util/find-doc-nits

   1 #! /usr/bin/env perl
   2 # Copyright 2002-2020 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 # Where to find openssl command
  32 my $openssl = "./util/opensslwrap.sh";
  33
  34 # Options.
  35 our($opt_d);
  36 our($opt_e);
  37 our($opt_s);
  38 our($opt_o);
  39 our($opt_h);
  40 our($opt_l);
  41 our($opt_n);
  42 our($opt_p);
  43 our($opt_u);
  44 our($opt_v);
  45 our($opt_c);
  46
  47 # Print usage message and exit.
  48 sub help {
  49     print <<EOF;
  50 Find small errors (nits) in documentation.  Options:
  51     -c List undocumented commands and options
  52     -d Detailed list of undocumented (implies -u)
  53     -e Detailed list of new undocumented (implies -v)
  54     -h Print this help message
  55     -l Print bogus links
  56     -n Print nits in POD pages
  57     -o Causes -e/-v to count symbols added since 1.1.1 as new (implies -v)
  58     -u Count undocumented functions
  59     -v Count new undocumented functions
  60 EOF
  61     exit;
  62 }
  63
  64 getopts('cdehlnouv');
  65
  66 help() if $opt_h;
  67 $opt_u = 1 if $opt_d;
  68 $opt_v = 1 if $opt_o || $opt_e;
  69 die "Cannot use both -u and -v"
  70     if $opt_u && $opt_v;
  71 die "Cannot use both -d and -e"
  72     if $opt_d && $opt_e;
  73
  74 # We only need to check c, l, n, u and v.
  75 # Options d, e, o imply one of the above.
  76 die "Need one of -[cdehlnouv] flags.\n"
  77     unless $opt_c or $opt_l or $opt_n or $opt_u or $opt_v;
  78
  79
  80 my $temp = '/tmp/docnits.txt';
  81 my $OUT;
  82 my $status = 0;
  83
  84 my @sections = ( 'man1', 'man3', 'man5', 'man7' );
  85 my %mandatory_sections = (
  86     '*' => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
  87     1   => [ 'SYNOPSIS', 'OPTIONS' ],
  88     3   => [ 'SYNOPSIS', 'RETURN VALUES' ],
  89     5   => [ ],
  90     7   => [ ]
  91                          );
  92
  93 # Symbols that we ignored.
  94 # They are internal macros that we currently don't document
  95 my $ignored = qr/(?| ^i2d_
  96                  |   ^d2i_
  97                  |   ^DEPRECATEDIN
  98                  |   \Q_fnsig(3)\E$
  99                  |   ^IMPLEMENT_
 100                  |   ^_?DECLARE_
 101                  |   ^sk_
 102                  |   ^SKM_DEFINE_STACK_OF_INTERNAL
 103                  |   ^lh_
 104                  |   ^DEFINE_LHASH_OF_INTERNAL
 105                  )/x;
 106
 107 # A common regexp for C symbol names
 108 my $C_symbol = qr/\b[[:alpha:]][_[:alnum:]]*\b/;
 109
 110 # Collect all POD files, both internal and public, and regardless of location
 111 # We collect them in a hash table with each file being a key, so we can attach
 112 # tags to them.  For example, internal docs will have the word "internal"
 113 # attached to them.
 114 my %files = ();
 115 # We collect files names on the fly, on known tag basis
 116 my %collected_tags = ();
 117 # We cache results based on tags
 118 my %collected_results = ();
 119
 120 # files OPTIONS
 121 #
 122 # Example:
 123 #
 124 #       files(TAGS => 'manual');
 125 #       files(TAGS => [ 'manual', 'man1' ]);
 126 #
 127 # This function returns an array of files corresponding to a set of tags
 128 # given with the options "TAGS".  The value of this option can be a single
 129 # word, or an array of several words, which work as inclusive or exclusive
 130 # selectors.  Inclusive selectors are used to add one more set of files to
 131 # the returned array, while exclusive selectors limit the set of files added
 132 # to the array.  The recognised tag values are:
 133 #
 134 # 'public_manual'       - inclusive selector, adds public manuals to the
 135 #                         returned array of files.
 136 # 'internal_manual'     - inclusive selector, adds internal manuals to the
 137 #                         returned array of files.
 138 # 'manual'              - inclusive selector, adds any manual to the returned
 139 #                         array of files.  This is really a shorthand for
 140 #                         'public_manual' and 'internal_manual' combined.
 141 # 'public_header'       - inclusive selector, adds public headers to the
 142 #                         returned array of files.
 143 # 'header'              - inclusive selector, adds any header file to the
 144 #                         returned array of files.  Since we currently only
 145 #                         care about public headers, this is exactly
 146 #                         equivalent to 'public_header', but is present for
 147 #                         consistency.
 148 #
 149 # 'man1', 'man3', 'man5', 'man7'
 150 #                       - exclusive selectors, only applicable together with
 151 #                         any of the manual selectors.  If any of these are
 152 #                         present, only the manuals from the given sections
 153 #                         will be include.  If none of these are present,
 154 #                         the manuals from all sections will be returned.
