util/find-doc-nits

   1 #! /usr/bin/env perl
   2 # Copyright 2002-2021 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 Carp qw(:DEFAULT cluck);
  15 use Pod::Checker;
  16 use File::Find;
  17 use File::Basename;
  18 use File::Spec::Functions;
  19 use Getopt::Std;
  20 use FindBin;
  21 use lib "$FindBin::Bin/perl";
  22
  23 use OpenSSL::Util::Pod;
  24
  25 use lib '.';
  26 use configdata;
  27
  28 # Set to 1 for debug output
  29 my $debug = 0;
  30
  31 # Options.
  32 our($opt_d);
  33 our($opt_e);
  34 our($opt_s);
  35 our($opt_o);
  36 our($opt_h);
  37 our($opt_l);
  38 our($opt_n);
  39 our($opt_p);
  40 our($opt_u);
  41 our($opt_v);
  42 our($opt_c);
  43
  44 # Print usage message and exit.
  45 sub help {
  46     print <<EOF;
  47 Find small errors (nits) in documentation.  Options:
  48     -c List undocumented commands and options
  49     -d Detailed list of undocumented (implies -u)
  50     -e Detailed list of new undocumented (implies -v)
  51     -h Print this help message
  52     -l Print bogus links
  53     -n Print nits in POD pages
  54     -o Causes -e/-v to count symbols added since 1.1.1 as new (implies -v)
  55     -u Count undocumented functions
  56     -v Count new undocumented functions
  57 EOF
  58     exit;
  59 }
  60
  61 getopts('cdehlnouv');
  62
  63 help() if $opt_h;
  64 $opt_u = 1 if $opt_d;
  65 $opt_v = 1 if $opt_o || $opt_e;
  66 die "Cannot use both -u and -v"
  67     if $opt_u && $opt_v;
  68 die "Cannot use both -d and -e"
  69     if $opt_d && $opt_e;
  70
  71 # We only need to check c, l, n, u and v.
  72 # Options d, e, o imply one of the above.
  73 die "Need one of -[cdehlnouv] flags.\n"
  74     unless $opt_c or $opt_l or $opt_n or $opt_u or $opt_v;
  75
  76
  77 my $temp = '/tmp/docnits.txt';
  78 my $OUT;
  79 my $status = 0;
  80
  81 my @sections = ( 'man1', 'man3', 'man5', 'man7' );
  82 my %mandatory_sections = (
  83     '*' => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
  84     1   => [ 'SYNOPSIS', 'OPTIONS' ],
  85     3   => [ 'SYNOPSIS', 'RETURN VALUES' ],
  86     5   => [ ],
  87     7   => [ ]
  88                          );
  89
  90 # Symbols that we ignored.
  91 # They are reserved macros that we currently don't document
  92 my $ignored = qr/(?| ^i2d_
  93                  |   ^d2i_
  94                  |   ^DEPRECATEDIN
  95                  |   ^OSSL_DEPRECATED
  96                  |   \Q_fnsig(3)\E$
  97                  |   ^IMPLEMENT_
  98                  |   ^_?DECLARE_
  99                  |   ^sk_
 100                  |   ^SKM_DEFINE_STACK_OF_INTERNAL
 101                  |   ^lh_
 102                  |   ^DEFINE_LHASH_OF_INTERNAL
 103                  )/x;
 104
 105 # A common regexp for C symbol names
 106 my $C_symbol = qr/\b[[:alpha:]][_[:alnum:]]*\b/;
 107
 108 # Collect all POD files, both internal and public, and regardless of location
 109 # We collect them in a hash table with each file being a key, so we can attach
 110 # tags to them.  For example, internal docs will have the word "internal"
 111 # attached to them.
 112 my %files = ();
 113 # We collect files names on the fly, on known tag basis
 114 my %collected_tags = ();
 115 # We cache results based on tags
 116 my %collected_results = ();
 117
 118 # files OPTIONS
 119 #
 120 # Example:
 121 #
 122 #       files(TAGS => 'manual');
 123 #       files(TAGS => [ 'manual', 'man1' ]);
 124 #
 125 # This function returns an array of files corresponding to a set of tags
 126 # given with the options "TAGS".  The value of this option can be a single
 127 # word, or an array of several words, which work as inclusive or exclusive
 128 # selectors.  Inclusive selectors are used to add one more set of files to
 129 # the returned array, while exclusive selectors limit the set of files added
 130 # to the array.  The recognised tag values are:
 131 #
 132 # 'public_manual'       - inclusive selector, adds public manuals to the
 133 #                         returned array of files.
 134 # 'internal_manual'     - inclusive selector, adds internal manuals to the
 135 #                         returned array of files.
 136 # 'manual'              - inclusive selector, adds any manual to the returned
 137 #                         array of files.  This is really a shorthand for
 138 #                         'public_manual' and 'internal_manual' combined.
 139 # 'public_header'       - inclusive selector, adds public headers to the
 140 #                         returned array of files.
 141 # 'header'              - inclusive selector, adds any header file to the
 142 #                         returned array of files.  Since we currently only
 143 #                         care about public headers, this is exactly
 144 #                         equivalent to 'public_header', but is present for
 145 #                         consistency.
 146 #
 147 # 'man1', 'man3', 'man5', 'man7'
 148 #                       - exclusive selectors, only applicable together with
 149 #                         any of the manual selectors.  If any of these are
 150 #                         present, only the manuals from the given sections
 151 #                         will be include.  If none of these are present,
 152 #                         the manuals from all sections will be returned.
 153 #
 154 # All returned manual files come from configdata.pm.
