Changeset 73 for trunk/gsdl/perllib/doc.pm

Timestamp:

1998-12-11T20:59:16+13:00 (25 years ago)

Author:

rjmcnab

Message:

Added support for UTF-8.

File:

• trunk/gsdl/perllib/doc.pm (modified) (5 diffs)

trunk/gsdl/perllib/doc.pm

@@ -3 +3 @@
 package doc;
 
+use unicode;
 use util;
 use html;
@@ -463 +464 @@
 # are for metadata which can have more than one value.
 
+# set_metadata_element assumes the value is in (extended) ascii form.
+# For text which hash been already converted to the UTF-8 format use
+# set_utf8_metadata_element.
 sub set_metadata_element {
     my $self = shift (@_);
     my ($section, $field, $value) = @_;
 
+    $self->set_utf8_metadata_element ($section, $field, 
+                      &unicode::ascii2utf8($value));
+}
+
+# set_utf8_metadata_element assumes the text has already been
+# converted to the UTF-8 encoding.
+sub set_utf8_metadata_element {
+    my $self = shift (@_);
+    my ($section, $field, $value) = @_;
+
     $self->delete_metadata ($section, $field);
-    $self->add_metadata ($section, $field, $value);
+    $self->add_utf8_metadata ($section, $field, $value);
 }
 
@@ -491 +505 @@
 }
 
+# add_metadata assumes the text is in (extended) ascii form. For
+# text which hash been already converted to the UTF-8 format use
+# add_utf8_metadata.
 sub add_metadata {
     my $self = shift (@_);
     my ($section, $field, $value) = @_;
 
-    my $section_ptr = $self->_lookup_section($section);
-    if (!defined $section_ptr) {
-    print STDERR "doc::add_metadata couldn't find section " .
+    $self->add_utf8_metadata ($section, $field,
+                  &unicode::ascii2utf8($value));
+}
+
+# add_utf8_metadata assumes the text has already been converted
+# to the UTF-8 encoding.
+sub add_utf8_metadata {
+    my $self = shift (@_);
+    my ($section, $field, $value) = @_;
+
+    my $section_ptr = $self->_lookup_section($section);
+    if (!defined $section_ptr) {
+    print STDERR "doc::add_utf8_metadata couldn't find section " .
         "$section\n";
     return;
@@ -618 +645 @@
 # methods for dealing with text
 
+# add_text assumes the text is in (extended) ascii form. For
+# text which has been already converted to the UTF-8 format
+# use add_utf8_text.
 sub add_text {
     my $self = shift (@_);
     my ($section, $text) = @_;
 
-    my $section_ptr = $self->_lookup_section($section);
-    if (!defined $section_ptr) {
-    print STDERR "doc::add_text couldn't find section " .
+    # convert the text to UTF-8 encoded unicode characters
+    # and add the text
+    $self->add_utf8_text($section, &unicode::ascii2utf8($text));
+}
+
+
+# add_utf8_text assumes the text to be added has already
+# been converted to the UTF-8 encoding. For ascii text use
+# add_text
+sub add_utf8_text {
+    my $self = shift (@_);
+    my ($section, $text) = @_;
+
+    my $section_ptr = $self->_lookup_section($section);
+    if (!defined $section_ptr) {
+    print STDERR "doc::add_utf8_text couldn't find section " .
         "$section\n";
     return;
@@ -647 +690 @@
 }
 
-# returns the length of the text for a section
+# returns the (utf-8 encoded) length of the text for a section
 sub get_text_length {
     my $self = shift (@_);