util/find-doc-nits

   1 #! /usr/bin/env perl
   2 # Copyright 2002-2016 The OpenSSL Project Authors. All Rights Reserved.
   3 #
   4 # Licensed under the OpenSSL license (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($opt_h);
  24 our($opt_l);
  25 our($opt_n);
  26 our($opt_p);
  27 our($opt_s);
  28 our($opt_u);
  29
  30 sub help()
  31 {
  32     print <<EOF;
  33 Find small errors (nits) in documentation.  Options:
  34     -d Detailed list of undocumented (implies -u)
  35     -l Print bogus links
  36     -n Print nits in POD pages
  37     -s Also print missing sections in POD pages (implies -n)
  38     -p Warn if non-public name documented (implies -n)
  39     -u List undocumented functions
  40     -h Print this help message
  41 EOF
  42     exit;
  43 }
  44
  45 my $temp = '/tmp/docnits.txt';
  46 my $OUT;
  47 my %public;
  48
  49 my %mandatory_sections =
  50     ( '*'    => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
  51       1      => [ 'SYNOPSIS', 'OPTIONS' ],
  52       3      => [ 'SYNOPSIS', 'RETURN VALUES' ],
  53       5      => [ ],
  54       7      => [ ] );
  55
  56 # Cross-check functions in the NAME and SYNOPSIS section.
  57 sub name_synopsis()
  58 {
  59     my $id = shift;
  60     my $filename = shift;
  61     my $contents = shift;
  62
  63     # Get NAME section and all words in it.
  64     return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
  65     my $tmp = $1;
  66     $tmp =~ tr/\n/ /;
  67     print "$id trailing comma before - in NAME\n" if $tmp =~ /, *-/;
  68     $tmp =~ s/ -.*//g;
  69     $tmp =~ s/  */ /g;
  70     print "$id missing comma in NAME\n" if $tmp =~ /[^,] /;
  71     $tmp =~ s/,//g;
  72
  73     my $dirname = dirname($filename);
  74     my $simplename = basename($filename);
  75     $simplename =~ s/.pod$//;
  76     my $foundfilename = 0;
  77     my %foundfilenames = ();
  78     my %names;
  79     foreach my $n ( split ' ', $tmp ) {
  80         $names{$n} = 1;
  81         $foundfilename++ if $n eq $simplename;
  82         $foundfilenames{$n} = 1
  83             if -f "$dirname/$n.pod" && $n ne $simplename;
  84     }
  85     print "$id the following exist as other .pod files:\n",
  86         join(" ", sort keys %foundfilenames), "\n"
  87         if %foundfilenames;
  88     print "$id $simplename (filename) missing from NAME section\n"
  89         unless $foundfilename;
  90     foreach my $n ( keys %names ) {
  91         print "$id $n is not public\n"
  92             if $opt_p and !defined $public{$n};
  93     }
  94
  95     # Find all functions in SYNOPSIS
  96     return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
  97     my $syn = $1;
  98     foreach my $line ( split /\n+/, $syn ) {
  99         my $sym;
 100         $line =~ s/STACK_OF\([^)]+\)/int/g;
 101         $line =~ s/__declspec\([^)]+\)//;
 102         if ( $line =~ /env (\S*)=/ ) {
 103             # environment variable env NAME=...
 104             $sym = $1;
 105         } elsif ( $line =~ /typedef.*\(\*(\S+)\)\(.*/ ) {
 106             # a callback function: typedef ... (*NAME)(...
 107             $sym = $1;
 108         } elsif ( $line =~ /typedef.* (\S+);/ ) {
 109             # a simple typedef: typedef ... NAME;
 110             $sym = $1;
 111         } elsif ( $line =~ /enum (\S*) \{/ ) {
 112             # an enumeration: enum ... {
 113             $sym = $1;
 114         } elsif ( $line =~ /#define ([A-Za-z0-9_]+)/ ) {
 115             $sym = $1;
 116         } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
 117             $sym = $1;
 118         }
 119         else {
 120             next;
 121         }
 122         print "$id $sym missing from NAME section\n"
 123             unless defined $names{$sym};
 124         $names{$sym} = 2;
 125
 126         # Do some sanity checks on the prototype.
