util/find-doc-nits

   1 #! /usr/bin/env perl
   2 # Copyright 2002-2019 The OpenSSL Project Authors. All Rights Reserved.
   3 #
   4 # Licensed under the Apache License 2.0 (the "License").  You may not use
   5 # this file except in compliance with the License.  You can obtain a copy
   6 # in the file LICENSE in the source distribution or at
   7 # https://www.openssl.org/source/license.html
   8
   9
  10 require 5.10.0;
  11 use warnings;
  12 use strict;
  13 use Pod::Checker;
  14 use File::Find;
  15 use File::Basename;
  16 use File::Spec::Functions;
  17 use Getopt::Std;
  18 use lib catdir(dirname($0), "perl");
  19 use OpenSSL::Util::Pod;
  20
  21 # Options.
  22 our($opt_d);
  23 our($opt_e);
  24 our($opt_s);
  25 our($opt_o);
  26 our($opt_h);
  27 our($opt_l);
  28 our($opt_n);
  29 our($opt_p);
  30 our($opt_u);
  31 our($opt_v);
  32 our($opt_c);
  33
  34 sub help()
  35 {
  36     print <<EOF;
  37 Find small errors (nits) in documentation.  Options:
  38     -d Detailed list of undocumented (implies -u)
  39     -e Detailed list of new undocumented (implies -v)
  40     -s Same as -e except no output is generated if nothing is undocumented
  41     -o Causes -e/-v to count symbols added since 1.1.1 as new (implies -v)
  42     -l Print bogus links
  43     -n Print nits in POD pages
  44     -p Warn if non-public name documented (implies -n)
  45     -u Count undocumented functions
  46     -v Count new undocumented functions
  47     -h Print this help message
  48     -c List undocumented commands and options
  49 EOF
  50     exit;
  51 }
  52
  53 my $temp = '/tmp/docnits.txt';
  54 my $OUT;
  55 my %public;
  56
  57 my %mandatory_sections =
  58     ( '*'    => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
  59       1      => [ 'SYNOPSIS', 'OPTIONS' ],
  60       3      => [ 'SYNOPSIS', 'RETURN VALUES' ],
  61       5      => [ ],
  62       7      => [ ] );
  63
  64 # Cross-check functions in the NAME and SYNOPSIS section.
  65 sub name_synopsis()
  66 {
  67     my $id = shift;
  68     my $filename = shift;
  69     my $contents = shift;
  70
  71     # Get NAME section and all words in it.
  72     return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
  73     my $tmp = $1;
  74     $tmp =~ tr/\n/ /;
  75     print "$id trailing comma before - in NAME\n" if $tmp =~ /, *-/;
  76     $tmp =~ s/ -.*//g;
  77     print "$id POD markup among the names in NAME\n" if $tmp =~ /[<>]/;
  78     $tmp =~ s/  */ /g;
  79     print "$id missing comma in NAME\n" if $tmp =~ /[^,] /;
  80
  81     my $dirname = dirname($filename);
  82     my $simplename = basename(basename($filename, ".in"), ".pod");
  83     my $foundfilename = 0;
  84     my %foundfilenames = ();
  85     my %names;
  86     foreach my $n ( split ',', $tmp ) {
  87         $n =~ s/^\s+//;
  88         $n =~ s/\s+$//;
  89         print "$id the name '$n' contains white-space\n"
  90             if $n =~ /\s/;
  91         $names{$n} = 1;
  92         $foundfilename++ if $n eq $simplename;
  93         $foundfilenames{$n} = 1
  94             if ((-f "$dirname/$n.pod.in" || -f "$dirname/$n.pod")
  95                 && $n ne $simplename);
  96     }
  97     print "$id the following exist as other .pod or .pod.in files:\n",
  98         join(" ", sort keys %foundfilenames), "\n"
  99         if %foundfilenames;
 100     print "$id $simplename (filename) missing from NAME section\n"
 101         unless $foundfilename;
 102     foreach my $n ( keys %names ) {
 103         print "$id $n is not public\n"
 104             if $opt_p and !defined $public{$n};
 105     }
 106
 107     # Find all functions in SYNOPSIS
 108     return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
 109     my $syn = $1;
 110     foreach my $line ( split /\n+/, $syn ) {
 111         next unless $line =~ /^\s/;
 112         my $sym;
 113         $line =~ s/STACK_OF\([^)]+\)/int/g;
 114         $line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
 115         $line =~ s/__declspec\([^)]+\)//;
 116         if ( $line =~ /env (\S*)=/ ) {
 117             # environment variable env NAME=...
