Changeset 29793 for main/trunk/gli/src

Timestamp:

2015-03-19T13:53:24+13:00 (9 years ago)

Author:

ak19

Message:

some more changes to make adding metadata to filenames with non latin1 characters work. previously, any char above 127 was being output as %XX. this doesn't work as higher values can end up like %101. but when you decode you only ever look for 2 digits after the %. so higher values we'll use entities like ā then had to modify the xml writing to not escape the text in hte filename element - otherwise get ā

Location:

main/trunk/gli/src/org/greenstone/gatherer/metadata

Files:

• FilenameEncoding.java (modified) (6 diffs)
• MetadataXMLFile.java (modified) (3 diffs)

main/trunk/gli/src/org/greenstone/gatherer/metadata/FilenameEncoding.java

@@ -32 +32 @@
 import java.util.*;
 import org.greenstone.gatherer.collection.CollectionManager;
-
+import org.greenstone.gatherer.DebugStream;
 
 /** Static access class that contains many of the methods used to work with filename encodings. 
@@ -347 +347 @@
             
             String filename_ascii = filename_uri.toASCIIString();
+            DebugStream.println("ascii = "+filename_ascii);
             String filename_raw_bytes = URLDecoder.decode(filename_ascii,"ISO-8859-1");
+            DebugStream.println("raw = "+filename_raw_bytes);
             filename_url_encoded = iso_8859_1_filename_to_url_encoded(filename_raw_bytes);
+            
         }
         catch (Exception e) {
@@ -355 +358 @@
             filename_url_encoded = file.getAbsolutePath(); 
         }
+            DebugStream.println("returning "+filename_url_encoded);
         return filename_url_encoded;
     }
@@ -368 +372 @@
     {
         String urlEncoded = "";
-
+DebugStream.println("in iso 8859 to url encoded, "+raw_bytes_filename);
         try {
             // By this point we have a UTF-8 encoded string that captures
@@ -386 +390 @@
             String unicode_filename = new String(raw_bytes,"UTF-8");
             
+                //urlEncoded = URLEncoder.encode(unicode_filename, "UTF-8");
             for(int i = 0; i < unicode_filename.length(); i++) {
             char charVal = unicode_filename.charAt(i);
-            if((int)charVal > 127) {
+            if ((int)charVal > 255) {
+                urlEncoded += String.format("&#x%02X;", (int)charVal);
+                    //urlEncoded += String.format("\\u%04X", (int)charVal);
+            }
+            else if((int)charVal > 127) {
                 urlEncoded += String.format("%%%02X", (int)charVal);
+                    //urlEncoded += Integer.toHexString((int)charVal);
             } else {
                 urlEncoded += String.format("%c", (char)charVal);
@@ -399 +409 @@
             throw(e);
         }
-
+DebugStream.println("returning "+urlEncoded);
         return urlEncoded;
     }

main/trunk/gli/src/org/greenstone/gatherer/metadata/MetadataXMLFile.java

@@ -45 +45 @@
     static final private String FILESET_ELEMENT = "FileSet";
     static final private String METADATA_ELEMENT = "Metadata";
-
+    static final private String[] nonEscapingElements = new String[]{FILENAME_ELEMENT};
+    
     /** Special metadata field: the filename encoding is a unique sort of metadata in
      * that it is not just information stored with a collection file, but also needs to
@@ -123 +124 @@
     // If no appropriate FileSet element exists create a new one for this file
     if (appropriate_fileset_element == null) {
-        DebugStream.println("Creating new FileSet element for file since none exists...");
+        DebugStream.println("Creating new FileSet element for file since none exists..."+file_path_regexp);
         appropriate_fileset_element = loaded_file_document.createElement(FILESET_ELEMENT);
 
@@ -605 +606 @@
     // If we have a file loaded into memory and it has been modified, save it now
     if (loaded_file != null && loaded_file_changed == true) {
-        XMLTools.writeXMLFile(loaded_file, loaded_file_document);
+        XMLTools.writeXMLFile(loaded_file, loaded_file_document, nonEscapingElements);
+    
+        
         loaded_file_changed = false;
     }