 155 #
 156 # All returned manual files come from configdata.pm.
 157 # All returned header files come from looking inside
 158 # "$config{sourcedir}/include/openssl"
 159 #
 160 sub files {
 161     my %opts = ( @_ );          # Make a copy of the arguments
 162
 163     $opts{TAGS} = [ $opts{TAGS} ] if ref($opts{TAGS}) eq '';
 164
 165     croak "No tags given, or not an array"
 166         unless exists $opts{TAGS} && ref($opts{TAGS}) eq 'ARRAY';
 167
 168     my %tags = map { $_ => 1 } @{$opts{TAGS}};
 169     $tags{public_manual} = 1
 170         if $tags{manual} && ($tags{public} // !$tags{internal});
 171     $tags{internal_manual} = 1
 172         if $tags{manual} && ($tags{internal} // !$tags{public});
 173     $tags{public_header} = 1
 174         if $tags{header} && ($tags{public} // !$tags{internal});
 175     delete $tags{manual};
 176     delete $tags{header};
 177     delete $tags{public};
 178     delete $tags{internal};
 179
 180     my $tags_as_key = join(':', sort keys %tags);
 181
 182     cluck  "DEBUG[files]: This is how we got here!" if $debug;
 183     print STDERR "DEBUG[files]: tags: $tags_as_key\n" if $debug;
 184
 185     my %tags_to_collect = ( map { $_ => 1 }
 186                             grep { !exists $collected_tags{$_} }
 187                             keys %tags );
 188
 189     if ($tags_to_collect{public_manual}) {
 190         print STDERR "DEBUG[files]: collecting public manuals\n"
 191             if $debug;
 192
 193         # The structure in configdata.pm is that $unified_info{mandocs}
 194         # contains lists of man files, and in turn, $unified_info{depends}
 195         # contains hash tables showing which POD file each of those man
 196         # files depend on.  We use that information to find the POD files,
 197         # and to attach the man section they belong to as tags
 198         foreach my $mansect ( @sections ) {
 199             foreach ( map { @{$unified_info{depends}->{$_}} }
 200                       @{$unified_info{mandocs}->{$mansect}} ) {
 201                 $files{$_} = { $mansect => 1, public_manual => 1 };
 202             }
 203         }
 204         $collected_tags{public_manual} = 1;
 205     }
 206
 207     if ($tags_to_collect{internal_manual}) {
 208         print STDERR "DEBUG[files]: collecting internal manuals\n"
 209             if $debug;
 210
 211         # We don't have the internal docs in configdata.pm.  However, they
 212         # are all in the source tree, so they're easy to find.
 213         foreach my $mansect ( @sections ) {
 214             foreach ( glob(catfile($config{sourcedir},
 215                                    'doc', 'internal', $mansect, '*.pod')) ) {
 216                 $files{$_} = { $mansect => 1, internal_manual => 1 };
 217             }
 218         }
 219         $collected_tags{internal_manual} = 1;
 220     }
 221
 222     if ($tags_to_collect{public_header}) {
 223         print STDERR "DEBUG[files]: collecting public headers\n"
 224             if $debug;
 225
 226         foreach ( glob(catfile($config{sourcedir},
 227                                'include', 'openssl', '*.h')) ) {
 228             $files{$_} = { public_header => 1 };
 229         }
 230     }
 231
 232     my @result = @{$collected_results{$tags_as_key} // []};
 233
 234     if (!@result) {
 235         # Produce a result based on caller tags
 236         foreach my $type ( ( 'public_manual', 'internal_manual' ) ) {
 237             next unless $tags{$type};
 238
 239             # If caller asked for specific sections, we care about sections.
 240             # Otherwise, we give back all of them.
 241             my @selected_sections =
 242                 grep { $tags{$_} } @sections;
 243             @selected_sections = @sections unless @selected_sections;
 244
 245             foreach my $section ( ( @selected_sections ) ) {
 246                 push @result,
 247                     ( sort { basename($a) cmp basename($b) }
 248                       grep { $files{$_}->{$type} && $files{$_}->{$section} }
 249                       keys %files );
 250             }
 251         }
 252         if ($tags{public_header}) {
 253             push @result,
 254                 ( sort { basename($a) cmp basename($b) }
 255                   grep { $files{$_}->{public_header} }
 256                   keys %files );
 257         }
 258
 259         if ($debug) {
 260             print STDERR "DEBUG[files]: result:\n";
 261             print STDERR "DEBUG[files]:     $_\n" foreach @result;
 262         }
 263         $collected_results{$tags_as_key} = [ @result ];
 264     }
 265
 266     return @result;
 267 }
 268
 269 # Print error message, set $status.
 270 sub err {
 271     print join(" ", @_), "\n";
 272     $status = 1
 273 }
 274
 275 # Cross-check functions in the NAME and SYNOPSIS section.
