Changeset 29810

Timestamp:

2015-03-26T14:33:03+13:00 (9 years ago)

Author:

ak19

Message:

trying to connect filenames got in perl to filenames in the metadata file. For windows, there are 2 apis to teh filenames. ansi codepage, and unicode. perl only uses the ansi version. If the unicode filename can't be converted to ansi then you get teh garbled M33~1 type filename. We do GetLongPathName to get the real filename which is returned in unicode. so now, if the short name is the same as teh long name, then we assume its in ansi, and decode using locale, otherwise, assume its in unicode and don't need to decode. this seems to work on windows for the files I have. ned to do a bit more testing, then go back to mac and linux and make sure I haven't mucked them up.

File:

• main/trunk/greenstone2/perllib/util.pm (modified) (2 diffs)

main/trunk/greenstone2/perllib/util.pm

@@ -168 +168 @@
 sub raw_filename_to_unicode
 {
-    my ($directory, $raw_file) = @_;
+    my ($directory, $raw_file, $filename_encoding ) = @_;
         
-    my $unicode_filename = &util::get_dirsep_tail(&util::upgrade_if_dos_filename(&FileUtils::filenameConcatenate($directory, $raw_file), 1));
-    $unicode_filename = &Encode::decode("utf8", $unicode_filename);
-    #$unicode_filename = &Encode::decode(locale => $unicode_filename);
-
+    my $unicode_filename = $raw_file;
+    print STDERR "&&&&&&&&&&&&&& raw file = $unicode_filename ".&unicode::debug_unicode_string($unicode_filename)."\n";
+    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
+        print STDERR "windows, upgrading...\n";
+        $unicode_filename = &util::get_dirsep_tail(&util::upgrade_if_dos_filename(&FileUtils::filenameConcatenate($directory, $raw_file), 0));
+            
+        
+            if ($unicode_filename eq $raw_file) {
+                print STDERR "long = original\n";
+                $unicode_filename = &Encode::decode(locale_fs => $unicode_filename);
+                } else {
+                    print STDERR "long not equal original\n";
+                    #$unicode_filename = &Encode::decode("UTF-16le" => $unicode_filename);
+                    }
+                    
+    } else {
+        print STDERR "not windows, not upgrading. decoding in utf8\n";
+        $unicode_filename = &Encode::decode("utf-8", $unicode_filename);
+    
+        
+        }
+    #
+    #$unicode_filename = &Encode::decode("ascii", $unicode_filename);
+    #$unicode_filename = &Encode::decode(locale_fs => $unicode_filename);
+    print STDERR "after decoding = $unicode_filename ".&unicode::debug_unicode_string($unicode_filename)."\n";
     #Does the filenaem have url encoded chars in it?
         if (&unicode::is_url_encoded($unicode_filename)) {
@@ -745 +766 @@
     my $long_filename = Win32::GetLongPathName($filename_full_path);
     if (defined $long_filename) {
+        
+print STDERR "long anme = $long_filename\n";
+                    
         $filename_full_path = $long_filename;
     }
     # Make sure initial drive letter is lower-case (to fit in with rest of Greenstone)
     $filename_full_path =~ s/^(.):/\u$1:/;
+    
     if ((defined $and_encode) && ($and_encode)) {
         $filename_full_path = encode("utf8",$filename_full_path);