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         my $is_prototype = 1;
 144         $line =~ s/STACK_OF\([^)]+\)/int/g;
 145         $line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
 146         $line =~ s/__declspec\([^)]+\)//;
 147         if ( $line =~ /env (\S*)=/ ) {
 148             # environment variable env NAME=...
 149             $sym = $1;
 150         } elsif ( $line =~ /typedef.*\(\*(\S+)\)\(.*/ ) {
 151             # a callback function pointer: typedef ... (*NAME)(...
 152             $sym = $1;
 153         } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
 154             # a callback function signature: typedef ... NAME(...
 155             $sym = $1;
 156         } elsif ( $line =~ /typedef.* (\S+);/ ) {
 157             # a simple typedef: typedef ... NAME;
 158             $is_prototype = 0;
 159             $sym = $1;
 160         } elsif ( $line =~ /enum (\S*) \{/ ) {
 161             # an enumeration: enum ... {
 162             $sym = $1;
 163         } elsif ( $line =~ /#(?:define|undef) ([A-Za-z0-9_]+)/ ) {
 164             $is_prototype = 0;
 165             $sym = $1;
 166         } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
 167             $sym = $1;
 168         }
 169         else {
 170             next;
 171         }
 172         err($id, "$sym missing from NAME section")
 173             unless defined $names{$sym};
 174         $names{$sym} = 2;
 175
 176         # Do some sanity checks on the prototype.
 177         err($id, "prototype missing spaces around commas: $line")
 178             if $is_prototype && $line =~ /[a-z0-9],[^ ]/;
 179     }
 180
 181     foreach my $n ( keys %names ) {
 182         next if $names{$n} == 2;
 183         err($id, "$n missing from SYNOPSIS")
 184     }
 185 }
 186
 187 # Check if SECTION ($3) is located before BEFORE ($4)
 188 sub check_section_location {
 189     my $id = shift;
 190     my $contents = shift;
 191     my $section = shift;
 192     my $before = shift;
 193
 194     return unless $contents =~ /=head1 $section/
 195         and $contents =~ /=head1 $before/;
 196     err($id, "$section should appear before $before section")
 197         if $contents =~ /=head1 $before.*=head1 $section/ms;
 198 }
 199
 200 # Check if a =head1 is duplicated, or a =headX is duplicated within a
 201 # =head1.  Treats =head2 =head3 as equivalent -- it doesn't reset the head3
 202 # sets if it finds a =head2 -- but that is good enough for now. Also check
 203 # for proper capitalization, trailing periods, etc.
 204 sub check_head_style {
 205     my $id = shift;
 206     my $contents = shift;
 207     my %head1;
 208     my %subheads;
 209
 210     foreach my $line ( split /\n+/, $contents ) {
 211         next unless $line =~ /^=head/;
 212         if ( $line =~ /head1/ ) {
 213             err($id, "duplicate section $line")
 214                 if defined $head1{$line};
 215             $head1{$line} = 1;
 216             %subheads = ();
 217         } else {
 218             err($id, "duplicate subsection $line")
 219                 if defined $subheads{$line};
 220             $subheads{$line} = 1;
 221         }
 222         err($id, "period in =head")
 223             if $line =~ /\.[^\w]/ or $line =~ /\.$/;
 224         err($id, "not all uppercase in =head1")
 225             if $line =~ /head1.*[a-z]/;
 226         err($id, "all uppercase in subhead")
 227             if $line =~ /head[234][ A-Z0-9]+$/;
 228     }
 229 }
 230
 231 # Because we have options and symbols with extra markup, we need
 232 # to take that into account, so we need a regexp that extracts
 233 # markup chunks, including recursive markup.
 234 # please read up on /(?R)/ in perlre(1)
 235 # (note: order is important, (?R) needs to come before .)
 236 # (note: non-greedy is important, or something like 'B<foo> and B<bar>'
 237 # will be captured as one item)
 238 my $markup_re =
 239     qr/(                        # Capture group
 240            [BIL]<               # The start of what we recurse on
 241            (?:(?-1)|.)*?        # recurse the whole regexp (referring to
 242                                 # the last opened capture group, i.e. the
 243                                 # start of this regexp), or pick next
 244                                 # character.  Do NOT be greedy!
