def testOverrideMetadataPrefix(self):
'''test that the metadataPrefix for an OAI feed can be overridden.
The extra_data for OAI can be either just a set spec or a html query
string of set= &metadataPrefix=
'''
httpretty.register_uri(httpretty.GET,
'http://content.cdlib.org/oai',
body=open(DIR_FIXTURES + '/testOAI.xml').read())
set_fetcher = fetcher.OAIFetcher('http://content.cdlib.org/oai',
'set=oac:images')
self.assertEqual(set_fetcher._set, 'oac:images')
rec = set_fetcher.next()
self.assertIsInstance(rec, dict)
self.assertIn('id', rec)
self.assertEqual(rec['id'], '13030/hb796nb5mn')
self.assertIn('datestamp', rec)
self.assertIn(rec['datestamp'], '2005-12-13')
self.assertEqual(
httpretty.last_request().querystring, {
u'verb': [u'ListRecords'],
u'set': [u'oac:images'],
u'metadataPrefix': [u'oai_dc']
})
httpretty.register_uri(httpretty.GET,
'http://content.cdlib.org/oai',
body=open(DIR_FIXTURES +
'/testOAI-didl.xml').read())
didl_fetcher = fetcher.OAIFetcher(
'http://content.cdlib.org/oai',
'set=oac:images&metadataPrefix=didl')
self.assertEqual(didl_fetcher._set, 'oac:images')
self.assertEqual(didl_fetcher._metadataPrefix, 'didl')
rec = didl_fetcher.next()
self.assertIsInstance(rec, dict)
self.assertIn('id', rec)
self.assertEqual(rec['id'], 'oai:ucispace-prod.lib.uci.edu:10575/25')
self.assertEqual(rec['title'], ['Schedule of lectures'])
self.assertIn('datestamp', rec)
self.assertEqual(rec['datestamp'], '2015-05-20T11:04:23Z')
self.assertEqual(
httpretty.last_request().querystring, {
u'verb': [u'ListRecords'],
u'set': [u'oac:images'],
u'metadataPrefix': [u'didl']
})
self.assertEqual(
rec['Resource']['@ref'],
'http://ucispace-prod.lib.uci.edu/xmlui/bitstream/' +
'10575/25/1/!COLLOQU.IA.pdf')
self.assertEqual(rec['Item']['@id'],
'uuid-640925bd-9cdf-46be-babb-b2138c3fce9c')
self.assertEqual(rec['Component']['@id'],
'uuid-897984d8-9392-4a68-912f-ffdf6fd7ce59')
self.assertIn('Descriptor', rec)
self.assertEqual(rec['Statement']['@mimeType'],
'application/xml; charset=utf-8')
self.assertEqual(
rec['DIDLInfo']['{urn:mpeg:mpeg21:2002:02-DIDL-NS}DIDLInfo'][0]
['text'], '2015-05-20T20:30:26Z')
del didl_fetcher
def setUp(self):
super(OAIFetcherTestCase, self).setUp()
httpretty.register_uri(httpretty.GET,
'http://content.cdlib.org/oai',
body=open(DIR_FIXTURES + '/testOAI.xml').read())
self.fetcher = fetcher.OAIFetcher('http://content.cdlib.org/oai',
'oac:images')
def test_getMetadataPrefix(self):
fmts = open(DIR_FIXTURES+'/oai-fmts.xml').read()
fmts_qdc = open(DIR_FIXTURES+'/oai-fmts-qdc.xml').read()
httpretty.register_uri(
httpretty.GET,
'http://xxxx.cdlib.org/oai?verb=ListMetadataFormats',
responses=[
httpretty.Response(body=fmts, status=200),
httpretty.Response(body=fmts, status=200),
httpretty.Response(body=fmts, status=200),
httpretty.Response(body=fmts_qdc, status=200),
])
set_fetcher = fetcher.OAIFetcher('http://xxxx.cdlib.org/oai',
'set=bogus')
self.assertEqual(set_fetcher._metadataPrefix, 'oai_dc')
prefix = set_fetcher.get_metadataPrefix('')
self.assertEqual(prefix, 'oai_dc')
prefix = set_fetcher.get_metadataPrefix('metadataPrefix=override')
self.assertEqual(prefix, 'override')
prefix = set_fetcher.get_metadataPrefix('')
self.assertEqual(prefix, 'oai_qdc')
def testDCTERMS(self):
'''Test the OAI fetcher when the source data has exteneded dcterms in
it.
'''
DCTERMS = (
'abstract', 'accessRights', 'accrualMethod',
'accrualPeriodicity', 'accrualPolicy', 'alternative',
'audience', 'available', 'bibliographicCitation', 'conformsTo',
'contributor', 'coverage', 'created', 'creator', 'date',
'dateAccepted', 'dateCopyrighted', 'dateSubmitted',
'description', 'educationLevel', 'extent', 'format',
'hasFormat', 'hasPart', 'hasVersion', 'identifier',
'instructionalMethod', 'isFormatOf', 'isPartOf',
'isReferencedBy', 'isReplacedBy', 'isRequiredBy', 'issued',
'isVersionOf', 'language', 'license', 'mediator', 'medium',
'modified', 'provenance', 'publisher', 'references',
'relation', 'replaces', 'requires', 'rights', 'rightsHolder',
'source', 'spatial', 'subject', 'tableOfContents', 'temporal',
'title', 'type', 'valid')
httpretty.register_uri(
httpretty.GET,
'http://content.cdlib.org/oai',
body=open(DIR_FIXTURES+'/oai-with-dcterms.xml').read())
set_fetcher = fetcher.OAIFetcher('http://content.cdlib.org/oai',
'set=bogus')
self.assertEqual(set_fetcher._set, 'bogus')
rec = set_fetcher.next()
self.assertIsInstance(rec, dict)
self.assertIn('id', rec)
self.assertEqual(rec['id'],
'oai:digitalcollections.usfca.edu:p264101coll6/188')
for term in DCTERMS: # dependent on data fed to sickle
self.assertIn(term, rec.keys())
self.assertEqual(rec['created'], ['TEST dcterms:created'])
self.assertEqual(rec['date'], ['1927; ', 'TEST dcterms:date'])