Changeset 29745 for main

Timestamp:

2015-02-16T14:41:02+13:00 (9 years ago)

Author:

kjdon

Message:

using Encode::decode to make the filenames 'unicode aware'. For file_block_read(), only done so that the print statement looks nice. But for read(), we are using the filename to look up in the extrametadata hash thing for metadata coming from metadata.xml. This is unicode aware, so our lookup name needs to be so too. Some debug and extra stuff left in, needs to be removed once windows testing done

File:

• main/trunk/greenstone2/perllib/plugins/DirectoryPlugin.pm (modified) (7 diffs)

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

@@ -332 +332 @@
 
     # Recursively read each $raw_subfile
-    print $outhandle "DirectoryPlugin block recurring: $raw_file_subfile\n" if ($verbosity > 2);
+    print $outhandle "DirectoryPlugin block recurring: ". Encode::decode("utf8", $raw_file_subfile) ."\n" if ($verbosity > 2);
     
     #$count += &plugin::file_block_read ($pluginfo, $this_file_base_dir,
@@ -407 +407 @@
     }
     @dir = sort readdir (DIR);
-    map { $_ = &unicode::raw_filename_to_url_encoded($_) } @dir;
+    map { $_ = &unicode::raw_filename_to_url_encoded($_); print STDERR "****$_\n"; } @dir;
     closedir (DIR);
 
@@ -541 +541 @@
     my $this_file_base_dir = $base_dir;
     my $raw_subfile = &unicode::url_encoded_to_raw_filename($subfile);
+    my $unicode_subfile = &Encode::decode("utf8", $raw_subfile);
 
     my $raw_file_subfile = &FileUtils::filenameConcatenate($file, $raw_subfile);
@@ -550 +551 @@
         next;
     }
-    #print STDERR "processing $raw_full_filename\n";
+    print STDERR "processing $raw_full_filename\n";
     # Follow Windows shortcuts
     if ($raw_subfile =~ m/(?i)\.lnk$/ && (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin"))) {
@@ -593 +594 @@
     }
 
-    # $subfile by this point is url-encoded => all ASCII chars => no need to encode as UTF8
-
+    ### $subfile by this point is url-encoded => all ASCII chars => no need to encode as UTF8
+    ###print STDERR "subfile = $subfile, raw_subfile = $raw_subfile\n";
+    ###print STDERR &unicode::debug_unicode_string("subfile = $subfile, raw_subfile = $raw_subfile\n");
+    # instead of using the subfile, we need unicode aware string
+    ###my $lookup_name = decode("utf8", $raw_subfile);
+    ####print STDERR "lookup nmae = $lookup_name\n";
     # Next add metadata read in XML files (if it is supplied)
     if ($additionalmetadata == 1) {
         foreach my $filespec (@extrametakeys) {
         ## use the url-encoded filename to do the filename comparison
-
-        if ($subfile =~ /^$filespec$/) {
-            print $outhandle "File \"$subfile\" matches filespec \"$filespec\"\n" 
+        print STDERR "comparing against filespec $filespec\n";
+        print STDERR &unicode::debug_unicode_string("comparing against filespec $filespec\n");
+        if ($unicode_subfile =~ /^$filespec$/) {
+        ###if ($lookup_name =~ /^$filespec$/) {
+            print $outhandle "File \"$unicode_subfile\" matches filespec \"$filespec\"\n" 
             if ($verbosity > 2);
             my $mdref = &extrametautil::getmetadata(\%extrametadata, $filespec);
@@ -634 +641 @@
             if (!$block_hash->{'reindex_files'}->{$raw_full_filename}) {
             # filename has been around for longer than inf_timestamp
-            print $outhandle "**** Skipping $subfile\n" if ($verbosity >3);
+            print $outhandle "**** Skipping $unicode_subfile\n" if ($verbosity >3);
             next;
             }
@@ -652 +659 @@
 
     # Recursively read each $subfile
-    print $outhandle "DirectoryPlugin recurring: $subfile\n" if ($verbosity > 2);
+    print $outhandle "DirectoryPlugin recurring: $unicode_subfile\n" if ($verbosity > 2);
     
     $count += &plugin::read ($pluginfo, $this_file_base_dir,