Home | History | Annotate | Line # | Download | only in util
find-doc-nits revision 1.1.1.3.2.1 
      1 #! /usr/bin/env perl
      2 # Copyright 2002-2019 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($opt_u);
     28 our($opt_c);
     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     -p Warn if non-public name documented (implies -n)
     38     -u List undocumented functions
     39     -h Print this help message
     40     -c List undocumented commands and options
     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 pointer: typedef ... (*NAME)(...
    107             $sym = $1;
    108         } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
    109             # a callback function signature: typedef ... NAME(...
    110             $sym = $1;
    111         } elsif ( $line =~ /typedef.* (\S+);/ ) {
    112             # a simple typedef: typedef ... NAME;
    113             $sym = $1;
    114         } elsif ( $line =~ /enum (\S*) \{/ ) {
    115             # an enumeration: enum ... {
    116             $sym = $1;
    117         } elsif ( $line =~ /#define ([A-Za-z0-9_]+)/ ) {
    118             $sym = $1;
    119         } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
    120             $sym = $1;
    121         }
    122         else {
    123             next;
    124         }
    125         print "$id $sym missing from NAME section\n"
    126             unless defined $names{$sym};
    127         $names{$sym} = 2;
    128 
    129         # Do some sanity checks on the prototype.
    130         print "$id prototype missing spaces around commas: $line\n"
    131             if ( $line =~ /[a-z0-9],[^ ]/ );
    132     }
    133 
    134     foreach my $n ( keys %names ) {
    135         next if $names{$n} == 2;
    136         print "$id $n missing from SYNOPSIS\n";
    137     }
    138 }
    139 
    140 # Check if SECTION ($3) is located before BEFORE ($4)
    141 sub check_section_location()
    142 {
    143     my $id = shift;
    144     my $contents = shift;
    145     my $section = shift;
    146     my $before = shift;
    147 
    148     return
    149         unless $contents =~ /=head1 $section/ and $contents =~ /=head1 $before/;
    150     print "$id $section should be placed before $before section\n"
    151         if $contents =~ /=head1 $before.*=head1 $section/ms;
    152 }
    153 
    154 sub check()
    155 {
    156     my $filename = shift;
    157     my $dirname = basename(dirname($filename));
    158 
    159     my $contents = '';
    160     {
    161         local $/ = undef;
    162         open POD, $filename or die "Couldn't open $filename, $!";
    163         $contents = <POD>;
    164         close POD;
    165     }
    166 
    167     my $id = "${filename}:1:";
    168 
    169     # Check ordering of some sections in man3
    170     if ( $filename =~ m|man3/| ) {
    171         &check_section_location($id, $contents, "RETURN VALUES", "EXAMPLES");
    172         &check_section_location($id, $contents, "SEE ALSO", "HISTORY");
    173         &check_section_location($id, $contents, "EXAMPLES", "SEE ALSO");
    174     }
    175 
    176     &name_synopsis($id, $filename, $contents)
    177         unless $contents =~ /=for comment generic/
    178             or $filename =~ m@man[157]/@;
    179 
    180     print "$id doesn't start with =pod\n"
    181         if $contents !~ /^=pod/;
    182     print "$id doesn't end with =cut\n"
    183         if $contents !~ /=cut\n$/;
    184     print "$id more than one cut line.\n"
    185         if $contents =~ /=cut.*=cut/ms;
    186     print "$id EXAMPLE not EXAMPLES section.\n"
    187         if $contents =~ /=head1 EXAMPLE[^S]/;
    188     print "$id WARNING not WARNINGS section.\n"
    189         if $contents =~ /=head1 WARNING[^S]/;
    190     print "$id missing copyright\n"
    191         if $contents !~ /Copyright .* The OpenSSL Project Authors/;
    192     print "$id copyright not last\n"
    193         if $contents =~ /head1 COPYRIGHT.*=head/ms;
    194     print "$id head2 in All uppercase\n"
    195         if $contents =~ /head2\s+[A-Z ]+\n/;
    196     print "$id extra space after head\n"
    197         if $contents =~ /=head\d\s\s+/;
    198     print "$id period in NAME section\n"
    199         if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
    200     print "$id POD markup in NAME section\n"
    201         if $contents =~ /=head1 NAME.*[<>].*=head1 SYNOPSIS/ms;
    202     print "$id Duplicate $1 in L<>\n"
    203         if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
    204     print "$id Bad =over $1\n"
    205         if $contents =~ /=over([^ ][^24])/;
    206     print "$id Possible version style issue\n"
    207         if $contents =~ /OpenSSL version [019]/;
    208 
    209     if ( $contents !~ /=for comment multiple includes/ ) {
    210         # Look for multiple consecutive openssl #include lines
    211         # (non-consecutive lines are okay; see man3/MD5.pod).