 155 # All returned header files come from looking inside
 156 # "$config{sourcedir}/include/openssl"
 157 #
 158 sub files {
 159     my %opts = ( @_ );          # Make a copy of the arguments
 160
 161     $opts{TAGS} = [ $opts{TAGS} ] if ref($opts{TAGS}) eq '';
 162
 163     croak "No tags given, or not an array"
 164         unless exists $opts{TAGS} && ref($opts{TAGS}) eq 'ARRAY';
 165
 166     my %tags = map { $_ => 1 } @{$opts{TAGS}};
 167     $tags{public_manual} = 1
 168         if $tags{manual} && ($tags{public} // !$tags{internal});
 169     $tags{internal_manual} = 1
 170         if $tags{manual} && ($tags{internal} // !$tags{public});
 171     $tags{public_header} = 1
 172         if $tags{header} && ($tags{public} // !$tags{internal});
 173     delete $tags{manual};
 174     delete $tags{header};
 175     delete $tags{public};
 176     delete $tags{internal};
 177
 178     my $tags_as_key = join(':', sort keys %tags);
 179
 180     cluck  "DEBUG[files]: This is how we got here!" if $debug;
 181     print STDERR "DEBUG[files]: tags: $tags_as_key\n" if $debug;
 182
 183     my %tags_to_collect = ( map { $_ => 1 }
 184                             grep { !exists $collected_tags{$_} }
 185                             keys %tags );
 186
 187     if ($tags_to_collect{public_manual}) {
 188         print STDERR "DEBUG[files]: collecting public manuals\n"
 189             if $debug;
 190
 191         # The structure in configdata.pm is that $unified_info{mandocs}
 192         # contains lists of man files, and in turn, $unified_info{depends}
 193         # contains hash tables showing which POD file each of those man
 194         # files depend on.  We use that information to find the POD files,
 195         # and to attach the man section they belong to as tags
 196         foreach my $mansect ( @sections ) {
 197             foreach ( map { @{$unified_info{depends}->{$_}} }
 198                       @{$unified_info{mandocs}->{$mansect}} ) {
 199                 $files{$_} = { $mansect => 1, public_manual => 1 };
 200             }
 201         }
 202         $collected_tags{public_manual} = 1;
 203     }
 204
 205     if ($tags_to_collect{internal_manual}) {
 206         print STDERR "DEBUG[files]: collecting internal manuals\n"
 207             if $debug;
 208
 209         # We don't have the internal docs in configdata.pm.  However, they
 210         # are all in the source tree, so they're easy to find.
 211         foreach my $mansect ( @sections ) {
 212             foreach ( glob(catfile($config{sourcedir},
 213                                    'doc', 'internal', $mansect, '*.pod')) ) {
 214                 $files{$_} = { $mansect => 1, internal_manual => 1 };
 215             }
 216         }
 217         $collected_tags{internal_manual} = 1;
 218     }
 219
 220     if ($tags_to_collect{public_header}) {
 221         print STDERR "DEBUG[files]: collecting public headers\n"
 222             if $debug;
 223
 224         foreach ( glob(catfile($config{sourcedir},
 225                                'include', 'openssl', '*.h')) ) {
 226             $files{$_} = { public_header => 1 };
 227         }
 228     }
 229
 230     my @result = @{$collected_results{$tags_as_key} // []};
 231
 232     if (!@result) {
 233         # Produce a result based on caller tags
 234         foreach my $type ( ( 'public_manual', 'internal_manual' ) ) {
 235             next unless $tags{$type};
 236
 237             # If caller asked for specific sections, we care about sections.
 238             # Otherwise, we give back all of them.
 239             my @selected_sections =
 240                 grep { $tags{$_} } @sections;
 241             @selected_sections = @sections unless @selected_sections;
 242
 243             foreach my $section ( ( @selected_sections ) ) {
 244                 push @result,
 245                     ( sort { basename($a) cmp basename($b) }
 246                       grep { $files{$_}->{$type} && $files{$_}->{$section} }
 247                       keys %files );
 248             }
 249         }
 250         if ($tags{public_header}) {
 251             push @result,
 252                 ( sort { basename($a) cmp basename($b) }
 253                   grep { $files{$_}->{public_header} }
 254                   keys %files );
 255         }
 256
 257         if ($debug) {
 258             print STDERR "DEBUG[files]: result:\n";
 259             print STDERR "DEBUG[files]:     $_\n" foreach @result;
 260         }
 261         $collected_results{$tags_as_key} = [ @result ];
 262     }
 263
 264     return @result;
 265 }
 266
 267 # Print error message, set $status.
 268 sub err {
 269     print join(" ", @_), "\n";
 270     $status = 1
 271 }
 272
 273 # Cross-check functions in the NAME and SYNOPSIS section.
 274 sub name_synopsis {
 275     my $id = shift;
 276     my $filename = shift;
 277     my $contents = shift;
 278
 279     # Get NAME section and all words in it.
 280     return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
 281     my $tmp = $1;
 282     $tmp =~ tr/\n/ /;
 283     err($id, "Trailing comma before - in NAME")
 284         if $tmp =~ /, *-/;
 285     $tmp =~ s/ -.*//g;
 286     err($id, "POD markup among the names in NAME")
 287         if $tmp =~ /[<>]/;
 288     $tmp =~ s/  */ /g;
 289     err($id, "Missing comma in NAME")
 290         if $tmp =~ /[^,] /;
 291
 292     my $dirname = dirname($filename);
 293     my $section = basename($dirname);
 294     my $simplename = basename($filename, ".pod");
 295     my $foundfilename = 0;
 296     my %foundfilenames = ();
 297     my %names;
 298     foreach my $n ( split ',', $tmp ) {
 299         $n =~ s/^\s+//;
 300         $n =~ s/\s+$//;
 301         err($id, "The name '$n' contains white-space")
 302             if $n =~ /\s/;
 303         $names{$n} = 1;
 304         $foundfilename++ if $n eq $simplename;
 305         $foundfilenames{$n} = 1
 306             if ( ( grep { basename($_) eq "$n.pod" }
 307                    files(TAGS => [ 'manual', $section ]) )
 308                  && $n ne $simplename );
 309     }
 310     err($id, "The following exist as other .pod files:",
 311          sort keys %foundfilenames)
 312         if %foundfilenames;
 313     err($id, "$simplename (filename) missing from NAME section")
 314         unless $foundfilename;
 315
 316     # Find all functions in SYNOPSIS
 317     return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
 318     my $syn = $1;
 319     my $ignore_until = undef;   # If defined, this is a regexp
 320     # Remove all non-code lines
 321     $syn =~ s/^(?:\s*?|\S.*?)$//msg;
 322     # Remove all comments
 323     $syn =~ s/\/\*.*?\*\///msg;
 324     while ( $syn ) {
 325         # "env" lines end at a newline.