 118             $sym = $1;
 119         } elsif ( $line =~ /typedef.*\(\*(\S+)\)\(.*/ ) {
 120             # a callback function pointer: typedef ... (*NAME)(...
 121             $sym = $1;
 122         } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
 123             # a callback function signature: typedef ... NAME(...
 124             $sym = $1;
 125         } elsif ( $line =~ /typedef.* (\S+);/ ) {
 126             # a simple typedef: typedef ... NAME;
 127             $sym = $1;
 128         } elsif ( $line =~ /enum (\S*) \{/ ) {
 129             # an enumeration: enum ... {
 130             $sym = $1;
 131         } elsif ( $line =~ /#(?:define|undef) ([A-Za-z0-9_]+)/ ) {
 132             $sym = $1;
 133         } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
 134             $sym = $1;
 135         }
 136         else {
 137             next;
 138         }
 139         print "$id $sym missing from NAME section\n"
 140             unless defined $names{$sym};
 141         $names{$sym} = 2;
 142
 143         # Do some sanity checks on the prototype.
 144         print "$id prototype missing spaces around commas: $line\n"
 145             if ( $line =~ /[a-z0-9],[^ ]/ );
 146     }
 147
 148     foreach my $n ( keys %names ) {
 149         next if $names{$n} == 2;
 150         print "$id $n missing from SYNOPSIS\n";
 151     }
 152 }
 153
 154 # Check if SECTION is located before BEFORE
 155 sub check_section_location()
 156 {
 157     my $filename = shift;
 158     my $contents = shift;
 159     my $section = shift;
 160     my $before = shift;
 161
 162     return unless $contents =~ /=head1 $section/
 163         and $contents =~ /=head1 $before/;
 164     print "$filename: $section should be placed before $before section\n"
 165         if $contents =~ /=head1 $before.*=head1 $section/ms;
 166 }
 167
 168 sub check()
 169 {
 170     my $filename = shift;
 171     my $dirname = basename(dirname($filename));
 172
 173     my $contents = '';
 174     {
 175         local $/ = undef;
 176         open POD, $filename or die "Couldn't open $filename, $!";
 177         $contents = <POD>;
 178         close POD;
 179     }
 180
 181     # Check if EXAMPLES is located after RETURN VALUES section.
