def test_uris_recognize_highwire_pdf(self):
doc = Document(
{'link': [{
'href': 'pdf-uri',
'type': 'application/pdf'
}]},
claimant='http://example.com')
expected = [
DocumentURI({
'claimant': 'http://example.com',
'uri': 'pdf-uri',
'type': 'highwire-pdf',
'content_type': 'application/pdf',
'created': None,
'updated': None
}),
DocumentURI({
'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': None,
'updated': None
})
]
assert sorted(doc.document_uris) == sorted(expected)
def test_uris_prefix_type_when_rel(self):
doc = Document(
{'link': [{
'href': 'https://example.com',
'rel': 'canonical'
}]},
claimant='http://example.com')
expected = [
DocumentURI({
'claimant': 'http://example.com',
'uri': 'https://example.com',
'type': 'rel-canonical',
'content_type': None,
'created': None,
'updated': None
}),
DocumentURI({
'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': None,
'updated': None
})
]
assert sorted(doc.document_uris) == sorted(expected)
def test_uris_only_one_self_claim(self):
doc = Document({'link': [{'href': 'http://example.com'}]},
claimant='http://example.com')
expected = [DocumentURI({'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': None, 'updated': None})]
assert doc.document_uris == expected
def test_uris_str_link(self):
doc = Document({'link': 'http://example.com'},
claimant='http://example.com',
created=datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
updated=datetime.datetime(2016, 2, 25, 16, 45, 23, 371849))
expected = [DocumentURI({'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
'updated': datetime.datetime(2016, 2, 25, 16, 45, 23, 371849)})]
assert doc.document_uris == expected
def test_uris_discard_self_claim_when_claimant_is_missing(self):
doc = Document({'link': [{'href': 'http://example.com'}]})
expected = [
DocumentURI({
'claimant': None,
'uri': 'http://example.com',
'type': None,
'content_type': None,
'created': None,
'updated': None
})
]
assert doc.document_uris == expected
def test_meta(self):
doc = Document(
{
'og': {
'title': ['Example Page'],
'url': ['http://example.com']
},
'title': ['Example Page'],
'link': ['http://example.com', 'https://example.com']
},
claimant='http://example.com',
created=datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
updated=datetime.datetime(2016, 2, 25, 16, 45, 23, 371849))
expected = [
DocumentMeta({
'type':
'og.title',
'value': ['Example Page'],
'claimant':
'http://example.com',
'created':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
'updated':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371849)
}),
DocumentMeta({
'type':
'og.url',
'value': ['http://example.com'],
'claimant':
'http://example.com',
'created':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
'updated':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371849)
}),
DocumentMeta({
'type':
'title',
'value': ['Example Page'],
'claimant':
'http://example.com',
'created':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
'updated':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371849)
})
]
assert sorted(doc.meta) == sorted(expected)
def test_updated(self):
doc = Document({},
updated=datetime.datetime(2016, 2, 25, 16, 45, 23,
371848))
assert doc.updated == datetime.datetime(2016, 2, 25, 16, 45, 23,
371848)
def test_uris_disregard_doi_links(self):
doc = Document({'link': [{'href': 'doi:foobar'}]})
assert len(doc.document_uris) == 0
def test_title_empty_array(self):
doc = Document({'title': []})
assert doc.title is None
def test_title(self):
doc = Document({'title': 'Example Page'})
assert doc.title == 'Example Page'
def test_title_array(self):
doc = Document({'title': ['Example Page']})
assert doc.title == 'Example Page'
def test_init_id(self):
"""It passes through other keyword arguments through to super class."""
doc = Document({}, id='id-value')
assert doc['id'] == 'id-value'
def test_uris_disregard_doi_links(self):
doc = Document({'link': [{'href': 'doi:foobar'}]})
# it always includes a self-claim, not removing doi links would result
# in a length of 2
assert len(doc.document_uris) == 1
def test_init_updated(self):
doc = Document({}, updated='updated-value')
assert doc.updated == 'updated-value'
def test_init_created(self):
doc = Document({}, created='created-value')
assert doc.created == 'created-value'
def test_init_claimant(self):
doc = Document({}, claimant='http://example.com')
assert doc.claimant == 'http://example.com'
def test_init(self):
doc = Document({'foo': 'bar'})
assert doc == {'foo': 'bar'}
class TestDocument(object):
def test_init(self):
doc = Document({'foo': 'bar'})
assert doc == {'foo': 'bar'}
def test_init_claimant(self):
doc = Document({}, claimant='http://example.com')
assert doc.claimant == 'http://example.com'
def test_init_created(self):
doc = Document({}, created='created-value')
assert doc.created == 'created-value'
def test_init_updated(self):
doc = Document({}, updated='updated-value')
assert doc.updated == 'updated-value'
def test_init_id(self):
"""It passes through other keyword arguments through to super class."""
