def testNearRealtimeOai(self):
self.run = True
portNumber = randint(50000, 60000)
suspendRegister = SuspendRegister()
oaiJazz = OaiJazz(join(self.tempdir, 'oai'))
oaiJazz.updateMetadataFormat(prefix="prefix", schema="", namespace="")
oaiJazz.addObserver(suspendRegister)
storageComponent = MultiSequentialStorage(join(self.tempdir,
'storage'))
self._addOaiRecords(storageComponent, oaiJazz, 3)
oaiPmhThread = Thread(
None, lambda: self.startOaiPmh(portNumber, oaiJazz,
storageComponent, suspendRegister))
observer = CallTrace("observer",
ignoredAttributes=["observer_init"],
methods={'add': lambda **kwargs: (x for x in [])})
harvestThread = Thread(
None, lambda: self.startOaiHarvester(portNumber, observer))
oaiPmhThread.start()
harvestThread.start()
try:
requests = 3
sleepWheel(1.0 + 1.0 * requests)
self.assertEqual([
'startOaiBatch', 'add', 'add', 'stopOaiBatch', 'startOaiBatch',
'add', 'stopOaiBatch'
], [m.name for m in observer.calledMethods])
ids = [
xpath(m.kwargs['lxmlNode'],
'//oai:header/oai:identifier/text()')
for m in observer.calledMethods if m.name == 'add'
]
self.assertEqual([['id0'], ['id1'], ['id2']], ids)
self.assertEqual(1, len(suspendRegister))
observer.calledMethods.reset()
requests += 1
storageComponent.addData(identifier="id3",
name="prefix",
data=b"<a>a3</a>")
oaiJazz.addOaiRecord(identifier="id3", metadataPrefixes=["prefix"])
sleepWheel(1)
self.assertEqual(0, len(suspendRegister))
self.assertEqual(['startOaiBatch', 'add', 'stopOaiBatch'],
[m.name for m in observer.calledMethods])
kwarg = lxmltostring(observer.calledMethods[1].kwargs['lxmlNode'])
self.assertTrue("id3" in kwarg, kwarg)
sleepWheel(1.0)
self.assertEqual(1, len(suspendRegister))
finally:
self.run = False
oaiPmhThread.join()
harvestThread.join()
oaiJazz.close()
def testRemoveSet(self):
oaiJazz = OaiJazz(self.tempdir)
oaiJazz.updateSet('a:b', 'set A/B')
oaiJazz.updateSet('a:c', 'set A/C')
oaiJazz.addOaiRecord('id:0', setSpecs=['a:b', 'a:c'], metadataFormats=[('prefix', '', '')])
oaiJazz.addOaiRecord('id:1', setSpecs=['a:b'], metadataFormats=[('prefix', '', '')])
oaiJazz.addOaiRecord('id:2', setSpecs=['a:c'], metadataFormats=[('prefix', '', '')])
self.assertEquals([
('id:0', set([u'a', u'a:b', u'a:c']), False),
('id:1', set([u'a', u'a:b']), False),
('id:2', set([u'a', u'a:c']), False),
],
[(r.identifier, r.sets, r.isDeleted) for r in oaiJazz.oaiSelect(prefix='prefix').records])
self.assertEquals(set(['a:b', 'a', 'a:c']), oaiJazz.getAllSets())
oaiJazz.close()
removeSetsFromOai(self.tempdir, sets=['a:b'], prefix='prefix', batchSize=1)
oaiJazz = OaiJazz(self.tempdir)
self.assertEquals([
('id:2', set([u'a', u'a:c']), False),
('id:0', set([u'a', u'a:c']), False),
('id:1', set([]), False), # remove hierarchical sets! if possible
],
[(r.identifier, r.sets, r.isDeleted) for r in oaiJazz.oaiSelect(prefix='prefix').records])
self.assertEquals(set(['a', 'a:c']), oaiJazz.getAllSets())
def testShouldRaiseExceptionOnSameRequestTwice(self):
self.run = True
portNumber = randint(50000, 60000)
oaiJazz = OaiJazz(join(self.tempdir, 'oai'))
suspendRegister = SuspendRegister()
oaiJazz.addObserver(suspendRegister)
storageComponent = MultiSequentialStorage(join(self.tempdir, 'storage'))
clientId = str(uuid4())
requests = []
def doOaiListRecord(port):
header, body = getRequest(port=portNumber, path="/", arguments={"verb": "ListRecords", "metadataPrefix": "prefix", "x-wait": "True"}, additionalHeaders={'X-Meresco-Oai-Client-Identifier': clientId}, parse=False)
requests.append((header, body))
oaiPmhThread = Thread(None, lambda: self.startOaiPmh(portNumber, oaiJazz, storageComponent, suspendRegister))
harvestThread1 = Thread(None, lambda: doOaiListRecord(portNumber))
harvestThread2 = Thread(None, lambda: doOaiListRecord(portNumber))
