fixed a bug with snippet-length

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@359 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/index.html

@@ -101,15 +101,8 @@ from 'late' peers to enrich this search result.
 
 #{results}#
 <!-- link begin -->
-<p><b>
-#[description]#
-</b><br>
-#(snippet)#
-::
-<i>
-#[text]#
-</i><br>
-#(/snippet)#
+<p><b>#[description]#</b><br>
+#(snippet)#::<i>#[text]#</i><br>#(/snippet)#
 <a href="#[url]#">#[urlname]#</a><br>
 #[date]#<br></p>
 <!-- link end -->

source/de/anomic/plasma/plasmaCondenser.java

@@ -444,9 +444,9 @@ public class plasmaCondenser {
     }
 
 
-    public static Enumeration wordTokenizer(String s) {
+    public static Enumeration wordTokenizer(String s, int minLength) {
 	try {
-	    return new sievedWordsEnum(new ByteArrayInputStream(s.getBytes()), 3);
+	    return new sievedWordsEnum(new ByteArrayInputStream(s.getBytes()), minLength);
 	} catch (Exception e) {
 	    return null;
 	}

source/de/anomic/plasma/plasmaSnippetCache.java

@@ -107,7 +107,8 @@ public class plasmaSnippetCache {
     }
     
     public boolean existsInCache(URL url, Set queryhashes) {
-        return retrieveFromCache(yacySearch.set2string(queryhashes), plasmaURL.urlHash(url)) != null;
+        String hashes = yacySearch.set2string(queryhashes);
+        return retrieveFromCache(hashes, plasmaURL.urlHash(url)) != null;
     }
     
     public result retrieve(URL url, Set queryhashes, boolean fetchOnline, int snippetMaxLength) {
@@ -237,7 +238,7 @@ public class plasmaSnippetCache {
         hs = hashSentence(result);
         j = queryhashes.iterator();
         Integer pos;
-        int p, minpos = maxLength, maxpos = -1;
+        int p, minpos = result.length(), maxpos = -1;
         while (j.hasNext()) {
             pos = (Integer) hs.get((String) j.next());
             if (pos != null) {
@@ -252,6 +253,24 @@ public class plasmaSnippetCache {
         if (maxpos > result.length()) maxpos = result.length();
         if (minpos < 0) minpos = 0;
         // we have a result, but is it short enough?
+        if (maxpos - minpos + 10 > maxLength) {
+            // the string is too long, even if we cut at both ends
+            // so cut here in the middle of the string
+            int lenb = result.length();
+            result = result.substring(0, (minpos + 20 > result.length()) ? result.length() : minpos + 20).trim() +
+                     " [..] " +
+                     result.substring((maxpos + 26 > result.length()) ? result.length() : maxpos + 26).trim();
+            maxpos = maxpos + lenb - result.length() + 6;
+        }
+        if (maxpos > maxLength) {
+            // the string is too long, even if we cut it at the end
+            // so cut it here at both ends at once
+            int newlen = maxpos - minpos + 10;
+            int around = (maxLength - newlen) / 2;
+            result = "[..] " + result.substring(minpos - around, maxpos + around).trim() + " [..]";
+            minpos = around;
+            maxpos = result.length() - around - 5;
+        }
         if (result.length() > maxLength) {
             // trim result, 1st step (cut at right side)
             result = result.substring(0, maxpos).trim() + " [..]";
@@ -277,7 +296,7 @@ public class plasmaSnippetCache {
     private HashMap hashSentence(String sentence) {
         // generates a word-wordPos mapping
         HashMap map = new HashMap();
-        Enumeration words = plasmaCondenser.wordTokenizer(sentence);
+        Enumeration words = plasmaCondenser.wordTokenizer(sentence, 0);
         int pos = 0;
         String word;
         while (words.hasMoreElements()) {

source/de/anomic/plasma/plasmaSwitchboard.java

@@ -373,12 +373,14 @@ public final class plasmaSwitchboard extends serverAbstractSwitch implements ser
         }
         
         // test routine for snippet fetch
-        // url = /www.heise.de/mobil/newsticker/meldung/mail/54980
         //Set query = new HashSet();
         //query.add(plasmaWordIndexEntry.word2hash("Weitergabe"));
         //query.add(plasmaWordIndexEntry.word2hash("Zahl"));
         //plasmaSnippetCache.result scr = snippetCache.retrieve(new URL("http://www.heise.de/mobil/newsticker/meldung/mail/54980"), query, true);
         //plasmaSnippetCache.result scr = snippetCache.retrieve(new URL("http://www.heise.de/security/news/foren/go.shtml?read=1&msg_id=7301419&forum_id=72721"), query, true);
+        //plasmaSnippetCache.result scr = snippetCache.retrieve(new URL("http://www.heise.de/kiosk/archiv/ct/2003/4/20"), query, true, 260);
+
+    
     }
     
     private static String ppRamString(int bytes) {

source/de/anomic/yacy/yacyCore.java

@@ -676,37 +676,4 @@ public class yacyCore {
         }
     }
 
-    private class vprobe implements Runnable {
-	public vprobe() {}
-	public final void run() {
-	    // read the probe URL
-	    String probeURL=switchboard.getConfig("onetimeAction", null);
-	    if ((probeURL == null) || (probeURL.length() == 0)) return; // not wanted
-	    // read version and date
-            String proxyHost = switchboard.getConfig("remoteProxyHost", "");
-            int proxyPort = Integer.parseInt(switchboard.getConfig("remoteProxyPort", "0"));
-            if (!(switchboard.getConfig("remoteProxyUse", "false").equals("true"))) {
-                proxyHost = null; proxyPort = 0;
-            }
-            String version = switchboard.getConfig("version", "");
-	    String date = switchboard.getConfig("vdate", "");
-	    probeURL = probeURL + "?version=" + version + "&date=" + date;
-	    // open new connection
-	    try {
-		latestVersion = new String(httpc.singleGET(new URL(probeURL), 10000, null, null, proxyHost, proxyPort)).trim();
-		float latest = Float.parseFloat(latestVersion);
-		float thisver = Float.parseFloat(version);
-		if (thisver > latest)  System.out.println("THIS SOFTWARE VERSION IS A PRE-RELEASE");
-		if (thisver < latest)  {
-		    log.logSystem("****************************************************************");
-		    log.logSystem("* THIS SOFTWARE VERSION IS OUTDATED.");
-		    log.logSystem("* PLEASE GO TO ANOMIC.DE AND DOWNLOAD THE LATEST VERSION " + latestVersion);
-		    log.logSystem("****************************************************************");
-		}
-	    } catch (Exception e) {
-		// we do nothing is this case
-	    }
-	}
-    }
-
 }