openssl/util/find-doc-nits

1099 lines
35 KiB
Plaintext
Raw Normal View History

#! /usr/bin/env perl
# Copyright 2002-2019 The OpenSSL Project Authors. All Rights Reserved.
#
# Licensed under the Apache License 2.0 (the "License"). You may not use
# this file except in compliance with the License. You can obtain a copy
# in the file LICENSE in the source distribution or at
# https://www.openssl.org/source/license.html
require 5.10.0;
use warnings;
use strict;
use Carp qw(:DEFAULT cluck);
use Pod::Checker;
use File::Find;
use File::Basename;
use File::Spec::Functions;
use Getopt::Std;
use FindBin;
use lib "$FindBin::Bin/perl";
use OpenSSL::Util::Pod;
use lib '.';
use configdata;
# Set to 1 for debug output
my $debug = 0;
# Where to find openssl command
my $openssl = "./util/opensslwrap.sh";
# Options.
our($opt_d);
our($opt_e);
our($opt_s);
our($opt_o);
our($opt_h);
our($opt_l);
our($opt_n);
our($opt_p);
our($opt_u);
our($opt_v);
our($opt_c);
# Print usage message and exit.
sub help {
print <<EOF;
Find small errors (nits) in documentation. Options:
-c List undocumented commands and options
-d Detailed list of undocumented (implies -u)
-e Detailed list of new undocumented (implies -v)
-h Print this help message
-l Print bogus links
-n Print nits in POD pages
-o Causes -e/-v to count symbols added since 1.1.1 as new (implies -v)
-u Count undocumented functions
-v Count new undocumented functions
EOF
exit;
}
getopts('cdehlnouv');
help() if $opt_h;
$opt_u = 1 if $opt_d;
$opt_v = 1 if $opt_o || $opt_e;
die "Cannot use both -u and -v"
if $opt_u && $opt_v;
die "Cannot use both -d and -e"
if $opt_d && $opt_e;
# We only need to check c, l, n, u and v.
# Options d, e, o imply one of the above.
die "Need one of -[cdehlnouv] flags.\n"
unless $opt_c or $opt_l or $opt_n or $opt_u or $opt_v;
my $temp = '/tmp/docnits.txt';
my $OUT;
my %public;
my $status = 0;
my @sections = ( 'man1', 'man3', 'man5', 'man7' );
my %mandatory_sections = (
'*' => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
1 => [ 'SYNOPSIS', 'OPTIONS' ],
3 => [ 'SYNOPSIS', 'RETURN VALUES' ],
5 => [ ],
7 => [ ]
);
# Collect all POD files, both internal and public, and regardless of location
# We collect them in a hash table with each file being a key, so we can attach
# tags to them. For example, internal docs will have the word "internal"
# attached to them.
my %files = ();
# We collect files names on the fly, on known tag basis
my %collected_tags = ();
# We cache results based on tags
my %collected_results = ();