 245            >                    # The end of what we recurse on
 246        )/x;                     # (the x allows this sort of split up regexp)
 247
 248 # Options must start with a dash, followed by a letter, possibly
 249 # followed by letters, digits, dashes and underscores, and the last
 250 # character must be a letter or a digit.
 251 # We do also accept the single -? or -n, where n is a digit
 252 my $option_re =
 253     qr/(?:
 254             \?                  # Single question mark
 255             |
 256             \d                  # Single digit
 257             |
 258             -                   # Single dash (--)
 259             |
 260             [[:alpha:]](?:[-_[:alnum:]]*?[[:alnum:]])?
 261        )/x;
 262
 263 # Helper function to check if a given $thing is properly marked up
 264 # option.  It returns one of these values:
 265 #     undef         if it's not an option
 266 #     ""            if it's a malformed option
 267 #     $unwrapped    the option with the outermost B<> wrapping removed.
 268 sub normalise_option {
 269     my $id = shift;
 270     my $filename = shift;
 271     my $thing = shift;
 272
 273     my $unwrapped = $thing;
 274     my $unmarked = $thing;
 275
 276     # $unwrapped is the option with the outer B<> markup removed
 277     $unwrapped =~ s/^B<//;
 278     $unwrapped =~ s/>$//;
 279     # $unmarked is the option with *all* markup removed
 280     $unmarked =~ s/[BIL]<|>//msg;
 281
 282
 283     # If we found an option, check it, collect it
 284     if ( $unwrapped =~ /^\s*-/ ) {
 285         return $unwrapped       # return option with outer B<> removed
 286             if $unmarked =~ /^-${option_re}$/;
 287         return "";              # Malformed option
 288     }
 289     return undef;               # Something else
 290 }
 291
 292 # Checks of command option (man1) formatting.  The man1 checks are
 293 # restricted to the SYNOPSIS and OPTIONS sections, the rest is too
 294 # free form, we simply cannot be too strict there.
 295
 296 sub option_check {
 297     my $id = shift;
 298     my $filename = shift;
 299     my $contents = shift;
 300
 301     my $synopsis = ($contents =~ /=head1\s+SYNOPSIS(.*?)=head1/s, $1);
 302
 303     # Some pages have more than one OPTIONS section, let's make sure
 304     # to get them all
 305     my $options = '';
 306     while ( $contents =~ /=head1\s+[A-Z ]*?OPTIONS$(.*?)(?==head1)/msg ) {
 307         $options .= $1;
 308     }
 309
 310     # Look for options with no or incorrect markup
 311     while ( $synopsis =~
 312             /(?<![-<[:alnum:]])-(?:$markup_re|.)*(?![->[:alnum:]])/msg ) {
 313         err($id, "Malformed option [1] in SYNOPSIS: $&");
 314     }
 315
 316     while ( $synopsis =~ /$markup_re/msg ) {
 317         my $found = $&;
 318         print STDERR "$id:DEBUG[option_check] SYNOPSIS: found $found\n"
 319             if $debug;
 320         my $option_uw = normalise_option($id, $filename, $found);
 321         err($id, "Malformed option [2] in SYNOPSIS: $found")
 322             if defined $option_uw && $option_uw eq '';
 323     }
 324
 325     # In OPTIONS, we look for =item paragraphs.
 326     # (?=^\s*$) detects an empty line.
 327     while ( $options =~ /=item\s+(.*?)(?=^\s*$)/msg ) {
 328         my $item = $&;
 329
 330         while ( $item =~ /(\[\s*)?($markup_re)/msg ) {
 331             my $found = $2;
 332             print STDERR "$id:DEBUG[option_check] OPTIONS: found $&\n"
 333                 if $debug;
 334             err($id, "Unexpected bracket in OPTIONS =item: $item")
 335                 if ($1 // '') ne '' && $found =~ /^B<\s*-/;
 336
 337             my $option_uw = normalise_option($id, $filename, $found);
 338             err($id, "Malformed option in OPTIONS: $found")
 339                 if defined $option_uw && $option_uw eq '';
 340         }
 341     }
 342 }
 343
 344 # Normal symbol form
 345 my $symbol_re = qr/[[:alpha:]_][_[:alnum:]]*?/;
 346
 347 # Checks of function name (man3) formatting.  The man3 checks are
 348 # easier than the man1 checks, we only check the names followed by (),
 349 # and only the names that have POD markup.