 276 sub name_synopsis {
 277     my $id = shift;
 278     my $filename = shift;
 279     my $contents = shift;
 280
 281     # Get NAME section and all words in it.
 282     return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
 283     my $tmp = $1;
 284     $tmp =~ tr/\n/ /;
 285     err($id, "Trailing comma before - in NAME")
 286         if $tmp =~ /, *-/;
 287     $tmp =~ s/ -.*//g;
 288     err($id, "POD markup among the names in NAME")
 289         if $tmp =~ /[<>]/;
 290     $tmp =~ s/  */ /g;
 291     err($id, "Missing comma in NAME")
 292         if $tmp =~ /[^,] /;
 293
 294     my $dirname = dirname($filename);
 295     my $section = basename($dirname);
 296     my $simplename = basename($filename, ".pod");
 297     my $foundfilename = 0;
 298     my %foundfilenames = ();
 299     my %names;
 300     foreach my $n ( split ',', $tmp ) {
 301         $n =~ s/^\s+//;
 302         $n =~ s/\s+$//;
 303         err($id, "The name '$n' contains white-space")
 304             if $n =~ /\s/;
 305         $names{$n} = 1;
 306         $foundfilename++ if $n eq $simplename;
 307         $foundfilenames{$n} = 1
 308             if ( ( grep { basename($_) eq "$n.pod" }
 309                    files(TAGS => [ 'manual', $section ]) )
 310                  && $n ne $simplename );
 311     }
 312     err($id, "The following exist as other .pod files:",
 313          sort keys %foundfilenames)
 314         if %foundfilenames;
 315     err($id, "$simplename (filename) missing from NAME section")
 316         unless $foundfilename;
 317
 318     # Find all functions in SYNOPSIS
 319     return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
 320     my $syn = $1;
 321     my $ignore_until = undef;   # If defined, this is a regexp
 322     # Remove all non-code lines
 323     $syn =~ s/^(?:\s*?|\S.*?)$//msg;
 324     # Remove all comments
 325     $syn =~ s/\/\*.*?\*\///msg;
 326     while ( $syn ) {
 327         # "env" lines end at a newline.
 328         # Preprocessor lines start with a # and end at a newline.
 329         # Other lines end with a semicolon, and may cover more than
 330         # one physical line.
 331         if ( $syn !~ /^ \s*(env .*?|#.*?|.*?;)\s*$/ms ) {
 332             err($id, "Can't parse rest of synopsis:\n$syn\n(declarations not ending with a semicolon (;)?)");
 333             last;
 334         }
 335         my $line = $1;
 336         $syn = $';
 337
 338         print STDERR "DEBUG[name_synopsis] \$line = '$line'\n" if $debug;
 339
 340         # Special code to skip over documented structures
 341         if ( defined $ignore_until) {
 342             next if $line !~ /$ignore_until/;
 343             $ignore_until = undef;
 344             next;
 345         }
 346         if ( $line =~ /^\s*(?:typedef\s+)?struct(?:\s+\S+)\s*\{/ ) {
 347             $ignore_until = qr/\}.*?;/;
 348             next;
 349         }
 350
 351         my $sym;
 352         my $is_prototype = 1;
 353         $line =~ s/STACK_OF\([^)]+\)/int/g;
 354         $line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
 355         $line =~ s/__declspec\([^)]+\)//;
 356
 357         ## We don't prohibit that space, to allow typedefs looking like
 358         ## this:
 359         ##
 360         ## typedef int (fantastically_long_name_breaks_80char_limit)
 361         ##     (fantastically_long_name_breaks_80char_limit *something);
 362         ##
 363         #if ( $line =~ /typedef.*\(\*?\S+\)\s+\(/ ) {
 364         #    # a callback function with whitespace before the argument list:
 365         #    # typedef ... (*NAME) (...
 366         #    # typedef ... (NAME) (...
 367         #    err($id, "Function typedef has space before arg list: $line");
 368         #}
 369
 370         if ( $line =~ /env (\S*)=/ ) {
 371             # environment variable env NAME=...
 372             $sym = $1;
 373         } elsif ( $line =~ /typedef.*\(\*?($C_symbol)\)\s*\(/ ) {
 374             # a callback function pointer: typedef ... (*NAME)(...
 375             # a callback function signature: typedef ... (NAME)(...
 376             $sym = $1;
 377         } elsif ( $line =~ /typedef.*($C_symbol)\s*\(/ ) {
 378             # a callback function signature: typedef ... NAME(...
