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
  14 use Pod::Checker;
  15 use File::Find;
  16 use File::Basename;
  17 use File::Spec::Functions;
  18 use Getopt::Std;
  19 use lib catdir(dirname($0), "perl");
  20 use OpenSSL::Util::Pod;
  21
  22 # Set to 1 for debug output
  23 my $debug = 0;
  24
  25 # Options.
  26 our($opt_d);
  27 our($opt_e);
  28 our($opt_s);
  29 our($opt_o);
  30 our($opt_h);
  31 our($opt_l);
  32 our($opt_n);
  33 our($opt_p);
  34 our($opt_u);
  35 our($opt_v);
  36 our($opt_c);
  37
  38 # Print usage message and exit.
  39 sub help {
  40     print <<EOF;
  41 Find small errors (nits) in documentation.  Options:
  42     -c List undocumented commands and options
  43     -d Detailed list of undocumented (implies -u)
  44     -e Detailed list of new undocumented (implies -v)
  45     -h Print this help message
  46     -l Print bogus links
  47     -n Print nits in POD pages
  48     -o Causes -e/-v to count symbols added since 1.1.1 as new (implies -v)
  49     -u Count undocumented functions
  50     -v Count new undocumented functions
  51 EOF
  52     exit;
  53 }
  54
  55 getopts('cdehlnouv');
  56
  57 help() if $opt_h;
  58 $opt_u = 1 if $opt_d;
  59 $opt_v = 1 if $opt_o || $opt_e;
  60 die "Cannot use both -u and -v"
  61     if $opt_u && $opt_v;
  62 die "Cannot use both -d and -e"
  63     if $opt_d && $opt_e;
  64
  65 # We only need to check c, l, n, u and v.
  66 # Options d, e, o imply one of the above.
  67 die "Need one of -[cdehlnouv] flags.\n"
  68     unless $opt_c or $opt_l or $opt_n or $opt_u or $opt_v;
  69
  70
  71 my $temp = '/tmp/docnits.txt';
  72 my $OUT;
  73 my %public;
  74 my $status = 0;
  75
  76 my %mandatory_sections = (
  77     '*' => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
  78     1   => [ 'SYNOPSIS', 'OPTIONS' ],
  79     3   => [ 'SYNOPSIS', 'RETURN VALUES' ],
  80     5   => [ ],
  81     7   => [ ]
  82 );
  83
  84
  85 # Print error message, set $status.
  86 sub err {
  87     print join(" ", @_), "\n";
  88     $status = 1
  89 }
  90
  91 # Cross-check functions in the NAME and SYNOPSIS section.
  92 sub name_synopsis {
  93     my $id = shift;
  94     my $filename = shift;
  95     my $contents = shift;
  96
  97     # Get NAME section and all words in it.
  98     return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
  99     my $tmp = $1;
 100     $tmp =~ tr/\n/ /;
 101     err($id, "trailing comma before - in NAME")
 102         if $tmp =~ /, *-/;
 103     $tmp =~ s/ -.*//g;
 104     err($id, "POD markup among the names in NAME")
 105         if $tmp =~ /[<>]/;
 106     $tmp =~ s/  */ /g;
 107     err($id, "missing comma in NAME")
 108         if $tmp =~ /[^,] /;
 109
 110     my $dirname = dirname($filename);
 111     my $simplename = basename(basename($filename, ".in"), ".pod");
 112     my $foundfilename = 0;
 113     my %foundfilenames = ();
 114     my %names;
 115     foreach my $n ( split ',', $tmp ) {
 116         $n =~ s/^\s+//;
 117         $n =~ s/\s+$//;
 118         err($id, "the name '$n' contains white-space")
 119             if $n =~ /\s/;
 120         $names{$n} = 1;
 121         $foundfilename++ if $n eq $simplename;
 122         $foundfilenames{$n} = 1
 123             if -f "$dirname/$n.pod" && $n ne $simplename;
 124     }
 125     err($id, "the following exist as other .pod files:",
 126          sort keys %foundfilenames)
 127         if %foundfilenames;
 128     err($id, "$simplename (filename) missing from NAME section")
 129         unless $foundfilename;
 130     if ( $filename !~ /internal/ ) {
 131         foreach my $n ( keys %names ) {
 132             err($id, "$n is not public")
 133                 if !defined $public{$n};
 134         }
 135     }
 136
 137     # Find all functions in SYNOPSIS
 138     return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
 139     my $syn = $1;
 140     foreach my $line ( split /\n+/, $syn ) {
 141         next unless $line =~ /^\s/;
 142         my $sym;
 143         $line =~ s/STACK_OF\([^)]+\)/int/g;
 144         $line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
 145         $line =~ s/__declspec\([^)]+\)//;
 146         if ( $line =~ /env (\S*)=/ ) {
 147             # environment variable env NAME=...
