Changeset 23352 for main/trunk/greenstone2/perllib/plugins/BasePlugin.pm

Timestamp:

2010-11-28T23:24:22+13:00 (13 years ago)

Author:

davidb

Message:

Modifications to code to support filename encoding issues when tested under Windows

File:

• main/trunk/greenstone2/perllib/plugins/BasePlugin.pm (modified) (9 diffs)

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

@@ -565 +565 @@
 
     elsif ($filename_encoding eq "auto-filesystem-encoding") 
-    {
+    {   
     # try locale
     $filename_encoding = $self->locale_encoding();
@@ -740 +740 @@
 
 # uses locale
-sub get_filesystem_encoding {
+sub get_filesystem_encoding 
+{
 
     my $self = shift(@_);
@@ -748 +749 @@
 
     eval {
+    # Works for Windows as well, returning the DOS code page in use 
     use POSIX qw(locale_h);
     
@@ -788 +790 @@
     
     }
+
     return $filesystem_encoding;
 }
@@ -840 +843 @@
     
     if (!defined $deduced_filename_encoding || ($deduced_filename_encoding =~ m/^\s*$/)) {
-        # 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
-                $deduced_filename_encoding = undef;
-            }
+
+    # Look to file system to provide a character encoding
+
+    # If Windows NTFS, then -- assuming we work with long file names got through
+    # Win32::GetLongFilePath() -- then the underlying file system is UTF16
+
+    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
+        # 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 = $self->locale_encoding();
+    
+        # if locale shows us filesystem is utf8, check to see filename is consistent
+        # => if not, then we have an "alien" filename on our hands
+        
+        if ($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*$/)) {
@@ -884 +900 @@
 sub set_Source_metadata {
     my $self = shift (@_);  
-    my ($doc_obj, $raw_file, $filename_encoding) = @_;
+    my ($doc_obj, $raw_filename, $filename_encoding) = @_;
 
     # 1. Sets the filename (Source) for display encoded as Unicode if possible,
@@ -890 +906 @@
     # 2. Sets the url ref (SourceFile) to the URL encoded version
     #    of filename for generated files
+    
+    my ($unused_full_rf, $raw_file) = &util::get_full_filenames("", $raw_filename);
 
     my $top_section = $doc_obj->get_top_section();
-    
+
+    my $octet_file = $raw_file;
+
     # UTF-8 version of filename
-    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
-        print STDERR "****** Setting Source Metadata given: $raw_file\n";
-    }
+    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
+    print STDERR "****** Setting Source Metadata given: $octet_file\n";
+    }
+    
+    # Deal with (on Windows) raw filenames that are in their
+    # abbreviated DOS form
+
+    if ($ENV{'GSDLOS'} =~ m/^windows$/i) {
+    if ((defined $filename_encoding) && ($filename_encoding eq "unicode")) {
+        if (-e $raw_filename) {
+        require Win32;
+        
+##      print STDERR "**** raw filename before LPN: $raw_filename\n";
+        my $unicode_filename = Win32::GetLongPathName($raw_filename);
+        
+        my $unused_full_uf;
+        ($unused_full_uf, $octet_file) = &util::get_full_filenames("", $unicode_filename);
+
+##      print STDERR "**** raw filename after LPN: $raw_filename\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);       
+    # => 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,$octet_file);     
     }
     else {
-        # 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";
-    }
-
+    # otherwise generate %xx encoded version of filename for char > 127
+    $url_encoded_filename = &unicode::raw_filename_to_url_encoded($octet_file);
+    }
+    
+    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
+    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", $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
@@ -926 +965 @@
                     $renamed_raw_url);
 
-    if ((defined $ENV{"DEBUG_UNICODE"}) && ($ENV{"DEBUG_UNICODE"})) {
-        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";
+    }
 }
    
@@ -988 +1027 @@
  
 
-    my $plugin_filename_encoding = $self->{'filename_encoding'};
+    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);
+    $self->set_Source_metadata($doc_obj,$filename_full_path,$filename_encoding,$filename_full_path);
 
     # plugin specific stuff - what args do we need here??