 379             $sym = $1;
 380         } elsif ( $line =~ /typedef.*($C_symbol);/ ) {
 381             # a simple typedef: typedef ... NAME;
 382             $is_prototype = 0;
 383             $sym = $1;
 384         } elsif ( $line =~ /enum ($C_symbol) \{/ ) {
 385             # an enumeration: enum ... {
 386             $sym = $1;
 387         } elsif ( $line =~ /#\s*(?:define|undef) ($C_symbol)/ ) {
 388             $is_prototype = 0;
 389             $sym = $1;
 390         } elsif ( $line =~ /^[^\(]*?\(\*($C_symbol)\s*\(/ ) {
 391             # a function returning a function pointer: TYPE (*NAME(args))(args)
 392             $sym = $1;
 393         } elsif ( $line =~ /^[^\(]*?($C_symbol)\s*\(/ ) {
 394             # a simple function declaration
 395             $sym = $1;
 396         }
 397         else {
 398             next;
 399         }
 400
 401         print STDERR "DEBUG[name_synopsis] \$sym = '$sym'\n" if $debug;
 402
 403         err($id, "$sym missing from NAME section")
 404             unless defined $names{$sym};
 405         $names{$sym} = 2;
 406
 407         # Do some sanity checks on the prototype.
 408         err($id, "Prototype missing spaces around commas: $line")
 409             if $is_prototype && $line =~ /[a-z0-9],[^\s]/;
 410     }
 411
 412     foreach my $n ( keys %names ) {
 413         next if $names{$n} == 2;
 414         err($id, "$n missing from SYNOPSIS")
 415     }
 416 }
 417
 418 # Check if SECTION ($3) is located before BEFORE ($4)
 419 sub check_section_location {
 420     my $id = shift;
 421     my $contents = shift;
 422     my $section = shift;
 423     my $before = shift;
 424
 425     return unless $contents =~ /=head1 $section/
 426         and $contents =~ /=head1 $before/;
 427     err($id, "$section should appear before $before section")
 428         if $contents =~ /=head1 $before.*=head1 $section/ms;
 429 }
 430
 431 # Check if a =head1 is duplicated, or a =headX is duplicated within a
 432 # =head1.  Treats =head2 =head3 as equivalent -- it doesn't reset the head3
 433 # sets if it finds a =head2 -- but that is good enough for now. Also check
 434 # for proper capitalization, trailing periods, etc.
 435 sub check_head_style {
 436     my $id = shift;
 437     my $contents = shift;
 438     my %head1;
 439     my %subheads;
 440
 441     foreach my $line ( split /\n+/, $contents ) {
 442         next unless $line =~ /^=head/;
 443         if ( $line =~ /head1/ ) {
 444             err($id, "Duplicate section $line")
 445                 if defined $head1{$line};
 446             $head1{$line} = 1;
 447             %subheads = ();
 448         } else {
 449             err($id, "Duplicate subsection $line")
 450                 if defined $subheads{$line};
 451             $subheads{$line} = 1;
 452         }
 453         err($id, "Period in =head")
 454             if $line =~ /\.[^\w]/ or $line =~ /\.$/;
 455         err($id, "not all uppercase in =head1")
 456             if $line =~ /head1.*[a-z]/;
 457         err($id, "All uppercase in subhead")
 458             if $line =~ /head[234][ A-Z0-9]+$/;
 459     }
 460 }
 461
 462 # Because we have options and symbols with extra markup, we need
 463 # to take that into account, so we need a regexp that extracts
 464 # markup chunks, including recursive markup.
 465 # please read up on /(?R)/ in perlre(1)
 466 # (note: order is important, (?R) needs to come before .)
 467 # (note: non-greedy is important, or something like 'B<foo> and B<bar>'
 468 # will be captured as one item)
 469 my $markup_re =
 470     qr/(                        # Capture group
 471            [BIL]<               # The start of what we recurse on
 472            (?:(?-1)|.)*?        # recurse the whole regexp (referring to
 473                                 # the last opened capture group, i.e. the
 474                                 # start of this regexp), or pick next
 475                                 # character.  Do NOT be greedy!
 476            >                    # The end of what we recurse on
 477        )/x;                     # (the x allows this sort of split up regexp)
 478
 479 # Options must start with a dash, followed by a letter, possibly
 480 # followed by letters, digits, dashes and underscores, and the last
 481 # character must be a letter or a digit.
 482 # We do also accept the single -? or -n, where n is a digit
 483 my $option_re =
 484     qr/(?:
 485             \?                  # Single question mark
 486             |
 487             \d                  # Single digit
 488             |
 489             -                   # Single dash (--)
 490             |
 491             [[:alpha:]](?:[-_[:alnum:]]*?[[:alnum:]])?
 492        )/x;
 493
 494 # Helper function to check if a given $thing is properly marked up
 495 # option.  It returns one of these values:
 496 #     undef         if it's not an option
 497 #     ""            if it's a malformed option
 498 #     $unwrapped    the option with the outermost B<> wrapping removed.
 499 sub normalise_option {
 500     my $id = shift;
 501     my $filename = shift;
 502     my $thing = shift;
 503
 504     my $unwrapped = $thing;
 505     my $unmarked = $thing;
 506
 507     # $unwrapped is the option with the outer B<> markup removed
 508     $unwrapped =~ s/^B<//;
 509     $unwrapped =~ s/>$//;
 510     # $unmarked is the option with *all* markup removed
 511     $unmarked =~ s/[BIL]<|>//msg;
 512
 513
 514     # If we found an option, check it, collect it
 515     if ( $unwrapped =~ /^\s*-/ ) {
 516         return $unwrapped       # return option with outer B<> removed
 517             if $unmarked =~ /^-${option_re}$/;
 518         return "";              # Malformed option
 519     }
 520     return undef;               # Something else
 521 }
 522
 523 # Checks of command option (man1) formatting.  The man1 checks are
 524 # restricted to the SYNOPSIS and OPTIONS sections, the rest is too
 525 # free form, we simply cannot be too strict there.
