#! /usr/bin/env perl
-# Copyright 2002-2019 The OpenSSL Project Authors. All Rights Reserved.
+# Copyright 2002-2021 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
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 lib catdir(dirname($0), "perl");
+use FindBin;
+use lib "$FindBin::Bin/perl";
+
use OpenSSL::Util::Pod;
+use lib '.';
+use configdata;
+
# Set to 1 for debug output
my $debug = 0;
our($opt_o);
our($opt_h);
our($opt_l);
+our($opt_m);
our($opt_n);
our($opt_p);
our($opt_u);
sub help {
print <<EOF;
Find small errors (nits) in documentation. Options:
- -c List undocumented commands and options
+ -c List undocumented commands, undocumented options and unimplemented 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
+ -m Name(s) of manuals to focus on. Default: man1,man3,man5,man7
-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
exit;
}
-getopts('cdehlnouv');
+getopts('cdehlm:nouv');
help() if $opt_h;
$opt_u = 1 if $opt_d;
my $temp = '/tmp/docnits.txt';
my $OUT;
-my %public;
my $status = 0;
+$opt_m = "man1,man3,man5,man7" unless $opt_m;
+die "Argument of -m option may contain only man1, man3, man5, and/or man7"
+ unless $opt_m =~ /^(man[1357][, ]?)*$/;
+my @sections = ( split /[, ]/, $opt_m );
+
my %mandatory_sections = (
'*' => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
1 => [ 'SYNOPSIS', 'OPTIONS' ],
3 => [ 'SYNOPSIS', 'RETURN VALUES' ],
5 => [ ],
7 => [ ]
-);
+ );
+
+# Symbols that we ignored.
+# They are reserved macros that we currently don't document
+my $ignored = qr/(?| ^i2d_
+ | ^d2i_
+ | ^DEPRECATEDIN
+ | ^OSSL_DEPRECATED
+ | \Q_fnsig(3)\E$
+ | ^IMPLEMENT_
+ | ^_?DECLARE_
+ | ^sk_
+ | ^SKM_DEFINE_STACK_OF_INTERNAL
+ | ^lh_
+ | ^DEFINE_LHASH_OF_(INTERNAL|DEPRECATED)
+ )/x;
+
+# A common regexp for C symbol names
+my $C_symbol = qr/\b[[:alpha:]][_[:alnum:]]*\b/;
+
+# 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 included. 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 {
return unless $contents =~ /=head1 NAME(.*)=head1 SYNOPSIS/ms;
my $tmp = $1;
$tmp =~ tr/\n/ /;
- err($id, "trailing comma before - in NAME")
+ 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")
+ err($id, "Missing comma in NAME")
if $tmp =~ /[^,] /;
my $dirname = dirname($filename);
- my $simplename = basename(basename($filename, ".in"), ".pod");
+ 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")
+ err($id, "The name '$n' contains white-space")
if $n =~ /\s/;
$names{$n} = 1;
$foundfilename++ if $n eq $simplename;
$foundfilenames{$n} = 1
- if -f "$dirname/$n.pod" && $n ne $simplename;
+ if ( ( grep { basename($_) eq "$n.pod" }
+ files(TAGS => [ 'manual', $section ]) )
+ && $n ne $simplename );
}
- err($id, "the following exist as other .pod files:",
+ 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 $ignore_until = undef; # If defined, this is a regexp
+ # Remove all non-code lines
+ $syn =~ s/^(?:\s*?|\S.*?)$//msg;
+ # Remove all comments
+ $syn =~ s/\/\*.*?\*\///msg;
+ while ( $syn ) {
+ # "env" lines end at a newline.
+ # Preprocessor lines start with a # and end at a newline.
+ # Other lines end with a semicolon, and may cover more than
+ # one physical line.