    212         if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
    213             my $count = 0;
    214             foreach my $line ( split /\n+/, $1 ) {
    215                 if ( $line =~ m@include <openssl/@ ) {
    216                     print "$id has multiple includes\n" if ++$count == 2;
    217                 } else {
    218                     $count = 0;
    219                 }
    220             }
    221         }
    222     }
    223 
    224     open my $OUT, '>', $temp
    225         or die "Can't open $temp, $!";
    226     podchecker($filename, $OUT);
    227     close $OUT;
    228     open $OUT, '<', $temp
    229         or die "Can't read $temp, $!";
    230     while ( <$OUT> ) {
    231         next if /\(section\) in.*deprecated/;
    232         print;
    233     }
    234     close $OUT;
    235     unlink $temp || warn "Can't remove $temp, $!";
    236 
    237     # Find what section this page is in; assume 3.
    238     my $section = 3;
    239     $section = $1 if $dirname =~ /man([1-9])/;
    240 
    241     foreach ((@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}})) {
    242         # Skip "return values" if not -s
    243         print "$id: missing $_ head1 section\n"
    244             if $contents !~ /^=head1\s+${_}\s*$/m;
    245     }
    246 }
    247 
    248 my %dups;
    249 
    250 sub parsenum()
    251 {
    252     my $file = shift;
    253     my @apis;
    254 
    255     open my $IN, '<', $file
    256         or die "Can't open $file, $!, stopped";
    257 
    258     while ( <$IN> ) {
    259         next if /^#/;
    260         next if /\bNOEXIST\b/;
    261         next if /\bEXPORT_VAR_AS_FUNC\b/;
    262         my @fields = split();
    263         die "Malformed line $_"
    264             if scalar @fields != 2 && scalar @fields != 4;
    265         push @apis, $fields[0];
    266     }
    267 
    268     close $IN;
    269 
    270     print "# Found ", scalar(@apis), " in $file\n" unless $opt_p;
    271     return sort @apis;
    272 }
    273 
    274 sub getdocced()
    275 {
    276     my $dir = shift;
    277     my %return;
    278 
    279     foreach my $pod ( glob("$dir/*.pod") ) {
    280         my %podinfo = extract_pod_info($pod);
    281         foreach my $n ( @{$podinfo{names}} ) {
    282             $return{$n} = $pod;
    283             print "# Duplicate $n in $pod and $dups{$n}\n"
    284                 if defined $dups{$n} && $dups{$n} ne $pod;
    285             $dups{$n} = $pod;
    286         }
    287     }
    288 
    289     return %return;
    290 }
    291 
    292 my %docced;
    293 
    294 sub checkmacros()
    295 {
    296     my $count = 0;
    297 
    298     print "# Checking macros (approximate)\n";
    299     foreach my $f ( glob('include/openssl/*.h') ) {
    300         # Skip some internals we don't want to document yet.