 526
 527 sub option_check {
 528     my $id = shift;
 529     my $filename = shift;
 530     my $contents = shift;
 531
 532     my $synopsis = ($contents =~ /=head1\s+SYNOPSIS(.*?)=head1/s, $1);
 533
 534     # Some pages have more than one OPTIONS section, let's make sure
 535     # to get them all
 536     my $options = '';
 537     while ( $contents =~ /=head1\s+[A-Z ]*?OPTIONS$(.*?)(?==head1)/msg ) {
 538         $options .= $1;
 539     }
 540
 541     # Look for options with no or incorrect markup
 542     while ( $synopsis =~
 543             /(?<![-<[:alnum:]])-(?:$markup_re|.)*(?![->[:alnum:]])/msg ) {
 544         err($id, "Malformed option [1] in SYNOPSIS: $&");
 545     }
 546
 547     while ( $synopsis =~ /$markup_re/msg ) {
 548         my $found = $&;
 549         print STDERR "$id:DEBUG[option_check] SYNOPSIS: found $found\n"
 550             if $debug;
 551         my $option_uw = normalise_option($id, $filename, $found);
 552         err($id, "Malformed option [2] in SYNOPSIS: $found")
 553             if defined $option_uw && $option_uw eq '';
 554     }
 555
 556     # In OPTIONS, we look for =item paragraphs.
 557     # (?=^\s*$) detects an empty line.
 558     while ( $options =~ /=item\s+(.*?)(?=^\s*$)/msg ) {
 559         my $item = $&;
 560
 561         while ( $item =~ /(\[\s*)?($markup_re)/msg ) {
 562             my $found = $2;
 563             print STDERR "$id:DEBUG[option_check] OPTIONS: found $&\n"
 564                 if $debug;
 565             err($id, "Unexpected bracket in OPTIONS =item: $item")
 566                 if ($1 // '') ne '' && $found =~ /^B<\s*-/;
 567
 568             my $option_uw = normalise_option($id, $filename, $found);
 569             err($id, "Malformed option in OPTIONS: $found")
 570                 if defined $option_uw && $option_uw eq '';
 571         }
 572     }
 573 }
 574
 575 # Normal symbol form
 576 my $symbol_re = qr/[[:alpha:]_][_[:alnum:]]*?/;
 577
 578 # Checks of function name (man3) formatting.  The man3 checks are
 579 # easier than the man1 checks, we only check the names followed by (),
 580 # and only the names that have POD markup.
 581 sub functionname_check {
 582     my $id = shift;
 583     my $filename = shift;
 584     my $contents = shift;
 585
 586     while ( $contents =~ /($markup_re)\(\)/msg ) {
 587         print STDERR "$id:DEBUG[functionname_check] SYNOPSIS: found $&\n"
 588             if $debug;
 589
 590         my $symbol = $1;
 591         my $unmarked = $symbol;
 592         $unmarked =~ s/[BIL]<|>//msg;
 593
 594         err($id, "Malformed symbol: $symbol")
 595             unless $symbol =~ /^B<.*?>$/ && $unmarked =~ /^${symbol_re}$/
 596     }
 597
 598     # We can't do the kind of collecting coolness that option_check()
 599     # does, because there are too many things that can't be found in
 600     # name repositories like the NAME sections, such as symbol names
 601     # with a variable part (typically marked up as B<foo_I<TYPE>_bar>
 602 }
 603
 604 # This is from http://man7.org/linux/man-pages/man7/man-pages.7.html
 605 my %preferred_words = (
 606     '16bit'         => '16-bit',
 607     'a.k.a.'        => 'aka',
 608     'bitmask'       => 'bit mask',
 609     'builtin'       => 'built-in',
 610    #'epoch'         => 'Epoch', # handled specially, below
 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, $!";
 780     podchecker($filename, $OUT);
 781     close $OUT;
 782     open $OUT, '<', $temp
 783         or die "Can't read $temp, $!";
 784     while ( <$OUT> ) {
 785         next if /\(section\) in.*deprecated/;
 786         print;
 787     }
 788     close $OUT;
 789     unlink $temp || warn "Can't remove $temp, $!";
 790
 791     # Find what section this page is in; assume 3.
 792     my $section = 3;
 793     $section = $1 if $dirname =~ /man([1-9])/;
 794
 795     foreach ( (@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}}) ) {
 796         err($id, "Missing $_ head1 section")
 797             if $contents !~ /^=head1\s+${_}\s*$/m;
 798     }
 799 }
 800
 801 # Information database ###############################################
 802
 803 # Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
 804 my %link_map = ();
 805 # Map of names in each POD file or from "missing" files; possible values are:
 806 # If found in a POD files, "name(s)" => filename
 807 # If found in a "missing" file or external, "name(s)" => ''
 808 my %name_map = ();
 809
 810 # State of man-page names.
