Check out our changes see if you like 'em.

.gitignore

@@ -0,0 +1 @@
+*.pyc

src/fcrepo/README.txt

@@ -17,7 +17,8 @@ complete REST API.
 On top of that a more highlevel abstraction is written, which will be
 demonstrated in this `doctest`_.
 This package has been written for FedoraCommons 3.3 and 3.4, it has not been 
-tested with older versions. REST API documentation can be found in the 
+tested with older versions.  It has been tested with newer Fedoras and will not
+work with fedora >= 3.6 due to WADL issues. REST API documentation can be found in the 
 `Fedora wiki`_.
 
 

src/fcrepo/client.py

@@ -4,6 +4,8 @@
 import urllib
 from collections import defaultdict
 
+from copy import copy
+
 from lxml import etree
 from lxml.builder import ElementMaker
 
@@ -24,12 +26,21 @@ def getNextPID(self, namespace, numPIDs=1, format=u'text/xml'):
                                   format=format)
         xml = response.read()
         response.close()
+        '''
+        The following use of default namespace is to make this code work in 3.4 and 3.5 where the namespacing is different
+        '''
         doc = etree.fromstring(xml)
-        ids = [id.decode('utf8') for id in doc.xpath('/pidList/pid/text()')]
+        fake_namespace_dictionary = {}
+        try:
+            fake_namespace_dictionary['default'] = doc.nsmap[None]
+            ids = [id.decode('utf8') for id in doc.xpath('/default:pidList/default:pid/text()', namespaces = fake_namespace_dictionary)]
+        except KeyError:
+            ids = [id.decode('utf8') for id in doc.xpath('/pidList/pid/text()')]
+
         if len(ids) == 1:
             return ids[0]
         return ids
-        
+
     def createObject(self, pid, label, state=u'A'):
         foxml = ElementMaker(namespace=NSMAP['foxml'], nsmap=NSMAP)
         foxml_state = {'A': u'Active',
@@ -133,30 +144,33 @@ def getDatastreamProfile(self, pid, dsid):
         response.close()
         doc = etree.fromstring(xml)
         result = {}
+        tags = {
+            'dsLabel': 'label',
+            'dsVerionId': 'versionId',
+            'dsCreateDate': 'createdDate',
+            'dsState': 'state',
+            'dsMIME': 'mimeType',
+            'dsFormatURI': 'formatURI',
+            'dsControlGroup': 'controlGroup',
+            'dsSize': 'size',
+            'dsVersionable': 'versionable',
+            'dsInfoType': 'infoType',
+            'dsLocation': 'location',
+            'dsLocationType': 'locationType',
+            'dsChecksum': 'checksum',
+            'dsChecksumType': 'checksumType'
+        }
+
         for child in doc:
             # rename elementnames to match property names in foxml
             # the xml data is namespaced in 3.4, but not in 3.3, so strip out
             # the namespace, to be compatible with both
-            name = {'dsLabel': 'label',
-                    'dsVerionId': 'versionId',
-                    'dsCreateDate': 'createdDate',
-                    'dsState': 'state',
-                    'dsMIME': 'mimeType',
-                    'dsFormatURI': 'formatURI',
-                    'dsControlGroup': 'controlGroup',
-                    'dsSize': 'size',
-                    'dsVersionable': 'versionable',
-                    'dsInfoType': 'infoType',
-                    'dsLocation': 'location',
-                    'dsLocationType': 'locationType',
-                    'dsChecksum': 'checksum',
-                    'dsChecksumType': 'checksumType'}.get(child.tag.split('}')[-1])
-            if name is None or child.text is None:
-                continue
-            value = child.text
-            if not isinstance(value, unicode):
-                value = value.decode('utf8')
-            result[name] = value
+            name = tags.get(child.tag.rpartition('}')[2], None)
+            if name is not None:
+                value = child.text
+                if value and not isinstance(value, unicode):
+                    value = value.decode('utf8')
+                result[name] = value
         return result
 
