Changeset 23347 for main/trunk/greenstone2/perllib/plugins/BasePlugin.pm

Timestamp:

2010-11-26T09:43:59+13:00 (13 years ago)

Author:

davidb

Message:

Tidy up of debugging statements for handline filename encodings, plus finishing off the 'deduce_filename_encoding' routine

File:

• main/trunk/greenstone2/perllib/plugins/BasePlugin.pm (modified) (8 diffs)

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

@@ -530 +530 @@
     # check if the filename is already in UTF8. If it is, then we're done.
     if($filename_encoding =~ m/auto/) {
-    if(&unicode::check_is_utf8($filemeta)) 
-    {
-        $filename_encoding = "utf8";
-        return $filemeta;
-    } 
+        if(&unicode::check_is_utf8($filemeta)) 
+        {
+            $filename_encoding = "utf8";
+            return $filemeta;
+        } 
     }
     
@@ -540 +540 @@
     if ($filename_encoding eq "auto") 
     {
-    # try textcat
-    $filename_encoding = $self->textcat_encoding($filemeta);
+        # try textcat
+        $filename_encoding = $self->textcat_encoding($filemeta);
     
-    # check the locale next
-    $filename_encoding = $self->locale_encoding() if $filename_encoding eq "undefined";
+        # check the locale next
+        $filename_encoding = $self->locale_encoding() if $filename_encoding eq "undefined";
     
-
-    # now try the encoding of the document, if available
-    if ($filename_encoding eq "undefined" && defined $file_encoding) {
-        $filename_encoding = $file_encoding;
-    }
+        
+        # now try the encoding of the document, if available
+        if ($filename_encoding eq "undefined" && defined $file_encoding) {
+            $filename_encoding = $file_encoding;
+        }
 
     }
@@ -633 +633 @@
 
     my $outhandle = $self->{'outhandle'};
+
+    print $outhandle "****!!!!**** BasePlugin::filename_to_utf8_metadata now deprecated\n";
+    my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(0);
+    print $outhandle "Calling method: $cfilename:$cline $cpackage->$csubr\n";
+
 
     my ($filemeta) = $file =~ /([^\\\/]+)$/; # getting the tail of the filepath (skips all string parts containing slashes upto the end)
@@ -791 +796 @@
 {
     my $self = shift (@_);  
-    my ($file,$metadata) = @_;
+    my ($file,$metadata,$plugin_filename_encoding) = @_;
 
     my $gs_filename_encoding = $metadata->{"gs.filename_encoding"};
@@ -798 +803 @@
     # 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 = $deduced_filename_encoding\n";
-    
+        if (ref ($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;
+        }
+    }
+        
     if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
-    # Look to see if plugin specifies this value
+        # Look to see if plugin specifies this value
+
+        if (defined $plugin_filename_encoding) {
+            # First look to see if we're using any of the "older" (i.e. deprecated auto-... plugin options)
+            if ($plugin_filename_encoding =~ m/^auto-.*$/) {
+                my $outhandle = $self->{'outhandle'};
+                print $outhandle "Warning: $plugin_filename_encoding is no longer supported\n";
+                print $outhandle "         default to 'auto'\n";
+                $self->{'filename_encoding'} = $plugin_filename_encoding = "auto";
+            }
+            
+            if ($plugin_filename_encoding ne "auto") {
+                # We've been given a specific filenamne encoding
+                # => so use it!
+                $deduced_filename_encoding = $plugin_filename_encoding;
+            }
+        }
     }
     
     if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
-    # See if we can determine the file system encoding through locale
-    # Unix only ?
+        # See if we can determine the file system encoding through locale
+        $deduced_filename_encoding = $self->locale_encoding();
+
+        # if locale shows us filesystem is utf8, check to see filename is consistent
+        # => if not, then we have an "alien" filename on our hands
+
+        if ($deduced_filename_encoding =~ m/^utf-?8$/i) {
+            if (!&unicode::check_is_utf8($file)) {
+                # "alien" filename, so revert
+                $deduced_filename_encoding = undef;
+            }
+        }
+    }
+    
+    
+#    if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
+#       # Last chance, apply textcat to deduce filename encoding
+#       $deduced_filename_encoding = $self->textcat_encoding($file);
+#    }
+
+    if ($self->{'verbosity'}>3) {
+        my $outhandle = $self->{'outhandle'};
+
+        if (defined $deduced_filename_encoding) {
+            print $outhandle "  Deduced filename encoding as: $deduced_filename_encoding\n";
+        }
+        else {
+            print $outhandle "  No filename encoding deduced\n";
+        }
+    }
     
-    # 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;
 }
@@ -861 +894 @@
     
     # UTF-8 version of filename
-    print STDERR "**** setting Source Metadata given: $raw_file\n";
-
-##    my $filemeta = $self->filename_to_utf8_metadata($raw_file, $filename_encoding);
+    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
+        print STDERR "****** Setting Source Metadata given: $raw_file\n";
+    }
 
     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";
+        # => 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);       
     }
     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";
+        # otherwise generate %xx encoded version of filename for char > 127
+        $url_encoded_filename = &unicode::raw_filename_to_url_encoded($raw_file);
+    }
+    
+    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
+        print STDERR "***** saving Source as:             $url_encoded_filename\n";
+    }
 
     
@@ -893 +926 @@
                     $renamed_raw_url);
 
-    print STDERR "***** saving SourceFile as:         $renamed_raw_url\n";
+    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
+        print STDERR "***** saving SourceFile as:         $renamed_raw_url\n";
+    }
 }
    
@@ -953 +988 @@
  
 
-    my $filename_encoding = $self->deduce_filename_encoding($file,$metadata);
+    my $plugin_filename_encoding = $self->{'filename_encoding'};
+    my $filename_encoding = $self->deduce_filename_encoding($file,$metadata,$plugin_filename_encoding);
     $self->set_Source_metadata($doc_obj,$filename_no_path,$filename_encoding);