#! /usr/bin/perl -w
+# -*- mode: perl; -*-
# file pocheck.pl
#
use strict;
use warnings;
use Getopt::Std;
+use Encode qw(encode decode);
+
+sub mylc($);
+sub replaceSynopsis($);
my $usage = <<EOT;
pocheck.pl [-acmpqst] po_file [po_file] ...
-This script performs some consistency checks on po files.
+This script performs some consistency checks on po files.
-We check for everything listed here, unless one or more of these
+We check for everything listed here, unless one or more of these
options is given, in which case we checks only for those requested.
-a: Check arguments, like %1\$s
-c: Check for colons at end
-s: Check for space at end
-t: Check for uniform translation
These options can be given with or without other options.
--f: Ignore fuzzy translations.
+-f: Ignore fuzzy translations
-w: Only report summary total of errors
+-i: Silent mode, report only errors
EOT
my %options;
-getopts(":hacfmpqstw", \%options);
+getopts(":hacfmpqstwi", \%options);
-if (defined($options{h})) {
- print $usage;
- exit 0;
+if (defined($options{h})) {
+ print $usage;
+ exit 0;
}
my $only_total = defined($options{w});
delete $options{w} if $only_total;
my $ignore_fuzzy = defined($options{f});
delete $options{f} if $ignore_fuzzy;
+my $silent_mode = defined($options{i});
+delete $options{i} if $silent_mode;
my $check_args = (!%options or defined($options{a}));
my $check_colons = (!%options or defined($options{c}));
-my $check_spaces = (!%options or defined($options{m}));
+my $check_spaces = (!%options or defined($options{s}));
my $check_periods = (!%options or defined($options{p}));
my $check_qt = (!%options or defined($options{q}));
-my $check_menu = (!%options or defined($options{s}));
+my $check_menu = (!%options or defined($options{m}));
my $check_trans = (!%options or defined($options{t}));
my %trans;
+my $total_warn = 0;
+
foreach my $pofilename ( @ARGV ) {
my %bad;
- print "Processing po file '$pofilename'...\n";
+ if (!$silent_mode) {
+ print "Processing po file '$pofilename'...\n";
+ }
open( INPUT, "<$pofilename" )
|| die "Cannot read po file '$pofilename'";
my $warn = 0;
my $i = 0;
- my ($msgid, $msgstr, $more);
+ my ($msgid, $msgid_trans, $msgstr, $more);
while ($i <= $noOfLines) {
my $linenum = $i;
# (surely that is always $msgstr?)
next if ($msgid eq "" or $msgstr eq "");
+ # discard [[...]] from the end of msgid, this is used only as hint to translation
+ $msgid_trans = $msgid; # used for uniform translation
+ $msgid =~ s/\[\[.*\]\]$//;
+
# Check for matching %1$s, etc.
if ($check_args) {
my @argstrs = ( $msgid =~ m/%(\d)\$s/g );
my $n = 0;
foreach my $arg (@argstrs) { $n = $arg if $arg > $n; }
if ($n <= 0) {
- print "Line $linenum: Problem finding arguments in:\n $msgid!\n"
+ print "$pofilename, line $linenum: Problem finding arguments in:\n $msgid!\n"
unless $only_total;
++$bad{"Missing arguments"};
$warn++;
foreach my $i (1..$n) {
my $arg = "%$i\\\$s";
if ( $msgstr !~ m/$arg/ ) {
- print "Line $linenum: Missing argument `$arg'\n '$msgid' ==> '$msgstr'\n"
+ print "$pofilename, line $linenum: Missing argument `$arg'\n '$msgid' ==> '$msgstr'\n"
unless $only_total;
++$bad{"Missing arguments"};
$warn++;
if ($check_periods) {
# Check period at the end of a message; uncomment code if you are paranoid
+ # Convert '...' to '…' first
+ $msgid = replaceSynopsis($msgid);
+ $msgstr = replaceSynopsis($msgstr);
if ( ( $msgid =~ m/\. *(\|.*)?$/ ) != ( $msgstr =~ m/\. *(\|.*)?$/ ) ) {
print "Line $linenum: Missing or unexpected period:\n '$msgid' => '$msgstr'\n"
unless $only_total;
if ($check_qt) {
# Check for "&" shortcuts
- if ( ( $msgid =~ m/&[^ ]/ ) != ( $msgstr =~ m/&[^ ]/ ) ) {
+ if ( ( $msgid =~ m/&[^ &]/ ) != ( $msgstr =~ m/&[^ &]/ ) ) {
print "Line $linenum: Missing or unexpected Qt shortcut:\n '$msgid' => '$msgstr'\n"
unless $only_total;
++$bad{"Bad Qt shortcuts"};
# we now collect these translations in a hash.
# this will allow us to check below if we have translated
# anything more than one way.
- my $msgid_clean = lc($msgid);
- my $msgstr_clean = lc($msgstr);
+ my $msgid_clean = lc($msgid_trans);
+ my $msgstr_clean = mylc($msgstr);
$msgid_clean =~ s/(.*)\|.*?$/$1/; # strip menu shortcuts
$msgstr_clean =~ s/(.*)\|.*?$/$1/;
# cleaned versions of ORIGINAL strings. the keys of the inner hash
# are the cleaned versions of their TRANSLATIONS. The value for the
# inner hash is an array of the orignal string and translation.
- $trans{$msgid_clean}{$msgstr_clean} = [ $msgid, $msgstr, $linenum ];
+ $trans{$msgid_clean}{$msgstr_clean} = [ $msgid_trans, $msgstr, $linenum ];
}
if ($check_trans) {
}
}
}
-
- if ($warn) {
- while (my ($k, $v) = each %bad) { print "$k: $v\n"; }
- if (scalar(keys %bad) > 1) {
- print "Total warnings: $warn\n";
+ if (!$silent_mode) {
+ if ($warn) {
+ while (my ($k, $v) = each %bad) { print "$k: $v\n"; }
+ if (scalar(keys %bad) > 1) {
+ print "Total warnings: $warn\n";
+ }
+ } else {
+ print "No warnings!\n";
}
- } else {
- print "No warnings!\n";
+ print "\n";
}
- print "\n";
+ $total_warn += $warn;
+}
+
+exit ($total_warn > 0);
+
+# Use lowercase also for non-ascii chars
+sub mylc($)
+{
+ my ($msg) = @_;
+ return(encode('utf-8',lc(decode('utf-8', $msg))));
+}
+
+sub replaceSynopsis($)
+{
+ my ($string) = @_;
+
+ return ($string) if ($string !~ /^(.*)\.\.\.(.*)$/);
+ my ($before, $after) = ($1, $2);
+ return $string if (($before =~ /\.$/) || ($after =~ /^\./));
+ return("$before…$after");
}
projects / lyx.git / blobdiff