 326         # Preprocessor lines start with a # and end at a newline.
 327         # Other lines end with a semicolon, and may cover more than
 328         # one physical line.
 329         if ( $syn !~ /^ \s*(env .*?|#.*?|.*?;)\s*$/ms ) {
 330             err($id, "Can't parse rest of synopsis:\n$syn\n(declarations not ending with a semicolon (;)?)");
 331             last;
 332         }
 333         my $line = $1;
 334         $syn = $';
 335
 336         print STDERR "DEBUG[name_synopsis] \$line = '$line'\n" if $debug;
 337
 338         # Special code to skip over documented structures
 339         if ( defined $ignore_until) {
 340             next if $line !~ /$ignore_until/;
 341             $ignore_until = undef;
 342             next;
 343         }
 344         if ( $line =~ /^\s*(?:typedef\s+)?struct(?:\s+\S+)\s*\{/ ) {
 345             $ignore_until = qr/\}.*?;/;
 346             next;
 347         }
 348
 349         my $sym;
 350         my $is_prototype = 1;
 351         $line =~ s/LHASH_OF\([^)]+\)/int/g;
 352         $line =~ s/STACK_OF\([^)]+\)/int/g;
 353         $line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
 354         $line =~ s/__declspec\([^)]+\)//;
 355
 356         ## We don't prohibit that space, to allow typedefs looking like
 357         ## this:
 358         ##
 359         ## typedef int (fantastically_long_name_breaks_80char_limit)
 360         ##     (fantastically_long_name_breaks_80char_limit *something);
 361         ##
 362         #if ( $line =~ /typedef.*\(\*?\S+\)\s+\(/ ) {
 363         #    # a callback function with whitespace before the argument list:
 364         #    # typedef ... (*NAME) (...
 365         #    # typedef ... (NAME) (...
 366         #    err($id, "Function typedef has space before arg list: $line");
 367         #}
 368
 369         if ( $line =~ /env (\S*)=/ ) {
 370             # environment variable env NAME=...
 371             $sym = $1;
 372         } elsif ( $line =~ /typedef.*\(\*?($C_symbol)\)\s*\(/ ) {
 373             # a callback function pointer: typedef ... (*NAME)(...
 374             # a callback function signature: typedef ... (NAME)(...
 375             $sym = $1;
 376         } elsif ( $line =~ /typedef.*($C_symbol)\s*\(/ ) {
 377             # a callback function signature: typedef ... NAME(...
 378             $sym = $1;
 379         } elsif ( $line =~ /typedef.*($C_symbol);/ ) {
 380             # a simple typedef: typedef ... NAME;
 381             $is_prototype = 0;
 382             $sym = $1;
 383         } elsif ( $line =~ /enum ($C_symbol) \{/ ) {
 384             # an enumeration: enum ... {
 385             $sym = $1;
 386         } elsif ( $line =~ /#\s*(?:define|undef) ($C_symbol)/ ) {
 387             $is_prototype = 0;
 388             $sym = $1;
 389         } elsif ( $line =~ /^[^\(]*?\(\*($C_symbol)\s*\(/ ) {
 390             # a function returning a function pointer: TYPE (*NAME(args))(args)
 391             $sym = $1;
 392         } elsif ( $line =~ /^[^\(]*?($C_symbol)\s*\(/ ) {
 393             # a simple function declaration
 394             $sym = $1;
 395         }
 396         else {
 397             next;
 398         }
 399
 400         print STDERR "DEBUG[name_synopsis] \$sym = '$sym'\n" if $debug;
 401
 402         err($id, "$sym missing from NAME section")
 403             unless defined $names{$sym};
 404         $names{$sym} = 2;
 405
 406         # Do some sanity checks on the prototype.
 407         err($id, "Prototype missing spaces around commas: $line")
 408             if $is_prototype && $line =~ /[a-z0-9],[^\s]/;
 409     }
 410
 411     foreach my $n ( keys %names ) {
 412         next if $names{$n} == 2;
 413         err($id, "$n missing from SYNOPSIS")
 414     }
 415 }
 416
 417 # Check if SECTION ($3) is located before BEFORE ($4)
 418 sub check_section_location {
 419     my $id = shift;
 420     my $contents = shift;
 421     my $section = shift;
 422     my $before = shift;
 423
 424     return unless $contents =~ /=head1 $section/
 425         and $contents =~ /=head1 $before/;
 426     err($id, "$section should appear before $before section")
 427         if $contents =~ /=head1 $before.*=head1 $section/ms;
 428 }
 429
 430 # Check if a =head1 is duplicated, or a =headX is duplicated within a
 431 # =head1.  Treats =head2 =head3 as equivalent -- it doesn't reset the head3
 432 # sets if it finds a =head2 -- but that is good enough for now. Also check
 433 # for proper capitalization, trailing periods, etc.
 434 sub check_head_style {
 435     my $id = shift;
 436     my $contents = shift;
 437     my %head1;
 438     my %subheads;
 439
 440     foreach my $line ( split /\n+/, $contents ) {
 441         next unless $line =~ /^=head/;
 442         if ( $line =~ /head1/ ) {
 443             err($id, "Duplicate section $line")
 444                 if defined $head1{$line};
 445             $head1{$line} = 1;
 446             %subheads = ();
 447         } else {
 448             err($id, "Duplicate subsection $line")
 449                 if defined $subheads{$line};
 450             $subheads{$line} = 1;
 451         }
 452         err($id, "Period in =head")
 453             if $line =~ /\.[^\w]/ or $line =~ /\.$/;
 454         err($id, "not all uppercase in =head1")
 455             if $line =~ /head1.*[a-z]/;
 456         err($id, "All uppercase in subhead")
 457             if $line =~ /head[234][ A-Z0-9]+$/;
 458     }
 459 }
 460
 461 # Because we have options and symbols with extra markup, we need
 462 # to take that into account, so we need a regexp that extracts
 463 # markup chunks, including recursive markup.
 464 # please read up on /(?R)/ in perlre(1)
 465 # (note: order is important, (?R) needs to come before .)
