From 8e476c74114f5af6ef8c25b13cfc8ae13e873ed5 Mon Sep 17 00:00:00 2001
From: Max Jakob <max.jakob@gmail.com>
Date: Tue, 16 Aug 2011 15:54:45 +0200
Subject: [PATCH] escape homepage URIs

---
 .../dbpedia/extraction/destinations/Quad.scala    | 15 +++++++++++++--
 dump/config.properties.default                    |  2 +-
 2 files changed, 14 insertions(+), 3 deletions(-)

diff --git a/core/src/main/scala/org/dbpedia/extraction/destinations/Quad.scala b/core/src/main/scala/org/dbpedia/extraction/destinations/Quad.scala
index 34fe086b75..93504d26be 100644
--- a/core/src/main/scala/org/dbpedia/extraction/destinations/Quad.scala
+++ b/core/src/main/scala/org/dbpedia/extraction/destinations/Quad.scala
@@ -8,6 +8,7 @@ import org.dbpedia.extraction.util.Language
 /**
  * Represents a statement in the N-Quads format (see: http://sw.deri.org/2008/07/n-quads/)
  */
+//TODO write out equivalent properties
 class Quad(	val language : Language,
             val dataset : Dataset,
             val subject : String,
@@ -66,13 +67,23 @@ class Quad(	val language : Language,
             {
                 sb append '"'
                 escapeString(sb, value) //sb append value //#int 
-                sb append "\"^^<" append datatype.uri append "> "
+                sb append "\"^^<"
+                escapeString(sb, datatype.uri)
+                sb append "> "
             }
         }
         else
         {
             sb append '<'
-            sb append value //this must not be escaped, it is a URI/IRI   //TODO URIs should be escaped if they contain unicode chars! (for example homepages) -> Max: maybe we should have DBpediaURI and OtherURI as well as Literal objects?
+
+            //HACK
+            //TODO find a good solution for this
+            //maybe we should have DBpediaURI and OtherURI as well as Literal objects?
+            if(predicate == "http://xmlns.com/foaf/0.1/homepage")
+                escapeString(sb, value) // escape unicode in homepage URI
+            else
+                sb append value //this must not be escaped, it is a URI/IRI
+
             sb append "> "
         }
 
diff --git a/dump/config.properties.default b/dump/config.properties.default
index aada63f74e..b770cafaee 100644
--- a/dump/config.properties.default
+++ b/dump/config.properties.default
@@ -74,7 +74,7 @@ extractors.ru=org.dbpedia.extraction.mappings.MappingExtractor \
               org.dbpedia.extraction.mappings.DisambiguationExtractor \
               org.dbpedia.extraction.mappings.HomepageExtractor \
               org.dbpedia.extraction.mappings.ImageExtractor \
-              org.dbpedia.extraction.mappings.InterLanguageLinksExtractorExtractor
+              org.dbpedia.extraction.mappings.InterLanguageLinksExtractor
 
 extractors.sl=org.dbpedia.extraction.mappings.MappingExtractor