Changeset 25249 for main/trunk/greenstone2

Timestamp:

2012-03-22T21:44:44+13:00 (12 years ago)

Author:

ak19

Message:

Added code to allow us to get all the up to date strings, instead of just all the strings (but unsorted), or just the strings that require translating/updating. We can create translation memory files (TMX) from all the up to date strings so that can translators using the Google Translator Toolkit can use the strings and string segments as a reference while translating the strings that still require updating/translating. If the TMX file is shared in the GTT, these strings and their segments will also be available to help translators of other application interfaces to use common and consistent terminology.

File:

• main/trunk/greenstone2/bin/script/gti.pl (modified) (4 diffs)

main/trunk/greenstone2/bin/script/gti.pl

@@ -131 +131 @@
         }
     }
+    elsif ($gti_command =~ /^get-uptodate-chunks$/i) {
+        if ($module eq "gs3interface") {       
+            print &get_uptodate_chunks_for_gs3(@gti_command_arguments);
+        } else {
+            print &get_uptodate_chunks(@gti_command_arguments);
+        }
+    }
     elsif ($gti_command =~ /^get-language-status$/i) {
         print &get_language_status(@gti_command_arguments);       
@@ -218 +225 @@
     my $xml_response = &create_xml_response_for_all_chunks($translation_file_key, $target_file, \%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping, \%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping);   
     
+    return $xml_response;
+}
+
+
+sub get_uptodate_chunks
+{
+    # The code of the target language (ensure it is lowercase)
+    my $target_language_code = lc(shift(@_));
+    # The key of the file to translate (ensure it is lowercase)
+    my $translation_file_key = lc(shift(@_));
+    
+    # Check that the necessary arguments were supplied
+    if (!$target_language_code || !$translation_file_key) {
+        &throw_fatal_error("Missing command argument.");
+    }
+    
+    # Get (and check) the translation configuration
+    my ($source_file, $target_file, $translation_file_type)
+    = &get_translation_configuration($target_language_code, $translation_file_key);
+    
+    # Parse the source language and target language files
+    my $source_file_path = &util::filename_cat($gsdl_root_directory, $source_file);
+    my @source_file_lines = &read_file_lines($source_file_path);
+    my %source_file_key_to_line_mapping = &build_key_to_line_mapping(\@source_file_lines, $translation_file_type);
+    
+    my $target_file_path = &util::filename_cat($gsdl_root_directory, $target_file);
+    my @target_file_lines = &read_file_lines($target_file_path);
+    my %target_file_key_to_line_mapping = &build_key_to_line_mapping(\@target_file_lines, $translation_file_type);
+    
+    # Filter out any automatically translated chunks
+    foreach my $chunk_key (keys(%source_file_key_to_line_mapping)) {
+        if (&is_chunk_automatically_translated($chunk_key, $translation_file_type)) {
+            delete $source_file_key_to_line_mapping{$chunk_key};
+            delete $target_file_key_to_line_mapping{$chunk_key};
+        }
+    }
+    
+    my %source_file_key_to_text_mapping = &build_key_to_text_mapping(\@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
+    my %target_file_key_to_text_mapping = &build_key_to_text_mapping(\@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type);
+    &log_message("Number of source chunks: " . scalar(keys(%source_file_key_to_text_mapping)));
+    &log_message("Number of target chunks: " . scalar(keys(%target_file_key_to_text_mapping)));
+    
+    my %source_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($source_file, \@source_file_lines, \%source_file_key_to_line_mapping, $translation_file_type);
+    my %target_file_key_to_last_update_date_mapping = &build_key_to_last_update_date_mapping($target_file, \@target_file_lines, \%target_file_key_to_line_mapping, $translation_file_type); 
+  
+
+    # Chunks needing updating are those in the target file that have been more recently edited in the source file
+    # All others are uptodate (which implies that they have certainly been translated at some point and would not be empty)
+    my @uptodate_target_file_keys = ();
+    foreach my $chunk_key (keys(%source_file_key_to_last_update_date_mapping)) {
+        my $source_chunk_last_update_date = $source_file_key_to_last_update_date_mapping{$chunk_key};
+        my $target_chunk_last_update_date = $target_file_key_to_last_update_date_mapping{$chunk_key};
+        
+        # print "key: $chunk_key\nsource date : $source_chunk_last_update_date\ntarget date : $target_chunk_last_update_date\nafter? ". &is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date) . "\n\n";        
+        
+        if (defined($target_chunk_last_update_date) && !&is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date)) {
+            # &log_message("Chunk with key $chunk_key needs updating.");
+            push(@uptodate_target_file_keys, $chunk_key);
+        }
+    }
+
+    my $xml_response = &create_xml_response_for_uptodate_chunks($translation_file_key, $target_file, \@uptodate_target_file_keys, \%source_file_key_to_text_mapping, \%target_file_key_to_text_mapping, \%source_file_key_to_last_update_date_mapping, \%target_file_key_to_last_update_date_mapping);   
+
     return $xml_response;
 }
@@ -1123 +1193 @@
 }
 