 148             $sym = $1;
 149         } elsif ( $line =~ /typedef.*\(\*(\S+)\)\(.*/ ) {
 150             # a callback function pointer: typedef ... (*NAME)(...
 151             $sym = $1;
 152         } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
 153             # a callback function signature: typedef ... NAME(...
 154             $sym = $1;
 155         } elsif ( $line =~ /typedef.* (\S+);/ ) {
 156             # a simple typedef: typedef ... NAME;
 157             $sym = $1;
 158         } elsif ( $line =~ /enum (\S*) \{/ ) {
 159             # an enumeration: enum ... {
 160             $sym = $1;
 161         } elsif ( $line =~ /#(?:define|undef) ([A-Za-z0-9_]+)/ ) {
 162             $sym = $1;
 163         } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
 164             $sym = $1;
 165         }
 166         else {
 167             next;
 168         }
 169         err($id, "$sym missing from NAME section")
 170             unless defined $names{$sym};
 171         $names{$sym} = 2;
 172
 173         # Do some sanity checks on the prototype.
 174         err($id, "prototype missing spaces around commas: $line")
 175             if ( $line =~ /[a-z0-9],[^ ]/ );
 176     }
 177
 178     foreach my $n ( keys %names ) {
 179         next if $names{$n} == 2;
 180         err($id, "$n missing from SYNOPSIS")
 181     }
 182 }
 183
 184 # Check if SECTION ($3) is located before BEFORE ($4)
 185 sub check_section_location {
 186     my $id = shift;
 187     my $contents = shift;
 188     my $section = shift;
 189     my $before = shift;
 190
 191     return unless $contents =~ /=head1 $section/
 192         and $contents =~ /=head1 $before/;
 193     err($id, "$section should appear before $before section")
 194         if $contents =~ /=head1 $before.*=head1 $section/ms;
 195 }
 196
 197 # Check if a =head1 is duplicated, or a =headX is duplicated within a
 198 # =head1.  Treats =head2 =head3 as equivalent -- it doesn't reset the head3
 199 # sets if it finds a =head2 -- but that is good enough for now. Also check
 200 # for proper capitalization, trailing periods, etc.
 201 sub check_head_style {
 202     my $id = shift;
 203     my $contents = shift;
 204     my %head1;
 205     my %subheads;
 206
 207     foreach my $line ( split /\n+/, $contents ) {
 208         next unless $line =~ /^=head/;
 209         if ( $line =~ /head1/ ) {
 210             err($id, "duplicate section $line")
 211                 if defined $head1{$line};
 212             $head1{$line} = 1;
 213             %subheads = ();
 214         } else {
 215             err($id, "duplicate subsection $line")
 216                 if defined $subheads{$line};
 217             $subheads{$line} = 1;
 218         }
 219         err($id, "period in =head")
 220             if $line =~ /\.[^\w]/ or $line =~ /\.$/;
 221         err($id, "not all uppercase in =head1")
 222             if $line =~ /head1.*[a-z]/;
 223         err($id, "all uppercase in subhead")
 224             if $line =~ /head[234][ A-Z0-9]+$/;
 225     }
 226 }
 227
 228 # Because we have options and symbols with extra markup, we need
 229 # to take that into account, so we need a regexp that extracts
 230 # markup chunks, including recursive markup.
 231 # please read up on /(?R)/ in perlre(1)
 232 # (note: order is important, (?R) needs to come before .)
 233 # (note: non-greedy is important, or something like 'B<foo> and B<bar>'
 234 # will be captured as one item)
 235 my $markup_re =
 236     qr/(                        # Capture group
 237            [BIL]<               # The start of what we recurse on
 238            (?:(?-1)|.)*?        # recurse the whole regexp (refering to
 239                                 # the last opened capture group, i.e. the
 240                                 # start of this regexp), or pick next
 241                                 # character.  Do NOT be greedy!