 466 # (note: non-greedy is important, or something like 'B<foo> and B<bar>'
 467 # will be captured as one item)
 468 my $markup_re =
 469     qr/(                        # Capture group
 470            [BIL]<               # The start of what we recurse on
 471            (?:(?-1)|.)*?        # recurse the whole regexp (referring to
 472                                 # the last opened capture group, i.e. the
 473                                 # start of this regexp), or pick next
 474                                 # character.  Do NOT be greedy!
 475            >                    # The end of what we recurse on
 476        )/x;                     # (the x allows this sort of split up regexp)
 477
 478 # Options must start with a dash, followed by a letter, possibly
 479 # followed by letters, digits, dashes and underscores, and the last
 480 # character must be a letter or a digit.
 481 # We do also accept the single -? or -n, where n is a digit
 482 my $option_re =
 483     qr/(?:
 484             \?                  # Single question mark
 485             |
 486             \d                  # Single digit
 487             |
 488             -                   # Single dash (--)
 489             |
 490             [[:alpha:]](?:[-_[:alnum:]]*?[[:alnum:]])?
 491        )/x;
 492
 493 # Helper function to check if a given $thing is properly marked up
 494 # option.  It returns one of these values:
 495 #     undef         if it's not an option
 496 #     ""            if it's a malformed option
 497 #     $unwrapped    the option with the outermost B<> wrapping removed.
 498 sub normalise_option {
 499     my $id = shift;
 500     my $filename = shift;
 501     my $thing = shift;
 502
 503     my $unwrapped = $thing;
 504     my $unmarked = $thing;
 505
 506     # $unwrapped is the option with the outer B<> markup removed
 507     $unwrapped =~ s/^B<//;
 508     $unwrapped =~ s/>$//;
 509     # $unmarked is the option with *all* markup removed
 510     $unmarked =~ s/[BIL]<|>//msg;
 511
 512
 513     # If we found an option, check it, collect it
 514     if ( $unwrapped =~ /^\s*-/ ) {
 515         return $unwrapped       # return option with outer B<> removed
 516             if $unmarked =~ /^-${option_re}$/;
 517         return "";              # Malformed option
 518     }
 519     return undef;               # Something else
 520 }
 521
 522 # Checks of command option (man1) formatting.  The man1 checks are
 523 # restricted to the SYNOPSIS and OPTIONS sections, the rest is too
 524 # free form, we simply cannot be too strict there.
 525
 526 sub option_check {
 527     my $id = shift;
 528     my $filename = shift;
 529     my $contents = shift;
 530
 531     my $synopsis = ($contents =~ /=head1\s+SYNOPSIS(.*?)=head1/s, $1);
 532
 533     # Some pages have more than one OPTIONS section, let's make sure
 534     # to get them all
 535     my $options = '';
 536     while ( $contents =~ /=head1\s+[A-Z ]*?OPTIONS$(.*?)(?==head1)/msg ) {
 537         $options .= $1;
 538     }
 539
 540     # Look for options with no or incorrect markup
 541     while ( $synopsis =~
 542             /(?<![-<[:alnum:]])-(?:$markup_re|.)*(?![->[:alnum:]])/msg ) {
 543         err($id, "Malformed option [1] in SYNOPSIS: $&");
 544     }
 545
 546     while ( $synopsis =~ /$markup_re/msg ) {
 547         my $found = $&;
 548         print STDERR "$id:DEBUG[option_check] SYNOPSIS: found $found\n"
 549             if $debug;
 550         my $option_uw = normalise_option($id, $filename, $found);
 551         err($id, "Malformed option [2] in SYNOPSIS: $found")
 552             if defined $option_uw && $option_uw eq '';
 553     }
 554
 555     # In OPTIONS, we look for =item paragraphs.
 556     # (?=^\s*$) detects an empty line.
 557     while ( $options =~ /=item\s+(.*?)(?=^\s*$)/msg ) {
 558         my $item = $&;
 559
 560         while ( $item =~ /(\[\s*)?($markup_re)/msg ) {
 561             my $found = $2;
 562             print STDERR "$id:DEBUG[option_check] OPTIONS: found $&\n"
 563                 if $debug;
 564             err($id, "Unexpected bracket in OPTIONS =item: $item")
 565                 if ($1 // '') ne '' && $found =~ /^B<\s*-/;
 566
 567             my $option_uw = normalise_option($id, $filename, $found);
 568             err($id, "Malformed option in OPTIONS: $found")
 569                 if defined $option_uw && $option_uw eq '';
 570         }
 571     }
 572 }
 573
 574 # Normal symbol form
 575 my $symbol_re = qr/[[:alpha:]_][_[:alnum:]]*?/;
 576
 577 # Checks of function name (man3) formatting.  The man3 checks are
 578 # easier than the man1 checks, we only check the names followed by (),
 579 # and only the names that have POD markup.
