def testIncrementalHarvestWithFromAfterSomePeriod(self):
observer = CallTrace(emptyGeneratorMethods=['add'])
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=False,
err=StringIO(),
incrementalHarvestSchedule=Schedule(period=10))
oaiDownloadProcessor._time = lambda: 1.0
oaiDownloadProcessor.addObserver(observer)
consume(
oaiDownloadProcessor.handle(parse(StringIO(LISTRECORDS_RESPONSE))))
self.assertEqual(None, oaiDownloadProcessor._resumptionToken)
self.assertEqual('2002-06-01T19:20:30Z', oaiDownloadProcessor._from)
self.assertEqual(None, oaiDownloadProcessor.buildRequest())
oaiDownloadProcessor._time = lambda: 6.0
self.assertEqual(None, oaiDownloadProcessor.buildRequest())
oaiDownloadProcessor._time = lambda: 10.0
self.assertEqual(None, oaiDownloadProcessor.buildRequest())
oaiDownloadProcessor._time = lambda: 11.1
request = oaiDownloadProcessor.buildRequest()
self.assertTrue(
request.startswith(
'GET /oai?verb=ListRecords&from=2002-06-01T19%3A20%3A30Z&metadataPrefix=oai_dc'
), request)
def testIncrementalHarvestScheduleNoneOverruledWithSetIncrementalHarvestSchedule(
self):
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=False,
err=StringIO(),
incrementalHarvestSchedule=None)
oaiDownloadProcessor._time = lambda: 10
consume(
oaiDownloadProcessor.handle(
parse(StringIO(LISTRECORDS_RESPONSE % ''))))
self.assertEqual(None, oaiDownloadProcessor._resumptionToken)
self.assertEqual('2002-06-01T19:20:30Z', oaiDownloadProcessor._from)
self.assertEqual(None, oaiDownloadProcessor._earliestNextRequestTime)
oaiDownloadProcessor.setIncrementalHarvestSchedule(schedule=Schedule(
period=3))
self.assertEqual(None, oaiDownloadProcessor.buildRequest())
self.assertEqual(None, oaiDownloadProcessor._earliestNextRequestTime)
oaiDownloadProcessor.scheduleNextRequest()
self.assertNotEqual(None, oaiDownloadProcessor.buildRequest())
self.assertEqual(0, oaiDownloadProcessor._earliestNextRequestTime)
consume(
oaiDownloadProcessor.handle(
parse(StringIO(LISTRECORDS_RESPONSE % ''))))
self.assertEqual(13, oaiDownloadProcessor._earliestNextRequestTime)
def testListRecordsRequestError(self):
resumptionToken = "u|c1286437597991025|mprefix|s|f"
with open(join(self.tempdir, 'harvester.state'), 'w') as f:
f.write("Resumptiontoken: %s\n" % resumptionToken)
observer = CallTrace()
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True,
err=StringIO())
oaiDownloadProcessor.addObserver(observer)
self.assertEqual(
'GET /oai?%s HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n'
% (urlencode([('verb', 'ListRecords'),
('resumptionToken', resumptionToken),
('x-wait', 'True')
]), oaiDownloadProcessor._identifier),
oaiDownloadProcessor.buildRequest())
consume(oaiDownloadProcessor.handle(parse(StringIO(ERROR_RESPONSE))))
self.assertEqual(0, len(observer.calledMethods))
self.assertEqual("someError: Some error occurred.\n",
oaiDownloadProcessor._err.getvalue())
self.assertEqual(
'GET /oai?%s HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n'
% (urlencode(
[('verb', 'ListRecords'), ('metadataPrefix', 'oai_dc'),
('x-wait', 'True')]), oaiDownloadProcessor._identifier),
oaiDownloadProcessor.buildRequest())
def testSetInRequest(self):
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
set="setName",
workingDirectory=self.tempdir,
xWait=True)
self.assertEqual(
"""GET /oai?verb=ListRecords&metadataPrefix=oai_dc&set=setName&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% oaiDownloadProcessor._identifier,
oaiDownloadProcessor.buildRequest())
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
set="set-_.!~*'()",
workingDirectory=self.tempdir,
xWait=True)
self.assertEqual(
"""GET /oai?verb=ListRecords&metadataPrefix=oai_dc&set=set-_.%%21~%%2A%%27%%28%%29&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% oaiDownloadProcessor._identifier,