    301         next if $f eq 'include/openssl/asn1.h';
    302         next if $f eq 'include/openssl/asn1t.h';
    303         next if $f eq 'include/openssl/err.h';
    304         open(IN, $f) || die "Can't open $f, $!";
    305         while ( <IN> ) {
    306             next unless /^#\s*define\s*(\S+)\(/;
    307             my $macro = $1;
    308             next if $docced{$macro};
    309             next if $macro =~ /i2d_/
    310                 || $macro =~ /d2i_/
    311                 || $macro =~ /DEPRECATEDIN/
    312                 || $macro =~ /IMPLEMENT_/
    313                 || $macro =~ /DECLARE_/;
    314             print "$f:$macro\n" if $opt_d;
    315             $count++;
    316         }
    317         close(IN);
    318     }
    319     print "# Found $count macros missing (not all should be documented)\n"
    320 }
    321 
    322 sub printem()
    323 {
    324     my $libname = shift;
    325     my $numfile = shift;
    326     my $count = 0;
    327 
    328     foreach my $func ( &parsenum($numfile) ) {
    329         next if $docced{$func};
    330 
    331         # Skip ASN1 utilities
    332         next if $func =~ /^ASN1_/;
    333 
    334         print "$libname:$func\n" if $opt_d;
    335         $count++;
    336     }
    337     print "# Found $count missing from $numfile\n\n";
    338 }
    339 
    340 
    341 # Collection of links in each POD file.
    342 # filename => [ "foo(1)", "bar(3)", ... ]
    343 my %link_collection = ();
    344 # Collection of names in each POD file.
    345 # "name(s)" => filename
    346 my %name_collection = ();
    347 
    348 sub collectnames {
    349     my $filename = shift;
    350     $filename =~ m|man(\d)/|;
    351     my $section = $1;
    352     my $simplename = basename($filename, ".pod");
    353     my $id = "${filename}:1:";
    354 
    355     my $contents = '';
    356     {
    357         local $/ = undef;
    358         open POD, $filename or die "Couldn't open $filename, $!";
    359         $contents = <POD>;
    360         close POD;
    361     }
    362 
    363     $contents =~ /=head1 NAME([^=]*)=head1 /ms;
    364     my $tmp = $1;
    365     unless (defined $tmp) {
    366         print "$id weird name section\n";
    367         return;
    368     }
    369     $tmp =~ tr/\n/ /;
    370     $tmp =~ s/-.*//g;
    371 
    372     my @names = map { s/\s+//g; $_ } split(/,/, $tmp);
    373     unless (grep { $simplename eq $_ } @names) {
    374         print "$id missing $simplename\n";
    375         push @names, $simplename;
    376     }
    377     foreach my $name (@names) {
    378         next if $name eq "";
    379         my $name_sec = "$name($section)";
    380         if (! exists $name_collection{$name_sec}) {
    381             $name_collection{$name_sec} = $filename;
    382         } else { #elsif ($filename ne $name_collection{$name_sec}) {
    383             print "$id $name_sec also in $name_collection{$name_sec}\n";
    384         }
    385     }
    386 
    387     my @foreign_names =
    388         map { map { s/\s+//g; $_ } split(/,/, $_) }
    389         $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
    390     foreach (@foreign_names) {
    391         $name_collection{$_} = undef; # It still exists!
    392     }
    393 
    394     my @links = $contents =~ /L<
    395                               # if the link is of the form L<something|name(s)>,
    396                               # then remove 'something'.  Note that 'something'
    397                               # may contain POD codes as well...
    398                               (?:(?:[^\|]|<[^>]*>)*\|)?
    399                               # we're only interested in references that have
    400                               # a one digit section number
    401                               ([^\/>\(]+\(\d\))
    402                              /gx;
    403     $link_collection{$filename} = [ @links ];
    404 }
    405 
    406 sub checklinks {
    407     foreach my $filename (sort keys %link_collection) {
    408         foreach my $link (@{$link_collection{$filename}}) {
    409             print "${filename}:1: reference to non-existing $link\n"
    410                 unless exists $name_collection{$link};
    411         }
    412     }
    413 }
    414 
    415 sub publicize() {
    416     foreach my $name ( &parsenum('util/libcrypto.num') ) {
    417         $public{$name} = 1;
    418     }
    419     foreach my $name ( &parsenum('util/libssl.num') ) {
    420         $public{$name} = 1;
    421     }
    422     foreach my $name ( &parsenum('util/private.num') ) {
    423         $public{$name} = 1;
    424     }
    425 }
    426 
    427 my %skips = (
    428     'aes128' => 1,
    429     'aes192' => 1,
    430     'aes256' => 1,
    431     'aria128' => 1,
    432     'aria192' => 1,
    433     'aria256' => 1,
    434     'camellia128' => 1,
    435     'camellia192' => 1,
    436     'camellia256' => 1,
    437     'des' => 1,
    438     'des3' => 1,
    439     'idea' => 1,
    440     '[cipher]' => 1,
    441     '[digest]' => 1,
    442 );
    443 
    444 sub checkflags() {
    445     my $cmd = shift;
    446     my %cmdopts;
    447     my %docopts;
    448     my $ok = 1;
    449 
    450     # Get the list of options in the command.
