Changeset 31445

Timestamp:

2017-02-27T14:37:11+13:00 (7 years ago)

Author:

ak19

Message:

added a method guessing_filesystem_encoding. use this to try and work out what encoding to use when turning import folder subdirs into perl native unicode strings

File:

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

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

@@ -836 +836 @@
 
 
+sub guess_filesystem_encoding
+{
+   my $self = shift (@_); 
+    # Look to file system to provide a character encoding
+   my $deduced_filename_encoding = "";
+    # If Windows NTFS, then -- assuming we work with long file names got through
+    # Win32::GetLongFilePath() -- then the underlying file system is UTF16
+
+    if (($ENV{'GSDLOS'} =~ m/^windows$/i) && ($^O ne "cygwin")) {
+        # Can do better than working with the DOS character encoding returned by locale     
+        $deduced_filename_encoding = "unicode";
+    }
+    else {
+        # Unix of some form or other
+
+        # See if we can determine the file system encoding through locale
+        $deduced_filename_encoding = $self->locale_encoding();
+        
+    }
+        print STDERR "guessing filesystem encoding is $deduced_filename_encoding\n";
+        return $deduced_filename_encoding;
+}
+
+
 
 sub deduce_filename_encoding
@@ -1192 +1216 @@
     return undef unless $self->can_process_this_file($filename_full_path);
     
+    #print STDERR "**** BEFORE READ INTO DOC OBJ: $file\n";
     my ($process_status,$doc_obj) = $self->read_into_doc_obj(@_);
-    
+    #print STDERR "**** AFTER READ INTO DOC OBJ: $file\n";
+    
     if ((defined $process_status) && ($process_status == 1)) {