# files OPTIONS
#
# Example:
#
# files(TAGS => 'manual');
# files(TAGS => [ 'manual', 'man1' ]);
#
# This function returns an array of files corresponding to a set of tags
# given with the options "TAGS". The value of this option can be a single
# word, or an array of several words, which work as inclusive or exclusive
# selectors. Inclusive selectors are used to add one more set of files to
# the returned array, while exclusive selectors limit the set of files added
# to the array. The recognised tag values are:
#
# 'public_manual' - inclusive selector, adds public manuals to the
# returned array of files.
# 'internal_manual' - inclusive selector, adds internal manuals to the
# returned array of files.
# 'manual' - inclusive selector, adds any manual to the returned
# array of files. This is really a shorthand for
# 'public_manual' and 'internal_manual' combined.
# 'public_header' - inclusive selector, adds public headers to the
# returned array of files.
# 'header' - inclusive selector, adds any header file to the
# returned array of files. Since we currently only
# care about public headers, this is exactly
# equivalent to 'public_header', but is present for
# consistency.
#
# 'man1', 'man3', 'man5', 'man7'
# - exclusive selectors, only applicable together with
# any of the manual selectors. If any of these are
# present, only the manuals from the given sections
# will be include. If none of these are present,
# the manuals from all sections will be returned.
#
# All returned manual files come from configdata.pm.
# All returned header files come from looking inside
# "$config{sourcedir}/include/openssl"
#
sub files {
my %opts = ( @_ ); # Make a copy of the arguments
$opts{TAGS} = [ $opts{TAGS} ] if ref($opts{TAGS}) eq '';
croak "No tags given, or not an array"
unless exists $opts{TAGS} && ref($opts{TAGS}) eq 'ARRAY';
my %tags = map { $_ => 1 } @{$opts{TAGS}};
$tags{public_manual} = 1
if $tags{manual} && ($tags{public} // !$tags{internal});
$tags{internal_manual} = 1
if $tags{manual} && ($tags{internal} // !$tags{public});
$tags{public_header} = 1
if $tags{header} && ($tags{public} // !$tags{internal});
delete $tags{manual};
delete $tags{header};
delete $tags{public};
delete $tags{internal};
my $tags_as_key = join(':', sort keys %tags);
cluck "DEBUG[files]: This is how we got here!" if $debug;
print STDERR "DEBUG[files]: tags: $tags_as_key\n" if $debug;
my %tags_to_collect = ( map { $_ => 1 }
grep { !exists $collected_tags{$_} }
keys %tags );
if ($tags_to_collect{public_manual}) {
print STDERR "DEBUG[files]: collecting public manuals\n"
if $debug;
# The structure in configdata.pm is that $unified_info{mandocs}
# contains lists of man files, and in turn, $unified_info{depends}
# contains hash tables showing which POD file each of those man
# files depend on. We use that information to find the POD files,
# and to attach the man section they belong to as tags
foreach my $mansect ( @sections ) {
foreach ( map { @{$unified_info{depends}->{$_}} }
@{$unified_info{mandocs}->{$mansect}} ) {
$files{$_} = { $mansect => 1, public_manual => 1 };
}
}
$collected_tags{public_manual} = 1;
}
if ($tags_to_collect{internal_manual}) {
print STDERR "DEBUG[files]: collecting internal manuals\n"
if $debug;
# We don't have the internal docs in configdata.pm. However, they
# are all in the source tree, so they're easy to find.
foreach my $mansect ( @sections ) {
foreach ( glob(catfile($config{sourcedir},
'doc', 'internal', $mansect, '*.pod')) ) {
$files{$_} = { $mansect => 1, internal_manual => 1 };
}
}
$collected_tags{internal_manual} = 1;
}
if ($tags_to_collect{public_header}) {
print STDERR "DEBUG[files]: collecting public headers\n"
if $debug;
foreach ( glob(catfile($config{sourcedir},
'include', 'openssl', '*.h')) ) {
$files{$_} = { public_header => 1 };
}
}
my @result = @{$collected_results{$tags_as_key} // []};
if (!@result) {
# Produce a result based on caller tags
foreach my $type ( ( 'public_manual', 'internal_manual' ) ) {
next unless $tags{$type};
# If caller asked for specific sections, we care about sections.
# Otherwise, we give back all of them.
my @selected_sections =
grep { $tags{$_} } @sections;
@selected_sections = @sections unless @selected_sections;
foreach my $section ( ( @selected_sections ) ) {
push @result,
( sort { basename($a) cmp basename($b) }
grep { $files{$_}->{$type} && $files{$_}->{$section} }
keys %files );
}
}
if ($tags{public_header}) {
push @result,
( sort { basename($a) cmp basename($b) }
grep { $files{$_}->{public_header} }
keys %files );
}
if ($debug) {
print STDERR "DEBUG[files]: result:\n";
print STDERR "DEBUG[files]: $_\n" foreach @result;
}
$collected_results{$tags_as_key} = [ @result ];
}
return @result;
}
# Print error message, set $status.
sub err {
print join(" ", @_), "\n";
$status = 1
}
# Cross-check functions in the NAME and SYNOPSIS section.
sub name_synopsis {
my $id = shift;
my $filename = shift;
my $contents = shift;
