[jsword-svn] r2112 - in trunk/jsword/src/main/java/org/crosswire: common/xml jsword/book/filter/thml

[dmsmith at crosswire.org]

Author: dmsmith
Date: 2011-03-10 18:29:39 -0700 (Thu, 10 Mar 2011)
New Revision: 2112

Modified:
   trunk/jsword/src/main/java/org/crosswire/common/xml/XMLUtil.java
   trunk/jsword/src/main/java/org/crosswire/jsword/book/filter/thml/IgnoreTag.java
   trunk/jsword/src/main/java/org/crosswire/jsword/book/filter/thml/THMLFilter.java
Log:


Modified: trunk/jsword/src/main/java/org/crosswire/common/xml/XMLUtil.java
===================================================================
--- trunk/jsword/src/main/java/org/crosswire/common/xml/XMLUtil.java	2011-03-08 14:05:54 UTC (rev 2111)
+++ trunk/jsword/src/main/java/org/crosswire/common/xml/XMLUtil.java	2011-03-11 01:29:39 UTC (rev 2112)
@@ -230,7 +230,7 @@
     }
 
     /**
-     * Remove all invalid characters in the input. XML has stringent
+     * Remove all invalid characters in the input, replacing them with a space. XML has stringent
      * requirements as to which characters are or are not allowed. The set of
      * allowable characters are:<br />
      * #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]<br/>
@@ -241,7 +241,7 @@
      * @return the cleaned string
      */
     public static String cleanAllCharacters(String broken) {
-        return invalidCharacterPattern.matcher(broken).replaceAll("");
+        return invalidCharacterPattern.matcher(broken).replaceAll(" ");
     }
 
     /**
@@ -321,10 +321,9 @@
                 }
             }
 
-            // So we have the end of the tag, delete it ...
-            // DataPolice.report("discarding tag: " + working.substring(lt, i +
-            // 1));
-            working = working.substring(0, lt) + working.substring(i + 1);
+            // So we have the end of the tag, delete it, but leave a space in it's place
+            // DataPolice.report("discarding tag: " + working.substring(lt, i + 1));
+            working = working.substring(0, lt) + " " + working.substring(i + 1);
         }
 
         return working;
@@ -350,7 +349,8 @@
             return replace;
         }
 
-        return "";
+        // replace unknown entities with a space
+        return " ";
     }
 
     // Map entities to their unicode equivalent

Modified: trunk/jsword/src/main/java/org/crosswire/jsword/book/filter/thml/IgnoreTag.java
===================================================================
--- trunk/jsword/src/main/java/org/crosswire/jsword/book/filter/thml/IgnoreTag.java	2011-03-08 14:05:54 UTC (rev 2111)
+++ trunk/jsword/src/main/java/org/crosswire/jsword/book/filter/thml/IgnoreTag.java	2011-03-11 01:29:39 UTC (rev 2112)
@@ -50,6 +50,6 @@
         // Replace the parent with this element
         Element parent = ele.getParentElement();
         parent.removeContent(ele);
-        parent.addContent(ele.getChildren());
+        parent.addContent(ele.removeContent());
     }
 }

Modified: trunk/jsword/src/main/java/org/crosswire/jsword/book/filter/thml/THMLFilter.java
===================================================================
--- trunk/jsword/src/main/java/org/crosswire/jsword/book/filter/thml/THMLFilter.java	2011-03-08 14:05:54 UTC (rev 2111)
+++ trunk/jsword/src/main/java/org/crosswire/jsword/book/filter/thml/THMLFilter.java	2011-03-11 01:29:39 UTC (rev 2112)
@@ -131,8 +131,8 @@
     private Element parse(Book book, Key key, String plain, String failMessage) {
         Exception ex = null;
         // We need to create a root element to house our document fragment
-        StringBuilder buf = new StringBuilder(15 + plain.length()); // 15 for the
-                                                                  // tags we add
+        // 15 for the tags we add
+        StringBuilder buf = new StringBuilder(15 + plain.length());
         buf.append('<').append(RootTag.TAG_ROOT).append('>').append(plain).append("</").append(RootTag.TAG_ROOT).append('>');
         finalInput = buf.toString();
         try {
@@ -152,7 +152,13 @@
             ex = e;
         } catch (ParserConfigurationException e) {
             ex = e;
-        }
+        } catch (IllegalArgumentException e) {
+            // JDOM has a few exceptions which are all derived from this.
+            ex = e;
+        } catch (RuntimeException e) {
+            // Catch everything else so that we handle the exception properly within a Sw*ng callback
+            ex = e;
+        } 
 
         errorMessage = failMessage;
         error = ex;