 580 sub functionname_check {
 581     my $id = shift;
 582     my $filename = shift;
 583     my $contents = shift;
 584
 585     while ( $contents =~ /($markup_re)\(\)/msg ) {
 586         print STDERR "$id:DEBUG[functionname_check] SYNOPSIS: found $&\n"
 587             if $debug;
 588
 589         my $symbol = $1;
 590         my $unmarked = $symbol;
 591         $unmarked =~ s/[BIL]<|>//msg;
 592
 593         err($id, "Malformed symbol: $symbol")
 594             unless $symbol =~ /^B<.*?>$/ && $unmarked =~ /^${symbol_re}$/
 595     }
 596
 597     # We can't do the kind of collecting coolness that option_check()
 598     # does, because there are too many things that can't be found in
 599     # name repositories like the NAME sections, such as symbol names
 600     # with a variable part (typically marked up as B<foo_I<TYPE>_bar>
 601 }
 602
 603 # This is from http://man7.org/linux/man-pages/man7/man-pages.7.html
 604 my %preferred_words = (
 605     '16bit'         => '16-bit',
 606     'a.k.a.'        => 'aka',
 607     'bitmask'       => 'bit mask',
 608     'builtin'       => 'built-in',
 609    #'epoch'         => 'Epoch', # handled specially, below
 610     'fall-back'     => 'fallback',
 611     'file name'     => 'filename',
 612     'file system'   => 'filesystem',
 613     'host name'     => 'hostname',
 614     'i-node'        => 'inode',
 615     'lower case'    => 'lowercase',
 616     'lower-case'    => 'lowercase',
 617     'manpage'       => 'man page',
 618     'non-blocking'  => 'nonblocking',
 619     'non-default'   => 'nondefault',
 620     'non-empty'     => 'nonempty',
 621     'non-negative'  => 'nonnegative',
 622     'non-zero'      => 'nonzero',
 623     'path name'     => 'pathname',
 624     'pre-allocated' => 'preallocated',
 625     'pseudo-terminal' => 'pseudoterminal',
 626     'real time'     => 'real-time',
 627     'realtime'      => 'real-time',
 628     'reserved port' => 'privileged port',
 629     'runtime'       => 'run time',
 630     'saved group ID'=> 'saved set-group-ID',
 631     'saved set-GID' => 'saved set-group-ID',
 632     'saved set-UID' => 'saved set-user-ID',
 633     'saved user ID' => 'saved set-user-ID',
 634     'set-GID'       => 'set-group-ID',
 635     'set-UID'       => 'set-user-ID',
 636     'setgid'        => 'set-group-ID',
 637     'setuid'        => 'set-user-ID',
 638     'sub-system'    => 'subsystem',
 639     'super block'   => 'superblock',
 640     'super-block'   => 'superblock',
 641     'super user'    => 'superuser',
 642     'super-user'    => 'superuser',
 643     'system port'   => 'privileged port',
 644     'time stamp'    => 'timestamp',
 645     'time zone'     => 'timezone',
 646     'upper case'    => 'uppercase',
 647     'upper-case'    => 'uppercase',
 648     'useable'       => 'usable',
 649     'user name'     => 'username',
 650     'userspace'     => 'user space',
 651     'zeroes'        => 'zeros'
 652 );
 653
 654 # Search manpage for words that have a different preferred use.
 655 sub wording {
 656     my $id = shift;
 657     my $contents = shift;
 658
 659     foreach my $k ( keys %preferred_words ) {
 660         # Sigh, trademark
 661         next if $k eq 'file system'
 662             and $contents =~ /Microsoft Encrypted File System/;
 663         err($id, "Found '$k' should use '$preferred_words{$k}'")
 664             if $contents =~ /\b\Q$k\E\b/i;
 665     }
 666     err($id, "Found 'epoch' should use 'Epoch'")
 667         if $contents =~ /\bepoch\b/;
 668     if ( $id =~ m@man1/@ ) {
 669         err($id, "found 'tool' in NAME, should use 'command'")
 670             if $contents =~ /=head1 NAME.*\btool\b.*=head1 SYNOPSIS/s;
 671         err($id, "found 'utility' in NAME, should use 'command'")
 672             if $contents =~ /NAME.*\butility\b.*=head1 SYNOPSIS/s;
 673
 674     }
 675 }
 676
 677 # Perform all sorts of nit/error checks on a manpage
 678 sub check {
 679     my %podinfo = @_;
 680     my $filename = $podinfo{filename};
 681     my $dirname = basename(dirname($filename));
 682     my $contents = $podinfo{contents};
 683
 684     my $id = "${filename}:1:";
 685     check_head_style($id, $contents);
 686
 687     # Check ordering of some sections in man3
 688     if ( $filename =~ m|man3/| ) {
 689         check_section_location($id, $contents, "RETURN VALUES", "EXAMPLES");
 690         check_section_location($id, $contents, "SEE ALSO", "HISTORY");
 691         check_section_location($id, $contents, "EXAMPLES", "SEE ALSO");
 692     }
 693
 694     # Make sure every link has a section.
 695     while ( $contents =~ /$markup_re/msg ) {
 696         my $target = $1;
 697         next unless $target =~ /^L<(.*)>$/;     # Skip if not L<...>
 698         $target = $1;                           # Peal away L< and >
 699         $target =~ s/\/[^\/]*$//;               # Peal away possible anchor
 700         $target =~ s/.*\|//g;                   # Peal away possible link text
 701         next if $target eq '';                  # Skip if links within page, or
 702         next if $target =~ /::/;                #   links to a Perl module, or
 703         next if $target =~ /^https?:/;          #   is a URL link, or
 704         next if $target =~ /\([1357]\)$/;       #   it has a section
 705         err($id, "Section missing in $target")
 706     }
 707     # Check for proper links to commands.
 708     while ( $contents =~ /L<([^>]*)\(1\)(?:\/.*)?>/g ) {
 709         my $target = $1;
 710         next if $target =~ /openssl-?/;
 711         next if ( grep { basename($_) eq "$target.pod" }
 712                   files(TAGS => [ 'manual', 'man1' ]) );
 713         # TODO: Filter out "foreign manual" links.
 714         next if $target =~ /ps|apropos|sha1sum|procmail|perl/;
 715         err($id, "Bad command link L<$target(1)>");
 716     }
 717     # Check for proper in-man-3 API links.
 718     while ( $contents =~ /L<([^>]*)\(3\)(?:\/.*)?>/g ) {
 719         my $target = $1;
 720         err($id, "Bad L<$target>")
 721             unless $target =~ /^[_[:alpha:]][_[:alnum:]]*$/
 722     }
 723
 724     unless ( $contents =~ /^=for openssl generic/ms ) {
 725         if ( $filename =~ m|man3/| ) {
 726             name_synopsis($id, $filename, $contents);
 727             functionname_check($id, $filename, $contents);
 728         } elsif ( $filename =~ m|man1/| ) {
 729             option_check($id, $filename, $contents)
 730         }
 731     }
 732
 733     wording($id, $contents);
 734
 735     err($id, "Doesn't start with =pod")
 736         if $contents !~ /^=pod/;
 737     err($id, "Doesn't end with =cut")
 738         if $contents !~ /=cut\n$/;
 739     err($id, "More than one cut line.")
 740         if $contents =~ /=cut.*=cut/ms;
 741     err($id, "EXAMPLE not EXAMPLES section.")
 742         if $contents =~ /=head1 EXAMPLE[^S]/;
 743     err($id, "WARNING not WARNINGS section.")
