Author: kwright
Date: Thu Sep 19 01:27:18 2013
New Revision: 1524609
URL: http://svn.apache.org/r1524609
Log:
Code up first rev of attachment finder.
Modified:
manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SPSProxyHelper.java
manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SharePointRepository.java
Modified:
manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SPSProxyHelper.java
URL:
http://svn.apache.org/viewvc/manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SPSProxyHelper.java?rev=1524609&r1=1524608&r2=1524609&view=diff
==============================================================================
---
manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SPSProxyHelper.java
(original)
+++
manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SPSProxyHelper.java
Thu Sep 19 01:27:18 2013
@@ -1531,6 +1531,127 @@ public class SPSProxyHelper {
}
}
+ /** Gets a list of attachment URLs, given a site, list name, and list item
ID. The returned URLs will be relative to the site.
+ */
+ public List<String> getAttachmentURLs( String site, String listName, String
itemID, boolean supportsAttachments )
+ throws ManifoldCFException, ServiceInterruption
+ {
+ long currentTime;
+ try
+ {
+ ArrayList<String> result = new ArrayList<String>();
+ if (supportsAttachments)
+ {
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("SharePoint: In getAttachmentURLs;
site='"+site+"', listName='"+listName+"', itemID='"+itemID+"'");
+
+ // The docLibrary must be a GUID, because we don't have title.
+
+ if ( site.compareTo( "/") == 0 )
+ site = "";
+ ListsWS listService = new ListsWS( baseUrl + site, userName, password,
configuration, httpClient );
+ ListsSoap listCall = listService.getListsSoapHandler();
+
+ GetAttachmentCollectionResponseGetAttachmentCollectionResult
listResponse =
+ listCall.getAttachmentCollection( listName, itemID );
+ org.apache.axis.message.MessageElement[] List = listResponse.get_any();
+
+ System.out.println(List[0].toString());
+
+ XMLDoc doc = new XMLDoc( List[0].toString() );
+ ArrayList nodeList = new ArrayList();
+
+ doc.processPath(nodeList, "*", null);
+ if (nodeList.size() != 1)
+ {
+ throw new ManifoldCFException("Bad xml - missing outer node - there
are "+Integer.toString(nodeList.size())+" nodes");
+ }
+
+ Object parent = nodeList.get(0);
+ if (!doc.getNodeName(parent).equals("ns1:List"))
+ throw new ManifoldCFException("Bad xml - outer node is '" +
doc.getNodeName(parent) + "' not 'ns1:List'");
+
+ nodeList.clear();
+ doc.processPath(nodeList, "*", parent); // <ns1:Attachments>
+
+ Object attachments = nodeList.get(0);
+ if ( !doc.getNodeName(attachments).equals("ns1:Attachments") )
+ throw new ManifoldCFException( "Bad xml - child node 0 '" +
doc.getNodeName(attachments) + "' is not 'ns1:Attachments'");
+
+ nodeList.clear();
+ doc.processPath(nodeList, "*", attachments);
+
+ int i = 0;
+ while (i < nodeList.size())
+ {
+ Object o = nodeList.get( i++ );
+ String attachmentURL = doc.getValue( o, "ns1:Attachment" );
+ if (attachmentURL != null)
+ result.add(attachmentURL);
+ }
+ }
+ return result;
+ }
+ catch (java.net.MalformedURLException e)
+ {
+ throw new ManifoldCFException("Bad SharePoint url: "+e.getMessage(),e);
+ }
+ catch (javax.xml.rpc.ServiceException e)
+ {
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("SharePoint: Got a service exception getting
attachments for site "+site+" listName "+listName+" itemID "+itemID+" -
retrying",e);
+ currentTime = System.currentTimeMillis();
+ throw new ServiceInterruption("Service exception: "+e.getMessage(), e,
currentTime + 300000L,
+ currentTime + 12 * 60 * 60000L,-1,true);
+ }
+ catch (org.apache.axis.AxisFault e)
+ {
+ if (e.getFaultCode().equals(new
javax.xml.namespace.QName("http://xml.apache.org/axis/","HTTP";)))
+ {
+ org.w3c.dom.Element elem = e.lookupFaultDetail(new
javax.xml.namespace.QName("http://xml.apache.org/axis/","HttpErrorCode";));
+ if (elem != null)
+ {
+ elem.normalize();
+ String httpErrorCode = elem.getFirstChild().getNodeValue().trim();
+ if (httpErrorCode.equals("404"))
+ return null;
+ else if (httpErrorCode.equals("403"))
+ throw new ManifoldCFException("Remote procedure exception:
"+e.getMessage(),e);
+ else if (httpErrorCode.equals("401"))
+ {
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("SharePoint: Crawl user does not have
sufficient privileges to get attachment list for site "+site+" listName
"+listName+" itemID "+itemID+" - skipping",e);
+ return null;
+ }
+ throw new ManifoldCFException("Unexpected http error code
"+httpErrorCode+" accessing SharePoint at "+baseUrl+site+": "+e.getMessage(),e);
+ }
+ throw new ManifoldCFException("Unknown http error occurred:
"+e.getMessage(),e);
+ }
+
+ if (e.getFaultCode().equals(new
javax.xml.namespace.QName("http://schemas.xmlsoap.org/soap/envelope/","Server.userException";)))
+ {
+ String exceptionName = e.getFaultString();
+ if (exceptionName.equals("java.lang.InterruptedException"))
+ throw new
ManifoldCFException("Interrupted",ManifoldCFException.INTERRUPTED);
+ }
+
+ // I don't know if this is what you get when the library is missing, but
here's hoping.
