Home | History | Annotate | Line # | Download | only in util
find-doc-nits revision 1.1.1.5 
      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 Count 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     my %seen;
    298 
    299     print "# Checking macros (approximate)\n";
    300     foreach my $f ( glob('include/openssl/*.h') ) {
    301         # Skip some internals we don't want to document yet.
    302         next if $f eq 'include/openssl/asn1.h';
    303         next if $f eq 'include/openssl/asn1t.h';
    304         next if $f eq 'include/openssl/err.h';
    305         open(IN, $f) || die "Can't open $f, $!";
    306         while ( <IN> ) {
    307             next unless /^#\s*define\s*(\S+)\(/;
    308             my $macro = $1;
    309             next if $docced{$macro} || defined $seen{$macro};
    310             next if $macro =~ /i2d_/
    311                 || $macro =~ /d2i_/
    312                 || $macro =~ /DEPRECATEDIN/
    313                 || $macro =~ /IMPLEMENT_/
    314                 || $macro =~ /DECLARE_/;
    315             print "$f:$macro\n" if $opt_d;
    316             $count++;
    317             $seen{$macro} = 1;
    318         }
    319         close(IN);
    320     }
    321     print "# Found $count macros missing (not all should be documented)\n"
    322 }
    323 
    324 sub printem()
    325 {
    326     my $libname = shift;
    327     my $numfile = shift;
    328     my $count = 0;
    329     my %seen;
    330 
    331     foreach my $func ( &parsenum($numfile) ) {
    332         next if $docced{$func} || defined $seen{$func};
    333 
    334         # Skip ASN1 utilities
    335         next if $func =~ /^ASN1_/;
    336 
    337         print "$libname:$func\n" if $opt_d;
    338         $count++;
    339         $seen{$func} = 1;
    340     }
    341     print "# Found $count missing from $numfile\n\n";
    342 }
    343 
    344 
    345 # Collection of links in each POD file.
    346 # filename => [ "foo(1)", "bar(3)", ... ]
    347 my %link_collection = ();
    348 # Collection of names in each POD file.
    349 # "name(s)" => filename
    350 my %name_collection = ();
    351 
    352 sub collectnames {
    353     my $filename = shift;
    354     $filename =~ m|man(\d)/|;
    355     my $section = $1;
    356     my $simplename = basename($filename, ".pod");
    357     my $id = "${filename}:1:";
    358 
    359     my $contents = '';
    360     {
    361         local $/ = undef;
    362         open POD, $filename or die "Couldn't open $filename, $!";
    363         $contents = <POD>;
    364         close POD;
    365     }
    366 
    367     $contents =~ /=head1 NAME([^=]*)=head1 /ms;
    368     my $tmp = $1;
    369     unless (defined $tmp) {
    370         print "$id weird name section\n";
    371         return;
    372     }
    373     $tmp =~ tr/\n/ /;
    374     $tmp =~ s/-.*//g;
    375 
    376     my @names = map { s/\s+//g; $_ } split(/,/, $tmp);
    377     unless (grep { $simplename eq $_ } @names) {
    378         print "$id missing $simplename\n";
    379         push @names, $simplename;
    380     }
    381     foreach my $name (@names) {
    382         next if $name eq "";
    383         my $name_sec = "$name($section)";
    384         if (! exists $name_collection{$name_sec}) {
    385             $name_collection{$name_sec} = $filename;
    386         } else { #elsif ($filename ne $name_collection{$name_sec}) {
    387             print "$id $name_sec also in $name_collection{$name_sec}\n";
    388         }
    389     }
    390 
    391     my @foreign_names =
    392         map { map { s/\s+//g; $_ } split(/,/, $_) }
    393         $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
    394     foreach (@foreign_names) {
    395         $name_collection{$_} = undef; # It still exists!
    396     }
    397 
    398     my @links = $contents =~ /L<
    399                               # if the link is of the form L<something|name(s)>,
    400                               # then remove 'something'.  Note that 'something'
    401                               # may contain POD codes as well...
    402                               (?:(?:[^\|]|<[^>]*>)*\|)?
