Changeset 23388 for main/trunk

Timestamp:

2010-12-06T13:15:57+13:00 (13 years ago)

Author:

davidb

Message:

Further changes to deal with documents that use different filename encodings on the file-system. Now sets UTF8URL metadata to perform the cross-document look up. Files stored in doc.pm as associated files are now always raw filenames (rather than potentially UTF8 encoded). Storing of filenames seen by HTMLPlug when scanning for files to block on is now done in Unicode aware strings rather than utf8 but unware strings.

File:

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

main/trunk/greenstone2/perllib/util.pm

@@ -530 +530 @@
     if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
     require Win32;
+
+    print STDERR "***** utf8 filename = $utf8_filename\n\n\n";
+
     my $unicode_filename = decode("utf8",$utf8_filename);
     $real_filename = Win32::GetShortPathName($unicode_filename);
@@ -1096 +1099 @@
 
     if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
-    require Win32;
     # Ensure any DOS-like filename, such as test~1.txt, has been upgraded
     # to its long (Windows) version
@@ -1105 +1107 @@
         $filename_full_path = encode("utf8",$filename_full_path);
     }
+    }
+
+    return $filename_full_path;
+}
+
+
+sub downgrade_if_dos_filename
+{
+    my ($filename_full_path) = @_;
+
+    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
+    require Win32;
+
+    # Ensure the given long Windows filename is in a form that can
+    # be opened by Perl => convert it to a short DOS-like filename
+
+##  print STDERR "**** asked to downgrade: $filename_full_path\n";
+
+    $filename_full_path = Win32::GetShortPathName($filename_full_path);
+
+##  print STDERR "**** now: $filename_full_path\n";
+
+    # Make sure initial drive letter is lower-case (to fit in with rest of Greenstone)
+    $filename_full_path =~ s/^(.)/\l$1/;
     }
 
@@ -1384 +1410 @@
     ($tailname,$dirname,$suffix) = File::Basename::fileparse($filename, "\\.(?:[^\\.]+?)\$");
     }
-    $suffix = "" if !$suffix;
+    if (!$suffix) {
+    $suffix = "";
+    }
+    else {
+    $suffix = lc($suffix);
+    }
 
     if ($rename_method eq "url") {