# Get NAME section and all words in it.
return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
my $tmp = $1;
$tmp =~ tr/\n/ /;
err($id, "trailing comma before - in NAME")
if $tmp =~ /, *-/;
$tmp =~ s/ -.*//g;
err($id, "POD markup among the names in NAME")
if $tmp =~ /[<>]/;
$tmp =~ s/ */ /g;
err($id, "missing comma in NAME")
if $tmp =~ /[^,] /;
my $dirname = dirname($filename);
my $section = basename($dirname);
my $simplename = basename($filename, ".pod");
my $foundfilename = 0;
my %foundfilenames = ();
my %names;
foreach my $n ( split ',', $tmp ) {
$n =~ s/^\s+//;
$n =~ s/\s+$//;
err($id, "the name '$n' contains white-space")
if $n =~ /\s/;
$names{$n} = 1;
$foundfilename++ if $n eq $simplename;
$foundfilenames{$n} = 1
if ( ( grep { basename($_) eq "$n.pod" }
files(TAGS => [ 'manual', $section ]) )
&& $n ne $simplename );
}
err($id, "the following exist as other .pod files:",
sort keys %foundfilenames)
if %foundfilenames;
err($id, "$simplename (filename) missing from NAME section")
unless $foundfilename;
if ( $filename !~ /internal/ ) {
foreach my $n ( keys %names ) {
err($id, "$n is not public")
if !defined $public{$n};
}
}
# Find all functions in SYNOPSIS
return unless $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms;
my $syn = $1;
foreach my $line ( split /\n+/, $syn ) {
next unless $line =~ /^\s/;
my $sym;
my $is_prototype = 1;
$line =~ s/STACK_OF\([^)]+\)/int/g;
$line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
$line =~ s/__declspec\([^)]+\)//;
if ( $line =~ /typedef.*\(\*\S+\)\s+\(/ ) {
# a callback function with whitespace before the argument list:
# typedef ... (*NAME) (...
err($id, "function typedef has space before arg list: $line");
}
if ( $line =~ /env (\S*)=/ ) {
# environment variable env NAME=...
$sym = $1;
} elsif ( $line =~ /typedef.*\(\*(\S+)\)\s*\(/ ) {
# a callback function pointer: typedef ... (*NAME)(...
$sym = $1;
} elsif ( $line =~ /typedef.* (\S+)\(/ ) {
# a callback function signature: typedef ... NAME(...
$sym = $1;
} elsif ( $line =~ /typedef.* (\S+);/ ) {
# a simple typedef: typedef ... NAME;
$is_prototype = 0;
$sym = $1;
} elsif ( $line =~ /enum (\S*) \{/ ) {
# an enumeration: enum ... {
$sym = $1;
} elsif ( $line =~ /#(?:define|undef) ([A-Za-z0-9_]+)/ ) {
$is_prototype = 0;
$sym = $1;
} elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
$sym = $1;
}
else {
next;
}
err($id, "$sym missing from NAME section")
unless defined $names{$sym};
$names{$sym} = 2;
# Do some sanity checks on the prototype.
err($id, "prototype missing spaces around commas: $line")
if $is_prototype && $line =~ /[a-z0-9],[^ ]/;
}
foreach my $n ( keys %names ) {
next if $names{$n} == 2;
err($id, "$n missing from SYNOPSIS")
}
}
# Check if SECTION ($3) is located before BEFORE ($4)
sub check_section_location {
my $id = shift;
my $contents = shift;
my $section = shift;
my $before = shift;
return unless $contents =~ /=head1 $section/
and $contents =~ /=head1 $before/;
err($id, "$section should appear before $before section")
if $contents =~ /=head1 $before.*=head1 $section/ms;
}
# Check if a =head1 is duplicated, or a =headX is duplicated within a
# =head1. Treats =head2 =head3 as equivalent -- it doesn't reset the head3
# sets if it finds a =head2 -- but that is good enough for now. Also check
# for proper capitalization, trailing periods, etc.
sub check_head_style {
my $id = shift;
my $contents = shift;
my %head1;
my %subheads;
foreach my $line ( split /\n+/, $contents ) {
next unless $line =~ /^=head/;
if ( $line =~ /head1/ ) {
err($id, "duplicate section $line")
if defined $head1{$line};
$head1{$line} = 1;
%subheads = ();
} else {
err($id, "duplicate subsection $line")
if defined $subheads{$line};
$subheads{$line} = 1;
}
err($id, "period in =head")
if $line =~ /\.[^\w]/ or $line =~ /\.$/;
err($id, "not all uppercase in =head1")
if $line =~ /head1.*[a-z]/;
err($id, "all uppercase in subhead")
if $line =~ /head[234][ A-Z0-9]+$/;