    403                               # we're only interested in references that have
    404                               # a one digit section number
    405                               ([^\/>\(]+\(\d\))
    406                              /gx;
    407     $link_collection{$filename} = [ @links ];
    408 }
    409 
    410 sub checklinks {
    411     foreach my $filename (sort keys %link_collection) {
    412         foreach my $link (@{$link_collection{$filename}}) {
    413             print "${filename}:1: reference to non-existing $link\n"
    414                 unless exists $name_collection{$link};
    415         }
    416     }
    417 }
    418 
    419 sub publicize() {
    420     foreach my $name ( &parsenum('util/libcrypto.num') ) {
    421         $public{$name} = 1;
    422     }
    423     foreach my $name ( &parsenum('util/libssl.num') ) {
    424         $public{$name} = 1;
    425     }
    426     foreach my $name ( &parsenum('util/private.num') ) {
    427         $public{$name} = 1;
    428     }
    429 }
    430 
    431 my %skips = (
    432     'aes128' => 1,
    433     'aes192' => 1,
    434     'aes256' => 1,
    435     'aria128' => 1,
    436     'aria192' => 1,
    437     'aria256' => 1,
    438     'camellia128' => 1,
    439     'camellia192' => 1,
    440     'camellia256' => 1,
    441     'des' => 1,
    442     'des3' => 1,
    443     'idea' => 1,
    444     '[cipher]' => 1,
    445     '[digest]' => 1,
    446 );
    447 
    448 sub checkflags() {
    449     my $cmd = shift;
    450     my %cmdopts;
    451     my %docopts;
    452     my $ok = 1;
    453 
    454     # Get the list of options in the command.
    455     open CFH, "./apps/openssl list --options $cmd|"
    456         || die "Can list options for $cmd, $!";
    457     while ( <CFH> ) {
    458         chop;
    459         s/ .$//;
    460         $cmdopts{$_} = 1;
    461     }
    462     close CFH;
    463 
    464     # Get the list of flags from the synopsis
    465     open CFH, "<doc/man1/$cmd.pod"
    466         || die "Can't open $cmd.pod, $!";
    467     while ( <CFH> ) {
    468         chop;
    469         last if /DESCRIPTION/;
    470         next unless /\[B<-([^ >]+)/;
    471         $docopts{$1} = 1;
    472     }
    473     close CFH;
    474 
    475     # See what's in the command not the manpage.
    476     my @undocced = ();
    477     foreach my $k ( keys %cmdopts ) {
    478         push @undocced, $k unless $docopts{$k};
    479     }
    480     if ( scalar @undocced > 0 ) {
    481         $ok = 0;
    482         foreach ( @undocced ) {
    483             print "doc/man1/$cmd.pod: Missing -$_\n";
    484         }
    485     }
    486 
    487     # See what's in the command not the manpage.
    488     my @unimpl = ();
    489     foreach my $k ( keys %docopts ) {
    490         push @unimpl, $k unless $cmdopts{$k};
    491     }
    492     if ( scalar @unimpl > 0 ) {
    493         $ok = 0;
    494         foreach ( @unimpl ) {
    495             next if defined $skips{$_};
    496             print "doc/man1/$cmd.pod: Not implemented -$_\n";
    497         }
    498     }
    499 
    500     return $ok;
    501 }
    502 
    503 getopts('cdlnphu');
    504 
    505 &help() if $opt_h;
    506 $opt_n = 1 if $opt_p;
    507 $opt_u = 1 if $opt_d;
    508 
    509 die "Need one of -[cdlnpu] flags.\n"
    510     unless $opt_c or $opt_l or $opt_n or $opt_u;
    511 
    512 if ( $opt_c ) {
    513     my $ok = 1;
    514     my @commands = ();
    515 
    516     # Get list of commands.
    517     open FH, "./apps/openssl list -1 -commands|"
    518         || die "Can't list commands, $!";
    519     while ( <FH> ) {
    520         chop;
    521         push @commands, $_;
    522     }
    523     close FH;
    524 
    525     # See if each has a manpage.
    526     foreach ( @commands ) {
    527         next if $_ eq 'help' || $_ eq 'exit';
    528         if ( ! -f "doc/man1/$_.pod" ) {
    529             print "doc/man1/$_.pod does not exist\n";
    530             $ok = 0;
    531         } else {
    532             $ok = 0 if not &checkflags($_);
    533         }
    534     }
    535 
    536     # See what help is missing.
    537     open FH, "./apps/openssl list --missing-help |"
    538         || die "Can't list missing help, $!";
    539     while ( <FH> ) {
    540         chop;
    541         my ($cmd, $flag) = split;
    542         print "$cmd has no help for -$flag\n";
    543         $ok = 0;
    544     }
    545     close FH;
    546 
    547     exit 1 if not $ok;
    548 }
    549 
    550 if ( $opt_l ) {
    551     foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
    552         collectnames($_);
    553     }
    554     checklinks();
    555 }
    556 
    557 if ( $opt_n ) {
    558     &publicize() if $opt_p;
    559     foreach (@ARGV ? @ARGV : glob('doc/*/*.pod')) {
    560         &check($_);
    561     }
    562 }
    563 
    564 if ( $opt_u ) {
    565     my %temp = &getdocced('doc/man3');
    566     foreach ( keys %temp ) {
    567         $docced{$_} = $temp{$_};
    568     }
    569     &printem('crypto', 'util/libcrypto.num');
    570     &printem('ssl', 'util/libssl.num');
    571     &checkmacros();
    572 }
    573 
    574 exit;
    575