 242            >                    # The end of what we recurse on
 243        )/x;                     # (the x allows this sort of split up regexp)
 244
 245 # Options must start with a dash, followed by a letter, possibly
 246 # followed by letters, digits, dashes and underscores, and the last
 247 # character must be a letter or a digit.
 248 # We do also accept the single -? or -n, where n is a digit
 249 my $option_re =
 250     qr/(?:
 251             \?                  # Single question mark
 252             |
 253             \d                  # Single digit
 254             |
 255             -                   # Single dash (--)
 256             |
 257             [[:alpha:]](?:[-_[:alnum:]]*?[[:alnum:]])?
 258        )/x;
 259
 260 # Helper function to check if a given $thing is properly marked up
 261 # option.  It returns one of these values:
 262 #     undef         if it's not an option
 263 #     ""            if it's a malformed option
 264 #     $unwrapped    the option with the outermost B<> wrapping removed.
 265 sub normalise_option {
 266     my $id = shift;
 267     my $filename = shift;
 268     my $thing = shift;
 269
 270     my $unwrapped = $thing;
 271     my $unmarked = $thing;
 272
 273     # $unwrapped is the option with the outer B<> markup removed
 274     $unwrapped =~ s/^B<//;
 275     $unwrapped =~ s/>$//;
 276     # $unmarked is the option with *all* markup removed
 277     $unmarked =~ s/[BIL]<|>//msg;
 278
 279
 280     # If we found an option, check it, collect it
 281     if ( $unwrapped =~ /^\s*-/ ) {
 282         return $unwrapped       # return option with outer B<> removed
 283             if $unmarked =~ /^-${option_re}$/;
 284         return "";              # Malformed option
 285     }
 286     return undef;               # Something else
 287 }
 288
 289 # Checks of command option (man1) formatting.  The man1 checks are
 290 # restricted to the SYNOPSIS and OPTIONS sections, the rest is too
 291 # free form, we simply cannot be too strict there.
 292
 293 sub option_check {
 294     my $id = shift;
 295     my $filename = shift;
 296     my $contents = shift;
 297
 298     my $synopsis = ($contents =~ /=head1\s+SYNOPSIS(.*?)=head1/s, $1);
 299
 300     # Some pages have more than one OPTIONS section, let's make sure
 301     # to get them all
 302     my $options = '';
 303     while ( $contents =~ /=head1\s+[A-Z ]*?OPTIONS$(.*?)(?==head1)/msg ) {
 304         $options .= $1;
 305     }
 306
 307     # Look for options with no or incorrect markup
 308     while ( $synopsis =~
 309             /(?<![-<[:alnum:]])-(?:$markup_re|.)*(?![->[:alnum:]])/msg ) {
 310         err($id, "Malformed option [1] in SYNOPSIS: $&");
 311     }
 312
 313     while ( $synopsis =~ /$markup_re/msg ) {
 314         my $found = $&;
 315         print STDERR "$id:DEBUG[option_check] SYNOPSIS: found $found\n"
 316             if $debug;
 317         my $option_uw = normalise_option($id, $filename, $found);
 318         err($id, "Malformed option [2] in SYNOPSIS: $found")
 319             if defined $option_uw && $option_uw eq '';
 320     }
 321
 322     # In OPTIONS, we look for =item paragraphs.
 323     # (?=^\s*$) detects an empty line.
 324     while ( $options =~ /=item\s+(.*?)(?=^\s*$)/msg ) {
 325         my $item = $&;
 326
 327         while ( $item =~ /(\[\s*)?($markup_re)/msg ) {
 328             my $found = $2;
 329             print STDERR "$id:DEBUG[option_check] OPTIONS: found $&\n"
 330                 if $debug;
 331             err($id, "Unexpected bracket in OPTIONS =item: $item")
 332                 if ($1 // '') ne '' && $found =~ /^B<\s*-/;
 333
 334             my $option_uw = normalise_option($id, $filename, $found);
 335             err($id, "Malformed option in OPTIONS: $found")
 336                 if defined $option_uw && $option_uw eq '';
 337         }
 338     }
 339 }
 340
 341 # Normal symbol form
 342 my $symbol_re = qr/[[:alpha:]_][_[:alnum:]]*?/;
 343
 344 # Checks of function name (man3) formatting.  The man3 checks are
 345 # easier than the man1 checks, we only check the names followed by (),
 346 # and only the names that have POD markup.
