def testAutoStartOrScheduleRequired(self):
reactor = CallTrace('reactor')
self.assertRaises(ValueError, lambda: PeriodicCall(reactor=reactor))
self.assertRaises(
ValueError, lambda: PeriodicCall(reactor=reactor, autoStart=True))
try:
PeriodicCall(reactor=reactor, autoStart=False)
except:
self.fail('Unexpected exception')
try:
PeriodicCall(reactor=reactor,
schedule=Schedule(period=1),
autoStart=False)
except:
self.fail('Unexpected exception')
try:
PeriodicCall(reactor=reactor,
schedule=Schedule(period=1),
autoStart=True)
except:
self.fail('Unexpected exception')
def testPausePausesOnStart(self):
# autoStart
reactor = CallTrace('reactor')
pc = PeriodicCall(reactor=reactor, autoStart=False)
pc.observer_init()
self.assertEquals([], reactor.calledMethodNames())
# explicit .pause()
pc = PeriodicCall(reactor=reactor, schedule=Schedule(period=1), autoStart=True)
pc.pause()
pc.observer_init()
self.assertEquals([], reactor.calledMethodNames())
def newDNA(self, **kwargs):
emptyGeneratorMethods = [kwargs.get('message', 'handle')]
self.reactor = CallTrace('Reactor', returnValues={'addTimer': 'TOKEN'})
self.observer = CallTrace('Observer', emptyGeneratorMethods=emptyGeneratorMethods, ignoredAttributes=['observer_init'])
self.pc = PeriodicCall(self.reactor, **kwargs)
self.dna = be((Observable(),
(self.pc,
(self.observer,),
),
))
def jvmMonitorTree(reactor, updatableGustosClient):
gustosInterval = 1 if __builtins__.get('__test__', False) else 60
return be((
PeriodicCall(reactor=reactor,
schedule=Schedule(period=gustosInterval)),
(
JvmMonitor(),
(updatableGustosClient, ),
),
))
def writerMain(writerReactor, statePath, luceneserverPort, gatewayPort, quickCommit=False):
http11Request = be(
(HttpRequest1_1(),
(SocketPool(reactor=writerReactor, unusedTimeout=5, limits=dict(totalSize=100, destinationSize=10)),),
)
)
indexCommitTimeout = 30
defaultLuceneSettings = LuceneSettings(
commitTimeout=indexCommitTimeout,
readonly=False,
)
luceneWriter = luceneAndReaderConfig(defaultLuceneSettings, http11Request, luceneserverPort)
periodicDownload = PeriodicDownload(
writerReactor,
host='localhost',
port=gatewayPort,
schedule=Schedule(period=1 if quickCommit else 10), # WST: Interval in seconds before sending a new request to the GATEWAY in case of an error while processing batch records.(default=1). IntegrationTests need 1 second! Otherwise tests will fail!
name='index',
autoStart=True)
oaiDownload = OaiDownloadProcessor(
path='/oaix',
metadataPrefix=NORMALISED_DOC_NAME,
workingDirectory=join(statePath, 'harvesterstate', 'gateway'),
userAgentAddition='idx-server',
xWait=True,
name='index',
autoCommit=False)
# Post commit naar Lucene(server):
scheduledCommitPeriodicCall = be(
(PeriodicCall(writerReactor, message='commit', name='Scheduled commit', schedule=Schedule(period=1 if quickCommit else 300), initialSchedule=Schedule(period=1)), # WST: Flushes data from memory to disk. IntegrationTests need 1 second! Otherwise tests will fail! (API).
(AllToDo(), # broadcast message to all components, despite of what kind of message...
# (periodicDownload,), # WST: periodicDownload does not do anything with a 'commit' message? So why send it to it???
(LuceneCommit(host='localhost', port=luceneserverPort,), # 'commit' message results in http post to /commit/ to Lucene server:
# (LogComponent("PERIODIC"),#), # [PERIODIC] httprequest1_1(*(), **{'body': None, 'host': 'localhost', 'request': '/commit/', 'port': 52501, 'method': 'POST'})
(http11Request,),
# ),
)
)
)
)
writerServer = \
(Observable(),
(scheduledCommitPeriodicCall,), # Stuur periodiek een 'Commit' naar de LuceneServer...
# (DebugPrompt(reactor=writerReactor, port=readerPort-1, globals=locals()),),
(periodicDownload, # Ga/connect (periodiek) naar de Gateway-server...
(XmlParseLxml(fromKwarg="data", toKwarg="lxmlNode", parseOptions=dict(huge_tree=True, remove_blank_text=True)),
(oaiDownload, # Haal OAI spulletjes van de Gateway...
(UpdateAdapterFromOaiDownloadProcessor(), # Maakt van een SRU update/delete bericht (lxmlNode) een relevante message: 'delete' of 'add' message.
# (LogComponent("SRU harvest van GATEWAY"),), #[SRU harvest van GATEWAY] add(*(), **{'partname': 'record', 'identifier': 'meresco:record:1', 'lxmlNode': '_ElementTree(<record xmlns="http://www.openarchives.org/OAI/2.0/" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"><header><identifier>meresco:record:1</identifier><datestamp>2016-07-13T15:31:10Z</datestamp></header><metadata><document xmlns="http://meres
(FilterMessages(allowed=['add']),
(XmlXPath(['/oai:record/oai:metadata/document:document'], fromKwarg='lxmlNode'),
# (LogComponent("NormdocToFieldsList"),),
(NormdocToFieldsList(), # Platte lijst met veldnamen en waardes...