     def modifyDatastream(self, pid, dsid, body='', **params):
@@ -252,15 +266,27 @@ def searchObjects(self, query, fields, terms=False, maxResults=10):
 
     def searchTriples(self, query, lang='sparql', format='Sparql',
                       limit=100, type='tuples', dt='on', flush=True):
+
         flush = str(flush).lower()
-        url = u'/risearch?%s' % urllib.urlencode({'query':query,
-                                                  'lang':lang,
-                                                  'flush': flush,
-                                                  'format':format,
-                                                  'limit':limit,
-                                                  'type':type,
-                                                  'dt':dt})
-        headers = {'Accept:': 'text/xml'}
+        URL_pramaters = {'query':query,
+                   'lang':lang,
+                   'flush': flush,
+                   'format':format,
+                   'type':type,
+                   'dt':dt}
+
+        #conditionaly set limit if there is one. (so it can be set to None)
+        if limit:
+            URL_pramaters['limit'] = limit
+
+        # Encode in utf8 to let unicode pass through urlencode
+        for key in URL_pramaters:
+            URL_pramaters[key] = URL_pramaters[key].encode('utf8')
+
+        url = u'/risearch?%s' % urllib.urlencode(URL_pramaters)
+        #Fedora started needing authentication in 3.5 for RI, tested in 3.4 as well
+        headers = copy(self.api.connection.form_headers)
+        headers['Accept:'] = 'text/xml'
         response = self.api.connection.open(url, '', headers, method='POST')
         xml = response.read()
         doc = etree.fromstring(xml)

src/fcrepo/connection.py

@@ -1,10 +1,8 @@
 # Copyright (c) 2010 Infrae / Technical University Delft. All rights reserved.
 # See also LICENSE.txt
-import StringIO
-import socket
-import httplib
-import urlparse
-import logging
+import StringIO, socket, httplib, urlparse, logging
+from time import sleep
+from copy import copy
 
 class APIException(Exception):
     """ An exception in the general usage of the API """
@@ -33,7 +31,7 @@ class Connection(object):
     """
     def __init__(self, url, debug=False,
                  username=None, password=None, 
-                 persistent=True):
+                 persistent=False):
         """
          url -- URI pointing to the Fedora server. eg.
 
@@ -65,18 +63,46 @@ def __init__(self, url, debug=False,
     def close(self):
         self.conn.close()
 
-    def open(self, url, body='', headers=None, method='GET',):
+    def open(self, url, body='', headers=None, method='GET'):
         if headers is None:
-            headers = {}
+            http_headers = {}
+        else:
+            # Copy because in an edge case later we may alter the headers.
+            http_headers = copy(headers)
         if url.startswith('/'):
             url = url[1:]
         url = '%s/%s' % (self.path, url)
+
+        # Fedora doesn't like a zero length message body when ingesting a datastream.
+        if body == '' and (method == 'PUT' or method == 'POST') and 'datastreams/' in url:
+            logging.debug('Body empty for HTTP request using'
+                          ' fake form for datastream ingest.')
+
+            # Establish the mime type and boundary.
+            parsed = urlparse.urlparse(url)
+            mime_type = urlparse.parse_qs(parsed.query)['mimeType']
+            if not mime_type:
+                mime_type = 'application/octet-stream'
+            boundary = '----------ThIs_Is_tHe_bouNdaRY_$'
+
+            # Build message body.
+            body = (('--{0}{1}Content-Disposition: form-data;'
+                    ' name="file"; filename="IGNORE THE HACK."{1}'
+                    'Content-Type: {2}{1}{1}{1}--{0}--{1}').format(boundary, '\r\n', mime_type))
+
+            # Change headers for new content type.
+            http_headers.update({
+                'User-Agent': 'INSERT USERAGENTNAME',
+                'Content-Type': 'multipart/form-data; boundary=%s' % boundary
+            })
+
+        # Send out the request.
         attempts = 3
-
         while attempts:
             try:
                 logging.debug('Trying %s on %s' % (method, url))
-                self.conn.request(method, url, body, headers)
+                # We can't have unicode characters floating around in the body.
+                self.conn.request(method, url, body, http_headers)
                 return check_response_status(self.conn.getresponse())
             except (socket.error,
                     httplib.ImproperConnectionState,
@@ -89,6 +115,15 @@ def open(self, url, body='', headers=None, method='GET',):
                 attempts -= 1
                 if not attempts:
                     raise
+            except FedoraConnectionException as e:
+                attempts -= 1
+                if not attempts or e.httpcode not in [409]:
+                    logging.exception('Got HTTP code %s in open...  Failure.' % e.httpcode)
+                    raise e
+                else:
+                    logging.exception('Got HTTP code %s in open... Retrying...' % e.httpcode)
+                    self._reconnect()
+                    sleep(5)
         if not self.persistent:
            self.close()
 