+ if ( $syn !~ /^ \s*(env .*?|#.*?|.*?;)\s*$/ms ) {
+ err($id, "Can't parse rest of synopsis:\n$syn\n(declarations not ending with a semicolon (;)?)");
+ last;
+ }
+ my $line = $1;
+ $syn = $';
+
+ print STDERR "DEBUG[name_synopsis] \$line = '$line'\n" if $debug;
+
+ # Special code to skip over documented structures
+ if ( defined $ignore_until) {
+ next if $line !~ /$ignore_until/;
+ $ignore_until = undef;
+ next;
+ }
+ if ( $line =~ /^\s*(?:typedef\s+)?struct(?:\s+\S+)\s*\{/ ) {
+ $ignore_until = qr/\}.*?;/;
+ next;
+ }
+
my $sym;
+ my $is_prototype = 1;
+ $line =~ s/LHASH_OF\([^)]+\)/int/g;
$line =~ s/STACK_OF\([^)]+\)/int/g;
$line =~ s/SPARSE_ARRAY_OF\([^)]+\)/int/g;
$line =~ s/__declspec\([^)]+\)//;
+
+ ## We don't prohibit that space, to allow typedefs looking like
+ ## this:
+ ##
+ ## typedef int (fantastically_long_name_breaks_80char_limit)
+ ## (fantastically_long_name_breaks_80char_limit *something);
+ ##
+ #if ( $line =~ /typedef.*\(\*?\S+\)\s+\(/ ) {
+ # # a callback function with whitespace before the argument list:
+ # # typedef ... (*NAME) (...
+ # # 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.*\(\*(\S+)\)\(.*/ ) {
+ } elsif ( $line =~ /typedef.*\(\*?($C_symbol)\)\s*\(/ ) {
# a callback function pointer: typedef ... (*NAME)(...
+ # a callback function signature: typedef ... (NAME)(...
$sym = $1;
- } elsif ( $line =~ /typedef.* (\S+)\(.*/ ) {
+ } elsif ( $line =~ /typedef.*($C_symbol)\s*\(/ ) {
# a callback function signature: typedef ... NAME(...
$sym = $1;
- } elsif ( $line =~ /typedef.* (\S+);/ ) {
+ } elsif ( $line =~ /typedef.*($C_symbol);/ ) {
# a simple typedef: typedef ... NAME;
+ $is_prototype = 0;
$sym = $1;
- } elsif ( $line =~ /enum (\S*) \{/ ) {
+ } elsif ( $line =~ /enum ($C_symbol) \{/ ) {
# an enumeration: enum ... {
$sym = $1;
- } elsif ( $line =~ /#(?:define|undef) ([A-Za-z0-9_]+)/ ) {
+ } elsif ( $line =~ /#\s*(?:define|undef) ($C_symbol)/ ) {
+ $is_prototype = 0;
$sym = $1;
- } elsif ( $line =~ /([A-Za-z0-9_]+)\(/ ) {
+ } elsif ( $line =~ /^[^\(]*?\(\*($C_symbol)\s*\(/ ) {
+ # a function returning a function pointer: TYPE (*NAME(args))(args)
+ $sym = $1;
+ } elsif ( $line =~ /^[^\(]*?($C_symbol)\s*\(/ ) {
+ # a simple function declaration
$sym = $1;
}
else {
next;
}
+
+ print STDERR "DEBUG[name_synopsis] \$sym = '$sym'\n" if $debug;
+
err($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 ( $line =~ /[a-z0-9],[^ ]/ );
+ err($id, "Prototype missing spaces around commas: $line")
+ if $is_prototype && $line =~ /[a-z0-9],[^\s]/;
}
foreach my $n ( keys %names ) {
foreach my $line ( split /\n+/, $contents ) {
next unless $line =~ /^=head/;
if ( $line =~ /head1/ ) {
- err($id, "duplicate section $line")
+ err($id, "Duplicate section $line")
if defined $head1{$line};
$head1{$line} = 1;
%subheads = ();
} else {
- err($id, "duplicate subsection $line")
+ err($id, "Duplicate subsection $line")
if defined $subheads{$line};
$subheads{$line} = 1;
}
- err($id, "period in =head")
+ err($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")
+ err($id, "All uppercase in subhead")
if $line =~ /head[234][ A-Z0-9]+$/;
}
}
my $markup_re =
qr/( # Capture group
[BIL]< # The start of what we recurse on
- (?:(?-1)|.)*? # recurse the whole regexp (refering to
+ (?:(?-1)|.)*? # recurse the whole regexp (referring to
# the last opened capture group, i.e. the
# start of this regexp), or pick next
# character. Do NOT be greedy!
err($id, "Malformed option [1] in SYNOPSIS: $&");
}
+ my @synopsis;
while ( $synopsis =~ /$markup_re/msg ) {
my $found = $&;
+ push @synopsis, $found if $found =~ /^B<-/;
print STDERR "$id:DEBUG[option_check] SYNOPSIS: found $found\n"
if $debug;
my $option_uw = normalise_option($id, $filename, $found);
