Changeset 34241 for main/trunk/gli/src/org/greenstone/gatherer/util

Timestamp:

2020-07-03T12:18:35+12:00 (4 years ago)

Author:

ak19

Message:

The previous commit fixed the issue where HTML in collection descriptions was not being preserved when GLI was not involved. This commit fixes the remaining problems with preserving HTML in coll descriptions when GLI is involved.

Location:

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

Files:

• Codec.java (modified) (2 diffs)
• XMLTools.java (modified) (1 diff)

main/trunk/gli/src/org/greenstone/gatherer/util/Codec.java

@@ -45 +45 @@
     static final public String GREENSTONE_TO_TEXT = "GREENSTONE_TO_TEXT";
     static final public String TEXT_TO_DOM = "TEXT_TO_DOM";
+    static final public String TEXT_TO_DOM_PRESERVE_TAGS = "TEXT_TO_DOM_PRESERVE_TAGS";
     static final public String TEXT_TO_GREENSTONE = "TEXT_TO_GREENSTONE";
     static final public String TEXT_TO_REGEXP = "TEXT_TO_REGEXP";
@@ -169 +170 @@
     TRANSFORMS.put(TEXT_TO_DOM, text_to_dom);
     text_to_dom = null;
+
+    // Same as above, but preserve html element tags
+    String[] text_to_dom_preserve_tags = {
+        "&", "&",
+        "\"", """,
+        "\'", "'"
+    };
+    TRANSFORMS.put(TEXT_TO_DOM_PRESERVE_TAGS, text_to_dom_preserve_tags);
+    text_to_dom_preserve_tags = null;
 
     // Unescape html (or xml) text

main/trunk/gli/src/org/greenstone/gatherer/util/XMLTools.java

@@ -957 +957 @@
             // And close.
             os.close();
+
+            // Useful for debugging unescaped chars that may otherwise break stuff
+            // w = new StringWriter();
+            // // Generate a new serializer from the above.
+            // s = new XMLSerializer(w, f);
+            // s.asDOMSerializer();
+            // // Finally serialize the document to file.
+            // s.serialize(document);
+
+            // System.err.println("@@@@ XMLTools: wrote out to doc:\n" + w.toString());
+            
         }
         catch (Exception exception)