Changeset 23335 for main/trunk/greenstone2/perllib/plugins/BasePlugin.pm

Timestamp:

2010-11-19T13:29:29+13:00 (13 years ago)

Author:

davidb

Message:

Work done on improving handing of filenames when the actualy filename encoding used is not necesarrily known. Tested for Linux. Work currently includes some debug statements that will be removed once testing for Windows and Mac is done.

File:

• main/trunk/greenstone2/perllib/plugins/BasePlugin.pm (modified) (5 diffs)

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

@@ -31 +31 @@
 
 use File::Basename;
+use Encode;
 
 use encodings;
@@ -785 +786 @@
 }
 
-# is there ever only one Source? Sometimes this will be called twice, for images etc that are converted.
+
+
+sub deduce_filename_encoding
+{
+    my $self = shift (@_);  
+    my ($file,$metadata) = @_;
+
+    my $gs_filename_encoding = $metadata->{"gs.filename_encoding"};
+    my $deduced_filename_encoding = undef;
+    
+    # Start by looking for manually assigned metadata
+    if (defined $gs_filename_encoding) {
+    if (ref ($gs_filename_encoding) eq "ARRAY") {
+        my $outhandle = $self->{'outhandle'};
+        
+        $deduced_filename_encoding = $gs_filename_encoding->[0];
+        
+        my $num_vals = scalar(@$gs_filename_encoding);
+        if ($num_vals>1) {
+        print $outhandle "Warning: gs.filename_encoding multiply defined for $file\n";
+        print $outhandle "         Selecting first value: $deduced_filename_encoding\n";
+        }
+    } 
+    else {
+        $deduced_filename_encoding = $gs_filename_encoding;
+    }
+    }
+    
+#   binmode(STDERR,":utf8");
+    
+#   print STDERR "**** file = $file\n";
+#   print STDERR "**** debug file = ", &unicode::debug_unicode_string($file),"\n";;
+    
+#   print STDERR "******* dfe = $deduced_filename_encoding\n";
+    
+    if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
+    # Look to see if plugin specifies this value
+    }
+    
+    if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
+    # See if we can determine the file system encoding through locale
+    # Unix only ?
+    
+    # if locale shows us filesystem is utf8, check to see filename is consistent
+    # => if not, then we have an "alien" filename on our hands
+    }
+    
+    
+    if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
+    # Last chance, apply textcat to deduce filename encoding
+    }
+    
+    return $deduced_filename_encoding;
+}
+
+
+
+
+# Notionally written to be called once for each document, it is however safe to
+# call multiple times (as in the case of ImagePlugin) which calls this later on
+# after the original image has potentially been converted to a *new* source image
+# format (e.g. TIFF to PNG)
+
 sub set_Source_metadata {
     my $self = shift (@_);  
-    my ($doc_obj, $filename_no_path, $file_encoding) = @_;
+    my ($doc_obj, $raw_file, $filename_encoding) = @_;
+
+    # 1. Sets the filename (Source) for display encoded as Unicode if possible,
+    #    and (as a fallback) using %xx if not for non-ascii chars
+    # 2. Sets the url ref (SourceFile) to the URL encoded version
+    #    of filename for generated files
 
     my $top_section = $doc_obj->get_top_section();
     
     # UTF-8 version of filename
-    my $filemeta = $self->filename_to_utf8_metadata($filename_no_path, $file_encoding);
+    print STDERR "**** setting Source Metadata given: $raw_file\n";
+
+##    my $filemeta = $self->filename_to_utf8_metadata($raw_file, $filename_encoding);
+
+    my $url_encoded_filename;
+    if (defined $filename_encoding) {
+    # => Generate a pretty print version of filename that is mapped to Unicode
+
+    # Use filename_encoding to map raw filename to a Perl unicode-aware string 
+    $url_encoded_filename = decode($filename_encoding,$raw_file);
+
+    print STDERR "@@@@ pretty print using $filename_encoding: ", encode("utf8",$url_encoded_filename),"\n";
+    }
+    else {
+    # otherwise generate %xx encoded version of filename for char > 127
+    $url_encoded_filename = &unicode::raw_filename_to_url_encoded($raw_file);
+    }
+
+    print STDERR "***** saving Source as:             $url_encoded_filename\n";
+
     
     # Source is the UTF8 display name - not necessarily the name of the file on the system
-    $doc_obj->set_utf8_metadata_element($top_section, "Source", $filemeta); 
-
-    $filemeta = &util::rename_file($filemeta, $self->{'file_rename_method'});
+    $doc_obj->set_utf8_metadata_element($top_section, "Source", $url_encoded_filename); 
+
+    my $renamed_raw_file = &util::rename_file($raw_file, $self->{'file_rename_method'});
     # If using URL encoding, then SourceFile is the url-reference to url-encoded
-    # filemeta: it's a url that refers to the actual file on the system
-    $filemeta = &unicode::filename_to_url($filemeta);
-
-    $doc_obj->set_utf8_metadata_element($top_section, "SourceFile", $filemeta);
+    # renamed_raw_url: it's a url that refers to the actual file on the system
+    my $renamed_raw_url = &unicode::filename_to_url($renamed_raw_file);
+
+    $doc_obj->set_utf8_metadata_element($top_section, "SourceFile",
+                    $renamed_raw_url);
+
+    print STDERR "***** saving SourceFile as:         $renamed_raw_url\n";
 }
    
@@ -863 +953 @@
  
 
-    # sets the UTF8 filename (Source) for display and sets the url ref to URL encoded version
-    # of the UTF8 filename (SourceFile) for generated files
-    $self->set_Source_metadata($doc_obj, $filename_no_path);
-
+    my $filename_encoding = $self->deduce_filename_encoding($file,$metadata);
+    $self->set_Source_metadata($doc_obj,$filename_no_path,$filename_encoding);
 
     # plugin specific stuff - what args do we need here??
@@ -1032 +1120 @@
     my $file_derived_title = $file;
     $file_derived_title =~ s/_/ /g;
-    $file_derived_title =~ s/\.[^.]$//;
+    $file_derived_title =~ s/\.[^.]+$//;
 
     return $file_derived_title;
@@ -1044 +1132 @@
 
     if (!defined $doc_obj->get_metadata_element ($section, "Title") 
-    or $doc_obj->get_metadata_element($section, "Title") eq "") {
-
-    my $file_derived_title = $self->filename_to_utf8_metadata($self->filename_based_title($file));
+    || $doc_obj->get_metadata_element($section, "Title") eq "") {
+
+    my $source_file = $doc_obj->get_metadata_element($section, "Source");
+    my $file_derived_title;
+    if (defined $source_file) {
+        $file_derived_title =  $self->filename_based_title($source_file);
+    }
+    else {
+        # pp = pretty print
+        my $pp_file = (defined $source_file) ? $source_file : $file;
+
+        my $raw_title = $self->filename_based_title($file);
+        my $file_derived_title = &unicode::raw_filename_to_url_encoded($raw_title);
+    }
+
+
     if (!defined $doc_obj->get_metadata_element ($section, "Title")) {
         $doc_obj->add_utf8_metadata ($section, "Title", $file_derived_title);