# In OPTIONS, we look for =item paragraphs.
# (?=^\s*$) detects an empty line.
+ my @options;
while ( $options =~ /=item\s+(.*?)(?=^\s*$)/msg ) {
my $item = $&;
my $option_uw = normalise_option($id, $filename, $found);
err($id, "Malformed option in OPTIONS: $found")
if defined $option_uw && $option_uw eq '';
+ if ($found =~ /^B<-/) {
+ push @options, $found;
+ err($id, "OPTIONS entry $found missing from SYNOPSIS")
+ unless (grep /^\Q$found\E$/, @synopsis)
+ || $id =~ /(openssl|-options)\.pod:1:$/;
+ }
}
}
+ foreach (@synopsis) {
+ my $option = $_;
+ err($id, "SYNOPSIS entry $option missing from OPTIONS")
+ unless (grep /^\Q$option\E$/, @options);
+ }
}
# Normal symbol form
$unmarked =~ s/[BIL]<|>//msg;
err($id, "Malformed symbol: $symbol")
- unless $symbol =~ /^B<.*>$/ && $unmarked =~ /^${symbol_re}$/
+ unless $symbol =~ /^B<.*?>$/ && $unmarked =~ /^${symbol_re}$/
}
# We can't do the kind of collecting coolness that option_check()
# This is from http://man7.org/linux/man-pages/man7/man-pages.7.html
my %preferred_words = (
+ '16bit' => '16-bit',
+ 'a.k.a.' => 'aka',
'bitmask' => 'bit mask',
'builtin' => 'built-in',
#'epoch' => 'Epoch', # handled specially, below
+ 'fall-back' => 'fallback',
'file name' => 'filename',
'file system' => 'filesystem',
'host name' => 'hostname',
'i-node' => 'inode',
'lower case' => 'lowercase',
'lower-case' => 'lowercase',
+ 'manpage' => 'man page',
+ 'non-blocking' => 'nonblocking',
+ 'non-default' => 'nondefault',
+ 'non-empty' => 'nonempty',
+ 'non-negative' => 'nonnegative',
'non-zero' => 'nonzero',
'path name' => 'pathname',
+ 'pre-allocated' => 'preallocated',
'pseudo-terminal' => 'pseudoterminal',
- 'reserved port' => 'privileged port',
- 'system port' => 'privileged port',
- 'realtime' => 'real-time',
'real time' => 'real-time',
+ 'realtime' => 'real-time',
+ 'reserved port' => 'privileged port',
'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',
+ 'saved user ID' => 'saved set-user-ID',
'set-GID' => 'set-group-ID',
- 'setgid' => 'set-group-ID',
'set-UID' => 'set-user-ID',
+ 'setgid' => 'set-group-ID',
'setuid' => 'set-user-ID',
- 'super user' => 'superuser',
- 'super-user' => 'superuser',
+ 'sub-system' => 'subsystem',
'super block' => 'superblock',
'super-block' => 'superblock',
+ 'super user' => 'superuser',
+ 'super-user' => 'superuser',
+ 'system port' => 'privileged port',
'time stamp' => 'timestamp',
'time zone' => 'timezone',
'upper case' => 'uppercase',
'upper-case' => 'uppercase',
'useable' => 'usable',
- 'userspace' => 'user space',
'user name' => 'username',
+ 'userspace' => 'user space',
'zeroes' => 'zeros'
);
# Sigh, trademark
next if $k eq 'file system'
and $contents =~ /Microsoft Encrypted File System/;
- err($id, "found '$k' should use '$preferred_words{$k}'")
+ err($id, "Found '$k' should use '$preferred_words{$k}'")
if $contents =~ /\b\Q$k\E\b/i;
}
- err($id, "found 'epoch' should use 'Epoch'")
+ err($id, "Found 'epoch' should use 'Epoch'")
if $contents =~ /\bepoch\b/;
+ if ( $id =~ m@man1/@ ) {
+ err($id, "found 'tool' in NAME, should use 'command'")
+ if $contents =~ /=head1 NAME.*\btool\b.*=head1 SYNOPSIS/s;
+ err($id, "found 'utility' in NAME, should use 'command'")
+ if $contents =~ /NAME.*\butility\b.*=head1 SYNOPSIS/s;
+
+ }
}
# Perform all sorts of nit/error checks on a manpage
sub check {
- my $filename = shift;
+ my %podinfo = @_;
+ my $filename = $podinfo{filename};
my $dirname = basename(dirname($filename));
-
- my $contents = '';
- {
- local $/ = undef;
- open POD, $filename or die "Couldn't open $filename, $!";
- $contents = <POD>;
- close POD;
- }
+ my $contents = $podinfo{contents};
my $id = "${filename}:1:";
check_head_style($id, $contents);
while ( $contents =~ /L<([^>]*)\(1\)(?:\/.*)?>/g ) {
my $target = $1;
next if $target =~ /openssl-?/;
- next if -f "doc/man1/$target.pod";
- # TODO: Filter out "foreign manual" links.