}
}
# Because we have options and symbols with extra markup, we need
# to take that into account, so we need a regexp that extracts
# markup chunks, including recursive markup.
# please read up on /(?R)/ in perlre(1)
# (note: order is important, (?R) needs to come before .)
# (note: non-greedy is important, or something like 'B<foo> and B<bar>'
# will be captured as one item)
my $markup_re =
qr/( # Capture group
[BIL]< # The start of what we recurse on
(?:(?-1)|.)*? # recurse the whole regexp (referring to
# the last opened capture group, i.e. the
# start of this regexp), or pick next
# character. Do NOT be greedy!
> # The end of what we recurse on
)/x; # (the x allows this sort of split up regexp)
# Options must start with a dash, followed by a letter, possibly
# followed by letters, digits, dashes and underscores, and the last
# character must be a letter or a digit.
# We do also accept the single -? or -n, where n is a digit
my $option_re =
qr/(?:
\? # Single question mark
|
\d # Single digit
|
- # Single dash (--)
|
[[:alpha:]](?:[-_[:alnum:]]*?[[:alnum:]])?
)/x;
# Helper function to check if a given $thing is properly marked up
# option. It returns one of these values:
# undef if it's not an option
# "" if it's a malformed option
# $unwrapped the option with the outermost B<> wrapping removed.
sub normalise_option {
my $id = shift;
my $filename = shift;
my $thing = shift;
my $unwrapped = $thing;
my $unmarked = $thing;
# $unwrapped is the option with the outer B<> markup removed
$unwrapped =~ s/^B<//;
$unwrapped =~ s/>$//;
# $unmarked is the option with *all* markup removed
$unmarked =~ s/[BIL]<|>//msg;
# If we found an option, check it, collect it
if ( $unwrapped =~ /^\s*-/ ) {
return $unwrapped # return option with outer B<> removed
if $unmarked =~ /^-${option_re}$/;
return ""; # Malformed option
}
return undef; # Something else
}
# Checks of command option (man1) formatting. The man1 checks are
# restricted to the SYNOPSIS and OPTIONS sections, the rest is too
# free form, we simply cannot be too strict there.
sub option_check {
my $id = shift;
my $filename = shift;
my $contents = shift;
my $synopsis = ($contents =~ /=head1\s+SYNOPSIS(.*?)=head1/s, $1);
# Some pages have more than one OPTIONS section, let's make sure
# to get them all
my $options = '';
while ( $contents =~ /=head1\s+[A-Z ]*?OPTIONS$(.*?)(?==head1)/msg ) {
$options .= $1;
}
# Look for options with no or incorrect markup
while ( $synopsis =~
/(?<![-<[:alnum:]])-(?:$markup_re|.)*(?![->[:alnum:]])/msg ) {
err($id, "Malformed option [1] in SYNOPSIS: $&");
}
while ( $synopsis =~ /$markup_re/msg ) {
my $found = $&;
print STDERR "$id:DEBUG[option_check] SYNOPSIS: found $found\n"
if $debug;
my $option_uw = normalise_option($id, $filename, $found);
err($id, "Malformed option [2] in SYNOPSIS: $found")
if defined $option_uw && $option_uw eq '';
}
# In OPTIONS, we look for =item paragraphs.
# (?=^\s*$) detects an empty line.
while ( $options =~ /=item\s+(.*?)(?=^\s*$)/msg ) {
my $item = $&;
while ( $item =~ /(\[\s*)?($markup_re)/msg ) {
my $found = $2;
print STDERR "$id:DEBUG[option_check] OPTIONS: found $&\n"
if $debug;
err($id, "Unexpected bracket in OPTIONS =item: $item")
if ($1 // '') ne '' && $found =~ /^B<\s*-/;
my $option_uw = normalise_option($id, $filename, $found);
err($id, "Malformed option in OPTIONS: $found")
if defined $option_uw && $option_uw eq '';
}
}
}
# Normal symbol form
my $symbol_re = qr/[[:alpha:]_][_[:alnum:]]*?/;