 744         if $contents =~ /=head1 WARNING[^S]/;
 745     err($id, "Missing copyright")
 746         if $contents !~ /Copyright .* The OpenSSL Project Authors/;
 747     err($id, "Copyright not last")
 748         if $contents =~ /head1 COPYRIGHT.*=head/ms;
 749     err($id, "head2 in All uppercase")
 750         if $contents =~ /head2\s+[A-Z ]+\n/;
 751     err($id, "Extra space after head")
 752         if $contents =~ /=head\d\s\s+/;
 753     err($id, "Period in NAME section")
 754         if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
 755     err($id, "Duplicate $1 in L<>")
 756         if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
 757     err($id, "Bad =over $1")
 758         if $contents =~ /=over([^ ][^24])/;
 759     err($id, "Possible version style issue")
 760         if $contents =~ /OpenSSL version [019]/;
 761
 762     if ( $contents !~ /=for openssl multiple includes/ ) {
 763         # Look for multiple consecutive openssl #include lines
 764         # (non-consecutive lines are okay; see man3/MD5.pod).
 765         if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
 766             my $count = 0;
 767             foreach my $line ( split /\n+/, $1 ) {
 768                 if ( $line =~ m@include <openssl/@ ) {
 769                     err($id, "Has multiple includes")
 770                         if ++$count == 2;
 771                 } else {
 772                     $count = 0;
 773                 }
 774             }
 775         }
 776     }
 777
 778     open my $OUT, '>', $temp
 779         or die "Can't open $temp, $!";
 780     err($id, "POD errors")
 781         if podchecker($filename, $OUT) != 0;
 782     close $OUT;
 783     open $OUT, '<', $temp
 784         or die "Can't read $temp, $!";
 785     while ( <$OUT> ) {
 786         next if /\(section\) in.*deprecated/;
 787         print;
 788     }
 789     close $OUT;
 790     unlink $temp || warn "Can't remove $temp, $!";
 791
 792     # Find what section this page is in; assume 3.
 793     my $section = 3;
 794     $section = $1 if $dirname =~ /man([1-9])/;
 795
 796     foreach ( (@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}}) ) {
 797         err($id, "Missing $_ head1 section")
 798             if $contents !~ /^=head1\s+${_}\s*$/m;
 799     }
 800 }
 801
 802 # Information database ###############################################
 803
 804 # Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
 805 my %link_map = ();
 806 # Map of names in each POD file or from "missing" files; possible values are:
 807 # If found in a POD files, "name(s)" => filename
 808 # If found in a "missing" file or external, "name(s)" => ''
 809 my %name_map = ();
 810
 811 # State of man-page names.
 812 # %state is affected by loading util/*.num and util/*.syms
 813 # Values may be one of:
 814 # 'crypto' : belongs in libcrypto (loaded from libcrypto.num)
 815 # 'ssl' : belongs in libssl (loaded from libssl.num)
 816 # 'other' : belongs in libcrypto or libssl (loaded from other.syms)
 817 # 'internal' : Internal
 818 # 'public' : Public (generic name or external documentation)
 819 # Any of these values except 'public' may be prefixed with 'missing_'
 820 # to indicate that they are known to be missing.
 821 my %state;
 822 # %missing is affected by loading util/missing*.txt.  Values may be one of:
 823 # 'crypto' : belongs in libcrypto (loaded from libcrypto.num)
 824 # 'ssl' : belongs in libssl (loaded from libssl.num)
 825 # 'other' : belongs in libcrypto or libssl (loaded from other.syms)
 826 # 'internal' : Internal
 827 my %missing;
 828
 829 # Parse libcrypto.num, etc., and return sorted list of what's there.
 830 sub loadnum ($;$) {
 831     my $file = shift;
 832     my $type = shift;
 833     my @symbols;
 834
 835     open my $IN, '<', catfile($config{sourcedir}, $file)
 836         or die "Can't open $file, $!, stopped";
 837
 838     while ( <$IN> ) {
 839         next if /^#/;
 840         next if /\bNOEXIST\b/;
 841         my @fields = split();
 842         die "Malformed line $. in $file: $_"
 843             if scalar @fields != 2 && scalar @fields != 4;
 844         $state{$fields[0].'(3)'} = $type // 'internal';
 845     }
 846     close $IN;
 847 }
 848
 849 # Load file of symbol names that we know aren't documented.
 850 sub loadmissing($;$)
 851 {
 852     my $missingfile = shift;
 853     my $type = shift;
 854
 855     open FH, catfile($config{sourcedir}, $missingfile)
 856         or die "Can't open $missingfile";
 857     while ( <FH> ) {
 858         chomp;
 859         next if /^#/;
 860         $missing{$_} = $type // 'internal';
 861     }
 862     close FH;
 863 }
 864
 865 # Check that we have consistent public / internal documentation and declaration
 866 sub checkstate () {
 867     # Collect all known names, no matter where they come from
 868     my %names = map { $_ => 1 } (keys %name_map, keys %state, keys %missing);
 869
 870     # Check section 3, i.e. functions and macros
 871     foreach ( grep { $_ =~ /\(3\)$/ } sort keys %names ) {
 872         next if ( $name_map{$_} // '') eq '' || $_ =~ /$ignored/;
 873
 874         # If a man-page isn't recorded public or if it's recorded missing
 875         # and internal, it's declared to be internal.
 876         my $declared_internal =
 877             ($state{$_} // 'internal') eq 'internal'
 878             || ($missing{$_} // '') eq 'internal';
 879         # If a man-page isn't recorded internal or if it's recorded missing
 880         # and not internal, it's declared to be public
 881         my $declared_public =
 882             ($state{$_} // 'internal') ne 'internal'
 883             || ($missing{$_} // 'internal') ne 'internal';
 884
 885         err("$_ is supposedly public but is documented as internal")
 886             if ( $declared_public && $name_map{$_} =~ /\/internal\// );
 887         err("$_ is supposedly internal (maybe missing from other.syms) but is documented as public")
 888             if ( $declared_internal && $name_map{$_} !~ /\/internal\// );
 889     }
 890 }
 891
 892 # Check for undocumented macros; ignore those in the "missing" file
 893 # and do simple check for #define in our header files.
