added option to re-index exported xml snapshot dumps to

HTCACHE/snapshots by just placing them in the SURROGATES/in path

source/net/yacy/cora/federate/solr/SchemaConfiguration.java

@@ -31,6 +31,8 @@ import java.util.Set;
 import org.apache.solr.common.SolrDocument;
 import org.apache.solr.common.SolrInputDocument;
 import org.apache.solr.common.SolrInputField;
+import org.apache.solr.common.params.MultiMapSolrParams;
+import org.apache.solr.common.util.NamedList;
 
 import net.yacy.cora.storage.Configuration;
 import net.yacy.cora.util.ConcurrentLog;
@@ -87,6 +89,16 @@ public class SchemaConfiguration extends Configuration implements Serializable {
         return sid;
     }
     
+    public SolrInputDocument toSolrInputDocument(final MultiMapSolrParams params) {
+        SolrInputDocument sid = new SolrInputDocument();
+        for (String name: params.getMap().keySet()) {
+            if (this.contains(name)) { // check each field if enabled in local Solr schema
+                sid.addField(name, params.getParams(name));
+            }
+        }
+        return sid;
+    }
+    
     public SolrDocument toSolrDocument(final SolrInputDocument doc, Set<String> omitFields) {
         SolrDocument sd = new SolrDocument();
         for (SolrInputField field: doc) {

source/net/yacy/document/content/DCEntry.java

@@ -127,6 +127,7 @@ public class DCEntry extends MultiMapSolrParams {
         // <dc:identifier>http://hdl.handle.net/2104/8302</dc:identifier>
         // <dc:identifier>10.1051/0004-6361/201117940</dc:identifier>
         String u = this.get("url");
+        if (u == null) u = this.get("sku");
         
         if (u == null) {
             final String[] urls = this.getParams("dc:identifier");
@@ -342,7 +343,7 @@ public class DCEntry extends MultiMapSolrParams {
             null,
             getDescriptions(),
             getLon(), getLat(),
-            "",
+            get("text_t", ""),
             null,
             null,
             null,

source/net/yacy/document/content/SurrogateReader.java

@@ -140,6 +140,7 @@ public class SurrogateReader extends DefaultHandler implements Runnable {
             this.surrogate = new DCEntry();
         } else if ("element".equals(tag) || "str".equals(tag) || "int".equals(tag) || "bool".equals(tag) || "long".equals(tag)) {
             this.elementName = atts.getValue("name");
+            this.parsingValue = true;
         } else if ("value".equals(tag)) {
             this.buffer.setLength(0);
             this.parsingValue = true;
@@ -174,6 +175,7 @@ public class SurrogateReader extends DefaultHandler implements Runnable {
             if (this.elementName != null) {
                 this.surrogate.getMap().put(this.elementName, new String[]{value});
             }
+            this.buffer.setLength(0);
             this.parsingValue = false;
         } else if ("value".equals(tag)) {
             //System.out.println("BUFFER-SIZE=" + buffer.length());

source/net/yacy/search/Switchboard.java

@@ -1942,6 +1942,7 @@ public final class Switchboard extends serverSwitch {
                             ConcurrentLog.logException(e);
                         }
                     }
+                    log.info("processed surrogate " + infile);
                 }
             }
             if (is != null) try {is.close();} catch (IOException e) {}
@@ -1969,23 +1970,27 @@ public final class Switchboard extends serverSwitch {
                 continue;
             }
 
-            // create a queue entry
-            final Document document = surrogate.document();
-            final Request request =
-                new Request(
-                    ASCII.getBytes(this.peers.mySeed().hash),
-                    surrogate.getIdentifier(true),
-                    null,
-                    "",
-                    surrogate.getDate(),
-                    this.crawler.defaultSurrogateProfile.handle(),
-                    0,
-                    this.crawler.defaultSurrogateProfile.timezoneOffset());
-            response = new Response(request, null, null, this.crawler.defaultSurrogateProfile, false, null);
-            final IndexingQueueEntry queueEntry =
-                new IndexingQueueEntry(response, new Document[] {document}, null);
-
-            this.indexingCondensementProcessor.enQueue(queueEntry);
+            if (surrogate.get("text_t") == null) {
+                // create a queue entry
+                final Document document = surrogate.document();
+                final Request request =
+                    new Request(
+                        ASCII.getBytes(this.peers.mySeed().hash),
+                        surrogate.getIdentifier(true),
+                        null,
+                        "",
+                        surrogate.getDate(),
+                        this.crawler.defaultSurrogateProfile.handle(),
+                        0,
+                        this.crawler.defaultSurrogateProfile.timezoneOffset());
+                response = new Response(request, null, null, this.crawler.defaultSurrogateProfile, false, null);
+                final IndexingQueueEntry queueEntry =
+                    new IndexingQueueEntry(response, new Document[] {document}, null);
+    
+                this.indexingCondensementProcessor.enQueue(queueEntry);
+            } else {
+                this.index.putDocument(this.index.fulltext().getDefaultConfiguration().toSolrInputDocument(surrogate));
+            }
             if (shallTerminate()) break;
         }
     }