# Checks of function name (man3) formatting. The man3 checks are
# easier than the man1 checks, we only check the names followed by (),
# and only the names that have POD markup.
sub functionname_check {
my $id = shift;
my $filename = shift;
my $contents = shift;
while ( $contents =~ /($markup_re)\(\)/msg ) {
print STDERR "$id:DEBUG[functionname_check] SYNOPSIS: found $&\n"
if $debug;
my $symbol = $1;
my $unmarked = $symbol;
$unmarked =~ s/[BIL]<|>//msg;
err($id, "Malformed symbol: $symbol")
unless $symbol =~ /^B<.*>$/ && $unmarked =~ /^${symbol_re}$/
}
# We can't do the kind of collecting coolness that option_check()
# does, because there are too many things that can't be found in
# name repositories like the NAME sections, such as symbol names
# with a variable part (typically marked up as B<foo_I<TYPE>_bar>
}
# This is from http://man7.org/linux/man-pages/man7/man-pages.7.html
my %preferred_words = (
'bitmask' => 'bit mask',
'builtin' => 'built-in',
#'epoch' => 'Epoch', # handled specially, below
'file name' => 'filename',
'file system' => 'filesystem',
'host name' => 'hostname',
'i-node' => 'inode',
'lower case' => 'lowercase',
'lower-case' => 'lowercase',
'non-zero' => 'nonzero',
'path name' => 'pathname',
'pseudo-terminal' => 'pseudoterminal',
'reserved port' => 'privileged port',
'system port' => 'privileged port',
'realtime' => 'real-time',
'real time' => 'real-time',
'runtime' => 'run time',
'saved group ID'=> 'saved set-group-ID',
'saved set-GID' => 'saved set-group-ID',
'saved user ID' => 'saved set-user-ID',
'saved set-UID' => 'saved set-user-ID',
'set-GID' => 'set-group-ID',
'setgid' => 'set-group-ID',
'set-UID' => 'set-user-ID',
'setuid' => 'set-user-ID',
'super user' => 'superuser',
'super-user' => 'superuser',
'super block' => 'superblock',
'super-block' => 'superblock',
'time stamp' => 'timestamp',
'time zone' => 'timezone',
'upper case' => 'uppercase',
'upper-case' => 'uppercase',
'useable' => 'usable',
'userspace' => 'user space',
'user name' => 'username',
'zeroes' => 'zeros'
);
# Search manpage for words that have a different preferred use.
sub wording {
my $id = shift;
my $contents = shift;
foreach my $k ( keys %preferred_words ) {
# Sigh, trademark
next if $k eq 'file system'
and $contents =~ /Microsoft Encrypted File System/;
err($id, "found '$k' should use '$preferred_words{$k}'")
if $contents =~ /\b\Q$k\E\b/i;
}
err($id, "found 'epoch' should use 'Epoch'")
if $contents =~ /\bepoch\b/;
}
# Perform all sorts of nit/error checks on a manpage
sub check {
my $filename = shift;
my $dirname = basename(dirname($filename));
my $contents = '';
{
local $/ = undef;
open POD, $filename or die "Couldn't open $filename, $!";
$contents = <POD>;
close POD;
}
my $id = "${filename}:1:";
check_head_style($id, $contents);
# Check ordering of some sections in man3
if ( $filename =~ m|man3/| ) {
check_section_location($id, $contents, "RETURN VALUES", "EXAMPLES");
check_section_location($id, $contents, "SEE ALSO", "HISTORY");
check_section_location($id, $contents, "EXAMPLES", "SEE ALSO");
}
# Make sure every link has a section.
while ( $contents =~ /$markup_re/msg ) {
my $target = $1;
next unless $target =~ /^L<(.*)>$/; # Skip if not L<...>
$target = $1; # Peal away L< and >
$target =~ s/\/[^\/]*$//; # Peal away possible anchor
$target =~ s/.*\|//g; # Peal away possible link text
next if $target eq ''; # Skip if links within page, or
next if $target =~ /::/; # links to a Perl module, or
next if $target =~ /^https?:/; # is a URL link, or
next if $target =~ /\([1357]\)$/; # it has a section
err($id, "Section missing in $target")
}
# Check for proper links to commands.
while ( $contents =~ /L<([^>]*)\(1\)(?:\/.*)?>/g ) {
my $target = $1;
next if $target =~ /openssl-?/;
next if ( grep { basename($_) eq "$target.pod" }
files(TAGS => [ 'manual', 'man1' ]) );
# TODO: Filter out "foreign manual" links.
next if $target =~ /ps|apropos|sha1sum|procmail|perl/;
err($id, "Bad command link L<$target(1)>");
}
# Check for proper in-man-3 API links.
while ( $contents =~ /L<([^>]*)\(3\)(?:\/.*)?>/g ) {
my $target = $1;
err($id, "Bad L<$target>")
unless $target =~ /^[_[:alpha:]][_[:alnum:]]*$/
}
unless ( $contents =~ /=for openssl generic/ ) {
if ( $filename =~ m|man3/| ) {
name_synopsis($id, $filename, $contents);
functionname_check($id, $filename, $contents);
} elsif ( $filename =~ m|man1/| ) {
option_check($id, $filename, $contents)
}
}
wording($id, $contents);
err($id, "doesn't start with =pod")
if $contents !~ /^=pod/;
err($id, "doesn't end with =cut")
if $contents !~ /=cut\n$/;
err($id, "more than one cut line.")