+ if (e.getMessage().indexOf("List does not exist") != -1)
+ return null;
+
+ if (Logging.connectors.isDebugEnabled())
+ Logging.connectors.debug("SharePoint: Got a remote exception getting
attachments for site "+site+" listName "+listName+" itemID "+itemID+" -
retrying",e);
+ currentTime = System.currentTimeMillis();
+ throw new ServiceInterruption("Remote procedure exception:
"+e.getMessage(), e, currentTime + 300000L,
+ currentTime + 3 * 60 * 60000L,-1,false);
+ }
+ catch (java.rmi.RemoteException e)
+ {
+ throw new ManifoldCFException("Unexpected remote exception occurred:
"+e.getMessage(),e);
+ }
+
+ }
+
/**
* Gets a list of field names of the given document library
* @param site
@@ -1550,8 +1671,9 @@ public class SPSProxyHelper {
// The docLibrary must be a GUID, because we don't have title.
- if ( site.compareTo( "/") == 0 ) site = "";
- ListsWS listService = new ListsWS( baseUrl + site, userName, password,
configuration, httpClient );
+ if ( site.compareTo( "/") == 0 )
+ site = "";
+ ListsWS listService = new ListsWS( baseUrl + site, userName, password,
configuration, httpClient );
ListsSoap listCall = listService.getListsSoapHandler();
GetListResponseGetListResult listResponse = listCall.getList( listName );
Modified:
manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SharePointRepository.java
URL:
http://svn.apache.org/viewvc/manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SharePointRepository.java?rev=1524609&r1=1524608&r2=1524609&view=diff
==============================================================================
---
manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SharePointRepository.java
(original)
+++
manifoldcf/branches/CONNECTORS-778/connectors/sharepoint/connector/src/main/java/org/apache/manifoldcf/crawler/connectors/sharepoint/SharePointRepository.java
Thu Sep 19 01:27:18 2013
@@ -1276,9 +1276,11 @@ public class SharePointRepository extend
String decodedListPath =
documentIdentifier.substring(0,dListSeparatorIndex);
String decodedItemPath = decodedListPath +
documentIdentifier.substring(dListSeparatorIndex+2);
+ // The item path may actually reference both an item number and
a binary attachment. Still trying to determine how to encode this.
+ // ???
+
int cutoff = decodedListPath.lastIndexOf("/");
-
- String encodedItemPath =
encodePath(decodedListPath.substring(cutoff) + "/Lists/" +
decodedItemPath.substring(cutoff+1));
+ String encodedItemPath =
encodePath(decodedListPath.substring(0,cutoff) + "/Lists/" +
decodedItemPath.substring(cutoff+1));
int listCutoff = decodedListPath.lastIndexOf( "/" );
String site = decodedListPath.substring(0,listCutoff);
@@ -1365,6 +1367,7 @@ public class SharePointRepository extend
{
String fieldName = (String)iter.next();
String fieldData = (String)values.get(fieldName);
+ System.out.println("Item '"+decodedItemPath+"' has
field '"+fieldName+"' value '"+fieldData+"'");
data.addField(fieldName,fieldData);
}
}
