Changeset 1838

Timestamp:

2001-01-16T11:48:04+13:00 (23 years ago)

Author:

sjboddie

Message:

Added support for Cyrillic languages (windows codepage 1251) - yet to be
tested by anyone who understands any such language though ;)

Location:

trunk/gsdl

Files:

• macros/pref.dm (modified) (1 diff)
• perllib/multiread.pm (modified) (4 diffs)
• perllib/plugins/BasPlug.pm (modified) (2 diffs)
• src/recpt/librarymain.cpp (modified) (1 diff)
• src/recpt/receptionist.cpp (modified) (1 diff)
• unicode/MAPPINGS/WINDOWS/1251.TXT (added)

trunk/gsdl/macros/pref.dm

@@ -349 +349 @@
   <option value="g" _If_("_cgiargw_" eq "g",selected)>GBK
   <option value="a" _If_("_cgiargw_" eq "a",selected)>Arabic (windows 1256)
+  <option value="c" _If_("_cgiargw_" eq "c",selected)>Cyrillic (windows 1251)
 </select>
 }

trunk/gsdl/perllib/multiread.pm

@@ -32 +32 @@
 # iso_8859_6   - 8 bit arabic (iso-8859-6)
 # windows_1256 - Windows codepage 1256 (Arabic)
+# windows_1251 - Windows codepage 1251 (Cyrillic)
 
 package multiread;
@@ -188 +189 @@
     }
 
+    if ($self->{'encoding'} eq "windows_1251") {
+    # Windows 1251 (Cyrillic)
+    return undef if (eof ($handle));
+    return &unicode::unicode2utf8(&unicode::windows2unicode ("1251", getc ($handle)));
+    }
+
     # unknown encoding
     return undef;
@@ -275 +282 @@
     return undef;
     }
+    
+    if ($self->{'encoding'} eq "windows_1251") {
+    # Windows 1251 (Cyrillic)
+    my $line = "";
+    if (defined ($line = <$handle>)) {
+        return &unicode::unicode2utf8(&unicode::windows2unicode ("1251", $line));
+    }
+    return undef;
+    }
 
     # unknown encoding
@@ -345 +361 @@
     return;
     }
+
+    if ($self->{'encoding'} eq "windows_1251") {
+    undef $/;
+    my $text = <$handle>;
+    $/ = "\n";
+    $$outputref .= &unicode::unicode2utf8(&unicode::windows2unicode ("1251", $text));
+    return;
+    }
 }
 

trunk/gsdl/perllib/plugins/BasPlug.pm

@@ -54 +54 @@
     print STDERR "                        unicode (just unicode -- doesn't currently do endian\n";
     print STDERR "                                 detection)\n";
+    print STDERR "                        windows_1251 (Windows codepage 1251 (Cyrillic))\n";
     print STDERR "   -process_exp      A perl regular expression to match against filenames.\n";
     print STDERR "                     Matching filenames will be processed by this plugin.\n";
@@ -87 +88 @@
 
     my $self = {};
-    my $encodings = "^(iso_8859_1|Latin1|ascii|gb|iso_8859_6|windows_1256|Arabic|utf8|unicode)\$";
+    my $encodings = "^(iso_8859_1|Latin1|ascii|gb|iso_8859_6|windows_1256|Arabic|utf8|unicode|windows_1251)\$";
     $self->{'outhandle'} = STDERR;
     my $year = (localtime)[5]+1900;

trunk/gsdl/src/recpt/librarymain.cpp

@@ -123 +123 @@
 
   // arabic
-  text_t armapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS");
-  armapfile = filename_cat (armapfile, "WINDOWS", "1256.TXT");
+  text_t mapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS");
+  mapfile = filename_cat (mapfile, "WINDOWS", "1256.TXT");
   simplemapinconvertclass arinconvert;
-  arinconvert.setmapfile (armapfile);
+  arinconvert.setmapfile (mapfile);
   simplemapoutconvertclass aroutconvert;
-  aroutconvert.setmapfile (armapfile);
+  aroutconvert.setmapfile (mapfile);
   recpt.add_converter ("a", &arinconvert, &aroutconvert);  
+
+  // cyrillic
+  mapfile = filename_cat (gsdlhome, "unicode", "MAPPINGS");
+  mapfile = filename_cat (mapfile, "WINDOWS", "1251.TXT");
+  simplemapinconvertclass cyinconvert;
+  cyinconvert.setmapfile (mapfile);
+  simplemapoutconvertclass cyoutconvert;
+  cyoutconvert.setmapfile (mapfile);
+  recpt.add_converter ("c", &cyinconvert, &cyoutconvert);  
   
   // hindi

trunk/gsdl/src/recpt/receptionist.cpp

@@ -750 +750 @@
     } else if (args["w"] == "a") {
       response_data += "; charset=windows-1256";
+    } else if (args["w"] == "c") {
+      response_data += "; charset=windows-1251";
     } else {
       response_data += "; charset=ISO-8859-1";