if $contents =~ /=cut.*=cut/ms;
err($id, "EXAMPLE not EXAMPLES section.")
if $contents =~ /=head1 EXAMPLE[^S]/;
err($id, "WARNING not WARNINGS section.")
if $contents =~ /=head1 WARNING[^S]/;
err($id, "missing copyright")
if $contents !~ /Copyright .* The OpenSSL Project Authors/;
err($id, "copyright not last")
if $contents =~ /head1 COPYRIGHT.*=head/ms;
err($id, "head2 in All uppercase")
if $contents =~ /head2\s+[A-Z ]+\n/;
err($id, "extra space after head")
if $contents =~ /=head\d\s\s+/;
err($id, "period in NAME section")
if $contents =~ /=head1 NAME.*\.\n.*=head1 SYNOPSIS/ms;
err($id, "Duplicate $1 in L<>")
if $contents =~ /L<([^>]*)\|([^>]*)>/ && $1 eq $2;
err($id, "Bad =over $1")
if $contents =~ /=over([^ ][^24])/;
err($id, "Possible version style issue")
if $contents =~ /OpenSSL version [019]/;
if ( $contents !~ /=for openssl multiple includes/ ) {
# Look for multiple consecutive openssl #include lines
# (non-consecutive lines are okay; see man3/MD5.pod).
if ( $contents =~ /=head1 SYNOPSIS(.*)=head1 DESCRIPTION/ms ) {
my $count = 0;
foreach my $line ( split /\n+/, $1 ) {
if ( $line =~ m@include <openssl/@ ) {
err($id, "has multiple includes")
if ++$count == 2;
} else {
$count = 0;
}
}
}
}
open my $OUT, '>', $temp
or die "Can't open $temp, $!";
podchecker($filename, $OUT);
close $OUT;
open $OUT, '<', $temp
or die "Can't read $temp, $!";
while ( <$OUT> ) {
next if /\(section\) in.*deprecated/;
print;
}
close $OUT;
unlink $temp || warn "Can't remove $temp, $!";
# Find what section this page is in; assume 3.
my $section = 3;
$section = $1 if $dirname =~ /man([1-9])/;
foreach ( (@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}}) ) {
err($id, "missing $_ head1 section")
if $contents !~ /^=head1\s+${_}\s*$/m;
}
}
# Parse libcrypto.num, etc., and return sorted list of what's there.
sub parsenum {
my $file = shift;
my @apis;
open my $IN, '<', catfile($config{sourcedir}, $file)
or die "Can't open $file, $!, stopped";
while ( <$IN> ) {
next if /^#/;
next if /\bNOEXIST\b/;
my @fields = split();
die "Malformed line $_"
if scalar @fields != 2 && scalar @fields != 4;
push @apis, $fields[0];
}
close $IN;
return sort @apis;
}
# Parse all the manpages, getting return map of what they document
# (by looking at their NAME sections).
# Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
my %link_map = ();
# Map of names in each POD file; "name(s)" => filename
my %name_map = ();
# Load file of symbol names that we know aren't documented.
sub loadmissing($)
{
my $missingfile = shift;
my @missing;
open FH, catfile($config{sourcedir}, $missingfile)
or die "Can't open $missingfile";
while ( <FH> ) {
chomp;
next if /^#/;
push @missing, $_;
}
close FH;
for (@missing) {
err("$missingfile:", "$_ is documented in $name_map{$_}")
if !$opt_o && exists $name_map{$_} && defined $name_map{$_};
}
return @missing;
}
# Check for undocumented macros; ignore those in the "missing" file
# and do simple check for #define in our header files.
sub checkmacros {
my $count = 0;
my %seen;
my @missing;
if ( $opt_o ) {
@missing = loadmissing('util/missingmacro111.txt');
} elsif ( $opt_v ) {
@missing = loadmissing('util/missingmacro.txt');
