fixes to web structure reference collection and url construction

git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5775 6c8d7289-2bf4-0310-a012-ef5d649a1542

source/de/anomic/plasma/plasmaWebStructure.java

@@ -259,34 +259,42 @@ public class plasmaWebStructure {
     }
     
     public structureEntry incomingReferences(final String domhash) {
+        String host = resolveDomHash2DomString(domhash);
+        if (host == null) return null;
         // collect the references
-        final Iterator<plasmaWebStructure.structureEntry> i = structureEntryIterator(false);
         plasmaWebStructure.structureEntry sentry;
         HashMap<String, Integer> domhashes = new HashMap<String, Integer>();
+        Iterator<plasmaWebStructure.structureEntry> i = structureEntryIterator(false);
         while (i.hasNext()) {
             sentry = i.next();
-            if (sentry.references.containsKey(domhash)) {
-                domhashes.put(sentry.domhash, sentry.references.get(domhash));
-            }
+            if (sentry.references.containsKey(domhash)) domhashes.put(sentry.domhash, sentry.references.get(domhash));
+        }
+        i = structureEntryIterator(true);
+        while (i.hasNext()) {
+            sentry = i.next();
+            if (sentry.references.containsKey(domhash)) domhashes.put(sentry.domhash, sentry.references.get(domhash));
         }
         // construct a new structureEntry Object
         return new structureEntry(
                 domhash,
-                resolveDomHash2DomString(domhash), 
+                host,
                 DateFormatter.formatShortDay(new Date()),
                 domhashes);
     }
     
     public HashMap<String, Integer> incomingDomains(final String domhash) {
         // collect the references
-        final Iterator<plasmaWebStructure.structureEntry> i = structureEntryIterator(false);
         plasmaWebStructure.structureEntry sentry;
         HashMap<String, Integer> domains = new HashMap<String, Integer>();
+        Iterator<plasmaWebStructure.structureEntry> i = structureEntryIterator(false);
         while (i.hasNext()) {
             sentry = i.next();
-            if (sentry.references.containsKey(domhash)) {
-                domains.put(sentry.domain, sentry.references.get(domhash));
-            }
+            if (sentry.references.containsKey(domhash)) domains.put(sentry.domain, sentry.references.get(domhash));
+        }
+        i = structureEntryIterator(true);
+        while (i.hasNext()) {
+            sentry = i.next();
+            if (sentry.references.containsKey(domhash)) domains.put(sentry.domain, sentry.references.get(domhash));
         }
         return domains;
     }

source/de/anomic/yacy/dht/Transmission.java

@@ -227,12 +227,12 @@ public class Transmission  {
         }
         
         public boolean isFinished() {
-        	System.out.println("canFinish: hit = " + this.hit + ", redundancy = " + seeds.redundancy() + ", targets.size() = " + targets.size());
+        	//System.out.println("canFinish: hit = " + this.hit + ", redundancy = " + seeds.redundancy() + ", targets.size() = " + targets.size());
             return this.hit >= seeds.redundancy();
         }
         
         public boolean canFinish() {
-        	System.out.println("canFinish: hit = " + this.hit + ", redundancy = " + seeds.redundancy() + ", targets.size() = " + targets.size());
+        	//System.out.println("canFinish: hit = " + this.hit + ", redundancy = " + seeds.redundancy() + ", targets.size() = " + targets.size());
             return this.targets.size() >= seeds.redundancy() - this.hit;
         }
 

source/de/anomic/yacy/yacyURL.java

@@ -79,12 +79,8 @@ public class yacyURL implements Serializable {
         url = url.trim();
         int p = url.indexOf(':');
         if (p < 0) {
-            if (url.startsWith("www.")) {
-                url = "http://" + url;
-                p = 4;
-            } else {
-                throw new MalformedURLException("protocol is not given in '" + url + "'");
-            }
+            url = "http://" + url;
+            p = 4;
         }
         this.protocol = url.substring(0, p).toLowerCase().trim();
         if (url.length() < p + 4) throw new MalformedURLException("URL not parseable: '" + url + "'");