+sub create_xml_response_for_uptodate_chunks
+{
+    my ($translation_file_key, $target_file, $uptodate_target_files_keys, $source_files_key_to_text_mapping, $target_files_key_to_text_mapping, $source_files_key_to_last_update_date_mapping, $target_files_key_to_last_update_date_mapping) = @_;
+    
+    # Form an XML response to the command
+    my $xml_response = "<?xml version=\"1.0\" encoding=\"UTF-8\" ?>\n";
+    $xml_response .= "<GTIResponse>\n";
+    $xml_response .= "  <TranslationFile"
+    . " key=\"" . $translation_file_key . "\""
+    . " target_file_path=\"" . $target_file . "\""
+    . " num_chunks_uptodate=\"" . scalar(@$uptodate_target_files_keys) . "\"\/>\n";
+    
+    
+    # Then do chunks requiring updating
+    $xml_response .= "  <UptodateChunks size=\"" . scalar(@$uptodate_target_files_keys) . "\">\n";
+    
+    
+    # foreach my $chunk_key (@uptodate_target_file_keys) {
+    my @sorted_chunk_keys = sort (@$uptodate_target_files_keys);
+    foreach my $chunk_key (@sorted_chunk_keys) {
+        
+        my $source_file_chunk_date = $source_files_key_to_last_update_date_mapping->{$chunk_key} || "";
+        my $source_file_chunk_text = &make_text_xml_safe($source_files_key_to_text_mapping->{$chunk_key});
+        my $target_file_chunk_date = $target_files_key_to_last_update_date_mapping->{$chunk_key} || "";
+        my $target_file_chunk_text = &make_text_xml_safe($target_files_key_to_text_mapping->{$chunk_key});
+        
+        $xml_response .= "    <Chunk key=\"" . &make_text_xml_safe($chunk_key) . "\">\n";   
+        $xml_response .= "      <SourceFileText date=\"$source_file_chunk_date\">$source_file_chunk_text</SourceFileText>\n";
+        $xml_response .= "      <TargetFileText date=\"$target_file_chunk_date\">$target_file_chunk_text</TargetFileText>\n";
+        $xml_response .= "    </Chunk>\n";
+
+    }
+    
+    $xml_response .= "  </UptodateChunks>\n";
+    
+    $xml_response .= "</GTIResponse>\n";
+    
+    return $xml_response;
+}
 
 sub create_xml_response_for_all_chunks
@@ -1716 +1825 @@
 }
 
+sub get_uptodate_chunks_for_gs3
+{
+    # The code of the target language (ensure it is lowercase)
+    my $target_language_code = lc(shift(@_));
+    # The key of the file to translate (ensure it is lowercase)
+    my $translation_file_key = lc(shift(@_));
+    # The number of chunks to return (defaults to one if not specified)
+    my $num_chunks_to_return = shift(@_) || "1";
+    
+    # Check that the necessary arguments were supplied
+    if (!$target_language_code || !$translation_file_key) {
+        &throw_fatal_error("Missing command argument.");
+    }
+    
+    my %source_files_key_to_text_mapping = ();
+    my %target_files_key_to_text_mapping = ();
+    my %source_files_key_to_last_update_date_mapping = ();
+    my %target_files_key_to_last_update_date_mapping = ();
+    
+    &build_gs3_configuration($target_language_code, \%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping, 
+    \%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
+    
+
+    # Chunks needing updating are those in the target file that have been more recently edited in the source file
+    # All others are uptodate (which implies that they have certainly been translated at some point and would not be empty)
+    my @uptodate_target_file_keys = ();
+    foreach my $chunk_key (keys(%source_files_key_to_last_update_date_mapping)) {
+        my $source_chunk_last_update_date = $source_files_key_to_last_update_date_mapping{$chunk_key};
+        my $target_chunk_last_update_date = $target_files_key_to_last_update_date_mapping{$chunk_key};
+        
+        # print "key: $chunk_key\nsource date : $source_chunk_last_update_date\ntarget date : $target_chunk_last_update_date\nafter? ". &is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date) . "\n\n";        
+        
+        if (defined($target_chunk_last_update_date) && !&is_date_after($source_chunk_last_update_date, $target_chunk_last_update_date)) {
+            # &log_message("Chunk with key $chunk_key needs updating.");
+            push(@uptodate_target_file_keys, $chunk_key);
+        }
+    }
+
+    my $xml_response = &create_xml_response_for_uptodate_chunks($translation_file_key, "", \@uptodate_target_file_keys, \%source_files_key_to_text_mapping, \%target_files_key_to_text_mapping, \%source_files_key_to_last_update_date_mapping, \%target_files_key_to_last_update_date_mapping);
+    
+    return $xml_response;
+}