 811 # %state is affected by loading util/*.num and util/*.syms
 812 # Values may be one of:
 813 # 'crypto' : belongs in libcrypto (loaded from libcrypto.num)
 814 # 'ssl' : belongs in libssl (loaded from libssl.num)
 815 # 'other' : belongs in libcrypto or libssl (loaded from other.syms)
 816 # 'internal' : Internal
 817 # 'public' : Public (generic name or external documentation)
 818 # Any of these values except 'public' may be prefixed with 'missing_'
 819 # to indicate that they are known to be missing.
 820 my %state;
 821 # %missing is affected by loading util/missing*.txt.  Values may be one of:
 822 # 'crypto' : belongs in libcrypto (loaded from libcrypto.num)
 823 # 'ssl' : belongs in libssl (loaded from libssl.num)
 824 # 'other' : belongs in libcrypto or libssl (loaded from other.syms)
 825 # 'internal' : Internal
 826 my %missing;
 827
 828 # Parse libcrypto.num, etc., and return sorted list of what's there.
 829 sub loadnum ($;$) {
 830     my $file = shift;
 831     my $type = shift;
 832     my @symbols;
 833
 834     open my $IN, '<', catfile($config{sourcedir}, $file)
 835         or die "Can't open $file, $!, stopped";
 836
 837     while ( <$IN> ) {
 838         next if /^#/;
 839         next if /\bNOEXIST\b/;
 840         my @fields = split();
 841         die "Malformed line $. in $file: $_"
 842             if scalar @fields != 2 && scalar @fields != 4;
 843         $state{$fields[0].'(3)'} = $type // 'internal';
 844     }
 845     close $IN;
 846 }
 847
 848 # Load file of symbol names that we know aren't documented.
 849 sub loadmissing($;$)
 850 {
 851     my $missingfile = shift;
 852     my $type = shift;
 853
 854     open FH, catfile($config{sourcedir}, $missingfile)
 855         or die "Can't open $missingfile";
 856     while ( <FH> ) {
 857         chomp;
 858         next if /^#/;
 859         $missing{$_} = $type // 'internal';
 860     }
 861     close FH;
 862 }
 863
 864 # Check that we have consistent public / internal documentation and declaration
 865 sub checkstate () {
 866     # Collect all known names, no matter where they come from
 867     my %names = map { $_ => 1 } (keys %name_map, keys %state, keys %missing);
 868
 869     # Check section 3, i.e. functions and macros
 870     foreach ( grep { $_ =~ /\(3\)$/ } sort keys %names ) {
 871         next if ( $name_map{$_} // '') eq '' || $_ =~ /$ignored/;
 872
 873         # If a man-page isn't recorded public or if it's recorded missing
 874         # and internal, it's declared to be internal.
 875         my $declared_internal =
 876             ($state{$_} // 'internal') eq 'internal'
 877             || ($missing{$_} // '') eq 'internal';
 878         # If a man-page isn't recorded internal or if it's recorded missing
 879         # and not internal, it's declared to be public
 880         my $declared_public =
 881             ($state{$_} // 'internal') ne 'internal'
 882             || ($missing{$_} // 'internal') ne 'internal';
 883
 884         err("$_ is supposedly public but is documented as internal")
 885             if ( $declared_public && $name_map{$_} =~ /\/internal\// );
 886         err("$_ is supposedly internal but is documented as public")
 887             if ( $declared_internal && $name_map{$_} !~ /\/internal\// );
 888     }
 889 }
 890
 891 # Check for undocumented macros; ignore those in the "missing" file
 892 # and do simple check for #define in our header files.
 893 sub checkmacros {
 894     my $count = 0;
 895     my %seen;
 896
 897     foreach my $f ( files(TAGS => 'public_header') ) {
 898         # Skip some internals we don't want to document yet.
 899         my $b = basename($f);
 900         next if $b eq 'asn1.h';
 901         next if $b eq 'asn1t.h';
 902         next if $b eq 'err.h';
 903         open(IN, $f)
 904             or die "Can't open $f, $!";
 905         while ( <IN> ) {
 906             next unless /^#\s*define\s*(\S+)\(/;
 907             my $macro = "$1(3)"; # We know they're all in section 3
 908             next if defined $name_map{$macro}
 909                 || defined $missing{$macro}
 910                 || defined $seen{$macro}
 911                 || $macro =~ /$ignored/;
 912
 913             err("$f:", "macro $macro undocumented")
 914                 if $opt_d || $opt_e;
 915             $count++;
 916             $seen{$macro} = 1;
 917         }
 918         close(IN);
 919     }
 920     err("# $count macros undocumented (count is approximate)")
 921         if $count > 0;
 922 }
 923
 924 # Find out what is undocumented (filtering out the known missing ones)
 925 # and display them.