 894 sub checkmacros {
 895     my $count = 0;
 896     my %seen;
 897
 898     foreach my $f ( files(TAGS => 'public_header') ) {
 899         # Skip some internals we don't want to document yet.
 900         my $b = basename($f);
 901         next if $b eq 'asn1.h';
 902         next if $b eq 'asn1t.h';
 903         next if $b eq 'err.h';
 904         open(IN, $f)
 905             or die "Can't open $f, $!";
 906         while ( <IN> ) {
 907             next unless /^#\s*define\s*(\S+)\(/;
 908             my $macro = "$1(3)"; # We know they're all in section 3
 909             next if defined $name_map{$macro}
 910                 || defined $missing{$macro}
 911                 || defined $seen{$macro}
 912                 || $macro =~ /$ignored/;
 913
 914             err("$f:", "macro $macro undocumented")
 915                 if $opt_d || $opt_e;
 916             $count++;
 917             $seen{$macro} = 1;
 918         }
 919         close(IN);
 920     }
 921     err("# $count macros undocumented (count is approximate)")
 922         if $count > 0;
 923 }
 924
 925 # Find out what is undocumented (filtering out the known missing ones)
 926 # and display them.
 927 sub printem ($) {
 928     my $type = shift;
 929     my $count = 0;
 930
 931     foreach my $func ( grep { $state{$_} eq $type } sort keys %state ) {
 932         next if defined $name_map{$func}
 933             || defined $missing{$func};
 934
 935         err("$type:", "function $func undocumented")
 936             if $opt_d || $opt_e;
 937         $count++;
 938     }
 939     err("# $count lib$type names are not documented")
 940         if $count > 0;
 941 }
 942
 943 # Collect all the names in a manpage.
 944 sub collectnames {
 945     my %podinfo = @_;
 946     my $filename = $podinfo{filename};
 947     $filename =~ m|man(\d)/|;
 948     my $section = $1;
 949     my $simplename = basename($filename, ".pod");
 950     my $id = "${filename}:1:";
 951     my $is_generic = $podinfo{contents} =~ /^=for openssl generic/ms;
 952
 953     unless ( grep { $simplename eq $_ } @{$podinfo{names}} ) {
 954         err($id, "$simplename not in NAME section");
 955         push @{$podinfo{names}}, $simplename;
 956     }
 957     foreach my $name ( @{$podinfo{names}} ) {
 958         next if $name eq "";
 959         err($id, "'$name' contains whitespace")
 960             if $name =~ /\s/;
 961         my $name_sec = "$name($section)";
 962         if ( !defined $name_map{$name_sec} ) {
 963             $name_map{$name_sec} = $filename;
 964             $state{$name_sec} //=
 965                 ( $filename =~ /\/internal\// ? 'internal' : 'public' )
 966                 if $is_generic;
 967         } elsif ( $filename eq $name_map{$name_sec} ) {
 968             err($id, "$name_sec duplicated in NAME section of",
 969                  $name_map{$name_sec});
 970         } elsif ( $name_map{$name_sec} ne '' ) {
 971             err($id, "$name_sec also in NAME section of",
 972                  $name_map{$name_sec});
 973         }
 974     }
 975
 976     if ( $podinfo{contents} =~ /=for openssl foreign manual (.*)\n/ ) {
 977         foreach my $f ( split / /, $1 ) {
 978             $name_map{$f} = ''; # It still exists!
 979             $state{$f} = 'public'; # We assume!
 980         }
 981     }
 982
 983     my @links =
 984         $podinfo{contents} =~ /L<
 985                               # if the link is of the form L<something|name(s)>,
 986                               # then remove 'something'.  Note that 'something'
 987                               # may contain POD codes as well...
 988                               (?:(?:[^\|]|<[^>]*>)*\|)?
 989                               # we're only interested in references that have
 990                               # a one digit section number
 991                               ([^\/>\(]+\(\d\))
 992                              /gx;
 993     $link_map{$filename} = [ @links ];
 994 }
 995
 996 # Look for L<> ("link") references that point to files that do not exist.
 997 sub checklinks {
 998     foreach my $filename ( sort keys %link_map ) {
 999         foreach my $link ( @{$link_map{$filename}} ) {
1000             err("${filename}:1:", "reference to non-existing $link")
1001                 unless defined $name_map{$link} || defined $missing{$link};
1002             err("${filename}:1:", "reference of internal $link in public documentation $filename")
1003                 if ( ( ($state{$link} // '') eq 'internal'
1004                        || ($missing{$link} // '') eq 'internal' )
1005                      && $filename !~ /\/internal\// );
1006         }
1007     }
1008 }
1009
1010 # Cipher/digests to skip if they show up as "not implemented"
1011 # because they are, via the "-*" construct.
1012 my %skips = (
1013     'aes128' => 1,
1014     'aes192' => 1,
1015     'aes256' => 1,
1016     'aria128' => 1,
1017     'aria192' => 1,
1018     'aria256' => 1,
1019     'camellia128' => 1,
1020     'camellia192' => 1,
1021     'camellia256' => 1,
1022     'des' => 1,
1023     'des3' => 1,
1024     'idea' => 1,
1025     'cipher' => 1,
1026     'digest' => 1,
1027 );
1028
1029 my %genopts; # generic options parsed from apps/include/opt.h
1030
1031 # Check the flags of a command and see if everything is in the manpage
1032 sub checkflags {
1033     my $cmd = shift;
1034     my $doc = shift;
1035     my @cmdopts;
1036     my %docopts;
1037
1038     # Get the list of options in the command source file.
