440     'aria192' => 1,
 441     'aria256' => 1,
 442     'camellia128' => 1,
 443     'camellia192' => 1,
 444     'camellia256' => 1,
 445     'des' => 1,
 446     'des3' => 1,
 447     'idea' => 1,
 448     '[cipher]' => 1,
 449     '[digest]' => 1,
 450 );
 451
 452 sub checkflags() {
 453     my $cmd = shift;
 454     my %cmdopts;
 455     my %docopts;
 456     my $ok = 1;
 457
 458     # Get the list of options in the command.
 459     open CFH, "./apps/openssl list --options $cmd|"
 460         || die "Can list options for $cmd, $!";
 461     while ( <CFH> ) {
 462         chop;
 463         s/ .$//;
 464         $cmdopts{$_} = 1;
 465     }
 466     close CFH;
 467
 468     # Get the list of flags from the synopsis
 469     open CFH, "<doc/man1/$cmd.pod"
 470         || die "Can't open $cmd.pod, $!";
 471     while ( <CFH> ) {
 472         chop;
 473         last if /DESCRIPTION/;
 474         next unless /\[B<-([^ >]+)/;
 475         $docopts{$1} = 1;
 476     }
 477     close CFH;
 478
 479     # See what's in the command not the manpage.
 480     my @undocced = ();
 481     foreach my $k ( keys %cmdopts ) {
 482         push @undocced, $k unless $docopts{$k};
 483     }
 484     if ( scalar @undocced > 0 ) {
 485         $ok = 0;
 486         foreach ( @undocced ) {
 487             print "doc/man1/$cmd.pod: Missing -$_\n";
 488         }
 489     }
 490
 491     # See what's in the command not the manpage.
 492     my @unimpl = ();
 493     foreach my $k ( keys %docopts ) {
 494         push @unimpl, $k unless $cmdopts{$k};
 495     }
 496     if ( scalar @unimpl > 0 ) {
 497         $ok = 0;
 498         foreach ( @unimpl ) {
 499             next if defined $skips{$_};
 500             print "doc/man1/$cmd.pod: Not implemented -$_\n";
 501         }
 502     }
 503
 504     return $ok;
 505 }
 506
 507 getopts('cdlnphu');
 508
 509 &help() if $opt_h;
 510 $opt_n = 1 if $opt_p;
 511 $opt_u = 1 if $opt_d;
 512
 513 die "Need one of -[cdlnpu] flags.\n"
 514     unless $opt_c or $opt_l or $opt_n or $opt_u;
 515
 516 if ( $opt_c ) {
 517     my $ok = 1;
 518     my @commands = ();
 519
 520     # Get list of commands.
 521     open FH, "./apps/openssl list -1 -commands|"
 522         || die "Can't list commands, $!";
 523     while ( <FH> ) {
 524         chop;
 525         push @commands, $_;
 526     }
 527     close FH;
 528
 529     # See if each has a manpage.
 530     foreach ( @commands ) {
 531         next if $_ eq 'help' || $_ eq 'exit';
 532         if ( ! -f "doc/man1/$_.pod" ) {
 533             print "doc/man1/$_.pod does not exist\n";
 534             $ok = 0;
 535         } else {
 536             $ok = 0 if not &checkflags($_);
 537         }
 538     }
 539
 540     # See what help is missing.
 541     open FH, "./apps/openssl list --missing-help |"
 542         || die "Can't list missing help, $!";
 543     while ( <FH> ) {
 544         chop;
 545         my ($cmd, $flag) = split;
 546         print "$cmd has no help for -$flag\n";
 547         $ok = 0;
 548     }
 549     close FH;
 550
 551     exit 1 if not $ok;
 552 }
 553
 554 if ( $opt_l ) {
 555     foreach (@ARGV ? @ARGV : (glob('doc/*/*.pod'),
 556                               glob('doc/internal/*/*.pod'))) {
 557         collectnames($_);
 558     }
 559     checklinks();
 560 }
 561
 562 if ( $opt_n ) {
 563     &publicize() if $opt_p;
 564     foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
 565         &check($_);
 566     }
 567     {
 568         local $opt_p = undef;
 569         foreach (@ARGV ? @ARGV : glob('doc/internal/*/*.pod')) {
 570             &check($_);
 571         }
 572     }
 573 }
 574
 575 if ( $opt_u ) {
 576     my %temp = getdocced('doc/man3');
 577     foreach ( keys %temp ) {
 578         $docced{$_} = $temp{$_};
 579     }
 580     &printem('crypto', 'util/libcrypto.num');
 581     &printem('ssl', 'util/libssl.num');
 582     &checkmacros();
 583 }
 584
 585 exit;