 347 sub functionname_check {
 348     my $id = shift;
 349     my $filename = shift;
 350     my $contents = shift;
 351
 352     while ( $contents =~ /($markup_re)\(\)/msg ) {
 353         print STDERR "$id:DEBUG[functionname_check] SYNOPSIS: found $&\n"
 354             if $debug;
 355
 356         my $symbol = $1;
 357         my $unmarked = $symbol;
 358         $unmarked =~ s/[BIL]<|>//msg;
 359
 360         err($id, "Malformed symbol: $symbol")
 361             unless $symbol =~ /^B<.*>$/ && $unmarked =~ /^${symbol_re}$/
 362     }
 363
 364     # We can't do the kind of collecting coolness that option_check()
 365     # does, because there are too many things that can't be found in
 366     # name repositories like the NAME sections, such as symbol names
 367     # with a variable part (typically marked up as B<foo_I<TYPE>_bar>
 368 }
 369
 370 # This is from http://man7.org/linux/man-pages/man7/man-pages.7.html
 371 my %preferred_words = (
 372     'bitmask'       => 'bit mask',
 373     'builtin'       => 'built-in',
 374    #'epoch'         => 'Epoch', # handled specially, below
 375     'file name'     => 'filename',
 376     'file system'   => 'filesystem',
 377     'host name'     => 'hostname',
 378     'i-node'        => 'inode',
 379     'lower case'    => 'lowercase',
 380     'lower-case'    => 'lowercase',
 381     'non-zero'      => 'nonzero',
 382     'path name'     => 'pathname',
 383     'pseudo-terminal' => 'pseudoterminal',
 384     'reserved port' => 'privileged port',
 385     'system port'   => 'privileged port',
 386     'realtime'      => 'real-time',
 387     'real time'     => 'real-time',
 388     'runtime'       => 'run time',
 389     'saved group ID'=> 'saved set-group-ID',
 390     'saved set-GID' => 'saved set-group-ID',
 391     'saved user ID' => 'saved set-user-ID',
 392     'saved set-UID' => 'saved set-user-ID',
 393     'set-GID'       => 'set-group-ID',
 394     'setgid'        => 'set-group-ID',
 395     'set-UID'       => 'set-user-ID',
 396     'setuid'        => 'set-user-ID',
 397     'super user'    => 'superuser',
 398     'super-user'    => 'superuser',
 399     'super block'   => 'superblock',
 400     'super-block'   => 'superblock',
 401     'time stamp'    => 'timestamp',
 402     'time zone'     => 'timezone',
 403     'upper case'    => 'uppercase',
 404     'upper-case'    => 'uppercase',
 405     'useable'       => 'usable',
 406     'userspace'     => 'user space',
 407     'user name'     => 'username',
 408     'zeroes'        => 'zeros'
 409 );
 410
 411 # Search manpage for words that have a different preferred use.
 412 sub wording {
 413     my $id = shift;
 414     my $contents = shift;
 415
 416     foreach my $k ( keys %preferred_words ) {
 417         # Sigh, trademark
 418         next if $k eq 'file system'
 419             and $contents =~ /Microsoft Encrypted File System/;
 420         err($id, "found '$k' should use '$preferred_words{$k}'")
 421             if $contents =~ /\b\Q$k\E\b/i;
 422     }
 423     err($id, "found 'epoch' should use 'Epoch'")
 424         if $contents =~ /\bepoch\b/;
 425 }
 426
 427 # Perform all sorts of nit/error checks on a manpage
 428 sub check {
 429     my $filename = shift;
 430     my $dirname = basename(dirname($filename));
 431
 432     my $contents = '';
 433     {
 434         local $/ = undef;
 435         open POD, $filename or die "Couldn't open $filename, $!";
 436         $contents = <POD>;
 437         close POD;
 438     }
 439
 440     my $id = "${filename}:1:";
 441     check_head_style($id, $contents);
 442
 443     # Check ordering of some sections in man3
 444     if ( $filename =~ m|man3/| ) {
 445         check_section_location($id, $contents, "RETURN VALUES", "EXAMPLES");
 446         check_section_location($id, $contents, "SEE ALSO", "HISTORY");
 447         check_section_location($id, $contents, "EXAMPLES", "SEE ALSO");
 448     }
 449
 450     # Make sure every link has a section.