oaiDownloadProcessor.buildRequest())
resumptionToken = "u|c1286437597991025|mprefix|s|f"
with open(join(self.tempdir, 'harvester.state'), 'w') as f:
f.write("Resumptiontoken: %s\n" % resumptionToken)
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
set="setName",
workingDirectory=self.tempdir,
xWait=True)
self.assertEqual(
"""GET /oai?verb=ListRecords&resumptionToken=u%%7Cc1286437597991025%%7Cmprefix%%7Cs%%7Cf&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% oaiDownloadProcessor._identifier,
oaiDownloadProcessor.buildRequest())
def testRequest(self):
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True)
self.assertEqual(
"""GET /oai?verb=ListRecords&metadataPrefix=oai_dc&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% oaiDownloadProcessor._identifier,
oaiDownloadProcessor.buildRequest())
def testRequestWithAdditionalHeaders(self):
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True)
request = oaiDownloadProcessor.buildRequest(
additionalHeaders={'Host': 'example.org'})
self.assertEqual(
"""GET /oai?verb=ListRecords&metadataPrefix=oai_dc&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nHost: example.org\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% oaiDownloadProcessor._identifier, request)
def testKeepResumptionTokenOnFailingAddCall(self):
resumptionToken = "u|c1286437597991025|mprefix|s|f"
with open(join(self.tempdir, 'harvester.state'), 'w') as f:
f.write("Resumptiontoken: %s\n" % resumptionToken)
observer = CallTrace()
observer.exceptions = {'add': Exception("Could be anything")}
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True,
err=StringIO())
oaiDownloadProcessor.addObserver(observer)
self.assertEqual(
'GET /oai?%s HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n'
% (urlencode([('verb', 'ListRecords'),
('resumptionToken', resumptionToken),
('x-wait', 'True')
]), oaiDownloadProcessor._identifier),
oaiDownloadProcessor.buildRequest())
self.assertRaises(
Exception, lambda: list(
compose(
oaiDownloadProcessor.handle(
parse(StringIO(LISTRECORDS_RESPONSE % RESUMPTION_TOKEN)
)))))
self.assertEqual(['startOaiBatch', 'add', 'stopOaiBatch'],
[m.name for m in observer.calledMethods])
errorOutput = oaiDownloadProcessor._err.getvalue()
self.assertTrue(errorOutput.startswith('Traceback'), errorOutput)
self.assertTrue(
'Exception: Could be anything\nWhile processing:\n<record xmlns="http://www.openarchives.org/OAI/2.0/"><header><identifier>oai:identifier:1'
in errorOutput, errorOutput)
self.assertEqual(
'GET /oai?%s HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n'
% (urlencode([('verb', 'ListRecords'),
('resumptionToken', resumptionToken),
('x-wait', 'True')
]), oaiDownloadProcessor._identifier),
oaiDownloadProcessor.buildRequest())
def testBuildRequestNoneWhenNoResumptionToken(self):
observer = CallTrace(emptyGeneratorMethods=['add'])
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=False,
err=StringIO())
oaiDownloadProcessor.addObserver(observer)
consume(
oaiDownloadProcessor.handle(parse(StringIO(LISTRECORDS_RESPONSE))))
self.assertEqual(None, oaiDownloadProcessor._resumptionToken)
self.assertEqual(None, oaiDownloadProcessor.buildRequest())
def testUpdateRequest(self):
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True)
oaiDownloadProcessor.setPath('/otherOai')
oaiDownloadProcessor.setMetadataPrefix('otherPrefix')
oaiDownloadProcessor.setSet('aSet')
oaiDownloadProcessor.setFrom('2014')
self.assertEqual(
"""GET /otherOai?verb=ListRecords&from=2014&metadataPrefix=otherPrefix&set=aSet&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% oaiDownloadProcessor._identifier,
oaiDownloadProcessor.buildRequest())
def testScheduleNextRequest(self):
oaiDownloadProcessor = OaiDownloadProcessor(
