Changeset 7818

Timestamp:

2004-07-23T12:58:37+12:00 (20 years ago)

Author:

jrm21

Message:

improvements to the handling of textcat's guessed encoding

File:

• trunk/gsdl/perllib/plugins/BasPlug.pm (modified) (1 diff)

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -680 +680 @@
     }
 
+
+    # check for equivalents where textcat doesn't have some encodings...
+    # eg MS versions of standard encodings
+    if ($encoding =~ /^iso_8859_(\d+)/) {
+    my $iso = $1; # which variant of the iso standard?
+    # iso-8859 sets don't use chars 0x80-0x9f, windows codepages do
+    if ($text =~ /[\x80-\x9f]/) {
+        # Western Europe
+        if ($iso == 1 or $iso == 15) { $encoding = 'windows_1252' }
+        elsif ($iso == 2) { $encoding = 'windows_1250' } # Central Europe
+        elsif ($iso == 5) { $encoding = 'windows_1251' } # Cyrillic
+        elsif ($iso == 6) { $encoding = 'windows_1256' } # Arabic
+        elsif ($iso == 7) { $encoding = 'windows_1253' } # Greek
+        elsif ($iso == 8) { $encoding = 'windows_1255' } # Hebrew
+        elsif ($iso == 9) { $encoding = 'windows_1254' } # Turkish
+    }
+    }
+
     if ($encoding !~ /^(ascii|utf8|unicode)$/ &&
     !defined $encodings::encodings->{$encoding}) {
     if ($self->{'verbosity'}) {
-        &gsprintf($outhandle, "BasPlug: {BasPlug.unsupported_encoding}\n", $filename, $encoding, $self->{'default_encoding'});
-        # print $outhandle "BasPlug: WARNING: $filename appears to be encoded in an unsupported encoding ($encoding) - ";
-        # print $outhandle "using $self->{'default_encoding'}\n";
+        gsprintf($outhandle, "BasPlug: {BasPlug.unsupported_encoding}\n",
+             $filename, $encoding, $self->{'default_encoding'});
     }
     $encoding = $self->{'default_encoding'};