 451     while ( $contents =~ /$markup_re/msg ) {
 452         my $target = $1;
 453         next unless $target =~ /^L</;           # Skip if not L<...>, or
 454         next if $target =~ /::/;                #   links to a Perl module, or
 455         next if $target =~ m@L</@;              #   links within the page, or
 456         next if $target =~ /^L<https?:/;        #   is a URL link, or
 457         next if $target =~ m@\([1357]\)>$@;     #   it has a section, or
 458         next if $target =~ m@\([1357]\)/.*>$@;  #   it has a section/anchor
 459         err($id, "Section missing in $target")
 460     }
 461     # Check for proper in-man-3 API links.
 462     while ( $contents =~ /L<([^>]*)\(3\)(?:\/.*)?>/g ) {
 463         my $target = $1;
 464         err($id, "Bad L<$target>")
 465             unless $target =~ /^[_[:alpha:]][_[:alnum:]]*$/
 466     }
 467
 468     unless ( $contents =~ /=for openssl generic/ ) {
 469         if ( $filename =~ m|man3/| ) {
 470             name_synopsis($id, $filename, $contents);
 471             functionname_check($id, $filename, $contents);
 472         } elsif ( $filename =~ m|man1/| ) {
 473             option_check($id, $filename, $contents)
 474         }
 475     }
 476
 477     wording($id, $contents);
 478
 479     err($id, "doesn't start with =pod")
 480         if $contents !~ /^=pod/;
 481     err($id, "doesn't end with =cut")
 482         if $contents !~ /=cut\n$/;
 483     err($id, "more than one cut line.")
 484         if $contents =~ /=cut.*=cut/ms;
 485     err($id, "EXAMPLE not EXAMPLES section.")
 486         if $contents =~ /=head1 EXAMPLE[^S]/;
 487     err($id, "WARNING not WARNINGS section.")
 488         if $contents =~ /=head1 WARNING[^S]/;
 489     err($id, "missing copyright")
 490         if $contents !~ /Copyright .* The OpenSSL Project Authors/;
 491     err($id, "copyright not last")
 492         if $contents =~ /head1 COPYRIGHT.*=head/ms;
 493     err($id, "head2 in All uppercase")
 494         if $contents =~ /head2\s+[A-Z ]+\n/;
 495     err($id, "extra space after head")
 496         if $contents =~ /=head\d\s\s+/;
 497     err($id, "period in NAME section")
 498         if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
 499     err($id, "Duplicate $1 in L<>")
 500         if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
 501     err($id, "Bad =over $1")
 502         if $contents =~ /=over([^ ][^24])/;
 503     err($id, "Possible version style issue")
 504         if $contents =~ /OpenSSL version [019]/;
 505
 506     if ( $contents !~ /=for openssl multiple includes/ ) {
 507         # Look for multiple consecutive openssl #include lines
 508         # (non-consecutive lines are okay; see man3/MD5.pod).
 509         if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
 510             my $count = 0;
 511             foreach my $line ( split /\n+/, $1 ) {
 512                 if ( $line =~ m@include <openssl/@ ) {
 513                     err($id, "has multiple includes")
 514                         if ++$count == 2;
 515                 } else {
 516                     $count = 0;
 517                 }
 518             }
 519         }
 520     }
 521
 522     open my $OUT, '>', $temp
 523         or die "Can't open $temp, $!";
 524     podchecker($filename, $OUT);
 525     close $OUT;
 526     open $OUT, '<', $temp
 527         or die "Can't read $temp, $!";
 528     while ( <$OUT> ) {
 529         next if /\(section\) in.*deprecated/;
 530         print;
 531     }
 532     close $OUT;
 533     unlink $temp || warn "Can't remove $temp, $!";
 534
 535     # Find what section this page is in; assume 3.
 536     my $section = 3;
 537     $section = $1 if $dirname =~ /man([1-9])/;
 538
 539     foreach ( (@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}}) ) {
 540         err($id, "missing $_ head1 section")
 541             if $contents !~ /^=head1\s+${_}\s*$/m;
 542     }
 543 }
 544
 545 # Parse libcrypto.num, etc., and return sorted list of what's there.