with stderr_replaced():
oaiPmhThread.start()
harvestThread1.start()
try:
while len(suspendRegister) == 0:
sleep(0.01)
harvest1Suspend = suspendRegister._suspendObject(clientId)
self.assertTrue(harvest1Suspend is not None)
harvestThread2.start()
while harvest1Suspend == suspendRegister._suspendObject(clientId):
sleep(0.01)
sleep(0.01)
self.assertTrue(clientId in suspendRegister)
self.assertTrue(harvest1Suspend != suspendRegister._suspendObject(clientId))
self.assertEquals(1, len(requests))
header, body = requests[0]
self.assertTrue('500' in header, header)
self.assertTrue(body.startswith('Aborting suspended request'), body)
storageComponent.addData(identifier="id1", name="prefix", data="<a>a1</a>")
oaiJazz.addOaiRecord(identifier="id1", sets=[], metadataFormats=[("prefix", "", "")])
sleep(0.1)
finally:
self.run = False
oaiPmhThread.join()
harvestThread1.join()
harvestThread2.join()
oaiJazz.close()
def testShouldNotStartToLoopLikeAMadMan(self):
self.run = True
portNumber = randint(50000, 60000)
oaiJazz = OaiJazz(join(self.tempdir, 'oai'))
suspendRegister = SuspendRegister(maximumSuspendedConnections=5)
oaiJazz.addObserver(suspendRegister)
storageComponent = MultiSequentialStorage(join(self.tempdir,
'storage'))
def doUrlOpenWithTimeout(port, basket):
try:
response = urlopen(
"http://localhost:%s/?verb=ListRecords&metadataPrefix=prefix&x-wait=True"
% port,
timeout=0.5)
basket.append(response.getcode())
except timeout as e:
self.assertTrue('timed out' in str(e), str(e))
oaiPmhThread = Thread(
None, lambda: self.startOaiPmh(portNumber, oaiJazz,
storageComponent, suspendRegister))
threads = []
todo = [doUrlOpenWithTimeout] * 7
statusCodes = []
oaiPmhThread.start()
with stderr_replaced():
while todo:
func = todo.pop()
harvestThread = Thread(None,
lambda: func(portNumber, statusCodes))
threads.append(harvestThread)
harvestThread.start()
try:
while len(suspendRegister) == 0:
sleep(0.01)
finally:
for t in threads:
t.join()
self.run = False
oaiPmhThread.join()
oaiJazz.close()
self.assertEqual([204] * 2, statusCodes)
def testNearRealtimeOai(self):
self.run = True
portNumber = randint(50000, 60000)
suspendRegister = SuspendRegister()
oaiJazz = OaiJazz(join(self.tempdir, 'oai'))
oaiJazz.addObserver(suspendRegister)
storageComponent = MultiSequentialStorage(join(self.tempdir, 'storage'))
self._addOaiRecords(storageComponent, oaiJazz, 3)
oaiPmhThread = Thread(None, lambda: self.startOaiPmh(portNumber, oaiJazz, storageComponent, suspendRegister))
observer = CallTrace("observer", ignoredAttributes=["observer_init"], methods={'add': lambda **kwargs: (x for x in [])})
harvestThread = Thread(None, lambda: self.startOaiHarvester(portNumber, observer))
oaiPmhThread.start()
harvestThread.start()
try:
requests = 3
sleepWheel(1.0 + 1.0 * requests)
self.assertEquals(['startOaiBatch', 'add', 'add', 'stopOaiBatch', 'startOaiBatch', 'add', 'stopOaiBatch'], [m.name for m in observer.calledMethods])
ids = [xpath(m.kwargs['lxmlNode'], '//oai:header/oai:identifier/text()') for m in observer.calledMethods if m.name == 'add']
self.assertEquals([['id0'],['id1'],['id2']], ids)
self.assertEquals(1, len(suspendRegister))
observer.calledMethods.reset()
requests += 1
storageComponent.addData(identifier="id3", name="prefix", data="<a>a3</a>")
oaiJazz.addOaiRecord(identifier="id3", sets=[], metadataFormats=[("prefix", "", "")])
sleepWheel(1)
self.assertEquals(0, len(suspendRegister))
self.assertEquals(['startOaiBatch', 'add', 'stopOaiBatch'], [m.name for m in observer.calledMethods])
kwarg = lxmltostring(observer.calledMethods[1].kwargs['lxmlNode'])
self.assertTrue("id3" in kwarg, kwarg)
sleepWheel(1.0)
self.assertEquals(1, len(suspendRegister))
finally:
