Hakase/openssl-custom
1
0 Fork 0
Code Issues Pull Requests Releases Wiki Activity

Latest update.

Browse Source
This commit is contained in:
Hakase
2019-10-17 23:54:38 +09:00
parent 41a23ae6f6
commit ee84d0dd84
1357 changed files with 41111 additions and 9603 deletions
Download Patch File Download Diff File Expand all files Collapse all files
util/find-doc-nits
+333 -109
View File
@@ -10,6 +10,7 @@
require 5.10.0;
use warnings;
use strict;
use Pod::Checker;
use File::Find;
use File::Basename;
@@ -18,6 +19,9 @@ use Getopt::Std;
use lib catdir(dirname($0), "perl");
use OpenSSL::Util::Pod;
# Set to 1 for debug output
my $debug = 0;
# Options.
our($opt_d);
our($opt_e);
@@ -31,35 +35,52 @@ our($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)
-s Same as -e except no output is generated if nothing is undocumented
-o Causes -e/-v to count symbols added since 1.1.1 as new (implies -v)
-h Print this help message
-l Print bogus links
-n Print nits in POD pages
-p Warn if non-public name documented (implies -n)
-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
-h Print this help message
-c List undocumented commands and options
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 %mandatory_sections =
( '*' => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
1 => [ 'SYNOPSIS', 'OPTIONS' ],
3 => [ 'SYNOPSIS', 'RETURN VALUES' ],
5 => [ ],
7 => [ ] );
my %mandatory_sections = (
'*' => [ 'NAME', 'DESCRIPTION', 'COPYRIGHT' ],
1 => [ 'SYNOPSIS', 'OPTIONS' ],
3 => [ 'SYNOPSIS', 'RETURN VALUES' ],
5 => [ ],
7 => [ ]
);
# Print error message, set $status.
sub err {
@@ -99,17 +120,18 @@ sub name_synopsis {
$names{$n} = 1;
$foundfilename++ if $n eq $simplename;
$foundfilenames{$n} = 1
if ((-f "$dirname/$n.pod.in" || -f "$dirname/$n.pod")
&& $n ne $simplename);
if -f "$dirname/$n.pod" && $n ne $simplename;
}
err($id, "the following exist as other .pod or .pod.in 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;
foreach my $n ( keys %names ) {
err($id, "$n is not public")
if $opt_p and !defined $public{$n};
if ( $filename !~ /internal/ ) {
foreach my $n ( keys %names ) {
err($id, "$n is not public")
if !defined $public{$n};
}
}
# Find all functions in SYNOPSIS
@@ -203,6 +225,206 @@ sub check_head_style {
}
}
# 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 (refering 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));
@@ -225,9 +447,16 @@ sub check {
check_section_location($id, $contents, "EXAMPLES", "SEE ALSO");
}
name_synopsis($id, $filename, $contents)
unless $contents =~ /=for comment generic/
or $filename =~ m@man[157]/@;
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 $contents !~ /^=pod/;
@@ -256,7 +485,7 @@ sub check {
err($id, "Possible version style issue")
if $contents =~ /OpenSSL version [019]/;
if ( $contents !~ /=for comment multiple includes/ ) {
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 ) {
@@ -289,15 +518,13 @@ sub check {
my $section = 3;
$section = $1 if $dirname =~ /man([1-9])/;
foreach ((@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}})) {
# Skip "return values" if not -s
foreach ( (@{$mandatory_sections{'*'}}, @{$mandatory_sections{$section}}) ) {
err($id, "missing $_ head1 section")
if $contents !~ /^=head1\s+${_}\s*$/m;
}
}
my %dups;
