Changeset 23835 for main/trunk/greenstone2/perllib/plugins/HTMLPlugin.pm

Timestamp:

2011-03-29T20:19:34+13:00 (13 years ago)

Author:

ak19

Message:

Dr Bainbridge fixed interlinking failure on Mac OS when filenames (and therefore links to files on the system) have characters that don't occur in English. The problem had to do with the URL obtained from the href in the HTML page not matching up with the URL encoded normalised decomposed URL stored in the doc.xml and the gdb database. The latter is the right form to store the URL in, since it refers accurately to the file as it exists on the system. So the former was changed by using normalization with decomposition on the href link in the text at the correct part of the replace_href_links of the HTMLPlugin.

File:

• main/trunk/greenstone2/perllib/plugins/HTMLPlugin.pm (modified) (4 diffs)

main/trunk/greenstone2/perllib/plugins/HTMLPlugin.pm

@@ -447 +447 @@
 #    my $utf8_file = $self->filename_to_utf8_metadata($file);
 #    $utf8_file =~ s/&\#095;/_/g;
-#    variable below used to be utf8_file    
+#    variable below used to be utf8_file
+
     my $url_encoded_file = &unicode::raw_filename_to_url_encoded($tailname);
     my $utf8_url_encoded_file = &unicode::raw_filename_to_utf8_url_encoded($tailname);
@@ -796 +797 @@
     my $self = shift (@_);
     my ($front, $link, $back, $base_dir, $file, $doc_obj, $section) = @_;
-    
+
     # remove quotes from link at start and end if necessary
     if ($link=~/^[\"\']/) {
@@ -821 +822 @@
     # is taken in to account
     my ($href, $hash_part, $rl) = $self->format_link ($link, $base_dir, $file);
- 
+
     # href may use '\'s where '/'s should be on Windows
     $href =~ s/\\/\//g;
@@ -853 +854 @@
         # If web page didn't give encoding, then default to utf8
         my $content_encoding= $self->{'content_encoding'} || "utf8";
-    
+
         if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
         print STDERR "**** Encoding with '$content_encoding', href: $href\n";
         }
 
+        # on Darwin, the unicode filenames are stored on the file
+        # system in decomposed form, so any href link (including when 
+        # URL-encoded) should refer to the decomposed name of the file
+        if ($ENV{'GSDLOS'} =~ /^darwin$/i) {
+        $href = normalize('D', $href); # Normalization Form D (decomposition) 
+        }
+
         $href = encode($content_encoding,$href);
     }
 
-    $href = &unicode::raw_filename_to_utf8_url_encoded($href);
+    $href = &unicode::raw_filename_to_utf8_url_encoded($href);  
     $href = &unicode::filename_to_url($href);