TIKA-4581 - fix faulty logic in PipesServer and add intermediate results for concatenated parsing

Author: tballison

tika-pipes/tika-pipes-core/src/main/java/org/apache/tika/pipes/core/server/ParseHandler.java

@@ -169,7 +169,7 @@ public List<Metadata> parseRecursive(FetchEmitTuple fetchEmitTuple,
 
     public List<Metadata> parseConcatenated(FetchEmitTuple fetchEmitTuple,
                                              HandlerConfig handlerConfig, TikaInputStream stream,
-                                             Metadata metadata, ParseContext parseContext) {
+                                             Metadata metadata, ParseContext parseContext) throws InterruptedException {
 
         ContentHandlerFactory contentHandlerFactory =
                 new BasicContentHandlerFactory(handlerConfig.getType(),
@@ -193,7 +193,9 @@ public boolean select(Metadata metadata) {
         String containerException = null;
         long start = System.currentTimeMillis();
         preParse(fetchEmitTuple, stream, metadata, parseContext);
-        //TODO -- add intermediate
+        //queue better be empty. we deserve an exception if not
+        intermediateResult.add(metadata);
+        countDownLatch.await();
         try {
             autoDetectParser.parse(stream, handler, metadata, parseContext);
         } catch (SAXException e) {

tika-pipes/tika-pipes-core/src/main/java/org/apache/tika/pipes/core/server/PipesServer.java

@@ -474,15 +474,11 @@ protected void initializeResources() throws TikaException, IOException, SAXExcep
             // User doesn't want container documents digested
             this.digester = null;
         }
-        if (this.digester != null) {
-            // If the user hasn't configured an embedded document extractor, set up the
-            // RUnpackExtractorFactory
-            if (autoDetectParser.getAutoDetectParserConfig()
-                    .getEmbeddedDocumentExtractorFactory() == null) {
-                autoDetectParser
-                        .getAutoDetectParserConfig().setEmbeddedDocumentExtractorFactory(
-                                new RUnpackExtractorFactory());
-            }
+
+        // If the user hasn't configured an embedded document extractor, set up the
+        // RUnpackExtractorFactory
+        if (autoDetectParser.getAutoDetectParserConfig().getEmbeddedDocumentExtractorFactory() == null) {
+                autoDetectParser.getAutoDetectParserConfig().setEmbeddedDocumentExtractorFactory(new RUnpackExtractorFactory());
         }
         this.detector = this.autoDetectParser.getDetector();
         this.rMetaParser = new RecursiveParserWrapper(autoDetectParser);