# Parse libcrypto.num, etc., and return sorted list of what's there.
sub parsenum {
my $file = shift;
my @apis;
@@ -316,20 +543,22 @@ sub parsenum {
close $IN;
print "# Found ", scalar(@apis), " in $file\n" unless $opt_p;
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"), glob("$dir/*.pod.in") ) {
foreach my $pod ( glob("$dir/*.pod") ) {
my %podinfo = extract_pod_info($pod);
foreach my $n ( @{$podinfo{names}} ) {
$return{$n} = $pod;
print "# Duplicate $n in $pod and $dups{$n}\n"
err("# Duplicate $n in $pod and $dups{$n}")
if defined $dups{$n} && $dups{$n} ne $pod;
$dups{$n} = $pod;
}
@@ -338,8 +567,14 @@ sub getdocced
return %return;
}
# Map of documented functions; function => manpage
my %docced;
# Map of links in each POD file; filename => [ "foo(1)", "bar(3)", ... ]
my %link_map = ();
# Map of names in each POD file; "name(s)" => filename
my %name_map = ();
# Load file of symbol names that we know aren't documented.
sub loadmissing($)
{
my $missingfile = shift;
@@ -357,19 +592,19 @@ sub loadmissing($)
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) {
if ( $opt_o ) {
@missing = loadmissing('util/missingmacro111.txt');
} elsif ($opt_v) {
} elsif ( $opt_v ) {
@missing = loadmissing('util/missingmacro.txt');
}
print "# Checking macros (approximate)\n"
if !$opt_s;
foreach my $f ( glob('include/openssl/*.h') ) {
# Skip some internals we don't want to document yet.
next if $f eq 'include/openssl/asn1.h';
@@ -389,17 +624,19 @@ sub checkmacros {
# Skip macros known to be missing
next if $opt_v && grep( /^$macro$/, @missing);
print "$f:$macro\n"
err("$f:", "macro $macro undocumented")
if $opt_d || $opt_e;
$count++;
$seen{$macro} = 1;
}
close(IN);
}
print "# Found $count macros missing\n"
if !$opt_s || $count > 0;
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;
@@ -407,7 +644,7 @@ sub printem {
my $count = 0;
my %seen;
my @missing = loadmissing($missingfile) if ($opt_v);
my @missing = loadmissing($missingfile) if ( $opt_v );
foreach my $func ( parsenum($numfile) ) {
next if $docced{$func} || defined $seen{$func};
@@ -418,28 +655,21 @@ sub printem {
# Skip functions known to be missing
next if $opt_v && grep( /^$func$/, @missing);
print "$libname:$func\n"
err("$libname:", "function $func undocumented")
if $opt_d || $opt_e;
$count++;
$seen{$func} = 1;
}
print "# Found $count missing from $numfile\n\n"
if !$opt_s || $count > 0;
err("# $count in $numfile are not documented")
if $count > 0;
}
# Collection of links in each POD file.
# filename => [ "foo(1)", "bar(3)", ... ]
my %link_collection = ();
# Collection of names in each POD file.
# "name(s)" => filename
my %name_collection = ();
# Collect all the names in a manpage.
sub collectnames {
my $filename = shift;
$filename =~ m|man(\d)/|;
my $section = $1;
my $simplename = basename(basename($filename, ".in"), ".pod");
my $simplename = basename($filename, ".pod");
my $id = "${filename}:1:";
my $contents = '';
@@ -452,7 +682,7 @@ sub collectnames {
$contents =~ /=head1 NAME([^=]*)=head1 /ms;
my $tmp = $1;
unless (defined $tmp) {
unless ( defined $tmp ) {
err($id, "weird name section");
return;
}
@@ -463,32 +693,32 @@ sub collectnames {
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) {
unless ( grep { $simplename eq $_ } @names ) {
err($id, "missing $simplename");
push @names, $simplename;
}
foreach my $name (@names) {
next if $name eq "";
if ($name =~ /\s/) {
if ( $name =~ /\s/ ) {
err($id, "'$name' contains white space")
}
my $name_sec = "$name($section)";
if (! exists $name_collection{$name_sec}) {
$name_collection{$name_sec} = $filename;
} elsif ($filename eq $name_collection{$name_sec}) {
if ( !exists $name_map{$name_sec} ) {
$name_map{$name_sec} = $filename;
} elsif ( $filename eq $name_map{$name_sec} ) {
err($id, "$name_sec repeated in NAME section of",
$name_collection{$name_sec});
$name_map{$name_sec});
} else {
err($id, "$name_sec also in NAME section of",
$name_collection{$name_sec});
$name_map{$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) {
$name_collection{$_} = undef; # It still exists!
foreach ( @foreign_names ) {
$name_map{$_} = undef; # It still exists!