 926 sub printem ($) {
 927     my $type = shift;
 928     my $count = 0;
 929
 930     foreach my $func ( grep { $state{$_} eq $type } sort keys %state ) {
 931         next if defined $name_map{$func}
 932             || defined $missing{$func};
 933
 934         err("$type:", "function $func undocumented")
 935             if $opt_d || $opt_e;
 936         $count++;
 937     }
 938     err("# $count lib$type names are not documented")
 939         if $count > 0;
 940 }
 941
 942 # Collect all the names in a manpage.
 943 sub collectnames {
 944     my %podinfo = @_;
 945     my $filename = $podinfo{filename};
 946     $filename =~ m|man(\d)/|;
 947     my $section = $1;
 948     my $simplename = basename($filename, ".pod");
 949     my $id = "${filename}:1:";
 950     my $is_generic = $podinfo{contents} =~ /^=for openssl generic/ms;
 951
 952     unless ( grep { $simplename eq $_ } @{$podinfo{names}} ) {
 953         err($id, "$simplename not in NAME section");
 954         push @{$podinfo{names}}, $simplename;
 955     }
 956     foreach my $name ( @{$podinfo{names}} ) {
 957         next if $name eq "";
 958         err($id, "'$name' contains whitespace")
 959             if $name =~ /\s/;
 960         my $name_sec = "$name($section)";
 961         if ( !defined $name_map{$name_sec} ) {
 962             $name_map{$name_sec} = $filename;
 963             $state{$name_sec} //=
 964                 ( $filename =~ /\/internal\// ? 'internal' : 'public' )
 965                 if $is_generic;
 966         } elsif ( $filename eq $name_map{$name_sec} ) {
 967             err($id, "$name_sec duplicated in NAME section of",
 968                  $name_map{$name_sec});
 969         } elsif ( $name_map{$name_sec} ne '' ) {
 970             err($id, "$name_sec also in NAME section of",
 971                  $name_map{$name_sec});
 972         }
 973     }
 974
 975     if ( $podinfo{contents} =~ /=for openssl foreign manual (.*)\n/ ) {
 976         foreach my $f ( split / /, $1 ) {
 977             $name_map{$f} = ''; # It still exists!
 978             $state{$f} = 'public'; # We assume!
 979         }
 980     }
 981
 982     my @links =
 983         $podinfo{contents} =~ /L<
 984                               # if the link is of the form L<something|name(s)>,
 985                               # then remove 'something'.  Note that 'something'
 986                               # may contain POD codes as well...
 987                               (?:(?:[^\|]|<[^>]*>)*\|)?
 988                               # we're only interested in references that have
 989                               # a one digit section number
 990                               ([^\/>\(]+\(\d\))
 991                              /gx;
 992     $link_map{$filename} = [ @links ];
 993 }
 994
 995 # Look for L<> ("link") references that point to files that do not exist.
 996 sub checklinks {
 997     foreach my $filename ( sort keys %link_map ) {
 998         foreach my $link ( @{$link_map{$filename}} ) {
 999             err("${filename}:1:", "reference to non-existing $link")
1000                 unless defined $name_map{$link} || defined $missing{$link};
1001             err("${filename}:1:", "reference of internal $link in public documentation $filename")
1002                 if ( ( ($state{$link} // '') eq 'internal'
1003                        || ($missing{$link} // '') eq 'internal' )
1004                      && $filename !~ /\/internal\// );
1005         }
1006     }
1007 }
1008
1009 # Cipher/digests to skip if they show up as "not implemented"
1010 # because they are, via the "-*" construct.
1011 my %skips = (
1012     'aes128' => 1,
1013     'aes192' => 1,
1014     'aes256' => 1,
1015     'aria128' => 1,
1016     'aria192' => 1,
1017     'aria256' => 1,
1018     'camellia128' => 1,
1019     'camellia192' => 1,
1020     'camellia256' => 1,
1021     'des' => 1,
1022     'des3' => 1,
1023     'idea' => 1,
1024     'cipher' => 1,
1025     'digest' => 1,
1026 );
1027
1028 # Check the flags of a command and see if everything is in the manpage
1029 sub checkflags {
1030     my $cmd = shift;
1031     my $doc = shift;
1032     my %cmdopts;
1033     my %docopts;
1034     my %localskips;
1035
1036     # Get the list of options in the command.
1037     open CFH, "$openssl list --options $cmd|"
1038         or die "Can list options for $cmd, $!";
1039     while ( <CFH> ) {
1040         chop;
1041         s/ .$//;
1042         $cmdopts{$_} = 1;
1043     }
1044     close CFH;
1045
1046     # Get the list of flags from the synopsis
1047     open CFH, "<$doc"
1048         or die "Can't open $doc, $!";
1049     while ( <CFH> ) {
1050         chop;
1051         last if /DESCRIPTION/;
1052         if ( /=for openssl ifdef (.*)/ ) {
1053             foreach my $f ( split / /, $1 ) {
1054                 $localskips{$f} = 1;
1055             }
1056             next;
1057         }
1058         my $opt;
1059         if ( /\[B<-([^ >]+)/ ) {
1060             $opt = $1;
1061         } elsif ( /^B<-([^ >]+)/ ) {
1062             $opt = $1;
1063         } else {
1064             next;
1065         }
1066         $opt = $1 if $opt =~ /I<(.*)/;
1067         $docopts{$1} = 1;
1068     }
1069     close CFH;
1070
1071     # See what's in the command not the manpage.