doc = Document({}, id='id-value')
assert doc['id'] == 'id-value'
def test_created(self):
doc = Document({},
created=datetime.datetime(2016, 2, 25, 16, 45, 23,
371848))
assert doc.created == datetime.datetime(2016, 2, 25, 16, 45, 23,
371848)
def test_updated(self):
doc = Document({},
updated=datetime.datetime(2016, 2, 25, 16, 45, 23,
371848))
assert doc.updated == datetime.datetime(2016, 2, 25, 16, 45, 23,
371848)
def test_title(self):
doc = Document({'title': 'Example Page'})
assert doc.title == 'Example Page'
def test_title_array(self):
doc = Document({'title': ['Example Page']})
assert doc.title == 'Example Page'
def test_title_empty_array(self):
doc = Document({'title': []})
assert doc.title is None
def test_meta(self):
doc = Document(
{
'og': {
'title': ['Example Page'],
'url': ['http://example.com']
},
'title': ['Example Page'],
'link': ['http://example.com', 'https://example.com']
},
claimant='http://example.com',
created=datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
updated=datetime.datetime(2016, 2, 25, 16, 45, 23, 371849))
expected = [
DocumentMeta({
'type':
'og.title',
'value': ['Example Page'],
'claimant':
'http://example.com',
'created':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
'updated':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371849)
}),
DocumentMeta({
'type':
'og.url',
'value': ['http://example.com'],
'claimant':
'http://example.com',
'created':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
'updated':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371849)
}),
DocumentMeta({
'type':
'title',
'value': ['Example Page'],
'claimant':
'http://example.com',
'created':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
'updated':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371849)
})
]
assert sorted(doc.meta) == sorted(expected)
def test_uris_only_one_self_claim(self):
doc = Document({'link': [{
'href': 'http://example.com'
}]},
claimant='http://example.com')
expected = [
DocumentURI({
'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': None,
'updated': None
})
]
assert doc.document_uris == expected
def test_uris_discard_self_claim_when_claimant_is_missing(self):
doc = Document({'link': [{'href': 'http://example.com'}]})
expected = [
DocumentURI({
'claimant': None,
'uri': 'http://example.com',
'type': None,
'content_type': None,
'created': None,
'updated': None
})
]
assert doc.document_uris == expected
def test_uris_disregard_doi_links(self):
doc = Document({'link': [{'href': 'doi:foobar'}]})
assert len(doc.document_uris) == 0
def test_uris_str_link(self):
doc = Document({'link': 'http://example.com'},
claimant='http://example.com',
created=datetime.datetime(2016, 2, 25, 16, 45, 23,
371848),
updated=datetime.datetime(2016, 2, 25, 16, 45, 23,
371849))
expected = [
DocumentURI({
'claimant':
'http://example.com',
'uri':
'http://example.com',
'type':
'self-claim',
'created':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371848),
'updated':
datetime.datetime(2016, 2, 25, 16, 45, 23, 371849)
})
]
assert doc.document_uris == expected
def test_uris_recognize_highwire_pdf(self):
doc = Document(
{'link': [{
'href': 'pdf-uri',
'type': 'application/pdf'
}]},
claimant='http://example.com')
expected = [
DocumentURI({
'claimant': 'http://example.com',
'uri': 'pdf-uri',
'type': 'highwire-pdf',
'content_type': 'application/pdf',
'created': None,
'updated': None
}),
DocumentURI({
'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': None,
'updated': None
})
]
assert sorted(doc.document_uris) == sorted(expected)
def test_uris_prefix_type_when_rel(self):
doc = Document(
{'link': [{
'href': 'https://example.com',
'rel': 'canonical'
}]},
claimant='http://example.com')
expected = [
DocumentURI({
'claimant': 'http://example.com',
'uri': 'https://example.com',
'type': 'rel-canonical',
'content_type': None,
'created': None,
'updated': None
}),
DocumentURI({
'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': None,
'updated': None
})
]
assert sorted(doc.document_uris) == sorted(expected)
@pytest.mark.parametrize('doc', [
Document({'highwire': {
'doi': ['foobar']
}},
claimant='http://example.com'),
Document({'highwire': {
'doi': ['doi:foobar']
}},
claimant='http://example.com')
])
def test_uris_generates_doi_uri_from_highwire_meta(self, doc):
expected = [
DocumentURI({
'claimant': 'http://example.com',
'uri': 'doi:foobar',
'type': 'highwire-doi',
'created': None,
'updated': None
}),
DocumentURI({
'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': None,
'updated': None
})
]
assert sorted(doc.document_uris) == sorted(expected)
@pytest.mark.parametrize('doc', [
Document({'dc': {
'identifier': ['foobar']
}},
claimant='http://example.com'),
Document({'dc': {
'identifier': ['doi:foobar']
}},
claimant='http://example.com')
])
def test_uris_generates_doi_uri_from_dc_meta(self, doc):
expected = [
DocumentURI({
'claimant': 'http://example.com',
'uri': 'doi:foobar',
'type': 'dc-doi',
'created': None,
'updated': None
}),
DocumentURI({
'claimant': 'http://example.com',
'uri': 'http://example.com',
'type': 'self-claim',
'created': None,
'updated': None
})
]
assert sorted(doc.document_uris) == sorted(expected)