git-svn-id: https://svn.berlios.de/svnroot/repos/yacy/trunk@5758 6c8d7289-2bf4-0310-a012-ef5d649a1542

htroot/yacysearch.java

@@ -103,13 +103,14 @@ public class yacysearch {
             prop.put("input_count", "10");
             prop.put("input_offset", "0");
             prop.put("input_resource", "global");
-            prop.put("input_urlmaskfilter", ".*");
-            prop.put("input_prefermaskfilter", "");
+            prop.put("input_urlmaskfilter", (post == null) ? ".*" : post.get("urlmaskfilter", ".*"));
+            prop.put("input_prefermaskfilter", (post == null) ? "" : post.get("prefermaskfilter", ""));
+            prop.put("input_tenant", (post == null) ? "" : post.get("tenant", ""));
             prop.put("input_indexof", "off");
             prop.put("input_constraint", "");
             prop.put("input_cat", "href");
             prop.put("input_depth", "0");
-            prop.put("input_verify", "true");
+            prop.put("input_verify", (post == null) ? "true" : post.get("verify", "true"));
             prop.put("input_contentdom", "text");
             prop.put("input_contentdomCheckText", "1");
             prop.put("input_contentdomCheckAudio", "0");
@@ -285,6 +286,10 @@ public class yacysearch {
                     }
                 }
             }
+            if (post.containsKey("tenant")) {
+                final String tenant = post.get("tenant");
+                if (urlmask == null) urlmask = ".*" + tenant + ".*"; else urlmask = ".*" + tenant + urlmask;
+            }
             if (urlmask == null || urlmask.length() == 0) urlmask = originalUrlMask; //if no urlmask was given
            
             // read the language from the language-restrict option 'lr'

source/de/anomic/crawler/Balancer.java

@@ -564,7 +564,9 @@ public class Balancer {
         final int s = urlFileIndex.size();
         Row.Entry rowEntry = urlFileIndex.remove(result.getBytes());
         if (rowEntry == null) {
-            throw new IOException("get() found a valid urlhash, but failed to fetch the corresponding url entry - total size = " + size() + ", fileStack.size() = " + urlFileStack.size() + ", ramStack.size() = " + urlRAMStack.size() + ", domainStacks.size() = " + domainStacks.size());
+            String error = "get() found a valid urlhash, but failed to fetch the corresponding url entry - total size = " + size() + ", fileStack.size() = " + urlFileStack.size() + ", ramStack.size() = " + urlRAMStack.size() + ", domainStacks.size() = " + domainStacks.size();
+            //this.clear();
+            throw new IOException(error + " - cleared the balancer");
         }
         assert urlFileIndex.size() + 1 == s : "urlFileIndex.size() = " + urlFileIndex.size() + ", s = " + s + ", result = " + result;
         final CrawlEntry crawlEntry = new CrawlEntry(rowEntry);

source/de/anomic/kelondro/blob/HeapWriter.java

@@ -103,14 +103,17 @@ public final class HeapWriter  {
     }
     
     protected static File fingerprintIndexFile(File f) {
+        assert f != null;
         return new File(f.getParentFile(), f.getName() + "." + fingerprintFileHash(f) + ".idx");
     }
     
     protected static File fingerprintGapFile(File f) {
+        assert f != null;
         return new File(f.getParentFile(), f.getName() + "." + fingerprintFileHash(f) + ".gap");
     }
     
     protected static String fingerprintFileHash(File f) {
+        assert f != null;
         return Digest.fastFingerprintB64(f, false).substring(0, 12);
     }
     

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

@@ -256,6 +256,7 @@ public class Dispatcher {
      * then no additional IO is necessary.
      */
     private void enqueueContainersToCloud(final ArrayList<ReferenceContainer>[] containers) {
+        if (transmissionCloud == null) return;
         ReferenceContainer lastContainer;
         String primaryTarget;
         Transmission.Chunk entry;