 182     &check_section_location($filename, $contents, "RETURN VALUES", "EXAMPLES") if $filename =~ m|man3/|;
 183     # Check if HISTORY is located after SEE ALSO
 184     &check_section_location($filename, $contents, "SEE ALSO", "HISTORY") if $filename =~ m|man3/|;
 185     # Check if SEE ALSO is located after EXAMPLES
 186     &check_section_location($filename, $contents, "EXAMPLES", "SEE ALSO") if $filename =~ m|man3/|;
 187
 188     my $id = "${filename}:1:";
 189
 190     &name_synopsis($id, $filename, $contents)
 191         unless $contents =~ /=for comment generic/
 192             or $filename =~ m@man[157]/@;
 193
 194     print "$id doesn't start with =pod\n"
 195         if $contents !~ /^=pod/;
 196     print "$id doesn't end with =cut\n"
 197         if $contents !~ /=cut\n$/;
 198     print "$id more than one cut line.\n"
 199         if $contents =~ /=cut.*=cut/ms;
 200     print "$id missing copyright\n"
 201         if $contents !~ /Copyright .* The OpenSSL Project Authors/;
 202     print "$id copyright not last\n"
 203         if $contents =~ /head1 COPYRIGHT.*=head/ms;
 204     print "$id head2 in All uppercase\n"
 205         if $contents =~ /head2\s+[A-Z ]+\n/;
 206     print "$id extra space after head\n"
 207         if $contents =~ /=head\d\s\s+/;
 208     print "$id period in NAME section\n"
 209         if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
 210     print "$id Duplicate $1 in L<>\n"
 211         if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
 212     print "$id Bad =over $1\n"
 213         if $contents =~ /=over([^ ][^24])/;
 214     print "$id Possible version style issue\n"
 215         if $contents =~ /OpenSSL version [019]/;
 216
 217     if ( $contents !~ /=for comment multiple includes/ ) {
 218         # Look for multiple consecutive openssl #include lines
 219         # (non-consecutive lines are okay; see man3/MD5.pod).
 220         if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
 221             my $count = 0;
 222             foreach my $line ( split /\n+/, $1 ) {
 223                 if ( $line =~ m@include <openssl/@ ) {
 224                     print "$id has multiple includes\n" if ++$count == 2;
 225                 } else {
 226                     $count = 0;
 227                 }
 228             }
 229         }
 230     }
 231
 232     open my $OUT, '>', $temp
 233         or die "Can't open $temp, $!";
 234     podchecker($filename, $OUT);
 235     close $OUT;
 236     open $OUT, '<', $temp
 237         or die "Can't read $temp, $!";
 238     while ( <$OUT> ) {
 239         next if /\(section\) in.*deprecated/;
 240         print;
 241     }
 242     close $OUT;
 243     unlink $temp || warn "Can't remove $temp, $!";
 244
 245     # Find what section this page is in; assume 3.
 246     my $section = 3;
 247     $section = $1 if $dirname =~ /man([1-9])/;
 248
 249     foreach ((@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}})) {
 250         # Skip "return values" if not -s
 251         print "$id: missing $_ head1 section\n"
 252             if $contents !~ /^=head1\s+${_}\s*$/m;
 253     }
 254 }
 255
 256 my %dups;
 257
 258 sub parsenum()
 259 {
 260     my $file = shift;
 261     my @apis;
 262
 263     open my $IN, '<', $file
 264         or die "Can't open $file, $!, stopped";
 265
 266     while ( <$IN> ) {
 267         next if /^#/;
 268         next if /\bNOEXIST\b/;
 269         my @fields = split();
 270         die "Malformed line $_"
 271             if scalar @fields != 2 && scalar @fields != 4;
 272         push @apis, $fields[0];
 273     }
 274
 275     close $IN;
 276
 277     print "# Found ", scalar(@apis), " in $file\n" unless $opt_p;
 278     return sort @apis;
 279 }
 280
 281 sub getdocced
 282 {
 283     my $dir = shift;
 284     my %return;
 285
 286     foreach my $pod ( glob("$dir/*.pod"), glob("$dir/*.pod.in") ) {
 287         my %podinfo = extract_pod_info($pod);
 288         foreach my $n ( @{$podinfo{names}} ) {
 289             $return{$n} = $pod;
 290             print "# Duplicate $n in $pod and $dups{$n}\n"
 291                 if defined $dups{$n} && $dups{$n} ne $pod;
 292             $dups{$n} = $pod;
 293         }
 294     }
 295
 296     return %return;
 297 }
 298
 299 my %docced;
 300
 301 sub loadmissing($)
 302 {
 303     my $missingfile = shift;
 304     my @missing;
 305
 306     open FH, $missingfile
 307         || die "Can't open $missingfile";
 308     while ( <FH> ) {
 309         chomp;
 310         next if /^#/;
 311         push @missing, $_;
 312     }
 313     close FH;
 314
 315     return @missing;
 316 }
 317
 318 sub checkmacros()
 319 {
 320     my $count = 0;
 321     my %seen;
 322     my @missing;
 323
 324     if ($opt_o) {
 325         @missing = loadmissing('util/missingmacro111.txt');
 326     } elsif ($opt_v) {
 327         @missing = loadmissing('util/missingmacro.txt');
 328     }
 329
 330     print "# Checking macros (approximate)\n" if !$opt_s;
 331     foreach my $f ( glob('include/openssl/*.h') ) {
 332         # Skip some internals we don't want to document yet.
