Changeset 4229

Timestamp:

2003-05-06T15:46:31+12:00 (21 years ago)

Author:

mdewsnip

Message:

Replaced unicode2koi8r with a more generic version: unicode2singlebyte. This can be used to convert unicode to any single byte encoding - (KOI8-R, the Kazakh encoding, etc.).

File:

• trunk/gsdl/perllib/unicode.pm (modified) (3 diffs)

trunk/gsdl/perllib/unicode.pm

@@ -230 +230 @@
 #
 # Examples of encodings that may be converted using singlebyte2unicode are
-# the iso-8859 and windows-125* series).
+# the iso-8859 and windows-125* series.
 sub singlebyte2unicode {
     my ($encodename, $textref) = @_;
@@ -385 +385 @@
 # @array256 is used for initialisation, there must be
 # a better way...
+# What about this?: @array256 = (0) x 256;
 @array256 = (0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
          0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
@@ -429 +430 @@
 }
 
-sub unicode2koi8r {
-    my ($uniref) = @_;
-    
+# unicode2singlebyte converts unicode to simple 8 bit encodings where
+# characters below 0x80 are normal ascii characters and the rest are encoded
+# using the appropriate mapping files.
+#
+# Examples of encodings that may be converted using unicode2singlebyte are
+# the iso-8859 and windows-125* series, KOI8-R (Russian), and the Kazakh encoding.
+sub unicode2singlebyte {
+    my ($uniref, $encoding) = @_;
+
     my $outtext = "";
-    my $encodename = "unicode-koi8_r";
-    my $enc_info = $encodings::encodings->{"koi8_r"};
+    my $encodename = "unicode-$encoding";
+    my $enc_info = $encodings::encodings->{"$encoding"};
     my $mapfile = &util::filename_cat($ENV{'GSDLHOME'}, "mappings",
                       "from_uc", $enc_info->{'mapfile'});