From 786be7d175460cdb5469f8ba59c91b9ff5b225e3 Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Wed, 20 Jun 2012 16:39:04 +0200
Subject: [PATCH] better integration of RDFaParser

---
 build.xml                                       | 7 +++++++
 source/net/yacy/document/Document.java          | 1 +
 source/net/yacy/document/TextParser.java        | 2 +-
 source/net/yacy/search/index/DocumentIndex.java | 1 +
 4 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/build.xml b/build.xml
index fc5c83f80..b786882fa 100644
--- a/build.xml
+++ b/build.xml
@@ -58,6 +58,7 @@
   <property name="release_mac" location="${release}/MAC"/>
   <property name="git" location=".git"/>
   <property name="defaults" location="defaults"/>
+  <property name="RDFaParser" location="RDFaParser"/>
 
   <!-- pseudo default releaseNr as fallback --> 
   <property name="releaseNr" value="9000" />
@@ -296,6 +297,12 @@
                includes="**/*"/>
     </copy>
 
+  	<!-- copy RDFaParser -->
+    <copy todir="${release_main}/RDFaParser">
+      <fileset dir="${RDFaParser}" 
+               includes="**/*"/>
+    </copy>
+
   	<!-- copy Unix wrappers -->
     <copy todir="${release_main}">
       <fileset dir=".">
diff --git a/source/net/yacy/document/Document.java b/source/net/yacy/document/Document.java
index 9bfecb957..42fb1f190 100644
--- a/source/net/yacy/document/Document.java
+++ b/source/net/yacy/document/Document.java
@@ -781,6 +781,7 @@ dc_rights
 
         for (final Document doc: docs) {
 
+        	if (doc == null) continue;
             final String author = doc.dc_creator();
             if (author.length() > 0) {
                 if (authors.length() > 0) authors.append(",");
diff --git a/source/net/yacy/document/TextParser.java b/source/net/yacy/document/TextParser.java
index d4a2ba225..da27a2fd5 100644
--- a/source/net/yacy/document/TextParser.java
+++ b/source/net/yacy/document/TextParser.java
@@ -261,7 +261,7 @@ public final class TextParser {
         if (log.isFine()) log.logInfo("Parsing " + location + " with mimeType '" + mimeType + "' and file extension '" + fileExt + "'.");
         try {
             final Document[] docs = parser.parse(location, mimeType, documentCharset, sourceStream);
-            for (final Document d: docs) { assert d.getText() != null; } // verify docs
+            for (final Document d: docs) { assert d != null && d.getText() != null; } // verify docs
             return docs;
         } catch (final Exception e) {
             throw new Parser.Failure("parser failed: " + parser.getName(), location);
diff --git a/source/net/yacy/search/index/DocumentIndex.java b/source/net/yacy/search/index/DocumentIndex.java
index a25763844..30af3a83b 100644
--- a/source/net/yacy/search/index/DocumentIndex.java
+++ b/source/net/yacy/search/index/DocumentIndex.java
@@ -157,6 +157,7 @@ public class DocumentIndex extends Segment
         final URIMetadataRow[] rows = new URIMetadataRow[documents.length];
         int c = 0;
         for ( final Document document : documents ) {
+        	if (document == null) continue;
             final Condenser condenser = new Condenser(document, true, true, LibraryProvider.dymLib, true);
             rows[c++] =
                 super.storeDocument(