1039     my $active = 0;
1040     my $expect_helpstr = "";
1041     open CFH, "apps/$cmd.c"
1042         or die "Can't open apps/$cmd.c to list options for $cmd, $!";
1043     while ( <CFH> ) {
1044         chop;
1045         if ($active) {
1046             last if m/^\s*};/;
1047             if ($expect_helpstr ne "") {
1048                 next if m/^\s*#\s*if/;
1049                 err("$cmd does not implement help for -$expect_helpstr") unless m/^\s*"/;
1050                 $expect_helpstr = "";
1051             } elsif (m/\{\s*"([^"]+)"\s*,\s*OPT_[A-Z0-9_]+\s*,\s*('[-\/:<>cEfFlMnNpsuU]'|0)\s*,(.*)$/
1052                        && !($cmd eq "s_client" && $1 eq "wdebug")) {
1053                 push @cmdopts, $1;
1054                 $expect_helpstr = $1;
1055                 $expect_helpstr = "" if $3 =~ m/^\s*"/;
1056             } elsif (m/[\s,](OPT_[A-Z]+_OPTIONS?)\s*(,|$)/) {
1057                 push @cmdopts, @{ $genopts{$1} };
1058             }
1059         } elsif (m/^const\s+OPTIONS\s*/) {
1060             $active = 1;
1061         }
1062     }
1063     close CFH;
1064
1065     # Get the list of flags from the synopsis
1066     open CFH, "<$doc"
1067         or die "Can't open $doc, $!";
1068     while ( <CFH> ) {
1069         chop;
1070         last if /DESCRIPTION/;
1071         my $opt;
1072         if ( /\[B<-([^ >]+)/ ) {
1073             $opt = $1;
1074         } elsif ( /^B<-([^ >]+)/ ) {
1075             $opt = $1;
1076         } else {
1077             next;
1078         }
1079         $opt = $1 if $opt =~ /I<(.*)/;
1080         $docopts{$1} = 1;
1081     }
1082     close CFH;
1083
1084     # See what's in the command not the manpage.
1085     my @undocced = sort grep { !defined $docopts{$_} } @cmdopts;
1086     foreach ( @undocced ) {
1087         next if $cmd eq "openssl" && $_ eq "help";
1088         err("$doc: undocumented option -$_");
1089     }
1090
1091     # See what's in the command not the manpage.
1092     my @unimpl = sort grep { my $e = $_; !(grep /^\Q$e\E$/, @cmdopts) } keys %docopts;
1093     foreach ( @unimpl ) {
1094         next if $_ eq "-"; # Skip the -- end-of-flags marker
1095         next if defined $skips{$_};
1096         err("$doc: $cmd does not implement -$_");
1097     }
1098 }
1099
1100 ##
1101 ##  MAIN()
1102 ##  Do the work requested by the various getopt flags.
1103 ##  The flags are parsed in alphabetical order, just because we have
1104 ##  to have *some way* of listing them.
1105 ##
1106
1107 if ( $opt_c ) {
1108     my @commands = ();
1109
1110     # Get the lists of generic options.
1111     my $active = "";
1112     open OFH, "apps/include/opt.h"
1113         or die "Can't open apps/include/opt.h to list generic options, $!";
1114     while ( <OFH> ) {
1115         chop;
1116         push @{ $genopts{$active} }, $1 if $active ne "" && m/^\s+\{\s*"([^"]+)"\s*,\s*OPT_/;
1117         $active = $1 if m/^\s*#\s*define\s+(OPT_[A-Z]+_OPTIONS?)\s*\\\s*$/;
1118         $active = "" if m/^\s*$/;
1119     }
1120     close OFH;
1121
1122     # Get list of commands.
1123     opendir(DIR, "apps");
1124     @commands = grep(/\.c$/, readdir(DIR));
1125     closedir(DIR);
1126
1127     # See if each has a manpage.
1128     foreach my $cmd ( @commands ) {
1129         $cmd =~ s/\.c$//;
1130         next if $cmd eq 'progs' || $cmd eq 'cmp_mock_srv' || $cmd eq 'vms_decc_init';
1131         my @doc = ( grep { basename($_) eq "openssl-$cmd.pod"
1132                            # For "tsget" and "CA.pl" pod pages
1133                            || basename($_) eq "$cmd.pod" }
1134                     files(TAGS => [ 'manual', 'man1' ]) );
1135         my $num = scalar @doc;
1136         if ($num > 1) {
1137             err("$num manuals for 'openssl $cmd': ".join(", ", @doc));
1138         } elsif ($num < 1) {
1139             err("no manual for 'openssl $cmd'");
1140         } else {
1141             checkflags($cmd, @doc);
1142         }
1143     }
1144
1145     exit $status;
1146 }
1147
1148 # Populate %state
1149 loadnum('util/libcrypto.num', 'crypto');
1150 loadnum('util/libssl.num', 'ssl');
1151 loadnum('util/other.syms', 'other');
1152 loadnum('util/other-internal.syms');
1153 if ( $opt_o ) {
1154     loadmissing('util/missingmacro111.txt', 'crypto');
1155     loadmissing('util/missingcrypto111.txt', 'crypto');
1156     loadmissing('util/missingssl111.txt', 'ssl');
1157 } elsif ( !$opt_u ) {
1158     loadmissing('util/missingmacro.txt', 'crypto');
1159     loadmissing('util/missingcrypto.txt', 'crypto');
1160     loadmissing('util/missingssl.txt', 'ssl');
1161     loadmissing('util/missingcrypto-internal.txt');
1162     loadmissing('util/missingssl-internal.txt');
1163 }
1164
1165 if ( $opt_n || $opt_l || $opt_u || $opt_v ) {
1166     my @files_to_read = ( $opt_n && @ARGV ) ? @ARGV : files(TAGS => 'manual');
1167
1168     foreach (@files_to_read) {
1169         my %podinfo = extract_pod_info($_, { debug => $debug });
1170
1171         collectnames(%podinfo)
1172             if ( $opt_l || $opt_u || $opt_v );
1173
1174         check(%podinfo)
1175             if ( $opt_n );
1176     }
1177 }
1178
1179 if ( $opt_l ) {
1180     checklinks();
1181 }
1182
1183 if ( $opt_n ) {
1184     # If not given args, check that all man1 commands are named properly.
1185     if ( scalar @ARGV == 0 ) {
1186         foreach ( files(TAGS => [ 'public_manual', 'man1' ]) ) {
1187             next if /openssl\.pod/
1188                 || /CA\.pl/ || /tsget\.pod/; # these commands are special cases
1189             err("$_ doesn't start with openssl-") unless /openssl-/;
1190         }
1191     }
1192 }
1193
1194 checkstate();
1195
1196 if ( $opt_u || $opt_v) {
1197     printem('crypto');
1198     printem('ssl');
1199     checkmacros();
1200 }
1201
1202 exit $status;