 350 sub functionname_check {
 351     my $id = shift;
 352     my $filename = shift;
 353     my $contents = shift;
 354
 355     while ( $contents =~ /($markup_re)\(\)/msg ) {
 356         print STDERR "$id:DEBUG[functionname_check] SYNOPSIS: found $&\n"
 357             if $debug;
 358
 359         my $symbol = $1;
 360         my $unmarked = $symbol;
 361         $unmarked =~ s/[BIL]<|>//msg;
 362
 363         err($id, "Malformed symbol: $symbol")
 364             unless $symbol =~ /^B<.*>$/ && $unmarked =~ /^${symbol_re}$/
 365     }
 366
 367     # We can't do the kind of collecting coolness that option_check()
 368     # does, because there are too many things that can't be found in
 369     # name repositories like the NAME sections, such as symbol names
 370     # with a variable part (typically marked up as B<foo_I<TYPE>_bar>
 371 }
 372
 373 # This is from http://man7.org/linux/man-pages/man7/man-pages.7.html
 374 my %preferred_words = (
 375     'bitmask'       => 'bit mask',
 376     'builtin'       => 'built-in',
 377    #'epoch'         => 'Epoch', # handled specially, below
 378     'file name'     => 'filename',
 379     'file system'   => 'filesystem',
 380     'host name'     => 'hostname',
 381     'i-node'        => 'inode',
 382     'lower case'    => 'lowercase',
 383     'lower-case'    => 'lowercase',
 384     'non-zero'      => 'nonzero',
 385     'path name'     => 'pathname',
 386     'pseudo-terminal' => 'pseudoterminal',
 387     'reserved port' => 'privileged port',
 388     'system port'   => 'privileged port',
 389     'realtime'      => 'real-time',
 390     'real time'     => 'real-time',
 391     'runtime'       => 'run time',
 392     'saved group ID'=> 'saved set-group-ID',
 393     'saved set-GID' => 'saved set-group-ID',
 394     'saved user ID' => 'saved set-user-ID',
 395     'saved set-UID' => 'saved set-user-ID',
 396     'set-GID'       => 'set-group-ID',
 397     'setgid'        => 'set-group-ID',
 398     'set-UID'       => 'set-user-ID',
 399     'setuid'        => 'set-user-ID',
 400     'super user'    => 'superuser',
 401     'super-user'    => 'superuser',
 402     'super block'   => 'superblock',
 403     'super-block'   => 'superblock',
 404     'time stamp'    => 'timestamp',
 405     'time zone'     => 'timezone',
 406     'upper case'    => 'uppercase',
 407     'upper-case'    => 'uppercase',
 408     'useable'       => 'usable',
 409     'userspace'     => 'user space',
 410     'user name'     => 'username',
 411     'zeroes'        => 'zeros'
 412 );
 413
 414 # Search manpage for words that have a different preferred use.
 415 sub wording {
 416     my $id = shift;
 417     my $contents = shift;
 418
 419     foreach my $k ( keys %preferred_words ) {
 420         # Sigh, trademark
 421         next if $k eq 'file system'
 422             and $contents =~ /Microsoft Encrypted File System/;
 423         err($id, "found '$k' should use '$preferred_words{$k}'")
 424             if $contents =~ /\b\Q$k\E\b/i;
 425     }
 426     err($id, "found 'epoch' should use 'Epoch'")
 427         if $contents =~ /\bepoch\b/;
 428 }
 429
 430 # Perform all sorts of nit/error checks on a manpage
 431 sub check {
 432     my $filename = shift;
 433     my $dirname = basename(dirname($filename));
 434
 435     my $contents = '';
 436     {
 437         local $/ = undef;
 438         open POD, $filename or die "Couldn't open $filename, $!";
 439         $contents = <POD>;
 440         close POD;
 441     }
 442
 443     my $id = "${filename}:1:";
 444     check_head_style($id, $contents);
 445
 446     # Check ordering of some sections in man3
 447     if ( $filename =~ m|man3/| ) {
 448         check_section_location($id, $contents, "RETURN VALUES", "EXAMPLES");
 449         check_section_location($id, $contents, "SEE ALSO", "HISTORY");
 450         check_section_location($id, $contents, "EXAMPLES", "SEE ALSO");
 451     }
 452
 453     # Make sure every link has a section.
