From cee7a05ff25d271a81b607826d50c16c769fbd99 Mon Sep 17 00:00:00 2001
From: orbiter <orbiter@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Thu, 15 Oct 2009 10:47:29 +0000
Subject: [PATCH] - de-serialized the pdf parser - added fail callback for file
 indexer

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@6415 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 .../de/anomic/document/parser/pdfParser.java  |  2 --
 source/de/anomic/search/DocumentIndex.java    | 20 +++++++++++++++----
 2 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/source/de/anomic/document/parser/pdfParser.java b/source/de/anomic/document/parser/pdfParser.java
index a391ab44b..bb446a4da 100644
--- a/source/de/anomic/document/parser/pdfParser.java
+++ b/source/de/anomic/document/parser/pdfParser.java
@@ -109,7 +109,6 @@ public class pdfParser extends AbstractParser implements Idiom {
             checkInterruption();
             
             // creating a text stripper
-            synchronized (SUPPORTED_MIME_TYPES) {
             final PDFTextStripper stripper = new PDFTextStripper();
             theDocument = parser.getPDDocument();
             
@@ -182,7 +181,6 @@ public class pdfParser extends AbstractParser implements Idiom {
             }
             
             return theDoc;
-            }
         }
         catch (final Exception e) {       
             if (e instanceof InterruptedException) throw (InterruptedException) e;
diff --git a/source/de/anomic/search/DocumentIndex.java b/source/de/anomic/search/DocumentIndex.java
index c2ceaa52c..727c406e4 100644
--- a/source/de/anomic/search/DocumentIndex.java
+++ b/source/de/anomic/search/DocumentIndex.java
@@ -77,12 +77,20 @@ public class DocumentIndex extends Segment {
     class Worker extends Thread {
         public void run() {
             File f;
+            URIMetadataRow resultRow;
             try {
                 while ((f = queue.take()) != poison) try {
-                    add(f);
-                    if (callback != null) callback.commitIndex(f);
+                    resultRow = add(f);
+                    if (callback != null) {
+                        if (resultRow == null) {
+                            callback.fail(f, "result is null");
+                        } else {
+                            callback.commit(f, resultRow);
+                        }
+                    }
                 } catch (IOException e) {
                     if (e.getMessage().indexOf("cannot parse") < 0) e.printStackTrace();
+                    callback.fail(f, e.getMessage());
                 }
             } catch (InterruptedException e) {}
         }
@@ -229,7 +237,8 @@ public class DocumentIndex extends Segment {
     }
     
     public interface CallbackListener {
-        public void commitIndex(File f);
+        public void commit(File f, URIMetadataRow resultRow);
+        public void fail(File f, String failReason);
     }
     
     public static void main(String[] args) {
@@ -245,9 +254,12 @@ public class DocumentIndex extends Segment {
         File segmentPath = new File(args[0]);
         System.out.println("using index files at " + segmentPath.getAbsolutePath());
         CallbackListener callback = new CallbackListener() {
-            public void commitIndex(File f) {
+            public void commit(File f, URIMetadataRow resultRow) {
                 System.out.println("indexed: " + f.toString());
             }
+            public void fail(File f, String failReason) {
+                System.out.println("not indexed " + f.toString() + ": " + failReason);
+            }
         };
         try {
             if (args[1].equals("add")) {