 127         print "$id prototype missing spaces around commas: $line\n"
 128             if ( $line =~ /[a-z0-9],[^ ]/ );
 129     }
 130
 131     foreach my $n ( keys %names ) {
 132         next if $names{$n} == 2;
 133         print "$id $n missing from SYNOPSIS\n";
 134     }
 135 }
 136
 137 sub check()
 138 {
 139     my $filename = shift;
 140     my $dirname = basename(dirname($filename));
 141
 142     my $contents = '';
 143     {
 144         local $/ = undef;
 145         open POD, $filename or die "Couldn't open $filename, $!";
 146         $contents = <POD>;
 147         close POD;
 148     }
 149
 150     my $id = "${filename}:1:";
 151
 152     &name_synopsis($id, $filename, $contents)
 153         unless $contents =~ /=for comment generic/
 154             or $filename =~ m@man[157]/@;
 155
 156     print "$id doesn't start with =pod\n"
 157         if $contents !~ /^=pod/;
 158     print "$id doesn't end with =cut\n"
 159         if $contents !~ /=cut\n$/;
 160     print "$id more than one cut line.\n"
 161         if $contents =~ /=cut.*=cut/ms;
 162     print "$id missing copyright\n"
 163         if $contents !~ /Copyright .* The OpenSSL Project Authors/;
 164     print "$id copyright not last\n"
 165         if $contents =~ /head1 COPYRIGHT.*=head/ms;
 166     print "$id head2 in All uppercase\n"
 167         if $contents =~ /head2\s+[A-Z ]+\n/;
 168     print "$id extra space after head\n"
 169         if $contents =~ /=head\d\s\s+/;
 170     print "$id period in NAME section\n"
 171         if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
 172     print "$id POD markup in NAME section\n"
 173         if $contents =~ /=head1 NAME.*[<>].*=head1 SYNOPSIS/ms;
 174     print "$id Duplicate $1 in L<>\n"
 175         if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
 176     print "$id Bad =over $1\n"
 177         if $contents =~ /=over([^ ][^24])/;
 178     print "$id Possible version style issue\n"
 179         if $contents =~ /OpenSSL version [019]/;
 180
 181     if ( $contents !~ /=for comment multiple includes/ ) {
 182         # Look for multiple consecutive openssl #include lines
 183         # (non-consecutive lines are okay; see man3/MD5.pod).
 184         if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
 185             my $count = 0;
 186             foreach my $line ( split /\n+/, $1 ) {
 187                 if ( $line =~ m@include <openssl/@ ) {
 188                     print "$id has multiple includes\n" if ++$count == 2;
 189                 } else {
 190                     $count = 0;
 191                 }
 192             }
 193         }
 194     }
 195
 196     open my $OUT, '>', $temp
 197         or die "Can't open $temp, $!";
 198     podchecker($filename, $OUT);
 199     close $OUT;
 200     open $OUT, '<', $temp
 201         or die "Can't read $temp, $!";
 202     while ( <$OUT> ) {
 203         next if /\(section\) in.*deprecated/;
 204         print;
 205     }
 206     close $OUT;
 207     unlink $temp || warn "Can't remove $temp, $!";
 208
 209     # Find what section this page is in; assume 3.
 210     my $section = 3;
 211     $section = $1 if $dirname =~ /man([1-9])/;
 212
 213     foreach ((@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}})) {
 214         # Skip "return values" if not -s
 215         next if $_ eq 'RETURN VALUES' and not $opt_s;
 216         print "$id: missing $_ head1 section\n"
 217             if $contents !~ /^=head1\s+${_}\s*$/m;
 218     }
 219 }
 220
 221 my %dups;
 222
 223 sub parsenum()
 224 {
 225     my $file = shift;
 226     my @apis;
 227
 228     open my $IN, '<', $file
 229         or die "Can't open $file, $!, stopped";
 230
 231     while ( <$IN> ) {
 232         next if /^#/;
 233         next if /\bNOEXIST\b/;
 234         next if /\bEXPORT_VAR_AS_FUNC\b/;
 235         my @fields = split();
 236         die "Malformed line $_"
 237             if scalar @fields != 2 && scalar @fields != 4;
 238         push @apis, $fields[0];
 239     }
 240
 241     close $IN;
 242
 243     print "# Found ", scalar(@apis), " in $file\n" unless $opt_p;
 244     return sort @apis;
 245 }
 246
 247 sub getdocced()
 248 {
 249     my $dir = shift;
 250     my %return;
 251
 252     foreach my $pod ( glob("$dir/*.pod") ) {
 253         my %podinfo = extract_pod_info($pod);
 254         foreach my $n ( @{$podinfo{names}} ) {
 255             $return{$n} = $pod;
 256             print "# Duplicate $n in $pod and $dups{$n}\n"
 257                 if defined $dups{$n} && $dups{$n} ne $pod;
 258             $dups{$n} = $pod;
 259         }
 260     }
 261
 262     return %return;
 263 }
 264
 265 my %docced;
 266
 267 sub checkmacros()
 268 {
 269     my $count = 0;
 270
 271     print "# Checking macros (approximate)\n";
 272     foreach my $f ( glob('include/openssl/*.h') ) {
 273         # Skip some internals we don't want to document yet.