 333         next if $f eq 'include/openssl/asn1.h';
 334         next if $f eq 'include/openssl/asn1t.h';
 335         next if $f eq 'include/openssl/err.h';
 336         open(IN, $f) || die "Can't open $f, $!";
 337         while ( <IN> ) {
 338             next unless /^#\s*define\s*(\S+)\(/;
 339             my $macro = $1;
 340             next if $docced{$macro} || defined $seen{$macro};
 341             next if $macro =~ /i2d_/
 342                 || $macro =~ /d2i_/
 343                 || $macro =~ /DEPRECATEDIN/
 344                 || $macro =~ /IMPLEMENT_/
 345                 || $macro =~ /DECLARE_/;
 346
 347             # Skip macros known to be missing
 348             next if $opt_v && grep( /^$macro$/, @missing);
 349
 350             print "$f:$macro\n" if $opt_d || $opt_e;
 351             $count++;
 352             $seen{$macro} = 1;
 353         }
 354         close(IN);
 355     }
 356     print "# Found $count macros missing\n" if !$opt_s || $count > 0;
 357 }
 358
 359 sub printem()
 360 {
 361     my $libname = shift;
 362     my $numfile = shift;
 363     my $missingfile = shift;
 364     my $count = 0;
 365     my %seen;
 366
 367     my @missing = loadmissing($missingfile) if ($opt_v);
 368
 369     foreach my $func ( &parsenum($numfile) ) {
 370         next if $docced{$func} || defined $seen{$func};
 371
 372         # Skip ASN1 utilities
 373         next if $func =~ /^ASN1_/;
 374
 375         # Skip functions known to be missing
 376         next if $opt_v && grep( /^$func$/, @missing);
 377
 378         print "$libname:$func\n" if $opt_d || $opt_e;
 379         $count++;
 380         $seen{$func} = 1;
 381     }
 382     print "# Found $count missing from $numfile\n\n" if !$opt_s || $count > 0;
 383 }
 384
 385
 386 # Collection of links in each POD file.
 387 # filename => [ "foo(1)", "bar(3)", ... ]
 388 my %link_collection = ();
 389 # Collection of names in each POD file.
 390 # "name(s)" => filename
 391 my %name_collection = ();
 392
 393 sub collectnames {
 394     my $filename = shift;
 395     $filename =~ m|man(\d)/|;
 396     my $section = $1;
 397     my $simplename = basename(basename($filename, ".in"), ".pod");
 398     my $id = "${filename}:1:";
 399
 400     my $contents = '';
 401     {
 402         local $/ = undef;
 403         open POD, $filename or die "Couldn't open $filename, $!";
 404         $contents = <POD>;
 405         close POD;
 406     }
 407
 408     $contents =~ /=head1 NAME([^=]*)=head1 /ms;
 409     my $tmp = $1;
 410     unless (defined $tmp) {
 411         print "$id weird name section\n";
 412         return;
 413     }
 414     $tmp =~ tr/\n/ /;
 415     $tmp =~ s/ -.*//g;
 416
 417     my @names =
 418         map { s|/|-|g; $_ }              # Treat slash as dash
 419         map { s/^\s+//g; s/\s+$//g; $_ } # Trim prefix and suffix blanks
 420         split(/,/, $tmp);
 421     unless (grep { $simplename eq $_ } @names) {
 422         print "$id missing $simplename\n";
 423         push @names, $simplename;
 424     }
 425     foreach my $name (@names) {
 426         next if $name eq "";
 427         if ($name =~ /\s/) {
 428             print "$id '$name' contains white space\n";
 429         }
 430         my $name_sec = "$name($section)";
 431         if (! exists $name_collection{$name_sec}) {
 432             $name_collection{$name_sec} = $filename;
 433         } elsif ($filename eq $name_collection{$name_sec}) {
 434             print "$id $name_sec repeated in NAME section of $name_collection{$name_sec}\n"
 435         } else {
 436             print "$id $name_sec also in NAME section of $name_collection{$name_sec}\n";
 437         }
 438     }
 439
 440     my @foreign_names =
 441         map { map { s/\s+//g; $_ } split(/,/, $_) }
 442         $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
 443     foreach (@foreign_names) {
 444         $name_collection{$_} = undef; # It still exists!