src/fcrepo/datastream.py

@@ -75,9 +75,7 @@ def _setProperty(self, name, value):
     state = property(lambda self: self._info['state'],
                            lambda self, value: self._setProperty('state',
                                                                  value))
-    checksumType = property(lambda self: self._info['checksumType'],
-                            lambda self, value: self._setProperty('checksumType',
-                                                                 value))
+
     versionId = property(lambda self: self._info['versionId'],
                         lambda self, value: self._setProperty('versionId',
                                                               value)) 
@@ -97,7 +95,8 @@ def _setProperty(self, name, value):
     createdDate = property(lambda self: self._info['createdDate'])
     controlGroup = property(lambda self: self._info['controlGroup'])
     size = typedproperty(lambda self: self._info['size'], pytype=int)
-    checksum = property(lambda self: self._info['formatURI'])
+    checksum = property(lambda self: self._info['checksum'])
+    checksumType = property(lambda self: self._info['checksumType'])
 
 
 

src/fcrepo/object.py

@@ -2,7 +2,10 @@
 # See also LICENSE.txt
 
 from fcrepo.datastream import FedoraDatastream, RELSEXTDatastream, DCDatastream
+from fcrepo.connection import FedoraConnectionException
+import logging
 
+logger = logging.getLogger('fcrepo.object.FedoraObject')
 class FedoraObject(object):
     def __init__(self, pid, client):
         self.pid = pid
@@ -31,7 +34,22 @@ def _setProperty(self, name, value):
 
     def datastreams(self):
         if self._dsids is None:
-            self._dsids = self.client.listDatastreams(self.pid)
+            '''
+            XXX: Wrapped in try/except, as it can cause race condition issues when
+            attempting to use the 'in' operator, when trying to determine if an
+            object contains a certain datastream.
+
+            Race condition:
+            1. We get the object (profile) as a FedoraObject
+            2. Another system purges the object
+            3. We got the object, so we try to enumate the DSs via the 'in'
+              operator, and fail.
+            '''
+            try:
+                self._dsids = self.client.listDatastreams(self.pid)
+            except FedoraConnectionException:
+                self._dsids = []
+                logger.debug('Unable to get the list of datastream! (perhaps the object has been purged?) Returning an empty list.')
         return self._dsids
 
     def __iter__(self):

src/fcrepo/wadl.py

@@ -71,6 +71,11 @@ def submit(self, body='', **params):
         for param, value in self.undocumented_params.items():
             if not param in qs:
                 qs[param] = value
+
+        #change to utf8 to let unicode pass through urlencode
+        for key in qs:
+            qs[key] = qs[key].encode('utf8')
+
         if qs:
             self.url = '%s?%s' % (self.url, urllib.urlencode(qs))
 
@@ -82,7 +87,8 @@ def submit(self, body='', **params):
 class API(object):
     def __init__(self, connection):
         self.connection = connection
-        fp = self.connection.open('/objects/application.wadl')
+        # hack for APIA auth
+        fp = self.connection.open('/objects/application.wadl', headers=self.connection.form_headers)
         wadl_xml = fp.read()
         fp.close()
         self.doc = etree.fromstring(wadl_xml)