*) Faster appearance of ConfigBasic by bypassing UPNP-scan in case of existing external connects

*) Marked two deprecated source-points
*) Added possibility to dump words from indexing to file. Should not affect performance in the current form.

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@3592 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/ConfigBasic.java

@@ -103,7 +103,10 @@ public class ConfigBasic {
         
         // scan for Upnp routers
         long begin = System.currentTimeMillis();
-        boolean upnpRouterFound = config.findUPnPRouter(3000);
+        boolean upnpRouterFound = false;
+        if (yacyCore.seedDB.mySeed.isVirgin() || yacyCore.seedDB.mySeed.isJunior()) {
+        	upnpRouterFound = config.findUPnPRouter(3000);
+        }
         long end = System.currentTimeMillis();
         
         // if the upnp router scan has taken less than 3 sec, we need to wait

htroot/WatchCrawler_p.java

@@ -266,6 +266,7 @@ public class WatchCrawler_p {
                                 HashMap hyperlinks = (HashMap) scraper.getAnchors();
                                 
                                 // creating a crawler profile
+                                																			/* file.toURL() marked as deprecated per Sun JRE 6 */
                                 plasmaCrawlProfile.entry profile = switchboard.profiles.newEntry(fileName, file.toURL().toString(), newcrawlingfilter, newcrawlingfilter, newcrawlingdepth, newcrawlingdepth, crawlingIfOlder, crawlingDomFilterDepth, crawlingDomMaxPages, crawlingQ, indexText, indexMedia, storeHTCache, true, crawlOrder, xsstopw, xdstopw, xpstopw);
                                 
                                 // loop through the contained links

source/de/anomic/net/ftpc.java

@@ -639,6 +639,7 @@ public class ftpc {
   }
 
   public boolean JAR() {
+	  /*Sun proprietary API may be removed in a future Java release*/
     sun.tools.jar.Main.main(shift(cmd));
     return true;
   }

source/de/anomic/plasma/plasmaCondenser.java

@@ -52,6 +52,7 @@ import java.io.FileNotFoundException;
 import java.io.IOException;
 import java.io.InputStream;
 import java.io.InputStreamReader;
+import java.io.RandomAccessFile;
 import java.io.Reader;
 import java.io.UnsupportedEncodingException;
 import java.util.Enumeration;
@@ -390,6 +391,21 @@ public final class plasmaCondenser {
             word = (new String((StringBuffer) wordenum.nextElement())).toLowerCase(); // TODO: does toLowerCase work for non ISO-8859-1 chars?
             //System.out.println("PARSED-WORD " + word);
             
+            //This is useful for testing what YaCy "sees" of a website.
+            if (false) { 	
+            	File f = new File("dump.txt");
+            	RandomAccessFile fa = null;
+            	try {
+					fa = new RandomAccessFile(f, "rw");
+					fa.seek(fa.length());
+					fa.writeBytes(word);
+					fa.write(160);
+					fa.close();
+				} catch (IOException e) {
+					e.printStackTrace();
+				}
+            }
+            
             // distinguish punctuation and words
             wordlen = word.length();
             if ((wordlen == 1) && (htmlFilterContentScraper.punctuation(word.charAt(0)))) {
@@ -450,7 +466,7 @@ public final class plasmaCondenser {
                 wordInSentenceCounter++;
             }
         }
-        // finnish last sentence
+        // finish last sentence
         if (sentence.length() > 0) {
             allsentencecounter++;
             sentence.insert(0, "."); // append at beginning