def _findFragmentNodesWithAboutUris(self, lxmlNode):
for descriptionNode in xpath(lxmlNode, "*[@rdf:about]"):
uri = str(descriptionNode.attrib[curieToTag("rdf:about")])
yield descriptionNode, uri
for statementNode in xpath(lxmlNode, "rdf:Statement"):
uri = str(xpathFirst(statementNode, 'rdf:subject/@rdf:resource'))
yield statementNode, uri
def testSortKeys(self):
body = self.doSruQuery('*', sortKeys='intfield1,,1')
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals(['record:%s' % i for i in xrange(1,11)], records)
body = self.doSruQuery('*', sortKeys='intfield1,,0')
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals(['record:%s' % i for i in xrange(100,90, -1)], records)
def testSortKeysWithMissingValues(self):
body = self.doSruQuery('*', sortKeys='field4,,1')
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals('record:1', records[0])
body = self.doSruQuery('*', sortKeys='field4,,0')
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals('record:1', records[0])
def testSortKeys(self):
body = self.doSruQuery('*', sortKeys='intfield1,,1')
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals(['record:%s' % i for i in xrange(1, 11)], records)
body = self.doSruQuery('*', sortKeys='intfield1,,0')
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals(['record:%s' % i for i in xrange(100, 90, -1)],
records)
def testMaximumRecords(self):
body = self.doSruQuery('*', maximumRecords=20)
records = xpath(body, '//srw:record')
self.assertEquals(20, len(records))
body = self.doSruQuery('*', maximumRecords=0, path='/via-remote-sru')
records = xpath(body, '//srw:record')
self.assertEquals(0, len(records))
diag = xpathFirst(body, '//diag:diagnostic/diag:details/text()')
self.assertEquals(None, diag)
def testMaximumRecords(self):
body = self.doSruQuery('*', maximumRecords=20)
records = xpath(body, '//srw:record')
self.assertEquals(20, len(records))
body = self.doSruQuery('*', maximumRecords=0, path='/via-remote-sru')
records = xpath(body, '//srw:record')
self.assertEquals(0, len(records))
diag = xpathFirst(body, '//diag:diagnostic/diag:details/text()')
self.assertEquals(None, diag)
def testGetRecordWithRepositoryIdentifierMissingExpectedPrefix(self):
oaigetrecord = OaiGetRecord(OaiRepository(identifier="example.org"))
result = asString(
oaigetrecord.getRecord(
arguments=dict(verb=["GetRecord"], metadataPrefix=["oai_dc"], identifier=["not:properly:prefixed:id0"]),
**self.httpkwargs
)
)
header, body = result.split("\r\n\r\n")
self.assertTrue(xpath(XML(body), '/oai:OAI-PMH/oai:error[@code="idDoesNotExist"]'), body)
def testAddTwoRecordsWithSameUriAndDeleteLast(self):
uri = "uri:someuri"
rdfNode, description = createRdfNode(uri)
createSubElement(description, "dc:title", text='One')
consume(self.dna.all.add(identifier="identifier1", partname="ignored", lxmlNode=rdfNode.getroot()))
rdfNode, description = createRdfNode(uri)
createSubElement(description, "dc:title", text='Two')
consume(self.dna.all.add(identifier="identifier2", partname="ignored", lxmlNode=rdfNode.getroot()))
consume(self.dna.all.delete(identifier="identifier2"))
record = self.oaiJazz.getRecord(identifier=uri)
self.assertEquals(['One'], xpath(XML(self.storage.getData(identifier=record.identifier, name='rdf')), '/rdf:RDF/rdf:Description/dc:title/text()'))
def testFacet(self):
body = self.doSruQuery('*', facet='untokenized.field2')
ddItems = xpath(
body,
"//drilldown:term-drilldown/drilldown:navigator[@name='untokenized.field2']/drilldown:item"
)
self.assertEquals(
set([('value0', '10'), ('value9', '10'), ('value8', '10'),
('value7', '10'), ('value6', '10'), ('value5', '10'),
('value4', '10'), ('value3', '10'), ('value2', '10'),
('value1', '9')]),
set([(i.attrib['value'], i.attrib['count']) for i in ddItems]))
def testGetRecordWithMultiSequentialStorage(self):
oaijazz = OaiJazz(self.tempdir + "/jazz")
storage = MultiSequentialStorage(self.tempdir + "/seq-store")
oairecord = OaiRecord()
oaigetrecord = be(
(
OaiGetRecord(repository=OaiRepository()),
(oaijazz,),
(oairecord, (RetrieveToGetDataAdapter(), (storage,))),
)
)
oaijazz.addOaiRecord(identifier="id0", sets=(), metadataFormats=[("oai_dc", "", "")])
storage.addData(identifier="id0", name="oai_dc", data="data01")
response = oaigetrecord.getRecord(
arguments=dict(verb=["GetRecord"], metadataPrefix=["oai_dc"], identifier=["id0"]), **self.httpkwargs
)
_, body = asString(response).split("\r\n\r\n")
self.assertEquals("data01", xpath(parse(StringIO(body)), "//oai:metadata")[0].text)
def testGetRecordWithMultiSequentialStorage(self):
oaijazz = OaiJazz(self.tempdir + '/jazz')
oaijazz.updateMetadataFormat(prefix="oai_dc", schema="", namespace="")
storage = MultiSequentialStorage(self.tempdir + "/seq-store")
oairecord = OaiRecord()
oaigetrecord = be(
(OaiGetRecord(repository=OaiRepository()), (oaijazz, ),
