From 89169d54fd1f7c7611ba1d347fbb49ed192a3483 Mon Sep 17 00:00:00 2001
From: orbiter <orbiter@6c8d7289-2bf4-0310-a012-ef5d649a1542>
Date: Sat, 2 Feb 2008 00:16:00 +0000
Subject: [PATCH] fixed search result preparation

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@4427 6c8d7289-2bf4-0310-a012-ef5d649a1542
---
 htroot/yacysearch.html                        |  2 +-
 .../de/anomic/plasma/plasmaSearchEvent.java   | 14 ++++++++++-
 .../de/anomic/plasma/plasmaSnippetCache.java  | 25 +++++++++++++++----
 3 files changed, 34 insertions(+), 7 deletions(-)

diff --git a/htroot/yacysearch.html b/htroot/yacysearch.html
index 4f0ddfd22..7b6ab5fa9 100644
--- a/htroot/yacysearch.html
+++ b/htroot/yacysearch.html
@@ -117,7 +117,7 @@ var progressbar = new Progressbar(#[results]#, document.getElementById("results"
 #{/results}#
 #(resultTable)#::</table>#(/resultTable)#
 <!-- linklist end -->
-
+<!--<div class="sidebar"><h3 class="head">Drill-Down</h3></div>-->
 <!-- attach the bottomline -->
 <!--#include virtual="yacysearchitem.html?bottomline=true&eventID=#[eventID]#" -->
 </body>
diff --git a/source/de/anomic/plasma/plasmaSearchEvent.java b/source/de/anomic/plasma/plasmaSearchEvent.java
index 5f783d174..43931b601 100644
--- a/source/de/anomic/plasma/plasmaSearchEvent.java
+++ b/source/de/anomic/plasma/plasmaSearchEvent.java
@@ -58,7 +58,7 @@ public final class plasmaSearchEvent {
     public static final String URLFETCH = "urlfetch";
     public static final String NORMALIZING = "normalizing";
     
-    public static int workerThreadCount = 8;
+    public static int workerThreadCount = 10;
     public static String lastEventID = "";
     private static HashMap<String, plasmaSearchEvent> lastEvents = new HashMap<String, plasmaSearchEvent>(); // a cache for objects from this class: re-use old search requests
     public static final long eventLifetime = 600000; // the time an event will stay in the cache, 10 Minutes
@@ -391,6 +391,17 @@ public final class plasmaSearchEvent {
         }
         return false;
     }
+
+    private int countFinishedWorkerThreads() {
+        if (this.workerThreads == null) return workerThreadCount;
+        int c = 0;
+        for (int i = 0; i < workerThreadCount; i++) {
+           if ((this.workerThreads[i] == null) ||
+               !(this.workerThreads[i].isAlive()) ||
+               (this.workerThreads[i].busytime() >= 3000)) c++;
+        }
+        return c;
+    }
     
     private boolean anyRemoteSearchAlive() {
         // check primary search threads
@@ -576,6 +587,7 @@ public final class plasmaSearchEvent {
     public ResultEntry oneResult(int item) {
         // first sleep a while to give accumulation threads a chance to work
         while (((localSearchThread != null) && (localSearchThread.isAlive())) ||
+                ((countFinishedWorkerThreads() <= item) && (item < workerThreadCount)) ||
                ((this.primarySearchThreads != null) && (this.primarySearchThreads.length > item) && (anyWorkerAlive()) && 
                 ((this.resultList.size() <= item) || (countFinishedRemoteSearch() <= item)))) {
             try {Thread.sleep(100);} catch (InterruptedException e) {}
diff --git a/source/de/anomic/plasma/plasmaSnippetCache.java b/source/de/anomic/plasma/plasmaSnippetCache.java
index 9c1587dfd..2494a17be 100644
--- a/source/de/anomic/plasma/plasmaSnippetCache.java
+++ b/source/de/anomic/plasma/plasmaSnippetCache.java
@@ -76,6 +76,7 @@ public class plasmaSnippetCache {
     public static final int SOURCE_CACHE = 0;
     public static final int SOURCE_FILE = 1;
     public static final int SOURCE_WEB = 2;
+    public static final int SOURCE_METADATA = 3;
     
     public static final int ERROR_NO_HASH_GIVEN = 11;
     public static final int ERROR_SOURCE_LOADING = 12;
@@ -280,11 +281,16 @@ public class plasmaSnippetCache {
                 if ((resContentLength > maxDocLen) && (!fetchOnline)) {
                     // content may be too large to be parsed here. To be fast, we omit calculation of snippet here
                     return new TextSnippet(url, null, ERROR_SOURCE_LOADING, queryhashes, "resource available, but too large: " + resContentLength + " bytes");
-                }/*
-            } else if (url.) {
+                }
+            } else if (containsAllHashes(comp.dc_title(), queryhashes)) {
                 // try to create the snippet from information given in the url itself
-                */
-                
+                return new TextSnippet(url, (comp.dc_subject().length() > 0) ? comp.dc_creator() : comp.dc_subject(), SOURCE_METADATA, null, null, faviconCache.get(url.hash()));
+            } else if (containsAllHashes(comp.dc_creator(), queryhashes)) {
+                // try to create the snippet from information given in the creator metadata
+                return new TextSnippet(url, comp.dc_creator(), SOURCE_METADATA, null, null, faviconCache.get(url.hash()));
+            } else if (containsAllHashes(comp.dc_subject(), queryhashes)) {
+                // try to create the snippet from information given in the subject metadata
+                return new TextSnippet(url, (comp.dc_creator().length() > 0) ? comp.dc_creator() : comp.dc_subject(), SOURCE_METADATA, null, null, faviconCache.get(url.hash()));
             } else if (fetchOnline) {
                 // if not found try to download it
                 
@@ -741,7 +747,16 @@ public class plasmaSnippetCache {
         }
         return map;
     }
-     
+    
+    private static boolean containsAllHashes(String sentence, Set<String> queryhashes) {
+        HashMap<String, Integer> m = hashSentence(sentence);
+        Iterator<String> i = queryhashes.iterator();
+        while (i.hasNext()) {
+            if (!(m.containsKey(i.next()))) return false;
+        }
+        return true;
+    }
+    
     public static plasmaParserDocument parseDocument(yacyURL url, long contentLength, InputStream resourceStream) throws ParserException {
         return parseDocument(url, contentLength, resourceStream, null);
     }