+ next if ( grep { basename($_) eq "$target.pod" }
+ files(TAGS => [ 'manual', 'man1' ]) );
next if $target =~ /ps|apropos|sha1sum|procmail|perl/;
- err($id, "Bad command link L<$target(1)>");
+ err($id, "Bad command link L<$target(1)>") if grep /man1/, @sections;
}
# Check for proper in-man-3 API links.
while ( $contents =~ /L<([^>]*)\(3\)(?:\/.*)?>/g ) {
unless $target =~ /^[_[:alpha:]][_[:alnum:]]*$/
}
- unless ( $contents =~ /=for openssl generic/ ) {
+ unless ( $contents =~ /^=for openssl generic/ms ) {
if ( $filename =~ m|man3/| ) {
name_synopsis($id, $filename, $contents);
functionname_check($id, $filename, $contents);
wording($id, $contents);
- err($id, "doesn't start with =pod")
+ err($id, "Doesn't start with =pod")
if $contents !~ /^=pod/;
- err($id, "doesn't end with =cut")
+ err($id, "Doesn't end with =cut")
if $contents !~ /=cut\n$/;
- err($id, "more than one cut line.")
+ err($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")
+ err($id, "Missing copyright")
if $contents !~ /Copyright .* The OpenSSL Project Authors/;
- err($id, "copyright not last")
+ 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")
+ err($id, "Extra space after head")
if $contents =~ /=head\d\s\s+/;
- err($id, "period in NAME section")
+ 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;
my $count = 0;
foreach my $line ( split /\n+/, $1 ) {
if ( $line =~ m@include <openssl/@ ) {
- err($id, "has multiple includes")
+ err($id, "Has multiple includes")
if ++$count == 2;
} else {
$count = 0;
open my $OUT, '>', $temp
or die "Can't open $temp, $!";
- podchecker($filename, $OUT);
+ err($id, "POD errors")
+ if podchecker($filename, $OUT) != 0;
close $OUT;
open $OUT, '<', $temp
or die "Can't read $temp, $!";
$section = $1 if $dirname =~ /man([1-9])/;
foreach ( (@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}}) ) {
- err($id, "missing $_ head1 section")
+ err($id, "Missing $_ head1 section")
if $contents !~ /^=head1\s+${_}\s*$/m;
}
}
+# Information database ###############################################
+
+# Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
+my %link_map = ();
+# Map of names in each POD file or from "missing" files; possible values are:
+# If found in a POD files, "name(s)" => filename
+# If found in a "missing" file or external, "name(s)" => ''
+my %name_map = ();
+
+# State of man-page names.
+# %state is affected by loading util/*.num and util/*.syms
+# Values may be one of:
+# 'crypto' : belongs in libcrypto (loaded from libcrypto.num)
+# 'ssl' : belongs in libssl (loaded from libssl.num)
+# 'other' : belongs in libcrypto or libssl (loaded from other.syms)
+# 'internal' : Internal
+# 'public' : Public (generic name or external documentation)
+# Any of these values except 'public' may be prefixed with 'missing_'
+# to indicate that they are known to be missing.