 546 sub parsenum {
 547     my $file = shift;
 548     my @apis;
 549
 550     open my $IN, '<', $file
 551         or die "Can't open $file, $!, stopped";
 552
 553     while ( <$IN> ) {
 554         next if /^#/;
 555         next if /\bNOEXIST\b/;
 556         my @fields = split();
 557         die "Malformed line $_"
 558             if scalar @fields != 2 && scalar @fields != 4;
 559         push @apis, $fields[0];
 560     }
 561
 562     close $IN;
 563
 564     return sort @apis;
 565 }
 566
 567 # Parse all the manpages, getting return map of what they document
 568 # (by looking at their NAME sections).
 569 sub getdocced
 570 {
 571     my $dir = shift;
 572     my %return;
 573     my %dups;
 574
 575     foreach my $pod ( glob("$dir/*.pod") ) {
 576         my %podinfo = extract_pod_info($pod);
 577         foreach my $n ( @{$podinfo{names}} ) {
 578             $return{$n} = $pod;
 579             err("# Duplicate $n in $pod and $dups{$n}")
 580                 if defined $dups{$n} && $dups{$n} ne $pod;
 581             $dups{$n} = $pod;
 582         }
 583     }
 584
 585     return %return;
 586 }
 587
 588 # Map of documented functions; function => manpage
 589 my %docced;
 590 # Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
 591 my %link_map = ();
 592 # Map of names in each POD file; "name(s)" => filename
 593 my %name_map = ();
 594
 595 # Load file of symbol names that we know aren't documented.
 596 sub loadmissing($)
 597 {
 598     my $missingfile = shift;
 599     my @missing;
 600
 601     open FH, $missingfile
 602         || die "Can't open $missingfile";
 603     while ( <FH> ) {
 604         chomp;
 605         next if /^#/;
 606         push @missing, $_;
 607     }
 608     close FH;
 609
 610     return @missing;
 611 }
 612
 613 # Check for undocumented macros; ignore those in the "missing" file
 614 # and do simple check for #define in our header files.
 615 sub checkmacros {
 616     my $count = 0;
 617     my %seen;
 618     my @missing;
 619
 620     if ( $opt_o ) {
 621         @missing = loadmissing('util/missingmacro111.txt');
 622     } elsif ( $opt_v ) {
 623         @missing = loadmissing('util/missingmacro.txt');
 624     }
 625
 626     foreach my $f ( glob('include/openssl/*.h') ) {
 627         # Skip some internals we don't want to document yet.
 628         next if $f eq 'include/openssl/asn1.h';
 629         next if $f eq 'include/openssl/asn1t.h';
 630         next if $f eq 'include/openssl/err.h';
 631         open(IN, $f) || die "Can't open $f, $!";
 632         while ( <IN> ) {
 633             next unless /^#\s*define\s*(\S+)\(/;
 634             my $macro = $1;
 635             next if $docced{$macro} || defined $seen{$macro};
 636             next if $macro =~ /i2d_/
 637                 || $macro =~ /d2i_/
 638                 || $macro =~ /DEPRECATEDIN/
 639                 || $macro =~ /IMPLEMENT_/
 640                 || $macro =~ /DECLARE_/;
 641
 642             # Skip macros known to be missing
 643             next if $opt_v && grep( /^$macro$/, @missing);
 644
 645             err("$f:", "macro $macro undocumented")
 646                 if $opt_d || $opt_e;
 647             $count++;
 648             $seen{$macro} = 1;
 649         }
 650         close(IN);
 651     }
 652     err("# $count macros undocumented (count is approximate)")
 653         if $count > 0;
 654 }
 655
 656 # Find out what is undocumented (filtering out the known missing ones)
 657 # and display them.
 658 sub printem {
 659     my $libname = shift;
 660     my $numfile = shift;
 661     my $missingfile = shift;
 662     my $count = 0;
 663     my %seen;
 664
 665     my @missing = loadmissing($missingfile) if ( $opt_v );
 666
 667     foreach my $func ( parsenum($numfile) ) {
 668         next if $docced{$func} || defined $seen{$func};
 669
 670         # Skip ASN1 utilities
 671         next if $func =~ /^ASN1_/;
 672
 673         # Skip functions known to be missing
 674         next if $opt_v && grep( /^$func$/, @missing);
 675
 676         err("$libname:", "function $func undocumented")
 677             if $opt_d || $opt_e;
 678         $count++;
 679         $seen{$func} = 1;
 680     }
 681     err("# $count in $numfile are not documented")
 682         if $count > 0;
 683 }
 684
 685 # Collect all the names in a manpage.