 454     while ( $contents =~ /$markup_re/msg ) {
 455         my $target = $1;
 456         next unless $target =~ /^L<(.*)>$/;     # Skip if not L<...>
 457         $target = $1;                           # Peal away L< and >
 458         $target =~ s/\/[^\/]*$//;               # Peal away possible anchor
 459         $target =~ s/.*\|//g;                   # Peal away possible link text
 460         next if $target eq '';                  # Skip if links within page, or
 461         next if $target =~ /::/;                #   links to a Perl module, or
 462         next if $target =~ /^https?:/;          #   is a URL link, or
 463         next if $target =~ /\([1357]\)$/;       #   it has a section
 464         err($id, "Section missing in $target")
 465     }
 466     # Check for proper links to commands.
 467     while ( $contents =~ /L<([^>]*)\(1\)(?:\/.*)?>/g ) {
 468         my $target = $1;
 469         next if $target =~ /openssl-?/;
 470         next if -f "doc/man1/$target.pod";
 471         # TODO: Filter out "foreign manual" links.
 472         next if $target =~ /ps|apropos|sha1sum|procmail|perl/;
 473         err($id, "Bad command link L<$target(1)>");
 474     }
 475     # Check for proper in-man-3 API links.
 476     while ( $contents =~ /L<([^>]*)\(3\)(?:\/.*)?>/g ) {
 477         my $target = $1;
 478         err($id, "Bad L<$target>")
 479             unless $target =~ /^[_[:alpha:]][_[:alnum:]]*$/
 480     }
 481
 482     unless ( $contents =~ /=for openssl generic/ ) {
 483         if ( $filename =~ m|man3/| ) {
 484             name_synopsis($id, $filename, $contents);
 485             functionname_check($id, $filename, $contents);
 486         } elsif ( $filename =~ m|man1/| ) {
 487             option_check($id, $filename, $contents)
 488         }
 489     }
 490
 491     wording($id, $contents);
 492
 493     err($id, "doesn't start with =pod")
 494         if $contents !~ /^=pod/;
 495     err($id, "doesn't end with =cut")
 496         if $contents !~ /=cut\n$/;
 497     err($id, "more than one cut line.")
 498         if $contents =~ /=cut.*=cut/ms;
 499     err($id, "EXAMPLE not EXAMPLES section.")
 500         if $contents =~ /=head1 EXAMPLE[^S]/;
 501     err($id, "WARNING not WARNINGS section.")
 502         if $contents =~ /=head1 WARNING[^S]/;
 503     err($id, "missing copyright")
 504         if $contents !~ /Copyright .* The OpenSSL Project Authors/;
 505     err($id, "copyright not last")
 506         if $contents =~ /head1 COPYRIGHT.*=head/ms;
 507     err($id, "head2 in All uppercase")
 508         if $contents =~ /head2\s+[A-Z ]+\n/;
 509     err($id, "extra space after head")
 510         if $contents =~ /=head\d\s\s+/;
 511     err($id, "period in NAME section")
 512         if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
 513     err($id, "Duplicate $1 in L<>")
 514         if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
 515     err($id, "Bad =over $1")
 516         if $contents =~ /=over([^ ][^24])/;
 517     err($id, "Possible version style issue")
 518         if $contents =~ /OpenSSL version [019]/;
 519
 520     if ( $contents !~ /=for openssl multiple includes/ ) {
 521         # Look for multiple consecutive openssl #include lines
 522         # (non-consecutive lines are okay; see man3/MD5.pod).
 523         if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
 524             my $count = 0;
 525             foreach my $line ( split /\n+/, $1 ) {
 526                 if ( $line =~ m@include <openssl/@ ) {
 527                     err($id, "has multiple includes")
 528                         if ++$count == 2;
 529                 } else {
 530                     $count = 0;
 531                 }
 532             }
 533         }
 534     }
 535
 536     open my $OUT, '>', $temp
 537         or die "Can't open $temp, $!";
 538     podchecker($filename, $OUT);
 539     close $OUT;
 540     open $OUT, '<', $temp
 541         or die "Can't read $temp, $!";
 542     while ( <$OUT> ) {
 543         next if /\(section\) in.*deprecated/;
 544         print;
 545     }
 546     close $OUT;
 547     unlink $temp || warn "Can't remove $temp, $!";
 548
 549     # Find what section this page is in; assume 3.
