Author: mattmann
Date: Tue Feb 17 21:22:44 2015
New Revision: 1660501

URL: http://svn.apache.org/r1660501
Log:
- fix for OODT-817 TikaCmdLineExtractor needs to add Filename and FileLocation 
fields

Modified:
    oodt/trunk/CHANGES.txt
    
oodt/trunk/metadata/src/main/java/org/apache/oodt/cas/metadata/extractors/TikaCmdLineMetExtractor.java

Modified: oodt/trunk/CHANGES.txt
URL: 
http://svn.apache.org/viewvc/oodt/trunk/CHANGES.txt?rev=1660501&r1=1660500&r2=1660501&view=diff
==============================================================================
--- oodt/trunk/CHANGES.txt (original)
+++ oodt/trunk/CHANGES.txt Tue Feb 17 21:22:44 2015
@@ -2,6 +2,8 @@ Apache OODT Change Log
 ======================
 Release 0.9 - Current Development
 
+* OODT-817 TikaCmdLineExtractor needs to add Filename and FileLocation fields 
(mattmann)
+
 * OODT-813 CAS-PGE incorrectly passes varargs to DynamicConfigFile 
implementations (mattmann)
 
 * OODT-812 RADIX ingest script (mattmann)

Modified: 
oodt/trunk/metadata/src/main/java/org/apache/oodt/cas/metadata/extractors/TikaCmdLineMetExtractor.java
URL: 
http://svn.apache.org/viewvc/oodt/trunk/metadata/src/main/java/org/apache/oodt/cas/metadata/extractors/TikaCmdLineMetExtractor.java?rev=1660501&r1=1660500&r2=1660501&view=diff
==============================================================================
--- 
oodt/trunk/metadata/src/main/java/org/apache/oodt/cas/metadata/extractors/TikaCmdLineMetExtractor.java
 (original)
+++ 
oodt/trunk/metadata/src/main/java/org/apache/oodt/cas/metadata/extractors/TikaCmdLineMetExtractor.java
 Tue Feb 17 21:22:44 2015
@@ -24,6 +24,9 @@ import java.io.InputStream;
 import java.util.Enumeration;
 import java.util.logging.Logger;
 
+//Apache imports
+import org.apache.commons.lang.StringEscapeUtils;
+
 //OODT imports
 import org.apache.oodt.cas.metadata.Metadata;
 import org.apache.oodt.cas.metadata.exceptions.MetExtractionException;
@@ -86,7 +89,7 @@ public class TikaCmdLineMetExtractor ext
 
             // copy tika met into oodt met
             for (String key : tikaMet.names()) {
-                met.addMetadata(key, tikaMet.get(key));
+                met.addMetadata(key, 
StringEscapeUtils.escapeXml(tikaMet.get(key)));
                 LOG.fine("Added tika met key [" + key + "] with value ["
                         + met.getMetadata(key) + "]");
             }
@@ -99,10 +102,14 @@ public class TikaCmdLineMetExtractor ext
                 String configMetKey = (String) configMetKeys.nextElement();
                 String configMetKeyVal = (String) myConfig.get(configMetKey);
 
-                met.addMetadata(configMetKey, configMetKeyVal);
+                met.addMetadata(configMetKey, 
StringEscapeUtils.escapeXml(configMetKeyVal));
                 LOG.fine("Added config file met key [" + configMetKey + 
                         "] with value [" + met.getMetadata(configMetKey) + 
"]");
             }
+            
+            // add standard OODT met
+            met.addMetadata("Filename", 
StringEscapeUtils.escapeXml(file.getName()));
+            met.addMetadata("FileLocation", 
StringEscapeUtils.escapeXml(file.getAbsolutePath()));
 
             return met;
 


Reply via email to