1072     my @undocced = sort grep { !defined $docopts{$_} } keys %cmdopts;
1073     foreach ( @undocced ) {
1074         next if /-/; # Skip the -- end-of-flags marker
1075         err("$doc: undocumented option -$_");
1076     }
1077
1078     # See what's in the command not the manpage.
1079     my @unimpl = sort grep { !defined $cmdopts{$_} } keys %docopts;
1080     foreach ( @unimpl ) {
1081         next if defined $skips{$_} || defined $localskips{$_};
1082         err("$doc: $cmd does not implement -$_");
1083     }
1084 }
1085
1086 ##
1087 ##  MAIN()
1088 ##  Do the work requested by the various getopt flags.
1089 ##  The flags are parsed in alphabetical order, just because we have
1090 ##  to have *some way* of listing them.
1091 ##
1092
1093 if ( $opt_c ) {
1094     my @commands = ();
1095
1096     # Get list of commands.
1097     open FH, "$openssl list -1 -commands|"
1098         or die "Can't list commands, $!";
1099     while ( <FH> ) {
1100         chop;
1101         push @commands, $_;
1102     }
1103     close FH;
1104
1105     # See if each has a manpage.
1106     foreach my $cmd ( @commands ) {
1107         next if $cmd eq 'help' || $cmd eq 'exit';
1108         my @doc = ( grep { basename($_) eq "openssl-$cmd.pod"
1109                            # For "tsget" and "CA.pl" pod pages
1110                            || basename($_) eq "$cmd.pod" }
1111                     files(TAGS => [ 'manual', 'man1' ]) );
1112         my $num = scalar @doc;
1113         if ($num > 1) {
1114             err("$num manuals for 'openssl $cmd': ".join(", ", @doc));
1115         } elsif ($num < 1) {
1116             err("no manual for 'openssl $cmd'");
1117         } else {
1118             checkflags($cmd, @doc);
1119         }
1120     }
1121
1122     # See what help is missing.
1123     open FH, "$openssl list --missing-help |"
1124         or die "Can't list missing help, $!";
1125     while ( <FH> ) {
1126         chop;
1127         my ($cmd, $flag) = split;
1128         err("$cmd has no help for -$flag");
1129     }
1130     close FH;
1131
1132     exit $status;
1133 }
1134
1135 # Populate %state
1136 loadnum('util/libcrypto.num', 'crypto');
1137 loadnum('util/libssl.num', 'ssl');
1138 loadnum('util/other.syms', 'other');
1139 loadnum('util/other-internal.syms');
1140 if ( $opt_o ) {
1141     loadmissing('util/missingmacro111.txt', 'crypto');
1142     loadmissing('util/missingcrypto111.txt', 'crypto');
1143     loadmissing('util/missingssl111.txt', 'ssl');
1144 } elsif ( !$opt_u ) {
1145     loadmissing('util/missingmacro.txt', 'crypto');
1146     loadmissing('util/missingcrypto.txt', 'crypto');
1147     loadmissing('util/missingssl.txt', 'ssl');
1148     loadmissing('util/missingcrypto-internal.txt');
1149     loadmissing('util/missingssl-internal.txt');
1150 }
1151
1152 if ( $opt_n || $opt_l || $opt_u || $opt_v ) {
1153     my @files_to_read = ( $opt_n && @ARGV ) ? @ARGV : files(TAGS => 'manual');
1154
1155     foreach (@files_to_read) {
1156         my %podinfo = extract_pod_info($_, { debug => $debug });
1157
1158         collectnames(%podinfo)
1159             if ( $opt_l || $opt_u || $opt_v );
1160
1161         check(%podinfo)
1162             if ( $opt_n );
1163     }
1164 }
1165
1166 if ( $opt_l ) {
1167     checklinks();
1168 }
1169
1170 if ( $opt_n ) {
1171     # If not given args, check that all man1 commands are named properly.
1172     if ( scalar @ARGV == 0 ) {
1173         foreach ( files(TAGS => [ 'public_manual', 'man1' ]) ) {
1174             next if /CA.pl/ || /openssl\.pod/ || /tsget\.pod/;
1175             err("$_ doesn't start with openssl-") unless /openssl-/;
1176         }
1177     }
1178 }
1179
1180 checkstate();
1181
1182 if ( $opt_u || $opt_v) {
1183     printem('crypto');
1184     printem('ssl');
1185     checkmacros();
1186 }
1187
1188 exit $status;