 550     my $section = 3;
 551     $section = $1 if $dirname =~ /man([1-9])/;
 552
 553     foreach ( (@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}}) ) {
 554         err($id, "missing $_ head1 section")
 555             if $contents !~ /^=head1\s+${_}\s*$/m;
 556     }
 557 }
 558
 559 # Parse libcrypto.num, etc., and return sorted list of what's there.
 560 sub parsenum {
 561     my $file = shift;
 562     my @apis;
 563
 564     open my $IN, '<', $file
 565         or die "Can't open $file, $!, stopped";
 566
 567     while ( <$IN> ) {
 568         next if /^#/;
 569         next if /\bNOEXIST\b/;
 570         my @fields = split();
 571         die "Malformed line $_"
 572             if scalar @fields != 2 && scalar @fields != 4;
 573         push @apis, $fields[0];
 574     }
 575
 576     close $IN;
 577
 578     return sort @apis;
 579 }
 580
 581 # Parse all the manpages, getting return map of what they document
 582 # (by looking at their NAME sections).
 583 # Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
 584 my %link_map = ();
 585 # Map of names in each POD file; "name(s)" => filename
 586 my %name_map = ();
 587
 588 # Load file of symbol names that we know aren't documented.
 589 sub loadmissing($)
 590 {
 591     my $missingfile = shift;
 592     my @missing;
 593
 594     open FH, $missingfile
 595         || die "Can't open $missingfile";
 596     while ( <FH> ) {
 597         chomp;
 598         next if /^#/;
 599         push @missing, $_;
 600     }
 601     close FH;
 602
 603     for (@missing) {
 604         err("$missingfile:", "$_ is documented in $name_map{$_}")
 605             if exists $name_map{$_} && defined $name_map{$_};
 606     }
 607
 608     return @missing;
 609 }
 610
 611 # Check for undocumented macros; ignore those in the "missing" file
 612 # and do simple check for #define in our header files.
 613 sub checkmacros {
 614     my $count = 0;
 615     my %seen;
 616     my @missing;
 617
 618     if ( $opt_o ) {
 619         @missing = loadmissing('util/missingmacro111.txt');
 620     } elsif ( $opt_v ) {
 621         @missing = loadmissing('util/missingmacro.txt');
 622     }
 623
 624     foreach my $f ( glob('include/openssl/*.h') ) {
 625         # Skip some internals we don't want to document yet.
 626         next if $f eq 'include/openssl/asn1.h';
 627         next if $f eq 'include/openssl/asn1t.h';
 628         next if $f eq 'include/openssl/err.h';
 629         open(IN, $f) || die "Can't open $f, $!";
 630         while ( <IN> ) {
 631             next unless /^#\s*define\s*(\S+)\(/;
 632             my $macro = "$1(3)"; # We know they're all in section 3
 633             next if exists $name_map{$macro} || defined $seen{$macro};
 634             next if $macro =~ /^i2d_/
 635                 || $macro =~ /^d2i_/
 636                 || $macro =~ /^DEPRECATEDIN/
 637                 || $macro =~ /\Q_fnsig(3)\E$/
 638                 || $macro =~ /^IMPLEMENT_/
 639                 || $macro =~ /^_?DECLARE_/;
 640
 641             # Skip macros known to be missing
 642             next if $opt_v && grep( /^\Q$macro\E$/, @missing);
 643
 644             err("$f:", "macro $macro undocumented")
 645                 if $opt_d || $opt_e;
 646             $count++;
 647             $seen{$macro} = 1;
 648         }
 649         close(IN);
 650     }
 651     err("# $count macros undocumented (count is approximate)")
 652         if $count > 0;
 653 }
 654
 655 # Find out what is undocumented (filtering out the known missing ones)
 656 # and display them.