+my %state;
+# %missing is affected by loading util/missing*.txt. Values may be one of:
+# 'crypto' : belongs in libcrypto (loaded from libcrypto.num)
+# 'ssl' : belongs in libssl (loaded from libssl.num)
+# 'other' : belongs in libcrypto or libssl (loaded from other.syms)
+# 'internal' : Internal
+my %missing;
+
# Parse libcrypto.num, etc., and return sorted list of what's there.
-sub parsenum {
+sub loadnum ($;$) {
my $file = shift;
- my @apis;
+ my $type = shift;
+ my @symbols;
- open my $IN, '<', $file
+ 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 $_"
+ die "Malformed line $. in $file: $_"
if scalar @fields != 2 && scalar @fields != 4;
- push @apis, $fields[0];
+ $state{$fields[0].'(3)'} = $type // 'internal';
}
-
close $IN;
-
- return sort @apis;
-}
-
-# Parse all the manpages, getting return map of what they document
-# (by looking at their NAME sections).
-sub getdocced
-{
- my $dir = shift;
- my %return;
- my %dups;
-
- foreach my $pod ( glob("$dir/*.pod") ) {
- my %podinfo = extract_pod_info($pod);
- foreach my $n ( @{$podinfo{names}} ) {
- $return{$n} = $pod;
- err("# Duplicate $n in $pod and $dups{$n}")
- if defined $dups{$n} && $dups{$n} ne $pod;
- $dups{$n} = $pod;
- }
- }
-
- return %return;
}
-# Map of documented functions; function => manpage
-my %docced;
-# Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
-my %link_map = ();
-# Map of names in each POD file; "name(s)" => filename
-my %name_map = ();
-
# Load file of symbol names that we know aren't documented.
-sub loadmissing($)
+sub loadmissing($;$)
{
my $missingfile = shift;
- my @missing;
+ my $type = shift;
- open FH, $missingfile
- || die "Can't open $missingfile";
+ open FH, catfile($config{sourcedir}, $missingfile)
+ or die "Can't open $missingfile";
while ( <FH> ) {
chomp;
next if /^#/;
- push @missing, $_;
+ $missing{$_} = $type // 'internal';
}
close FH;
+}
- return @missing;
+# Check that we have consistent public / internal documentation and declaration
+sub checkstate () {
+ # Collect all known names, no matter where they come from
+ my %names = map { $_ => 1 } (keys %name_map, keys %state, keys %missing);
+
+ # Check section 3, i.e. functions and macros
+ foreach ( grep { $_ =~ /\(3\)$/ } sort keys %names ) {
+ next if ( $name_map{$_} // '') eq '' || $_ =~ /$ignored/;
+
+ # If a man-page isn't recorded public or if it's recorded missing
+ # and internal, it's declared to be internal.
+ my $declared_internal =
+ ($state{$_} // 'internal') eq 'internal'
+ || ($missing{$_} // '') eq 'internal';
+ # If a man-page isn't recorded internal or if it's recorded missing
+ # and not internal, it's declared to be public
+ my $declared_public =
+ ($state{$_} // 'internal') ne 'internal'
+ || ($missing{$_} // 'internal') ne 'internal';
+
+ err("$_ is supposedly public but is documented as internal")
+ if ( $declared_public && $name_map{$_} =~ /\/internal\// );
+ err("$_ is supposedly internal (maybe missing from other.syms) but is documented as public")
+ if ( $declared_internal && $name_map{$_} !~ /\/internal\// );
+ }
}
# Check for undocumented macros; ignore those in the "missing" file
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 ( glob('include/openssl/*.h') ) {
+ foreach my $f ( files(TAGS => 'public_header') ) {
# Skip some internals we don't want to document yet.
- next if $f eq 'include/openssl/asn1.h';
- next if $f eq 'include/openssl/asn1t.h';
- next if $f eq 'include/openssl/err.h';
- open(IN, $f) || die "Can't open $f, $!";
+ 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;
- next if $docced{$macro} || defined $seen{$macro};
- next if $macro =~ /i2d_/
- || $macro =~ /d2i_/
- || $macro =~ /DEPRECATEDIN/
- || $macro =~ /IMPLEMENT_/
- || $macro =~ /DECLARE_/;
-
- # Skip macros known to be missing
- next if $opt_v && grep( /^$macro$/, @missing);
-
+ my $macro = "$1(3)"; # We know they're all in section 3
+ next if defined $name_map{$macro}
+ || defined $missing{$macro}
+ || defined $seen{$macro}
+ || $macro =~ /$ignored/;
+
err("$f:", "macro $macro undocumented")
if $opt_d || $opt_e;
$count++;