path='/p', metadataPrefix='p', workingDirectory=self.tempdir)
oaiDownloadProcessor._time = lambda: 17
consume(
oaiDownloadProcessor.handle(
parse(StringIO(LISTRECORDS_RESPONSE % ''))))
self.assertTrue(oaiDownloadProcessor._earliestNextRequestTime > 17)
oaiDownloadProcessor.scheduleNextRequest()
self.assertEqual(0, oaiDownloadProcessor._earliestNextRequestTime)
self.assertEqual(True, oaiDownloadProcessor._timeForNextRequest())
self.assertNotEqual(None, oaiDownloadProcessor.buildRequest())
oaiDownloadProcessor.scheduleNextRequest(Schedule(period=0))
self.assertEqual(17, oaiDownloadProcessor._earliestNextRequestTime)
self.assertEqual(True, oaiDownloadProcessor._timeForNextRequest())
self.assertNotEqual(None, oaiDownloadProcessor.buildRequest())
oaiDownloadProcessor.scheduleNextRequest(Schedule(period=120))
self.assertEqual(137, oaiDownloadProcessor._earliestNextRequestTime)
self.assertEqual(False, oaiDownloadProcessor._timeForNextRequest())
self.assertEqual(None, oaiDownloadProcessor.buildRequest())
def testUpdateRequestAfterSetResumptionToken(self):
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
set="aSet",
workingDirectory=self.tempdir,
xWait=False)
oaiDownloadProcessor.setPath('/otherOai')
oaiDownloadProcessor.setFrom('2014')
oaiDownloadProcessor.setResumptionToken('ReSumptionToken')
self.assertEqual(
"""GET /otherOai?verb=ListRecords&resumptionToken=ReSumptionToken HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% oaiDownloadProcessor._identifier,
oaiDownloadProcessor.buildRequest())
def testPersistentIdentifier(self):
identifierFilepath = join(self.tempdir, 'harvester.identifier')
self.assertFalse(isfile(identifierFilepath))
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True)
currentIdentifier = oaiDownloadProcessor._identifier
self.assertTrue(isfile(identifierFilepath))
with open(identifierFilepath) as f:
self.assertEqual(currentIdentifier, f.read())
self.assertEqual(
"""GET /oai?verb=ListRecords&metadataPrefix=oai_dc&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% currentIdentifier, oaiDownloadProcessor.buildRequest())
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True)
self.assertEqual(
"""GET /oai?verb=ListRecords&metadataPrefix=oai_dc&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n"""
% currentIdentifier, oaiDownloadProcessor.buildRequest())
def testRestartAfterFinish(self):
observer = CallTrace(emptyGeneratorMethods=['add'])
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=False,
err=StringIO(),
restartAfterFinish=True)
oaiDownloadProcessor.addObserver(observer)
consume(
oaiDownloadProcessor.handle(parse(StringIO(LISTRECORDS_RESPONSE))))
self.assertEqual(None, oaiDownloadProcessor._resumptionToken)
request = oaiDownloadProcessor.buildRequest()
self.assertTrue(
request.startswith(
'GET /oai?verb=ListRecords&metadataPrefix=oai_dc HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: '
), request)
def testUseResumptionToken(self):
observer = CallTrace(emptyGeneratorMethods=['add'])
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True,
err=StringIO())
oaiDownloadProcessor.addObserver(observer)
consume(
oaiDownloadProcessor.handle(
parse(StringIO(LISTRECORDS_RESPONSE % RESUMPTION_TOKEN))))
self.assertEqual('x?y&z', oaiDownloadProcessor._resumptionToken)
self.assertEqual(
'GET /oai?verb=ListRecords&resumptionToken=x%%3Fy%%26z&x-wait=True HTTP/1.0\r\nX-Meresco-Oai-Client-Identifier: %s\r\nUser-Agent: Meresco-Oai-DownloadProcessor/5.x\r\n\r\n'
% oaiDownloadProcessor._identifier,
oaiDownloadProcessor.buildRequest())
oaiDownloadProcessor = OaiDownloadProcessor(
path="/oai",
metadataPrefix="oai_dc",
workingDirectory=self.tempdir,
xWait=True,
err=StringIO())
self.assertEqual('x?y&z', oaiDownloadProcessor._resumptionToken)