 657 sub printem {
 658     my $libname = shift;
 659     my $numfile = shift;
 660     my $missingfile = shift;
 661     my $count = 0;
 662     my %seen;
 663
 664     my @missing = loadmissing($missingfile) if ( $opt_v );
 665
 666     foreach my $func ( parsenum($numfile) ) {
 667         $func .= '(3)';         # We know they're all in section 3
 668         next if exists $name_map{$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 $opt_v && grep( /^\Q$func\E$/, @missing);
 675
 676         err("$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     my %podinfo = extract_pod_info($filename, { debug => $debug });
 693
 694     unless ( grep { $simplename eq $_ } @{$podinfo{names}} ) {
 695         err($id, "$simplename not in NAME section");
 696         push @{$podinfo{names}}, $simplename;
 697     }
 698     foreach my $name (@{$podinfo{names}}) {
 699         next if $name eq "";
 700         err($id, "'$name' contains white space")
 701             if $name =~ /\s/;
 702         my $name_sec = "$name($section)";
 703         if ( !exists $name_map{$name_sec} ) {
 704             $name_map{$name_sec} = $filename;
 705         } elsif ( $filename eq $name_map{$name_sec} ) {
 706             err($id, "$name_sec duplicated in NAME section of",
 707                  $name_map{$name_sec});
 708         } else {
 709             err($id, "$name_sec also in NAME section of",
 710                  $name_map{$name_sec});
 711         }
 712     }
 713
 714     my @foreign_names =
 715         map { map { s/\s+//g; $_ } split(/,/, $_) }
 716         $podinfo{contents} =~ /=for\s+openssl\s+foreign\s+manuals:\s*(.*)\n\n/;
 717     foreach ( @foreign_names ) {
 718         $name_map{$_} = undef; # It still exists!
 719     }
 720
 721     my @links =
 722         $podinfo{contents} =~ /L<
 723                               # if the link is of the form L<something|name(s)>,
 724                               # then remove 'something'.  Note that 'something'
 725                               # may contain POD codes as well...
 726                               (?:(?:[^\|]|<[^>]*>)*\|)?
 727                               # we're only interested in references that have
 728                               # a one digit section number
 729                               ([^\/>\(]+\(\d\))
 730                              /gx;
 731     $link_map{$filename} = [ @links ];
 732 }
 733
 734 # Look for L<> ("link") references that point to files that do not exist.
 735 sub checklinks {
 736     foreach my $filename (sort keys %link_map) {
 737         foreach my $link (@{$link_map{$filename}}) {
 738             err("${filename}:1:", "reference to non-existing $link")
 739                 unless exists $name_map{$link};
 740         }
 741     }
 742 }
 743
 744 # Load the public symbol/macro names
 745 sub publicize {
 746     foreach my $name ( parsenum('util/libcrypto.num') ) {
 747         $public{$name} = 1;
 748     }
 749     foreach my $name ( parsenum('util/libssl.num') ) {
 750         $public{$name} = 1;
 751     }
 752     foreach my $name ( parsenum('util/other.syms') ) {
 753         $public{$name} = 1;
 754     }
 755 }
 756
 757 # Cipher/digests to skip if they show up as "not implemented"
 758 # because they are, via the "-*" construct.
 759 my %skips = (
 760     'aes128' => 1,
 761     'aes192' => 1,
 762     'aes256' => 1,
 763     'aria128' => 1,
 764     'aria192' => 1,
 765     'aria256' => 1,
 766     'camellia128' => 1,
 767     'camellia192' => 1,
 768     'camellia256' => 1,
 769     'des' => 1,
 770     'des3' => 1,
 771     'idea' => 1,
 772     'cipher' => 1,
 773     'digest' => 1,
 774 );
 775
 776 # Check the flags of a command and see if everything is in the manpage
 777 sub checkflags {
 778     my $cmd = shift;
 779     my $doc = shift;
 780     my %cmdopts;
 781     my %docopts;
 782     my %localskips;
 783
 784     # Get the list of options in the command.