 686 sub collectnames {
 687     my $filename = shift;
 688     $filename =~ m|man(\d)/|;
 689     my $section = $1;
 690     my $simplename = basename($filename, ".pod");
 691     my $id = "${filename}:1:";
 692
 693     my $contents = '';
 694     {
 695         local $/ = undef;
 696         open POD, $filename or die "Couldn't open $filename, $!";
 697         $contents = <POD>;
 698         close POD;
 699     }
 700
 701     $contents =~ /=head1 NAME([^=]*)=head1 /ms;
 702     my $tmp = $1;
 703     unless ( defined $tmp ) {
 704         err($id, "weird name section");
 705         return;
 706     }
 707     $tmp =~ tr/\n/ /;
 708     $tmp =~ s/ -.*//g;
 709
 710     my @names =
 711         map { s|/|-|g; $_ }              # Treat slash as dash
 712         map { s/^\s+//g; s/\s+$//g; $_ } # Trim prefix and suffix blanks
 713         split(/,/, $tmp);
 714     unless ( grep { $simplename eq $_ } @names ) {
 715         err($id, "missing $simplename");
 716         push @names, $simplename;
 717     }
 718     foreach my $name (@names) {
 719         next if $name eq "";
 720         if ( $name =~ /\s/ ) {
 721             err($id, "'$name' contains white space")
 722         }
 723         my $name_sec = "$name($section)";
 724         if ( !exists $name_map{$name_sec} ) {
 725             $name_map{$name_sec} = $filename;
 726         } elsif ( $filename eq $name_map{$name_sec} ) {
 727             err($id, "$name_sec repeated in NAME section of",
 728                  $name_map{$name_sec});
 729         } else {
 730             err($id, "$name_sec also in NAME section of",
 731                  $name_map{$name_sec});
 732         }
 733     }
 734
 735     my @foreign_names =
 736         map { map { s/\s+//g; $_ } split(/,/, $_) }
 737         $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
 738     foreach ( @foreign_names ) {
 739         $name_map{$_} = undef; # It still exists!
 740     }
 741
 742     my @links = $contents =~ /L<
 743                               # if the link is of the form L<something|name(s)>,
 744                               # then remove 'something'.  Note that 'something'
 745                               # may contain POD codes as well...
 746                               (?:(?:[^\|]|<[^>]*>)*\|)?
 747                               # we're only interested in references that have
 748                               # a one digit section number
 749                               ([^\/>\(]+\(\d\))
 750                              /gx;
 751     $link_map{$filename} = [ @links ];
 752 }
 753
 754 # Look for L<> ("link") references that point to files that do not exist.
 755 sub checklinks {
 756     foreach my $filename (sort keys %link_map) {
 757         foreach my $link (@{$link_map{$filename}}) {
 758             err("${filename}:1:", "reference to non-existing $link")
 759                 unless exists $name_map{$link};
 760         }
 761     }
 762 }
 763
 764 # Load the public symbol/macro names
 765 sub publicize {
 766     foreach my $name ( parsenum('util/libcrypto.num') ) {
 767         $public{$name} = 1;
 768     }
 769     foreach my $name ( parsenum('util/libssl.num') ) {
 770         $public{$name} = 1;
 771     }
 772     foreach my $name ( parsenum('util/other.syms') ) {
 773         $public{$name} = 1;
 774     }
 775 }
 776
 777 # Cipher/digests to skip if they show up as "not implemented"
 778 # because they are, via the "-*" construct.
 779 my %skips = (
 780     'aes128' => 1,
 781     'aes192' => 1,
 782     'aes256' => 1,
 783     'aria128' => 1,
 784     'aria192' => 1,
 785     'aria256' => 1,
 786     'camellia128' => 1,
 787     'camellia192' => 1,
 788     'camellia256' => 1,
 789     'des' => 1,
 790     'des3' => 1,
 791     'idea' => 1,
 792     'cipher' => 1,
 793     'digest' => 1,
 794 );
 795
 796 # Check the flags of a command and see if everything is in the manpage
 797 sub checkflags {
 798     my $cmd = shift;
 799     my $doc = shift;
 800     my %cmdopts;
 801     my %docopts;
 802     my %localskips;
 803
 804     # Get the list of options in the command.