 274         next if $f eq 'include/openssl/asn1.h';
 275         next if $f eq 'include/openssl/asn1t.h';
 276         next if $f eq 'include/openssl/err.h';
 277         open(IN, $f) || die "Can't open $f, $!";
 278         while ( <IN> ) {
 279             next unless /^#\s*define\s*(\S+)\(/;
 280             my $macro = $1;
 281             next if $docced{$macro};
 282             next if $macro =~ /i2d_/
 283                 || $macro =~ /d2i_/
 284                 || $macro =~ /DEPRECATEDIN/
 285                 || $macro =~ /IMPLEMENT_/
 286                 || $macro =~ /DECLARE_/;
 287             print "$f:$macro\n" if $opt_d;
 288             $count++;
 289         }
 290         close(IN);
 291     }
 292     print "# Found $count macros missing (not all should be documented)\n"
 293 }
 294
 295 sub printem()
 296 {
 297     my $libname = shift;
 298     my $numfile = shift;
 299     my $count = 0;
 300
 301     foreach my $func ( &parsenum($numfile) ) {
 302         next if $docced{$func};
 303
 304         # Skip ASN1 utilities
 305         next if $func =~ /^ASN1_/;
 306
 307         print "$libname:$func\n" if $opt_d;
 308         $count++;
 309     }
 310     print "# Found $count missing from $numfile\n\n";
 311 }
 312
 313
 314 # Collection of links in each POD file.
 315 # filename => [ "foo(1)", "bar(3)", ... ]
 316 my %link_collection = ();
 317 # Collection of names in each POD file.
 318 # "name(s)" => filename
 319 my %name_collection = ();
 320
 321 sub collectnames {
 322     my $filename = shift;
 323     $filename =~ m|man(\d)/|;
 324     my $section = $1;
 325     my $simplename = basename($filename, ".pod");
 326     my $id = "${filename}:1:";
 327
 328     my $contents = '';
 329     {
 330         local $/ = undef;
 331         open POD, $filename or die "Couldn't open $filename, $!";
 332         $contents = <POD>;
 333         close POD;
 334     }
 335
 336     $contents =~ /=head1 NAME([^=]*)=head1 /ms;
 337     my $tmp = $1;
 338     unless (defined $tmp) {
 339         print "$id weird name section\n";
 340         return;
 341     }
 342     $tmp =~ tr/\n/ /;
 343     $tmp =~ s/-.*//g;
 344
 345     my @names = map { s/\s+//g; $_ } split(/,/, $tmp);
 346     unless (grep { $simplename eq $_ } @names) {
 347         print "$id missing $simplename\n";
 348         push @names, $simplename;
 349     }
 350     foreach my $name (@names) {
 351         next if $name eq "";
 352         my $name_sec = "$name($section)";
 353         if (! exists $name_collection{$name_sec}) {
 354             $name_collection{$name_sec} = $filename;
 355         } else { #elsif ($filename ne $name_collection{$name_sec}) {
 356             print "$id $name_sec also in $name_collection{$name_sec}\n";
 357         }
 358     }
 359
 360     my @foreign_names =
 361         map { map { s/\s+//g; $_ } split(/,/, $_) }
 362         $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
 363     foreach (@foreign_names) {
 364         $name_collection{$_} = undef; # It still exists!
 365     }
 366
 367     my @links = $contents =~ /L<
 368                               # if the link is of the form L<something|name(s)>,
 369                               # then remove 'something'.  Note that 'something'
 370                               # may contain POD codes as well...
 371                               (?:(?:[^\|]|<[^>]*>)*\|)?
 372                               # we're only interested in referenses that have
 373                               # a one digit section number
 374                               ([^\/>\(]+\(\d\))
 375                              /gx;
 376     $link_collection{$filename} = [ @links ];
 377 }
 378
 379 sub checklinks {
 380     foreach my $filename (sort keys %link_collection) {
 381         foreach my $link (@{$link_collection{$filename}}) {
 382             print "${filename}:1: reference to non-existing $link\n"
 383                 unless exists $name_collection{$link};
 384         }
 385     }
 386 }
 387
 388 sub publicize() {
 389     foreach my $name ( &parsenum('util/libcrypto.num') ) {
 390         $public{$name} = 1;
 391     }
 392     foreach my $name ( &parsenum('util/libssl.num') ) {
 393         $public{$name} = 1;
 394     }
 395     foreach my $name ( &parsenum('util/private.num') ) {
 396         $public{$name} = 1;
 397     }
 398 }
 399
 400 getopts('dlnsphu');
 401
 402 &help() if $opt_h;
 403 $opt_n = 1 if $opt_s or $opt_p;
 404 $opt_u = 1 if $opt_d;
 405
 406 die "Need one of -[dlnspu] flags.\n"
 407     unless $opt_l or $opt_n or $opt_u;
 408
 409
 410 if ( $opt_n ) {
 411     &publicize() if $opt_p;
 412     foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
 413         &check($_);
 414     }
 415 }
 416
 417 if ( $opt_l ) {
 418     foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
 419         collectnames($_);
 420     }
 421     checklinks();
 422 }
 423
 424 if ( $opt_u ) {
 425     my %temp = &getdocced('doc/man3');
 426     foreach ( keys %temp ) {
 427         $docced{$_} = $temp{$_};
 428     }
 429     &printem('crypto', 'util/libcrypto.num');
 430     &printem('ssl', 'util/libssl.num');
 431     &checkmacros();
 432 }
 433
 434 exit;