# Find out what is undocumented (filtering out the known missing ones)
# and display them.
-sub printem {
- my $libname = shift;
- my $numfile = shift;
- my $missingfile = shift;
+sub printem ($) {
+ my $type = shift;
my $count = 0;
- my %seen;
- my @missing = loadmissing($missingfile) if ( $opt_v );
+ foreach my $func ( grep { $state{$_} eq $type } sort keys %state ) {
+ next if defined $name_map{$func}
+ || defined $missing{$func};
- foreach my $func ( parsenum($numfile) ) {
- next if $docced{$func} || defined $seen{$func};
-
- # Skip ASN1 utilities
- next if $func =~ /^ASN1_/;
-
- # Skip functions known to be missing
- next if $opt_v && grep( /^$func$/, @missing);
-
- err("$libname:", "function $func undocumented")
+ err("$type:", "function $func undocumented")
if $opt_d || $opt_e;
$count++;
- $seen{$func} = 1;
}
- err("# $count in $numfile are not documented")
+ err("# $count lib$type names are not documented")
if $count > 0;
}
# Collect all the names in a manpage.
sub collectnames {
- my $filename = shift;
+ my %podinfo = @_;
+ my $filename = $podinfo{filename};
$filename =~ m|man(\d)/|;
my $section = $1;
my $simplename = basename($filename, ".pod");
my $id = "${filename}:1:";
+ my $is_generic = $podinfo{contents} =~ /^=for openssl generic/ms;
- my $contents = '';
- {
- local $/ = undef;
- open POD, $filename or die "Couldn't open $filename, $!";
- $contents = <POD>;
- close POD;
- }
-
- $contents =~ /=head1 NAME([^=]*)=head1 /ms;
- my $tmp = $1;
- unless ( defined $tmp ) {
- err($id, "weird name section");
- return;
- }
- $tmp =~ tr/\n/ /;
- $tmp =~ s/ -.*//g;
-
- my @names =
- map { s|/|-|g; $_ } # Treat slash as dash
- map { s/^\s+//g; s/\s+$//g; $_ } # Trim prefix and suffix blanks
- split(/,/, $tmp);
- unless ( grep { $simplename eq $_ } @names ) {
- err($id, "missing $simplename");
- push @names, $simplename;
+ unless ( grep { $simplename eq $_ } @{$podinfo{names}} ) {
+ err($id, "$simplename not in NAME section");
+ push @{$podinfo{names}}, $simplename;
}
- foreach my $name (@names) {
+ foreach my $name ( @{$podinfo{names}} ) {
next if $name eq "";
- if ( $name =~ /\s/ ) {
- err($id, "'$name' contains white space")
- }
+ err($id, "'$name' contains whitespace")
+ if $name =~ /\s/;
my $name_sec = "$name($section)";
- if ( !exists $name_map{$name_sec} ) {
+ if ( !defined $name_map{$name_sec} ) {
$name_map{$name_sec} = $filename;
+ $state{$name_sec} //=
+ ( $filename =~ /\/internal\// ? 'internal' : 'public' )
+ if $is_generic;
} elsif ( $filename eq $name_map{$name_sec} ) {
- err($id, "$name_sec repeated in NAME section of",
+ err($id, "$name_sec duplicated in NAME section of",
$name_map{$name_sec});
- } else {
+ } elsif ( $name_map{$name_sec} ne '' ) {
err($id, "$name_sec also in NAME section of",
$name_map{$name_sec});
}
}
- my @foreign_names =
- map { map { s/\s+//g; $_ } split(/,/, $_) }
- $contents =~ /=for\s+comment\s+foreign\s+manuals:\s*(.*)\n\n/;
- foreach ( @foreign_names ) {
- $name_map{$_} = undef; # It still exists!
+ if ( $podinfo{contents} =~ /=for openssl foreign manual (.*)\n/ ) {
+ foreach my $f ( split / /, $1 ) {
+ $name_map{$f} = ''; # It still exists!