}
foreach my $f ( files(TAGS => 'public_header') ) {
# Skip some internals we don't want to document yet.
my $b = basename($f);
next if $b eq 'asn1.h';
next if $b eq 'asn1t.h';
next if $b eq 'err.h';
open(IN, $f)
or die "Can't open $f, $!";
while ( <IN> ) {
next unless /^#\s*define\s*(\S+)\(/;
my $macro = "$1(3)"; # We know they're all in section 3
next if exists $name_map{$macro} || defined $seen{$macro};
next if $macro =~ /^i2d_/
|| $macro =~ /^d2i_/
|| $macro =~ /^DEPRECATEDIN/
|| $macro =~ /\Q_fnsig(3)\E$/
|| $macro =~ /^IMPLEMENT_/
|| $macro =~ /^_?DECLARE_/;
# Skip macros known to be missing
next if $opt_v && grep( /^\Q$macro\E$/, @missing);
err("$f:", "macro $macro undocumented")
if $opt_d || $opt_e;
$count++;
$seen{$macro} = 1;
}
close(IN);
}
err("# $count macros undocumented (count is approximate)")
if $count > 0;
}
# Find out what is undocumented (filtering out the known missing ones)
# and display them.
sub printem {
my $libname = shift;
my $numfile = shift;
my $missingfile = shift;
my $count = 0;
my %seen;
my @missing = loadmissing($missingfile) if $opt_v;
foreach my $func ( parsenum($numfile) ) {
$func .= '(3)'; # We know they're all in section 3
next if exists $name_map{$func} || defined $seen{$func};
# Skip functions known to be missing.
next if $opt_v && grep( /^\Q$func\E$/, @missing);
err("$libname:", "function $func undocumented")
if $opt_d || $opt_e;
$count++;
$seen{$func} = 1;
}
err("# $count in $numfile are not documented")
if $count > 0;
}
# Collect all the names in a manpage.
sub collectnames {
my $filename = shift;
$filename =~ m|man(\d)/|;
my $section = $1;
my $simplename = basename($filename, ".pod");
my $id = "${filename}:1:";
my %podinfo = extract_pod_info($filename, { debug => $debug });
unless ( grep { $simplename eq $_ } @{$podinfo{names}} ) {
err($id, "$simplename not in NAME section");
push @{$podinfo{names}}, $simplename;
}
foreach my $name ( @{$podinfo{names}} ) {
next if $name eq "";
err($id, "'$name' contains white space")
if $name =~ /\s/;
my $name_sec = "$name($section)";
if ( !exists $name_map{$name_sec} ) {
$name_map{$name_sec} = $filename;
} elsif ( $filename eq $name_map{$name_sec} ) {
err($id, "$name_sec duplicated in NAME section of",
$name_map{$name_sec});
} else {
err($id, "$name_sec also in NAME section of",
$name_map{$name_sec});
}
}
if ( $podinfo{contents} =~ /=for openssl foreign manual (.*)\n/ ) {
foreach my $f ( split / /, $1 ) {
$name_map{$f} = undef; # It still exists!
}
}
my @links =
$podinfo{contents} =~ /L<
# if the link is of the form L<something|name(s)>,
# then remove 'something'. Note that 'something'
# may contain POD codes as well...
(?:(?:[^\|]|<[^>]*>)*\|)?
# we're only interested in references that have
# a one digit section number
([^\/>\(]+\(\d\))
/gx;
$link_map{$filename} = [ @links ];
}
# Look for L<> ("link") references that point to files that do not exist.
sub checklinks {
foreach my $filename ( sort keys %link_map ) {
foreach my $link ( @{$link_map{$filename}} ) {
err("${filename}:1:", "reference to non-existing $link")
unless exists $name_map{$link};
}
}
}
# Load the public symbol/macro names
sub publicize {
foreach my $name ( parsenum('util/libcrypto.num') ) {
$public{$name} = 1;
}
foreach my $name ( parsenum('util/libssl.num') ) {
$public{$name} = 1;
}
foreach my $name ( parsenum('util/other.syms') ) {
$public{$name} = 1;
