Author: kwright
Date: Fri May 30 08:05:37 2014
New Revision: 1598484
URL: http://svn.apache.org/r1598484
Log:
Make pipeline check available in IOutputAddActivity
Modified:
manifoldcf/branches/CONNECTORS-946/framework/agents/src/main/java/org/apache/manifoldcf/agents/interfaces/IOutputAddActivity.java
manifoldcf/branches/CONNECTORS-946/framework/pull-agent/src/main/java/org/apache/manifoldcf/crawler/system/WorkerThread.java
Modified:
manifoldcf/branches/CONNECTORS-946/framework/agents/src/main/java/org/apache/manifoldcf/agents/interfaces/IOutputAddActivity.java
URL:
http://svn.apache.org/viewvc/manifoldcf/branches/CONNECTORS-946/framework/agents/src/main/java/org/apache/manifoldcf/agents/interfaces/IOutputAddActivity.java?rev=1598484&r1=1598483&r2=1598484&view=diff
==============================================================================
---
manifoldcf/branches/CONNECTORS-946/framework/agents/src/main/java/org/apache/manifoldcf/agents/interfaces/IOutputAddActivity.java
(original)
+++
manifoldcf/branches/CONNECTORS-946/framework/agents/src/main/java/org/apache/manifoldcf/agents/interfaces/IOutputAddActivity.java
Fri May 30 08:05:37 2014
@@ -24,7 +24,7 @@ import org.apache.manifoldcf.agents.inte
/** This interface abstracts from the activities that an output connector can
do
when adding or replacing documents.
*/
-public interface IOutputAddActivity extends IOutputHistoryActivity
+public interface IOutputAddActivity extends
IOutputHistoryActivity,IOutputCheckActivity
{
public static final String _rcsid = "@(#)$Id: IOutputAddActivity.java 988245
2010-08-23 18:39:35Z kwright $";
Modified:
manifoldcf/branches/CONNECTORS-946/framework/pull-agent/src/main/java/org/apache/manifoldcf/crawler/system/WorkerThread.java
URL:
http://svn.apache.org/viewvc/manifoldcf/branches/CONNECTORS-946/framework/pull-agent/src/main/java/org/apache/manifoldcf/crawler/system/WorkerThread.java?rev=1598484&r1=1598483&r2=1598484&view=diff
==============================================================================
---
manifoldcf/branches/CONNECTORS-946/framework/pull-agent/src/main/java/org/apache/manifoldcf/crawler/system/WorkerThread.java
(original)
+++
manifoldcf/branches/CONNECTORS-946/framework/pull-agent/src/main/java/org/apache/manifoldcf/crawler/system/WorkerThread.java
Fri May 30 08:05:37 2014
@@ -2496,6 +2496,51 @@ public class WorkerThread extends Thread
return IPipelineConnector.DOCUMENTSTATUS_REJECTED;
}
+ /** Detect if a mime type is acceptable downstream or not. This method is
used to determine whether it makes sense to fetch a document
+ * in the first place.
+ *@param mimeType is the mime type of the document.
+ *@return true if the mime type can be accepted by the downstream
connection.
+ */
+ public boolean checkMimeTypeIndexable(String mimeType)
+ throws ManifoldCFException, ServiceInterruption
+ {
+ return false;
+ }
+
+ /** Pre-determine whether a document (passed here as a File object) is
acceptable downstream. This method is
+ * used to determine whether a document needs to be actually transferred.
This hook is provided mainly to support
+ * search engines that only handle a small set of accepted file types.
+ *@param localFile is the local file to check.
+ *@return true if the file is acceptable by the downstream connection.
+ */
+ public boolean checkDocumentIndexable(File localFile)
+ throws ManifoldCFException, ServiceInterruption
+ {
+ return false;
+ }
+
+ /** Pre-determine whether a document's length is acceptable downstream.
This method is used
+ * to determine whether to fetch a document in the first place.
+ *@param length is the length of the document.
+ *@return true if the file is acceptable by the downstream connection.
+ */
+ public boolean checkLengthIndexable(long length)
+ throws ManifoldCFException, ServiceInterruption
+ {
+ return false;
+ }
+
+ /** Pre-determine whether a document's URL is acceptable downstream. This
method is used
+ * to help filter out documents that cannot be indexed in advance.
+ *@param url is the URL of the document.
+ *@return true if the file is acceptable by the downstream connection.
+ */
+ public boolean checkURLIndexable(String url)
+ throws ManifoldCFException, ServiceInterruption
+ {
+ return false;
+ }
+
}
}
