- fix for snippets that are too short

- added keyword to snippet fetch to suppres removal of not-found snippet words (for debugging)


git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3009 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/js/yacysearch.js

@@ -13,7 +13,7 @@ function AllSnippets() {
 
 function requestSnippet(url, query){
 	var request=createRequestObject();
-	request.open('get', '/xml/snippet.xml?url=' + escape(url) + '&search=' + escape(query),true);
+	request.open('get', '/xml/snippet.xml?url=' + escape(url) + '&search=' + escape(query) + '&remove=true',true);
 	request.onreadystatechange = function () {handleState(request)};
 	request.send(null);
 }

htroot/xml/snippet.java

@@ -25,6 +25,9 @@ public class snippet {
         String urlString = post.get("url", "");
         URL url = new URL(urlString);
         
+        // if 'remove' is set to true, then RWI references to URLs that do not have the snippet are removed
+        boolean remove = post.get("remove", "false").equals("true");
+        
         String querystring = post.get("search", "").trim();
         if ((querystring.length() > 2) && (querystring.charAt(0) == '"') && (querystring.charAt(querystring.length() - 1) == '"')) {
             querystring = querystring.substring(1, querystring.length() - 1).trim();
@@ -47,7 +50,7 @@ public class snippet {
             prop.put("text", (snippet.exists()) ? "<![CDATA["+snippet.getLineMarked(queryHashes)+"]]>" : "unknown"); 
         } else {
             String error = snippet.getError();
-            if (error.equals("no matching snippet found")) {
+            if ((remove) && (error.equals("no matching snippet found"))) {
                 switchboard.removeReferences(plasmaURL.urlHash(url), query);
             }
             prop.put("text", error);

source/de/anomic/plasma/plasmaSnippetCache.java

@@ -269,7 +269,7 @@ public class plasmaSnippetCache {
          * COMPUTE SNIPPET
          * =========================================================================== */        
         // we have found a parseable non-empty file: use the lines
-        line = computeSnippet(sentences, queryhashes, 8 + 6 * queryhashes.size(), snippetMaxLength);
+        line = computeSnippet(sentences, queryhashes, 3 * queryhashes.size(), snippetMaxLength);
         //System.out.println("loaded snippet for URL " + url + ": " + line);
         if (line == null) return new Snippet(null, ERROR_NO_MATCH, "no matching snippet found");
         if (line.length() > snippetMaxLength) line = line.substring(0, snippetMaxLength);
@@ -378,7 +378,7 @@ public class plasmaSnippetCache {
             String sentence;
             while (sentences.hasMoreElements()) {
                 sentence = (String) sentences.nextElement();
-                //System.out.println("Sentence " + i + ": " + sentences[i]);
+                //System.out.println("Snippet-Sentence :" + sentence); // DEBUG
                 if (sentence.length() > minLength) {
                     hs = hashSentence(sentence);
                     j = queryhashes.iterator();