    451     open CFH, "./apps/openssl list --options $cmd|"
    452         || die "Can list options for $cmd, $!";
    453     while ( <CFH> ) {
    454         chop;
    455         s/ .$//;
    456         $cmdopts{$_} = 1;
    457     }
    458     close CFH;
    459 
    460     # Get the list of flags from the synopsis
    461     open CFH, "<doc/man1/$cmd.pod"
    462         || die "Can't open $cmd.pod, $!";
    463     while ( <CFH> ) {
    464         chop;
    465         last if /DESCRIPTION/;
    466         next unless /\[B<-([^ >]+)/;
    467         $docopts{$1} = 1;
    468     }
    469     close CFH;
    470 
    471     # See what's in the command not the manpage.
    472     my @undocced = ();
    473     foreach my $k ( keys %cmdopts ) {
    474         push @undocced, $k unless $docopts{$k};
    475     }
    476     if ( scalar @undocced > 0 ) {
    477         $ok = 0;
    478         foreach ( @undocced ) {
    479             print "doc/man1/$cmd.pod: Missing -$_\n";
    480         }
    481     }
    482 
    483     # See what's in the command not the manpage.
    484     my @unimpl = ();
    485     foreach my $k ( keys %docopts ) {
    486         push @unimpl, $k unless $cmdopts{$k};
    487     }
    488     if ( scalar @unimpl > 0 ) {
    489         $ok = 0;
    490         foreach ( @unimpl ) {
    491             next if defined $skips{$_};
    492             print "doc/man1/$cmd.pod: Not implemented -$_\n";
    493         }
    494     }
    495 
    496     return $ok;
    497 }
    498 
    499 getopts('cdlnphu');
    500 
    501 &help() if $opt_h;
    502 $opt_n = 1 if $opt_p;
    503 $opt_u = 1 if $opt_d;
    504 
    505 die "Need one of -[cdlnpu] flags.\n"
    506     unless $opt_c or $opt_l or $opt_n or $opt_u;
    507 
    508 if ( $opt_c ) {
    509     my $ok = 1;
    510     my @commands = ();
    511 
    512     # Get list of commands.
    513     open FH, "./apps/openssl list -1 -commands|"
    514         || die "Can't list commands, $!";
    515     while ( <FH> ) {
    516         chop;
    517         push @commands, $_;
    518     }
    519     close FH;
    520 
    521     # See if each has a manpage.
    522     foreach ( @commands ) {
    523         next if $_ eq 'help' || $_ eq 'exit';
    524         if ( ! -f "doc/man1/$_.pod" ) {
    525             print "doc/man1/$_.pod does not exist\n";
    526             $ok = 0;
    527         } else {
    528             $ok = 0 if not &checkflags($_);
    529         }
    530     }
    531 
    532     # See what help is missing.
    533     open FH, "./apps/openssl list --missing-help |"
    534         || die "Can't list missing help, $!";
    535     while ( <FH> ) {
    536         chop;
    537         my ($cmd, $flag) = split;
    538         print "$cmd has no help for -$flag\n";
    539         $ok = 0;
    540     }
    541     close FH;
    542 
    543     exit 1 if not $ok;
    544 }
    545 
    546 if ( $opt_l ) {
    547     foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
    548         collectnames($_);
    549     }
    550     checklinks();
    551 }
    552 
    553 if ( $opt_n ) {
    554     &publicize() if $opt_p;
    555     foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
    556         &check($_);
    557     }
    558 }
    559 
    560 if ( $opt_u ) {
    561     my %temp = &getdocced('doc/man3');
    562     foreach ( keys %temp ) {
    563         $docced{$_} = $temp{$_};
    564     }
    565     &printem('crypto', 'util/libcrypto.num');
    566     &printem('ssl', 'util/libssl.num');
    567     &checkmacros();
    568 }
    569 
    570 exit;
    571