add embedding depth to rpw handler

nddipiazza · Nov 22, 2019 · 3d2afec · 3d2afec
1 parent cb3c4ba
commit 3d2afec
Showing 1 changed file with 6 additions and 0 deletions.
diff --git a/tika-core/src/main/java/org/apache/tika/sax/AbstractRecursiveParserWrapperHandler.java b/tika-core/src/main/java/org/apache/tika/sax/AbstractRecursiveParserWrapperHandler.java
@@ -54,6 +54,10 @@ public abstract class AbstractRecursiveParserWrapperHandler extends DefaultHandl
     public final static Property EMBEDDED_RESOURCE_PATH =
             Property.internalText(TikaCoreProperties.TIKA_META_PREFIX+"embedded_resource_path");
 
+    public final static Property EMBEDDED_DEPTH =
+            Property.internalInteger(TikaCoreProperties.TIKA_META_PREFIX+"embedded_depth");
+
+
     private final ContentHandlerFactory contentHandlerFactory;
 
     private static final int MAX_DEPTH = 100;
@@ -93,6 +97,7 @@ public void startEmbeddedDocument(ContentHandler contentHandler, Metadata metada
         if (embeddedDepth >= MAX_DEPTH) {
             throw new SAXException("Max embedded depth reached: "+embeddedDepth);
         }
+        metadata.set(EMBEDDED_DEPTH, embeddedDepth);
     }
     /**
      * This is called after parsing each embedded document.  Override this
@@ -120,6 +125,7 @@ public void endDocument(ContentHandler contentHandler, Metadata metadata) throws
         if (hasHitMaximumEmbeddedResources()) {
             metadata.set(EMBEDDED_RESOURCE_LIMIT_REACHED, "true");
         }
+        metadata.set(EMBEDDED_DEPTH, 0);
     }
 
     /**