 785     open CFH, "./apps/openssl list --options $cmd|"
 786         || die "Can list options for $cmd, $!";
 787     while ( <CFH> ) {
 788         chop;
 789         s/ .$//;
 790         $cmdopts{$_} = 1;
 791     }
 792     close CFH;
 793
 794     # Get the list of flags from the synopsis
 795     open CFH, "<$doc"
 796         || die "Can't open $doc, $!";
 797     while ( <CFH> ) {
 798         chop;
 799         last if /DESCRIPTION/;
 800         if ( /=for openssl ifdef (.*)/ ) {
 801             foreach my $f ( split / /, $1 ) {
 802                 $localskips{$f} = 1;
 803             }
 804             next;
 805         }
 806         next unless /\[B<-([^ >]+)/;
 807         my $opt = $1;
 808         $opt = $1 if $opt =~ /I<(.*)/;
 809         $docopts{$1} = 1;
 810     }
 811     close CFH;
 812
 813     # See what's in the command not the manpage.
 814     my @undocced = sort grep { !defined $docopts{$_} } keys %cmdopts;
 815     foreach ( @undocced ) {
 816         next if /-/; # Skip the -- end-of-flags marker
 817         err("$doc: undocumented option -$_");
 818     }
 819
 820     # See what's in the command not the manpage.
 821     my @unimpl = sort grep { !defined $cmdopts{$_} } keys %docopts;
 822     foreach ( @unimpl ) {
 823         next if defined $skips{$_} || defined $localskips{$_};
 824         err("$cmd documented but not implemented -$_");
 825     }
 826 }
 827
 828 ##
 829 ##  MAIN()
 830 ##  Do the work requested by the various getopt flags.
 831 ##  The flags are parsed in alphabetical order, just because we have
 832 ##  to have *some way* of listing them.
 833 ##
 834
 835 if ( $opt_c ) {
 836     my @commands = ();
 837
 838     # Get list of commands.
 839     open FH, "./apps/openssl list -1 -commands|"
 840         || die "Can't list commands, $!";
 841     while ( <FH> ) {
 842         chop;
 843         push @commands, $_;
 844     }
 845     close FH;
 846
 847     # See if each has a manpage.
 848     foreach my $cmd ( @commands ) {
 849         next if $cmd eq 'help' || $cmd eq 'exit';
 850         my $doc = "doc/man1/$cmd.pod";
 851         $doc = "doc/man1/openssl-$cmd.pod" if -f "doc/man1/openssl-$cmd.pod";
 852         if ( ! -f "$doc" ) {
 853             err("$doc does not exist");
 854         } else {
 855             checkflags($cmd, $doc);
 856         }
 857     }
 858
 859     # See what help is missing.
 860     open FH, "./apps/openssl list --missing-help |"
 861         || die "Can't list missing help, $!";
 862     while ( <FH> ) {
 863         chop;
 864         my ($cmd, $flag) = split;
 865         err("$cmd has no help for -$flag");
 866     }
 867     close FH;
 868
 869     exit $status;
 870 }
 871
 872 # Preparation for some options, populate %name_map and %link_map
 873 if ( $opt_l || $opt_u || $opt_v ) {
 874     foreach ( glob('doc/*/*.pod doc/internal/*/*.pod') ) {
 875         collectnames($_);
 876     }
 877 }
 878
 879 if ( $opt_l ) {
 880     checklinks();
 881 }
 882
 883 if ( $opt_n ) {
 884     publicize();
 885     foreach ( @ARGV ? @ARGV : glob('doc/*/*.pod doc/internal/*/*.pod') ) {
 886         check($_);
 887     }
 888
 889     # If not given args, check that all man1 commands are named properly.
 890     if ( scalar @ARGV == 0 ) {
 891         foreach (glob('doc/man1/*.pod')) {
 892             next if /CA.pl/ || /openssl\.pod/ || /tsget\.pod/;
 893             err("$_ doesn't start with openssl-") unless /openssl-/;
 894         }
 895     }
 896 }
 897
 898 if ( $opt_u || $opt_v) {
 899     if ( $opt_o ) {
 900         printem('crypto', 'util/libcrypto.num', 'util/missingcrypto111.txt');
 901         printem('ssl', 'util/libssl.num', 'util/missingssl111.txt');
 902     } else {
 903         printem('crypto', 'util/libcrypto.num', 'util/missingcrypto.txt');
 904         printem('ssl', 'util/libssl.num', 'util/missingssl.txt');
 905     }
 906     checkmacros();
 907 }
 908
 909 exit $status;