(RecordPidToAuthNid(),),
# (LogComponent("NormdocToFieldsList"),), # [DcToFieldsList] add(*(), **{'fieldslist': [('dc:identifier', 'http://meresco.com?record=1'), ('dc:description', 'This is an example program about Search with Meresco'), ('dc:title', 'Example Program 1'), ('dc:creator', 'Seecr'), ('dc:publisher', 'Seecr'), ('dc:date', '2016'), ('dc:type', 'Example'), ('dc:subject', 'Search'), ('dc:language', 'en'), ('dc:rights', 'Open Source')], 'partname': 'record', 'identifier': 'meresco:record:1'})
# [NormdocToFieldsList] lookupNameIds(*(set(['doi:10.1002/lno.10611', 'wos:000423029300003']),), **{})
(FieldsListToLuceneDocument( # Maakt addDocument messege + creeert de facet/drilldown velden waarvan de value's tot max. 256 chars getruncated worden.
fieldRegistry=luceneWriter.settings.fieldRegistry, # o.a. drilldownfields definitie
untokenizedFieldnames=untokenizedFieldnames, # untokenized fields
indexFieldFactory=DcFields, # Creeert een "__all__", veldnaam en optioneel "untokenized.veldnaam"...
#rewriteIdentifier=(lambda idee: idee.split(':', 1)[-1]) # meresco:record:1' => 'record:1'
),
# (LogComponent("FieldsListToLuceneDocument"),), # [LUCENE_WRITER] addDocument(*(), **{'fields': [{'type': 'TextField', 'name': '__all__', 'value': 'http://meresco.com?record=1'}, {'type': 'TextField', 'name': 'dc:identifier', 'value': 'http://meresco.com?record=1'}, {'type': 'StringField', 'name': 'untokenized.dc:identifier', 'value': 'http://meresco.com?record=1'}, {'type': 'TextField', 'name': '__all__', 'value': 'This is an example program about Search with Meresco'}, {'type': 'TextField', 'name': 'dc:description', 'value': 'This is an example program about Search with Meresco'}, {'type': 'TextField', 'name': '__all__', 'value': 'Example Program 1'}, {'type': 'TextField', 'name': 'dc:title', 'value': 'Example Program 1'}, {'type': 'TextField', 'name': '__all__', 'value': 'Seecr'}, {'type': 'TextField', 'name': 'dc:creator', 'value': 'Seecr'}, {'type': 'TextField', 'name': '__all__', 'value': 'Seecr'}, {'type': 'TextField', 'name': 'dc:publisher', 'value': 'Seecr'}, {'type': 'TextField', 'name': '__all__', 'value': '2016'}, {'type': 'TextField', 'name': 'dc:date', 'value': '2016'}, {'path': ['2016'], 'type': 'FacetField', 'name': 'untokenized.dc:date'}, {'type': 'TextField', 'name': '__all__', 'value': 'Example'}, {'type': 'TextField', 'name': 'dc:type', 'value': 'Example'}, {'type': 'TextField', 'name': '__all__', 'value': 'Search'}, {'type': 'TextField', 'name': 'dc:subject', 'value': 'Search'}, {'path': ['Search'], 'type': 'FacetField', 'name': 'untokenized.dc:subject'}, {'type': 'TextField', 'name': '__all__', 'value': 'en'}, {'type': 'TextField', 'name': 'dc:language', 'value': 'en'}, {'type': 'TextField', 'name': '__all__', 'value': 'Open Source'}, {'type': 'TextField', 'name': 'dc:rights', 'value': 'Open Source'}], 'identifier': 'meresco:record:1'})
# [####LUCENE_WRITER] addDocument(*(), **{'fields': [{'type': 'TextField', 'name': '__all__', 'value': 'knaw'}, {'type': 'TextField', 'name': 'meta:id', 'value': 'knaw'}, {'type': 'TextField', 'name': '__all__', 'value': 'olddata'}, {'type': 'TextField', 'name': 'meta:set', 'value': 'olddata'}, {'type': 'TextField', 'name': '__all__', 'value': 'http://oai.knaw.nl/oai'}, {'type': 'TextField', 'name': 'meta:baseurl', 'value': 'http://oai.knaw.nl/oai'}, {'type': 'TextField', 'name': '__all__', 'value': 'knaw'}, {'type': 'TextField', 'name': 'meta:repositoryGroupId', 'value': 'knaw'}, {'type': 'TextField', 'name': '__all__', 'value': 'nl_didl'}, {'type': 'TextField', 'name': 'meta:metadataPrefix', 'value': 'nl_didl'}, {'type': 'TextField', 'name': '__all__', 'value': 'publication'}, {'type': 'TextField', 'name': 'meta_collection', 'value': 'publication'}, {'path': ['publication'], 'type': 'FacetField', 'name': 'untokenized.meta_collection'}], 'identifier': 'knaw:record:3'})
(luceneWriter,),
# ),
)
)
# )
# )
)
),
(FilterMessages(allowed=['delete']),
(luceneWriter,),
)
)
)
)
)
)
return writerServer
