X-Git-Url: https://git.lyx.org/gitweb/?a=blobdiff_plain;f=po%2Fpocheck.pl;h=fcc5dc11dd2ab395654a51c3286a4dbb93feaa39;hb=d64c9a9d00c175c8859ce94dfcf7b39d6293132d;hp=33f0e94a22ca0b3e3fed099be4f9c20e312573e9;hpb=2fd9eec0dae1fa96900e83a62e0223c8899e658a;p=lyx.git

diff --git a/po/pocheck.pl b/po/pocheck.pl
index 33f0e94a22..fcc5dc11dd 100755
--- a/po/pocheck.pl
+++ b/po/pocheck.pl
@@ -1,4 +1,5 @@
 #! /usr/bin/perl -w
+# -*- mode: perl; -*-
 
 # file pocheck.pl
 #
@@ -11,13 +12,17 @@
 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
@@ -26,34 +31,44 @@ options is given, in which case we checks only for those requested.
 -q: Check Qt shortcuts
 -s: Check for space at end
 -t: Check for uniform translation
-This option can be given with or without other options.
+These options can be given with or without other options.
+-f: Ignore fuzzy translations
 -w: Only report summary total of errors
+-i: Silent mode, report only errors
 EOT
 
 my %options;
-getopts(":hacmpqstw", \%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'";
@@ -68,13 +83,17 @@ foreach my $pofilename ( @ARGV ) {
   my $warn = 0;
 
   my $i = 0;
-  my ($msgid, $msgstr, $more);
+  my ($msgid, $msgid_trans, $msgstr, $more);
 
   while ($i <= $noOfLines) {
     my $linenum = $i;
     ( $msgid ) = ( $pofile[$i] =~ m/^msgid "(.*)"/ );
     $i++;
     next unless $msgid;
+    if ($ignore_fuzzy) {
+      my $previous = $pofile[$i - 2];
+      next if $previous =~ m/#,.*fuzzy/;
+    }
     
     # some msgid's are more than one line long, so add those.
     while ( ( $more ) = $pofile[$i] =~ m/^"(.*)"/ ) {
@@ -96,6 +115,10 @@ foreach my $pofilename ( @ARGV ) {
     # (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 );
@@ -103,7 +126,7 @@ foreach my $pofilename ( @ARGV ) {
         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++;
@@ -111,7 +134,7 @@ foreach my $pofilename ( @ARGV ) {
           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++;
@@ -133,6 +156,9 @@ foreach my $pofilename ( @ARGV ) {
 
     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;
@@ -142,8 +168,18 @@ foreach my $pofilename ( @ARGV ) {
     }
 
     if ($check_spaces) {
-      # Check space at the end of a message
-      if ( ( $msgid =~ m/  *?(\|.*)?$/ ) != ( $msgstr =~ m/  *?(\|.*)?$/ ) ) {
+      # Check space at the end of a message (if not a shortcut)
+      my ($msgid1, $msgstr1) = ($msgid, $msgstr);
+      $msgid1 =~ s/\|.$//;
+      if ($msgstr =~ /^(.*)\|(.+)$/) {
+	my ($msg, $shortcut) = ($1, $2);
+	# Check for unicode char
+	my $u = decode('utf-8', $shortcut);
+	if (length($u) == 1) {
+	  $msgstr1 = $msg;
+	}
+      }
+      if (($msgid1 =~ / $/) != ($msgstr1 =~ / $/)) {
         print "Line $linenum: Missing or unexpected space:\n  '$msgid' => '$msgstr'\n"
           unless $only_total;
         ++$bad{"Bad spaces"};
@@ -153,7 +189,7 @@ foreach my $pofilename ( @ARGV ) {
 
     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"};
@@ -162,8 +198,16 @@ foreach my $pofilename ( @ARGV ) {
     }
 
     if ($check_menu) {
-      # Check for "|..." shortcuts
-      if ( ( $msgid =~ m/\|[^ ]/ ) != ( $msgstr =~ m/\|[^ ]/ ) ) {
+      # Check for "|..." shortcuts (space shortcut allowed)
+      # Shortcut is either 1 char (ascii in msgid) or utf8 char (in msgstr)
+      my ($s1, $s2) = (0,0);
+      $s1 = 1 if ($msgid =~ /\|(.)$/);
+      if ($msgstr =~ /.*\|(.+)$/) {
+	my $chars = $1;
+	my $u = decode('utf-8', $chars);
+	$s2 = 1 if (length($u) == 1);
+      }
+      if($s1 != $s2) {
         print "Line $linenum: Missing or unexpected menu shortcut:\n  '$msgid' => '$msgstr'\n"
           unless $only_total;
         ++$bad{"Bad menu shortcuts"};
@@ -176,8 +220,8 @@ foreach my $pofilename ( @ARGV ) {
     # 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/;
@@ -188,7 +232,7 @@ foreach my $pofilename ( @ARGV ) {
     # 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) {
@@ -213,14 +257,35 @@ foreach my $pofilename ( @ARGV ) {
       }
     }
   }
-
-  if ($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");
 }