 445     }
 446
 447     my @links = $contents =~ /L<
 448                               # if the link is of the form L<something|name(s)>,
 449                               # then remove 'something'.  Note that 'something'
 450                               # may contain POD codes as well...
 451                               (?:(?:[^\|]|<[^>]*>)*\|)?
 452                               # we're only interested in references that have
 453                               # a one digit section number
 454                               ([^\/>\(]+\(\d\))
 455                              /gx;
 456     $link_collection{$filename} = [ @links ];
 457 }
 458
 459 sub checklinks {
 460     foreach my $filename (sort keys %link_collection) {
 461         foreach my $link (@{$link_collection{$filename}}) {
 462             print "${filename}:1: reference to non-existing $link\n"
 463                 unless exists $name_collection{$link};
 464         }
 465     }
 466 }
 467
 468 sub publicize() {
 469     foreach my $name ( &parsenum('util/libcrypto.num') ) {
 470         $public{$name} = 1;
 471     }
 472     foreach my $name ( &parsenum('util/libssl.num') ) {
 473         $public{$name} = 1;
 474     }
 475     foreach my $name ( &parsenum('util/private.num') ) {
 476         $public{$name} = 1;
 477     }
 478 }
 479
 480 my %skips = (
 481     'aes128' => 1,
 482     'aes192' => 1,
 483     'aes256' => 1,
 484     'aria128' => 1,
 485     'aria192' => 1,
 486     'aria256' => 1,
 487     'camellia128' => 1,
 488     'camellia192' => 1,
 489     'camellia256' => 1,
 490     'des' => 1,
 491     'des3' => 1,
 492     'idea' => 1,
 493     '[cipher]' => 1,
 494     '[digest]' => 1,
 495 );
 496
 497 sub checkflags() {
 498     my $cmd = shift;
 499     my %cmdopts;
 500     my %docopts;
 501     my $ok = 1;
 502
 503     # Get the list of options in the command.
 504     open CFH, "./apps/openssl list --options $cmd|"
 505         || die "Can list options for $cmd, $!";
 506     while ( <CFH> ) {
 507         chop;
 508         s/ .$//;
 509         $cmdopts{$_} = 1;
 510     }
 511     close CFH;
 512
 513     # Get the list of flags from the synopsis
 514     open CFH, "<doc/man1/$cmd.pod"
 515         || die "Can't open $cmd.pod, $!";
 516     while ( <CFH> ) {
 517         chop;
 518         last if /DESCRIPTION/;
 519         next unless /\[B<-([^ >]+)/;
 520         $docopts{$1} = 1;
 521     }
 522     close CFH;
 523
 524     # See what's in the command not the manpage.
 525     my @undocced = ();
 526     foreach my $k ( keys %cmdopts ) {
 527         push @undocced, $k unless $docopts{$k};
 528     }
 529     if ( scalar @undocced > 0 ) {
 530         $ok = 0;
 531         foreach ( @undocced ) {
 532             print "doc/man1/$cmd.pod: Missing -$_\n";
 533         }
 534     }
 535
 536     # See what's in the command not the manpage.