+ $state{$f} = 'public'; # We assume!
+ }
}
- my @links = $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;
+ my @links = ();
+ # Don't use this regexp directly on $podinfo{contents}, as it causes
+ # a regexp recursion, which fails on really big PODs. Instead, use
+ # $markup_re to pick up general markup, and use this regexp to check
+ # that the markup that was found is indeed a link.
+ my $linkre = qr/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
+ ([^\/>\(]+\(\d\))
+ /x;
+ while ( $podinfo{contents} =~ /$markup_re/msg ) {
+ my $x = $1;
+
+ if ($x =~ $linkre) {
+ push @links, $1;
+ }
+ }
$link_map{$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}}) {
+ 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};
+ unless defined $name_map{$link} || defined $missing{$link};
+ err("${filename}:1:", "reference of internal $link in public documentation $filename")
+ if ( ( ($state{$link} // '') eq 'internal'
+ || ($missing{$link} // '') eq 'internal' )
+ && $filename !~ /\/internal\// );
}
}
}
-# 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 = (
'digest' => 1,
);
+my %genopts; # generic options parsed from apps/include/opt.h
+
# 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 @cmdopts;
my %docopts;
- my %localskips;
- # Get the list of options in the command.
- open CFH, "./apps/openssl list --options $cmd|"
- || die "Can list options for $cmd, $!";
+ # Get the list of options in the command source file.
+ my $active = 0;
+ my $expect_helpstr = "";
+ open CFH, "apps/$cmd.c"
+ or die "Can't open apps/$cmd.c to list options for $cmd, $!";
while ( <CFH> ) {
chop;
- s/ .$//;
- $cmdopts{$_} = 1;
+ if ($active) {
+ last if m/^\s*};/;
+ if ($expect_helpstr ne "") {
+ next if m/^\s*#\s*if/;
+ err("$cmd does not implement help for -$expect_helpstr") unless m/^\s*"/;
+ $expect_helpstr = "";
+ }
+ if (m/\{\s*"([^"]+)"\s*,\s*OPT_[A-Z0-9_]+\s*,\s*('[-\/:<>cEfFlMnNpsuU]'|0)(.*)$/
+ && !($cmd eq "s_client" && $1 eq "wdebug")) {
+ push @cmdopts, $1;
+ $expect_helpstr = $1;
+ $expect_helpstr = "" if $3 =~ m/^\s*,\s*"/;
+ } elsif (m/[\s,](OPT_[A-Z]+_OPTIONS?)\s*(,|$)/) {
+ push @cmdopts, @{ $genopts{$1} };
+ }
+ } elsif (m/^const\s+OPTIONS\s*/) {
+ $active = 1;
+ }
}
close CFH;
# Get the list of flags from the synopsis
open CFH, "<$doc"
- || die "Can't open $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;
- }
+ my $opt;
+ if ( /\[B<-([^ >]+)/ ) {
+ $opt = $1;
+ } elsif ( /^B<-([^ >]+)/ ) {
+ $opt = $1;
+ } else {
next;
}
- next unless /\[B<-([^ >]+)/;
- my $opt = $1;
$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;
+ my @undocced = sort grep { !defined $docopts{$_} } @cmdopts;
foreach ( @undocced ) {
- next if /-/; # Skip the -- end-of-flags marker
- err("$doc: undocumented option -$_");
+ err("$doc: undocumented $cmd option -$_");
}
# See what's in the command not the manpage.
- my @unimpl = sort grep { !defined $cmdopts{$_} } keys %docopts;
+ my @unimpl = sort grep { my $e = $_; !(grep /^\Q$e\E$/, @cmdopts) } keys %docopts;
foreach ( @unimpl ) {
- next if defined $skips{$_} || defined $localskips{$_};
- err("$cmd documented but not implemented -$_");
+ next if $_ eq "-"; # Skip the -- end-of-flags marker
+ next if defined $skips{$_};
+ err("$doc: $cmd does not implement -$_");
}
}
if ( $opt_c ) {
my @commands = ();
- # Get list of commands.
- open FH, "./apps/openssl list -1 -commands|"
- || die "Can't list commands, $!";
- while ( <FH> ) {
+ # Get the lists of generic options.
