Changeset 28503 for main/trunk/greenstone2/bin/script

Timestamp:

2013-10-21T20:57:25+13:00 (11 years ago)

Author:

ak19

Message:

1. The process of submitting translation spreadsheets produced some errors in borderline cases saying that the source string of the spreadsheet didn't match the one in the original translation file owing to newline entity differences and differences as to when entities got expanded during the check for whether the source strings still matched. These cases are now handled. 2. Added debugging statements in key positions (but commented out).

Location:

main/trunk/greenstone2/bin/script

Files:

• gti-process-spreadsheet.pl (modified) (3 diffs)
• gti.pl (modified) (1 diff)

main/trunk/greenstone2/bin/script/gti-process-spreadsheet.pl

@@ -83 +83 @@
     $source_file_chunk =~ /^(\S+)\s+((.|\n)*)$/;
     my $source_file_chunk_key = $1;
+#   print STDERR "******** key: |$source_file_chunk_key| ";
     my $source_file_chunk_text = $2;
+#   print STDERR "******** text: |$source_file_chunk_text|\n";
     $target_file_chunk =~ /^(\S+)\s+((.|\n)*)$/;
     my $target_file_chunk_key = $1;
@@ -90 +92 @@
     # Remove the quotes around multiline chunks
     if ($source_file_chunk_text =~ /^\"/ && $source_file_chunk_text =~ /\"$/) {
+#       print STDERR "******** source text: |$source_file_chunk_text| \n";
         $source_file_chunk_text =~ s/^\"//;
         $source_file_chunk_text =~ s/\"$//;
     }
     if ($target_file_chunk_text =~ /^\"/ && $target_file_chunk_text =~ /\"$/) {
+#       print STDERR "******** target text: |$target_file_chunk_text| \n";
         $target_file_chunk_text =~ s/^\"//;
         $target_file_chunk_text =~ s/\"$//;
     }
+#   else {
+#       print STDERR "******** !target text: |$target_file_chunk_text| \n";
+#   }
 
         # Remove the blank space Excel adds at the start of each line
@@ -106 +113 @@
     $target_file_chunk_text =~ s/\"\"/\"/g;
 
+    # ensure newline html entities in the unicode txt file version of the spreadsheet are replaced with newlines
+    $source_file_chunk_text =~ s/
 /\n/g;
+    $target_file_chunk_text =~ s/
 /\n/g;
+
     print "<SourceFileText key=\"" . $source_file_chunk_key . "\">\n" . $source_file_chunk_text . "\n</SourceFileText>\n";
     print "<TargetFileText key=\"" . $target_file_chunk_key . "\">\n" . $target_file_chunk_text . "\n</TargetFileText>\n";

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

@@ -669 +669 @@
         
         # Make sure the submitted source chunk matches the source file chunk
-        if ($source_file_key_to_submission_mapping{$chunk_key} ne &unmake_text_xml_safe($source_file_key_to_text_mapping{$chunk_key})) {
+        if (&unmake_text_xml_safe($source_file_key_to_submission_mapping{$chunk_key}) ne &unmake_text_xml_safe($source_file_key_to_text_mapping{$chunk_key})) {
+                    #print STDERR "**** $source_file_key_to_submission_mapping{$chunk_key}\n";
+                #print STDERR "**** " . &unmake_text_xml_safe($source_file_key_to_text_mapping{$chunk_key}) ."\n";
+
             &log_message("Warning: Source chunk $chunk_key has changed (ignoring submission).");
             &log_message("Submission source: $source_file_key_to_submission_mapping{$chunk_key}");