 805     open CFH, "./apps/openssl list --options $cmd|"
 806         || die "Can list options for $cmd, $!";
 807     while ( <CFH> ) {
 808         chop;
 809         s/ .$//;
 810         $cmdopts{$_} = 1;
 811     }
 812     close CFH;
 813
 814     # Get the list of flags from the synopsis
 815     open CFH, "<$doc"
 816         || die "Can't open $doc, $!";
 817     while ( <CFH> ) {
 818         chop;
 819         last if /DESCRIPTION/;
 820         if ( /=for openssl ifdef (.*)/ ) {
 821             foreach my $f ( split / /, $1 ) {
 822                 $localskips{$f} = 1;
 823             }
 824             next;
 825         }
 826         next unless /\[B<-([^ >]+)/;
 827         my $opt = $1;
 828         $opt = $1 if $opt =~ /I<(.*)/;
 829         $docopts{$1} = 1;
 830     }
 831     close CFH;
 832
 833     # See what's in the command not the manpage.
 834     my @undocced = sort grep { !defined $docopts{$_} } keys %cmdopts;
 835     foreach ( @undocced ) {
 836         next if /-/; # Skip the -- end-of-flags marker
 837         err("$doc: undocumented option -$_");
 838     }
 839
 840     # See what's in the command not the manpage.
 841     my @unimpl = sort grep { !defined $cmdopts{$_} } keys %docopts;
 842     foreach ( @unimpl ) {
 843         next if defined $skips{$_} || defined $localskips{$_};
 844         err("$cmd documented but not implemented -$_");
 845     }
 846 }
 847
 848 ##
 849 ##  MAIN()
 850 ##  Do the work requested by the various getopt flags.
 851 ##  The flags are parsed in alphabetical order, just because we have
 852 ##  to have *some way* of listing them.
 853 ##
 854
 855 if ( $opt_c ) {
 856     my @commands = ();
 857
 858     # Get list of commands.
 859     open FH, "./apps/openssl list -1 -commands|"
 860         || die "Can't list commands, $!";
 861     while ( <FH> ) {
 862         chop;
 863         push @commands, $_;
 864     }
 865     close FH;
 866
 867     # See if each has a manpage.
 868     foreach my $cmd ( @commands ) {
 869         next if $cmd eq 'help' || $cmd eq 'exit';
 870         my $doc = "doc/man1/$cmd.pod";
 871         $doc = "doc/man1/openssl-$cmd.pod" if -f "doc/man1/openssl-$cmd.pod";
 872         if ( ! -f "$doc" ) {
 873             err("$doc does not exist");
 874         } else {
 875             checkflags($cmd, $doc);
 876         }
 877     }
 878
 879     # See what help is missing.
 880     open FH, "./apps/openssl list --missing-help |"
 881         || die "Can't list missing help, $!";
 882     while ( <FH> ) {
 883         chop;
 884         my ($cmd, $flag) = split;
 885         err("$cmd has no help for -$flag");
 886     }
 887     close FH;
 888
 889     exit $status;
 890 }
 891
 892 if ( $opt_l ) {
 893     foreach ( glob('doc/*/*.pod doc/internal/*/*.pod') ) {
 894         collectnames($_);
 895     }
 896     checklinks();
 897 }
 898
 899 if ( $opt_n ) {
 900     publicize();
 901     foreach ( @ARGV ? @ARGV : glob('doc/*/*.pod doc/internal/*/*.pod') ) {
 902         check($_);
 903     }
 904
 905     # If not given args, check that all man1 commands are named properly.
 906     if ( scalar @ARGV == 0 ) {
 907         foreach (glob('doc/man1/*.pod')) {
 908             next if /CA.pl/ || /openssl\.pod/ || /tsget\.pod/;
 909             err("$_ doesn't start with openssl-") unless /openssl-/;
 910         }
 911     }
 912 }
 913
 914 if ( $opt_u || $opt_v) {
 915     my %temp = getdocced('doc/man3');
 916     foreach ( keys %temp ) {
 917         $docced{$_} = $temp{$_};
 918     }
 919     if ( $opt_o ) {
 920         printem('crypto', 'util/libcrypto.num', 'util/missingcrypto111.txt');
 921         printem('ssl', 'util/libssl.num', 'util/missingssl111.txt');
 922     } else {
 923         printem('crypto', 'util/libcrypto.num', 'util/missingcrypto.txt');
 924         printem('ssl', 'util/libssl.num', 'util/missingssl.txt');
 925     }
 926     checkmacros();
 927 }
 928
 929 exit $status;