self.run = False
oaiPmhThread.join()
harvestThread.join()
oaiJazz.close()
def testRemoveSet(self):
oaiJazz = OaiJazz(self.tempdir)
oaiJazz.updateSet('a:b', 'set A/B')
oaiJazz.updateSet('a:c', 'set A/C')
oaiJazz.updateMetadataFormat(prefix="prefix", schema="", namespace="")
oaiJazz.addOaiRecord('id:0',
setSpecs=['a:b', 'a:c'],
metadataPrefixes=['prefix'])
oaiJazz.addOaiRecord('id:1',
setSpecs=['a:b'],
metadataPrefixes=['prefix'])
oaiJazz.addOaiRecord('id:2',
setSpecs=['a:c'],
metadataPrefixes=['prefix'])
self.assertEqual([
('id:0', set(['a', 'a:b', 'a:c']), False),
('id:1', set(['a', 'a:b']), False),
('id:2', set(['a', 'a:c']), False),
], [(r.identifier, r.sets, r.isDeleted)
for r in oaiJazz.oaiSelect(prefix='prefix').records])
self.assertEqual(set(['a:b', 'a', 'a:c']), oaiJazz.getAllSets())
oaiJazz.close()
removeSetsFromOai(self.tempdir,
sets=['a:b'],
prefix='prefix',
batchSize=1)
oaiJazz = OaiJazz(self.tempdir)
self.assertEqual([
('id:2', set(['a', 'a:c']), False),
('id:0', set(['a', 'a:c']), False),
('id:1', set([]), False),
], [(r.identifier, r.sets, r.isDeleted)
for r in oaiJazz.oaiSelect(prefix='prefix').records])
self.assertEqual(set(['a', 'a:c']), oaiJazz.getAllSets())
def _convert(self, verbose=False):
with open(join(self._dataDir, 'converting'), 'w') as f:
f.write('CONVERTING')
try:
with open(join(self._dataDir, 'oai.version'), 'w') as v:
v.write('12')
o = OaiJazz(self._dataDir)
try:
continueAfter = 0
while continueAfter is not None:
if continueAfter == 0:
continueAfter = None
result = o.oaiSelect(prefix=None, continueAfter=continueAfter)
continueAfter = result.continueAfter
for record in result.records:
if record.isDeleted and record.prefixes != record.deletedPrefixes:
if verbose:
print 'Converting', record.identifier
o.deleteOaiRecord(identifier=record.identifier)
finally:
o.close()
finally:
remove(join(self._dataDir, 'converting'))
def testShouldRaiseExceptionOnSameRequestTwice(self):
self.run = True
portNumber = randint(50000, 60000)
oaiJazz = OaiJazz(join(self.tempdir, 'oai'))
oaiJazz.updateMetadataFormat(prefix="prefix", schema="", namespace="")
suspendRegister = SuspendRegister()
oaiJazz.addObserver(suspendRegister)
storageComponent = MultiSequentialStorage(join(self.tempdir,
'storage'))
clientId = str(uuid4())
responses = []
def doOaiListRecord(port):
header, body = getRequest(port=portNumber,
path="/",
arguments={
"verb": "ListRecords",
"metadataPrefix": "prefix",
"x-wait": "True"
},
additionalHeaders={
'X-Meresco-Oai-Client-Identifier':
clientId
},
parse=False)
responses.append((header, body))
oaiPmhThread = Thread(
None, lambda: self.startOaiPmh(portNumber, oaiJazz,
storageComponent, suspendRegister))
harvestThread1 = Thread(None, lambda: doOaiListRecord(portNumber))
harvestThread2 = Thread(None, lambda: doOaiListRecord(portNumber))
with stderr_replaced():
oaiPmhThread.start()
harvestThread1.start()
try:
while len(suspendRegister) == 0:
sleep(0.01)
harvest1Suspend = suspendRegister._suspendObject(clientId)
self.assertTrue(harvest1Suspend is not None)
harvestThread2.start()
while harvest1Suspend == suspendRegister._suspendObject(
clientId):
sleep(0.01)
sleep(0.01)
self.assertTrue(clientId in suspendRegister)
self.assertTrue(
harvest1Suspend != suspendRegister._suspendObject(clientId)
)
self.assertEqual(1, len(responses))
statusAndHeader, body = responses[0]
self.assertEqual("204", statusAndHeader['StatusCode'])
self.assertTrue(body.startswith(b'Aborting suspended request'),
body)
storageComponent.addData(identifier="id1",
name="prefix",
data=b"<a>a1</a>")
oaiJazz.addOaiRecord(identifier="id1",
metadataPrefixes=["prefix"])
sleep(0.1)
finally:
self.run = False
oaiPmhThread.join()
harvestThread1.join()
harvestThread2.join()
oaiJazz.close()