(oairecord, (RetrieveToGetDataAdapter(), (storage, )))))
oaijazz.addOaiRecord(identifier="id0", metadataPrefixes=['oai_dc'])
storage.addData(identifier="id0", name="oai_dc", data=b"data01")
response = oaigetrecord.getRecord(arguments=dict(
verb=['GetRecord'],
metadataPrefix=['oai_dc'],
identifier=['id0'],
),
**self.httpkwargs)
_, body = asString(response).split("\r\n\r\n")
self.assertEqual(
"data01",
xpath(parse(BytesIO(body.encode())), '//oai:metadata')[0].text)
def testSortOnTimestamp(self):
body = self.doSruQuery('*',
sortKeys='__timestamp__,,1',
maximumRecords=100)
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals(['record:%s' % i for i in xrange(1, 101)], records)
def add(self, identifier, lxmlNode, *args, **kwargs):
if xpath(lxmlNode, '/oai:record/oai:header[@status="deleted"]'):
yield self.all.delete(identifier=identifier)
else:
yield self.all.add(identifier=identifier, partname='record', lxmlNode=lxmlNode)
def testFacet(self):
body = self.doSruQuery('*', facet='untokenized.field2')
ddItems = xpath(body, "//drilldown:term-drilldown/drilldown:navigator[@name='untokenized.field2']/drilldown:item")
self.assertEquals(
set([('value0', '10'), ('value9', '10'), ('value8', '10'), ('value7', '10'), ('value6', '10'), ('value5', '10'), ('value4', '10'), ('value3', '10'), ('value2', '10'), ('value1', '9')]),
set([(i.attrib['value'], i.attrib['count']) for i in ddItems]))
def testReificationWithRdfID(self):
testNamespaces = namespaces.copyUpdate(dict(test="urn:test#"))
xpathFirst = testNamespaces.xpathFirst
rdfXml = '''<rdf:RDF %(xmlns_rdf)s %(xmlns_test)s>
<rdf:Description rdf:about="some:uri">
<test:relation rdf:ID="_987">object</test:relation>
</rdf:Description>
<rdf:Statement rdf:about="#_987">
<test:reificationRelation>reification object</test:reificationRelation>
</rdf:Statement>
</rdf:RDF>''' % testNamespaces
graph = RDFParser().parse(XML(rdfXml))
result = Triples2RdfXml(namespaces=testNamespaces).asRdfXml(graph)
self.assertEquals('object', xpathFirst(result, '/rdf:RDF/rdf:Description[@rdf:about="some:uri"]/test:relation[@rdf:ID="_987"]/text()'))
self.assertEquals('reification object', xpathFirst(result, '/rdf:RDF/rdf:Statement[@rdf:about="#_987"]/test:reificationRelation/text()'))
self.assertEquals(['rdf:Description', 'rdf:Statement'], [tagToCurie(node.tag) for node in xpath(result, '/rdf:RDF/*')])
self.assertXmlEquals(rdfXml, result)
def __iter__(self):
for record in xpath(
self._response,
"/srw:searchRetrieveResponse/srw:records/srw:record"):
yield SruRecord(record)
def testReificationWithBlankNodeSubject(self):
testNamespaces = namespaces.copyUpdate(dict(test="urn:test#"))
xpathFirst = testNamespaces.xpathFirst
rdfXml = '''<rdf:RDF %(xmlns_rdf)s %(xmlns_test)s>
<rdf:Description>
<test:relation rdf:ID="_987">object</test:relation>
</rdf:Description>
<rdf:Statement rdf:about="#_987">
<test:reificationRelation>reification object</test:reificationRelation>
</rdf:Statement>
</rdf:RDF>''' % testNamespaces
graph = RDFParser().parse(XML(rdfXml))
result = Triples2RdfXml(namespaces=testNamespaces).asRdfXml(graph)
self.assertEquals('reification object', xpathFirst(result, '/rdf:RDF/rdf:Statement[@rdf:about="#_987"]/test:reificationRelation/text()'))
self.assertEquals(None, xpathFirst(result, '/rdf:RDF/rdf:Statement[@rdf:about="#_987"]/rdf:subject')) # rdf:Statement with bnode subject should not actually refer to it...!!
self.assertEquals('object', xpathFirst(result, '/rdf:RDF/rdf:Description/test:relation[@rdf:ID="_987"]/text()'))
self.assertEquals(['rdf:Description', 'rdf:Statement'], [tagToCurie(node.tag) for node in xpath(result, '/rdf:RDF/*')])
def testStartRecord(self):
body = self.doSruQuery('*', maximumRecords=100)
records = xpath(body, '//srw:recordIdentifier/text()')
body = self.doSruQuery('*', maximumRecords=10, startRecord=51)
self.assertEquals(records[50:60], xpath(body, '//srw:recordIdentifier/text()'))
def __iter__(self):
for record in xpath(self._response, "/srw:searchRetrieveResponse/srw:records/srw:record"):
yield SruRecord(record)
def testRecordIds(self):
body = self.doSruQuery('*', maximumRecords=100)
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals(100, len(records))
self.assertEquals(set(['record:%s' % i for i in xrange(1,101)]), set(records))
def testRecordIds(self):
body = self.doSruQuery('*', maximumRecords=100)
records = xpath(body, '//srw:recordIdentifier/text()')
self.assertEquals(100, len(records))
self.assertEquals(set(['record:%s' % i for i in xrange(1, 101)]),
set(records))
def testStartRecord(self):
body = self.doSruQuery('*', maximumRecords=100)
records = xpath(body, '//srw:recordIdentifier/text()')
body = self.doSruQuery('*', maximumRecords=10, startRecord=51)
self.assertEquals(records[50:60],
xpath(body, '//srw:recordIdentifier/text()'))