+ my $active = "";
+ open OFH, catdir($config{sourcedir}, "apps/include/opt.h")
+ or die "Can't open apps/include/opt.h to list generic options, $!";
+ while ( <OFH> ) {
chop;
- push @commands, $_;
+ push @{ $genopts{$active} }, $1 if $active ne "" && m/^\s+\{\s*"([^"]+)"\s*,\s*OPT_/;
+ $active = $1 if m/^\s*#\s*define\s+(OPT_[A-Z]+_OPTIONS?)\s*\\\s*$/;
+ $active = "" if m/^\s*$/;
}
- close FH;
+ close OFH;
+
+ # Get list of commands.
+ opendir(DIR, "apps");
+ @commands = grep(/\.c$/, readdir(DIR));
+ closedir(DIR);
# See if each has a manpage.
foreach my $cmd ( @commands ) {
- next if $cmd eq 'help' || $cmd eq 'exit';
- my $doc = "doc/man1/$cmd.pod";
- $doc = "doc/man1/openssl-$cmd.pod" if -f "doc/man1/openssl-$cmd.pod";
- if ( ! -f "$doc" ) {
- err("$doc does not exist");
+ $cmd =~ s/\.c$//;
+ next if $cmd eq 'progs' || $cmd eq 'vms_decc_init';
+ my @doc = ( grep { basename($_) 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, $doc);
+ checkflags($cmd, @doc);
}
}
+}
- # See what help is missing.
- open FH, "./apps/openssl list --missing-help |"
- || die "Can't list missing help, $!";
- while ( <FH> ) {
- chop;
- my ($cmd, $flag) = split;
- err("$cmd has no help for -$flag");
- }
- close FH;
+# Populate %state
+loadnum('util/libcrypto.num', 'crypto');
+loadnum('util/libssl.num', 'ssl');
+loadnum('util/other.syms', 'other');
+loadnum('util/other-internal.syms');
+if ( $opt_o ) {
+ loadmissing('util/missingmacro111.txt', 'crypto');
+ loadmissing('util/missingcrypto111.txt', 'crypto');
+ loadmissing('util/missingssl111.txt', 'ssl');
+} elsif ( !$opt_u ) {
+ loadmissing('util/missingmacro.txt', 'crypto');
+ loadmissing('util/missingcrypto.txt', 'crypto');
+ loadmissing('util/missingssl.txt', 'ssl');
+ loadmissing('util/missingcrypto-internal.txt');
+ loadmissing('util/missingssl-internal.txt');
+}
+
+if ( $opt_n || $opt_l || $opt_u || $opt_v ) {
+ my @files_to_read = ( $opt_n && @ARGV ) ? @ARGV : files(TAGS => 'manual');
- exit $status;
+ foreach (@files_to_read) {
+ my %podinfo = extract_pod_info($_, { debug => $debug });
+
+ collectnames(%podinfo)
+ if ( $opt_l || $opt_u || $opt_v );
+
+ check(%podinfo)
+ if ( $opt_n );
+ }
}
if ( $opt_l ) {
- foreach ( glob('doc/*/*.pod doc/internal/*/*.pod') ) {
- collectnames($_);
- }
checklinks();
}
if ( $opt_n ) {
- publicize();
- foreach ( @ARGV ? @ARGV : glob('doc/*/*.pod doc/internal/*/*.pod') ) {
- check($_);
- }
-
# If not given args, check that all man1 commands are named properly.
- if ( scalar @ARGV == 0 ) {
- foreach (glob('doc/man1/*.pod')) {
- next if /CA.pl/ || /openssl\.pod/ || /tsget\.pod/;
+ if ( scalar @ARGV == 0 && grep /man1/, @sections ) {
+ foreach ( files(TAGS => [ 'public_manual', 'man1' ]) ) {
+ next if /openssl\.pod/
+ || /CA\.pl/ || /tsget\.pod/; # these commands are special cases
err("$_ doesn't start with openssl-") unless /openssl-/;
}
}
}
+checkstate();
+
if ( $opt_u || $opt_v) {
- my %temp = getdocced('doc/man3');
- foreach ( keys %temp ) {
- $docced{$_} = $temp{$_};
- }
- 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');
- }
+ printem('crypto');
+ printem('ssl');
checkmacros();
}