Changeset 29763 for main/trunk/greenstone2/perllib

Timestamp:

2015-02-26T12:30:42+13:00 (9 years ago)

Author:

ak19

Message:

on macos, accented chars in filenames are in decomposed form, eg the letter plus the accent. Convert to canonical composed form for looking up in the extra metadata table.

File:

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

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

@@ -44 +44 @@
 use Encode::Locale;
 use Encode;
+use Unicode::Normalize;
 
 BEGIN {
@@ -545 +546 @@
     #my $unicode_subfile = &Encode::decode("utf8", $raw_subfile);
     my $unicode_subfile = &Encode::decode(locale => $raw_subfile);
-    
+
+    # Normalise the filename to canonical composition - on mac, filenames use decopmposed form for accented chars
+    $unicode_subfile = NFC($unicode_subfile);
+
     my $raw_file_subfile = &FileUtils::filenameConcatenate($file, $raw_subfile);
     my $raw_full_filename 
@@ -598 +602 @@
 
     ### $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");
+       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);
@@ -607 +611 @@
         foreach my $filespec (@extrametakeys) {
         ## use the url-encoded filename to do the filename comparison
-        print STDERR "comparing against filespec $filespec (";
+        print STDERR "### comparing against filespec $filespec (";
         print STDERR &unicode::debug_unicode_string("$filespec");
         print STDERR ")\n";