TIKA-4303: Handle OneNotePropertyEnum.CachedTitleString as RichEditTe…

…xtUnicode (#2098)
apache · Jan 27, 2025 · 7f94520 · 7f94520
1 parent 1b8c1e2
commit 7f94520
Show file tree

Hide file tree

Showing 4 changed files with 16 additions and 4 deletions.
diff --git a/...soft-module/src/main/java/org/apache/tika/parser/microsoft/onenote/OneNoteTreeWalker.java b/...soft-module/src/main/java/org/apache/tika/parser/microsoft/onenote/OneNoteTreeWalker.java
@@ -393,7 +393,8 @@ private List<Map<String, Object>> processPropertySet(PropertySet propertySet,
     private boolean propertyIsBinary(OneNotePropertyEnum property) {
         return property == OneNotePropertyEnum.RgOutlineIndentDistance ||
                 property == OneNotePropertyEnum.NotebookManagementEntityGuid ||
-                property == OneNotePropertyEnum.RichEditTextUnicode;
+                property == OneNotePropertyEnum.RichEditTextUnicode ||
+                property == OneNotePropertyEnum.CachedTitleString;
     }
 
     /**
@@ -508,7 +509,9 @@ private Map<String, Object> processPropertyValue(PropertyValue propertyValue,
                                     dif.size());
                 }
                 if (propertyValue.propertyId.propertyEnum ==
-                        OneNotePropertyEnum.RichEditTextUnicode) {
+                        OneNotePropertyEnum.RichEditTextUnicode
+                        || propertyValue.propertyId.propertyEnum ==
+                        OneNotePropertyEnum.CachedTitleString) {
                     if (!options.isOnlyLatestRevision()
                             || (parentPropertyId != null &&
                             parentPropertyId.propertyEnum != OneNotePropertyEnum.ElementChildNodesOfVersionHistory)) {

diff --git a/...dule/src/main/java/org/apache/tika/parser/microsoft/onenote/OneNoteTreeWalkerOptions.java b/...dule/src/main/java/org/apache/tika/parser/microsoft/onenote/OneNoteTreeWalkerOptions.java
@@ -29,8 +29,7 @@ public class OneNoteTreeWalkerOptions implements Serializable {
     private boolean crawlAllFileNodesFromRoot = true;
     private boolean onlyLatestRevision = true;
     private Set<OneNotePropertyEnum> utf16PropertiesToPrint = new HashSet<>(
-            Arrays.asList(OneNotePropertyEnum.ImageFilename, OneNotePropertyEnum.Author,
-                    OneNotePropertyEnum.CachedTitleString));
+            Arrays.asList(OneNotePropertyEnum.ImageFilename, OneNotePropertyEnum.Author));
 
     /**
      * Do this to ignore revisions and just parse all file nodes from the root recursively.

diff --git a/...soft-module/src/test/java/org/apache/tika/parser/microsoft/onenote/OneNoteParserTest.java b/...soft-module/src/test/java/org/apache/tika/parser/microsoft/onenote/OneNoteParserTest.java
@@ -288,4 +288,14 @@ public void testDupeText() throws Exception {
 
         assertEquals(1, StringUtils.countMatches(txt, "Sunday morning"));
     }
+
+    /**
+     * TIKA-4303 - test extract Chinese
+     */
+    @Test
+    public void testExtractChinese() throws Exception {
+        Metadata metadata = new Metadata();
+        XMLResult xml = getXML("test-tika-4303-Chinese-notes.one", metadata);
+        assertContains("<p>中文标题</p>", xml.xml);
+    }
 }
diff --git a/...arser-microsoft-module/src/test/resources/test-documents/test-tika-4303-Chinese-notes.one b/...arser-microsoft-module/src/test/resources/test-documents/test-tika-4303-Chinese-notes.one