 537     my @unimpl = ();
 538     foreach my $k ( keys %docopts ) {
 539         push @unimpl, $k unless $cmdopts{$k};
 540     }
 541     if ( scalar @unimpl > 0 ) {
 542         $ok = 0;
 543         foreach ( @unimpl ) {
 544             next if defined $skips{$_};
 545             print "doc/man1/$cmd.pod: Not implemented -$_\n";
 546         }
 547     }
 548
 549     return $ok;
 550 }
 551
 552 getopts('cdesolnphuv');
 553
 554 &help() if $opt_h;
 555
 556 $opt_n = 1 if $opt_p;
 557 $opt_u = 1 if $opt_d;
 558 $opt_e = 1 if $opt_s;
 559 $opt_v = 1 if $opt_o || $opt_e;
 560
 561 die "Cannot use both -u and -v" if $opt_u && $opt_v;
 562 die "Cannot use both -d and -e" if $opt_d && $opt_e;
 563
 564 # We only need to check c, l, n, u and v.
 565 # Options d, e, s, o and p imply one of the above.
 566 die "Need one of -[cdesolnpuv] flags.\n"
 567     unless $opt_c or $opt_l or $opt_n or $opt_u or $opt_v;
 568
 569 if ( $opt_c ) {
 570     my $ok = 1;
 571     my @commands = ();
 572
 573     # Get list of commands.
 574     open FH, "./apps/openssl list -1 -commands|"
 575         || die "Can't list commands, $!";
 576     while ( <FH> ) {
 577         chop;
 578         push @commands, $_;
 579     }
 580     close FH;
 581
 582     # See if each has a manpage.
 583     foreach ( @commands ) {
 584         next if $_ eq 'help' || $_ eq 'exit';
 585         if ( ! -f "doc/man1/$_.pod" ) {
 586             print "doc/man1/$_.pod does not exist\n";
 587             $ok = 0;
 588         } else {
 589             $ok = 0 if not &checkflags($_);
 590         }
 591     }
 592
 593     # See what help is missing.
 594     open FH, "./apps/openssl list --missing-help |"
 595         || die "Can't list missing help, $!";
 596     while ( <FH> ) {
 597         chop;
 598         my ($cmd, $flag) = split;
 599         print "$cmd has no help for -$flag\n";
 600         $ok = 0;
 601     }
 602     close FH;
 603
 604     exit 1 if not $ok;
 605 }
 606
 607 if ( $opt_l ) {
 608     foreach (@ARGV ? @ARGV : (glob('doc/*/*.pod'), glob('doc/*/*.pod.in'),
 609                               glob('doc/internal/*/*.pod'))) {
 610         collectnames($_);
 611     }
 612     checklinks();
 613 }
 614
 615 if ( $opt_n ) {
 616     &publicize() if $opt_p;
 617     foreach (@ARGV ? @ARGV : (glob('doc/*/*.pod'), glob('doc/*/*.pod.in'))) {
 618         &check($_);
 619     }
 620     {
 621         local $opt_p = undef;
 622         foreach (@ARGV ? @ARGV : glob('doc/internal/*/*.pod')) {
 623             &check($_);
 624         }
 625     }
 626 }
 627
 628 if ( $opt_u || $opt_v) {
 629     my %temp = getdocced('doc/man3');
 630     foreach ( keys %temp ) {
 631         $docced{$_} = $temp{$_};
 632     }
 633     if ($opt_o) {
 634         &printem('crypto', 'util/libcrypto.num', 'util/missingcrypto111.txt');
 635         &printem('ssl', 'util/libssl.num', 'util/missingssl111.txt');
 636     } else {
 637         &printem('crypto', 'util/libcrypto.num', 'util/missingcrypto.txt');
 638         &printem('ssl', 'util/libssl.num', 'util/missingssl.txt');
 639     }
 640     &checkmacros();
 641 }
 642
 643 exit;