}
my @links = $contents =~ /L<
@@ -500,18 +730,20 @@ sub collectnames {
# a one digit section number
([^\/>\(]+\(\d\))
/gx;
$link_collection{$filename} = [ @links ];
$link_map{$filename} = [ @links ];
}
# Look for L<> ("link") references that point to files that do not exist.
sub checklinks {
foreach my $filename (sort keys %link_collection) {
foreach my $link (@{$link_collection{$filename}}) {
foreach my $filename (sort keys %link_map) {
foreach my $link (@{$link_map{$filename}}) {
err("${filename}:1:", "reference to non-existing $link")
unless exists $name_collection{$link};
unless exists $name_map{$link};
}
}
}
# Load the public symbol/macro names
sub publicize {
foreach my $name ( parsenum('util/libcrypto.num') ) {
$public{$name} = 1;
@@ -519,11 +751,13 @@ sub publicize {
foreach my $name ( parsenum('util/libssl.num') ) {
$public{$name} = 1;
}
foreach my $name ( parsenum('util/private.num') ) {
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,
@@ -537,15 +771,17 @@ my %skips = (
'des' => 1,
'des3' => 1,
'idea' => 1,
'[cipher]' => 1,
'[digest]' => 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, "./apps/openssl list --options $cmd|"
@@ -563,53 +799,40 @@ sub checkflags {
while ( <CFH> ) {
chop;
last if /DESCRIPTION/;
if ( /=for openssl ifdef (.*)/ ) {
foreach my $f ( split / /, $1 ) {
$localskips{$f} = 1;
}
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 = ();
foreach my $k ( keys %cmdopts ) {
push @undocced, $k unless $docopts{$k};
}
if ( scalar @undocced > 0 ) {
foreach ( @undocced ) {
err("doc/man1/$cmd.pod: Missing -$_");
}
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 = ();
foreach my $k ( keys %docopts ) {
push @unimpl, $k unless $cmdopts{$k};
}
if ( scalar @unimpl > 0 ) {
foreach ( @unimpl ) {
next if defined $skips{$_};
err("doc/man1/$cmd.pod: Not implemented -$_");
}
my @unimpl = sort grep { !defined $cmdopts{$_} } keys %docopts;
foreach ( @unimpl ) {
next if defined $skips{$_} || defined $localskips{$_};
err("$cmd documented but not implemented -$_");
}
}
getopts('cdesolnphuv');
help() if $opt_h;
$opt_n = 1 if $opt_p;
$opt_u = 1 if $opt_d;
$opt_e = 1 if $opt_s;
$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, s, o and p imply one of the above.
die "Need one of -[cdesolnpuv] flags.\n"
unless $opt_c or $opt_l or $opt_n or $opt_u or $opt_v;
##
## 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 = ();
@@ -649,22 +872,23 @@ if ( $opt_c ) {
}
if ( $opt_l ) {
foreach (@ARGV ? @ARGV : (glob('doc/*/*.pod'), glob('doc/*/*.pod.in'),
glob('doc/internal/*/*.pod'))) {
foreach ( @ARGV ? @ARGV : glob('doc/*/*.pod doc/internal/*/*.pod') ) {
collectnames($_);
}
checklinks();
}
if ( $opt_n ) {
publicize() if $opt_p;
foreach (@ARGV ? @ARGV : (glob('doc/*/*.pod'), glob('doc/*/*.pod.in'))) {
publicize();
foreach ( @ARGV ? @ARGV : glob('doc/*/*.pod doc/internal/*/*.pod') ) {
check($_);
}
{
local $opt_p = undef;
foreach (@ARGV ? @ARGV : glob('doc/internal/*/*.pod')) {
check($_);
# 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/;
err("$_ doesn't start with openssl-") unless /openssl-/;
}
}
}
@@ -674,7 +898,7 @@ if ( $opt_u || $opt_v) {
foreach ( keys %temp ) {
$docced{$_} = $temp{$_};
}
if ($opt_o) {
if ( $opt_o ) {
printem('crypto', 'util/libcrypto.num', 'util/missingcrypto111.txt');
printem('ssl', 'util/libssl.num', 'util/missingssl111.txt');
} else {