doc-check

#!/usr/bin/perl -w
#
# This script checks if the translations of the documents are up to date.
#

use Getopt::Std;
use File::Find;
$opt_d = $opt_h = $opt_r = $opt_s = $opt_v = $opt_V = 0;
getopts('hr:dsvV');
# You may set this to your default language code
$lang = shift || "pl";
if ($opt_r and $opt_r !~ /[0-9]+/) {
	warn "Please enter a revision number for parameter -r.\n";
	exit 1
}

sub usage
{
	printf <<HERE, $lang;
Usage: $0 [options] [<language>]

  -h        print this help message.
  -d        print what has changed in the original since last translation.
  -r <rev>  take the diff against the revision number <rev> instead of the
	    latest revision of the English original.
  -s        show also all files marked untranslated.
  -v        be verbose.
  -V        be slightly more verbose.

  Current <language> is '%s'.
HERE
	exit 0;
}

sub checkdiff
{
	my ($plfname, $enfname) = (@_);
	my ($plrev, $enrev, $untrans) = getrev($plfname, $enfname);
	$plrev and $enrev or return;
	if ( "$plrev" ne "$enrev" ) {
		if ($opt_d and $opt_r and $opt_r < $enrev) {
			$enrev = $opt_r;
		}
		if ($untrans) {
			print "$enfname : $plrev -> $enrev (untranslated)\n";
		} else {
			print "$enfname : $plrev -> $enrev\n";
		}
		if ($opt_d) {
			my $s = "svn diff -r $plrev:$enrev $enfname";
			warn "running $s:\n" if ($opt_V);
			system($s);
		}
	} else {
		if ($untrans && $opt_s) {
			print "$plfname: untranslated\n";
		}
	}
}

sub getrev
{
	my ($plfname, $enfname) = (@_);
	my ($plrev, $enrev, $untrans, $notconverted) = (0, 0, 0, 0);
	my $line = 0;

	warn "checking $plfname:\n" if $opt_v;
	open FILE, $plfname or warn "$plfname: $!\n" and return;
	while (<FILE>) {
		$line++;

		# Also check for revision comments of original documents
		# These may be used in combination with a translation comment in
		# which case the translation comment overrules what's found here
		if (($plrev == 0) && (/<!--\s*\$Id: \S+ (\d+) /)) {
			$plrev = $1;
			$notconverted = 1;
			$untrans = 1;
		}
		if (/<!--\s*original version\D*([\d\.]+)\s*-->/) {
			$plrev = $1;
			$untrans = 0;
			$notconverted = 0;
			last;
		}
		if (/<!--\s*original version\D*(\d+)\s*untranslated\s*-->/) {
			$plrev = $1;
			$untrans = 1;
			$notconverted = 0;
			last;
		}
		# Also support CVS style revision comments (depreciated)
		if (/<!--\s*original document: en\/\S+, revision ([\d\.]+)\s*-->/) {
			$plrev = $1;
			last;
		}
		if ($line++ > 10) {
			last;
		}
	}
	warn "checking $enfname:\n" if $opt_v;
	open FILE, $enfname or warn "$enfname: $!\n" and return;
	while (<FILE>) {
		if (/\$Id: \S+ (\d+) /) {
			$enrev = $1;
			last;
		}
		# Also support CVS style revision comments (depreciated)
		if (/\$Revision: (\d+) \$/) {
			$enrev = $1;
			last;
		}
	}
	close FILE;
	warn "failed to find revision for $plfname\n" unless $plrev;
	warn "failed to find revision for $enfname\n" unless $enrev;
	if ($notconverted) {
		warn "$plfname: contains revision comment for original document\n";
		warn "   use 'rev-update' to convert\n";
	}
	return ($plrev, $enrev, $untrans);
}

sub process
{
	my $enfname = $File::Find::name;
	return unless $enfname =~ m/\.xml$/;
	my $plfname = $enfname;
	$plfname =~ s,^en/,$lang/,;
	checkdiff($plfname, $enfname);
}

sub process_obsolete
{
	my $plfname = $File::Find::name;
	return unless $plfname =~ m/\.xml$/;
	my $enfname = $plfname;
	$enfname =~ s,^$lang/,en/,;
	open FILE, $enfname or warn "$plfname: Probably obsoleted\n" and return;
	close FILE;
}

usage() if $opt_h;

warn "\nChecking for outdated or missing translations...\n" if $opt_v;
File::Find::find({ wanted => \&process, no_chdir => 1 }, 'en');
warn "\nChecking for obsoleted translations...\n" if $opt_v;
File::Find::find({ wanted => \&process_obsolete, no_chdir => 1 }, $lang);

#checkdiff("build/install.$lang.xml", "build/install.en.xml");
#checkdiff("release-notes.$lang.sgml","release-notes.sgml");
#checkdiff("index.$lang.html.m4","index.en.html.m4");
#checkdiff("dselect-beginner.$lang.sgml","dselect-beginner.sgml");

warn "All done.\n" if $opt_v;