}
}
# Cipher/digests to skip if they show up as "not implemented"
# because they are, via the "-*" construct.
my %skips = (
'aes128' => 1,
'aes192' => 1,
'aes256' => 1,
'aria128' => 1,
'aria192' => 1,
'aria256' => 1,
'camellia128' => 1,
'camellia192' => 1,
'camellia256' => 1,
'des' => 1,
'des3' => 1,
'idea' => 1,
'cipher' => 1,
'digest' => 1,
);
# Check the flags of a command and see if everything is in the manpage
sub checkflags {
my $cmd = shift;
my $doc = shift;
my %cmdopts;
my %docopts;
my %localskips;
# Get the list of options in the command.
open CFH, "$openssl list --options $cmd|"
or die "Can list options for $cmd, $!";
while ( <CFH> ) {
chop;
s/ .$//;
$cmdopts{$_} = 1;
}
close CFH;
# Get the list of flags from the synopsis
open CFH, "<$doc"
or die "Can't open $doc, $!";
while ( <CFH> ) {
chop;
last if /DESCRIPTION/;
if ( /=for openssl ifdef (.*)/ ) {
foreach my $f ( split / /, $1 ) {
$localskips{$f} = 1;
}
next;
}
my $opt;
if ( /\[B<-([^ >]+)/ ) {
$opt = $1;
} elsif ( /^B<-([^ >]+)/ ) {
$opt = $1;
} else {
next;
}
$opt = $1 if $opt =~ /I<(.*)/;
$docopts{$1} = 1;
}
close CFH;
# See what's in the command not the manpage.
my @undocced = sort grep { !defined $docopts{$_} } keys %cmdopts;
foreach ( @undocced ) {
next if /-/; # Skip the -- end-of-flags marker
err("$doc: undocumented option -$_");
}
# See what's in the command not the manpage.
my @unimpl = sort grep { !defined $cmdopts{$_} } keys %docopts;
foreach ( @unimpl ) {
next if defined $skips{$_} || defined $localskips{$_};
err("$doc: $cmd does not implement -$_");
}
}
##
## MAIN()
## Do the work requested by the various getopt flags.
## The flags are parsed in alphabetical order, just because we have
## to have *some way* of listing them.
##
if ( $opt_c ) {
my @commands = ();
# Get list of commands.
open FH, "$openssl list -1 -commands|"
or die "Can't list commands, $!";
while ( <FH> ) {
chop;
push @commands, $_;
}
close FH;
# See if each has a manpage.
foreach my $cmd ( @commands ) {
next if $cmd eq 'help' || $cmd eq 'exit';
my @doc = ( grep { basename($_) eq "openssl-$cmd.pod"
# For "tsget" and "CA.pl" pod pages
|| basename($_) eq "$cmd.pod" }
files(TAGS => [ 'manual', 'man1' ]) );
my $num = scalar @doc;
if ($num > 1) {
err("$num manuals for 'openssl $cmd': ".join(", ", @doc));
} elsif ($num < 1) {
err("no manual for 'openssl $cmd'");
} else {
checkflags($cmd, @doc);
}
}
# See what help is missing.
open FH, "$openssl list --missing-help |"
or die "Can't list missing help, $!";
while ( <FH> ) {
chop;
my ($cmd, $flag) = split;
err("$cmd has no help for -$flag");
}
close FH;
exit $status;
}
# Preparation for some options, populate %name_map and %link_map
if ( $opt_l || $opt_u || $opt_v ) {
foreach ( files(TAGS => 'manual') ) {
collectnames($_);
}
}
if ( $opt_l ) {
foreach my $func ( loadmissing("util/missingcrypto.txt") ) {
$name_map{$func} = undef;
}
checklinks();
}
if ( $opt_n ) {
publicize();
foreach ( @ARGV ? @ARGV : files(TAGS => 'manual') ) {
check($_);
}
# If not given args, check that all man1 commands are named properly.
if ( scalar @ARGV == 0 ) {
foreach ( files(TAGS => [ 'public_manual', 'man1' ]) ) {
next if /CA.pl/ || /openssl\.pod/ || /tsget\.pod/;
err("$_ doesn't start with openssl-") unless /openssl-/;
}
}
}
if ( $opt_u || $opt_v) {
if ( $opt_o ) {
printem('crypto', 'util/libcrypto.num', 'util/missingcrypto111.txt');
printem('ssl', 'util/libssl.num', 'util/missingssl111.txt');
} else {
printem('crypto', 'util/libcrypto.num', 'util/missingcrypto.txt');
printem('ssl', 'util/libssl.num', 'util/missingssl.txt');
}
checkmacros();
}
exit $status;