def test_calling_document_uris_from_dc_when_no_dc(self, document_uris_from_dc):
document_data = {} # No 'dc' key.
claimant = "http://localhost:5000/docs/help"
document_claims.document_uris_from_data(
document_data=document_data, claimant=claimant
)
document_uris_from_dc.assert_called_once_with({}, claimant)
def test_calling_document_uris_from_highwire_doi_when_no_highwire(
self, document_uris_from_highwire_doi
):
document_data = {} # No 'highwire' key.
claimant = "http://localhost:5000/docs/help"
document_claims.document_uris_from_data(
document_data=document_data, claimant=claimant
)
document_uris_from_highwire_doi.assert_called_once_with({}, claimant)
def test_calling_document_uris_from_links_when_no_links(
self, document_uris_from_links
):
document_data = {} # No 'link' key.
claimant = "http://localhost:5000/docs/help"
document_claims.document_uris_from_data(
document_data=document_data, claimant=claimant
)
document_uris_from_links.assert_called_once_with([], claimant)
def test_it_gets_self_claim_document_uris(self, document_uri_self_claim):
claimant = "http://example.com/claimant"
document_uris = document_claims.document_uris_from_data({}, claimant)
document_uri_self_claim.assert_called_once_with(claimant)
assert document_uri_self_claim.return_value in document_uris
def test_it_gets_document_uris_from_links(self, document_uris_from_links):
document_data = {
"link": [
# In production these would be link dicts not strings.
"link_dict_1",
"link_dict_2",
"link_dict_3",
]
}
claimant = "http://localhost:5000/docs/help"
document_uris_from_links.return_value = [
{"uri": "uri_1"},
{"uri": "uri_2"},
{"uri": "uri_3"},
]
document_uris = document_claims.document_uris_from_data(
document_data=document_data, claimant=claimant
)
document_uris_from_links.assert_called_once_with(
document_data["link"], claimant
)
for document_uri in document_uris_from_links.return_value:
assert document_uri in document_uris
def test_it_ignores_whitespace_only_self_claim_uris(self, document_uri_self_claim):
for uri in (" ", "\n ", "\r\n", " \t"):
document_uri_self_claim.return_value = {"uri": uri}
document_uris = document_claims.document_uris_from_data(
{}, "http://example.com/claimant"
)
assert document_uris == []
def test_it_strips_whitespace_from_self_claim_uris(
self, document_uris_from_links, document_uri_self_claim
):
for uri in (" self_claim", "self_claim ", " self_claim "):
document_uris_from_links.return_value = []
document_uri_self_claim.return_value = {"uri": uri}
document_uris = document_claims.document_uris_from_data(
{}, "http://example.com/claimant"
)
assert document_uris == [{"uri": uri.strip()}]
def test_it_strips_whitespace_from_uris(
self,
document_uris_from_links,
document_uris_from_highwire_pdf,
document_uris_from_highwire_doi,
document_uris_from_dc,
document_uri_self_claim,
matchers,
):
document_uris_from_links.return_value = [
{"uri": " from_link_1"},
{"uri": "from_link_2 "},
{"uri": " from_link_3 "},
]
document_uris_from_highwire_pdf.return_value = [
{"uri": " highwire_1"},
{"uri": "highwire_2 "},
{"uri": " highwire_3 "},
]
document_uris_from_highwire_doi.return_value = [
{"uri": " doi_1"},
{"uri": "doi_2 "},
{"uri": " doi_3 "},
]
document_uris_from_dc.return_value = [
{"uri": " dc_1"},
{"uri": "dc_2 "},
{"uri": " dc_3 "},
]
document_uris = document_claims.document_uris_from_data(
{}, "http://example.com/claimant"
)
assert document_uris == matchers.UnorderedList(
[
{"uri": "from_link_1"},
{"uri": "from_link_2"},
{"uri": "from_link_3"},
{"uri": "highwire_1"},
{"uri": "highwire_2"},
{"uri": "highwire_3"},
{"uri": "doi_1"},
{"uri": "doi_2"},
{"uri": "doi_3"},
{"uri": "dc_1"},
{"uri": "dc_2"},
{"uri": "dc_3"},
document_uri_self_claim.return_value,
]
)
def test_it_gets_documents_uris_from_dc(self, document_uris_from_dc):
document_data = {"dc": {"identifier": ["doi_1", "doi_2", "doi_3"]}}
claimant = "http://localhost:5000/docs/help"
document_uris_from_dc.return_value = [
{"uri": "uri_1"},
{"uri": "uri_2"},
{"uri": "uri_3"},
]
document_uris = document_claims.document_uris_from_data(
document_data=document_data, claimant=claimant
)
document_uris_from_dc.assert_called_once_with(document_data["dc"], claimant)
for document_uri in document_uris_from_dc.return_value:
assert document_uri in document_uris
def _document(document, claimant):
"""
Return document meta and document URI data from the given document dict.
Transforms the "document" dict that the client posts into a convenient
format for creating DocumentURI and DocumentMeta objects later.
"""
document = document or {}
document_uri_dicts = document_claims.document_uris_from_data(
copy.deepcopy(document), claimant=claimant)
document_meta_dicts = document_claims.document_metas_from_data(
copy.deepcopy(document), claimant=claimant)
return {
"document_uri_dicts": document_uri_dicts,
"document_meta_dicts": document_meta_dicts,
}
def test_it_ignores_whitespace_only_uris(
self,
document_uris_from_links,
document_uris_from_highwire_pdf,
document_uris_from_highwire_doi,
document_uris_from_dc,
document_uri_self_claim,
):
uris = [" ", "\n ", "\r\n", " \t"]
document_uris_from_links.return_value = [{"uri": u} for u in uris]
document_uris_from_highwire_pdf.return_value = [{"uri": u} for u in uris]
document_uris_from_highwire_doi.return_value = [{"uri": u} for u in uris]
document_uris_from_dc.return_value = [{"uri": u} for u in uris]
document_uris = document_claims.document_uris_from_data(
{}, "http://example.com/claimant"
)
assert document_uris == [document_uri_self_claim.return_value]
def test_it_ignores_empty_string_uris(
self,
document_uris_from_links,
document_uris_from_highwire_pdf,
document_uris_from_highwire_doi,
document_uris_from_dc,
document_uri_self_claim,
):
document_uris_from_links.return_value = [{"uri": ""}]
document_uris_from_highwire_pdf.return_value = [{"uri": ""}]
document_uris_from_highwire_doi.return_value = [{"uri": ""}]
document_uris_from_dc.return_value = [{"uri": ""}]
document_uri_self_claim.return_value = {"uri": ""}
document_uris = document_claims.document_uris_from_data(
{}, "http://example.com/claimant"
)
assert document_uris == []
def test_it_gets_documents_uris_from_highwire_pdf(
self, document_uris_from_highwire_pdf
):
document_data = {"highwire": {"pdf": ["pdf_1", "pdf_2", "pdf_3"]}}
claimant = "http://localhost:5000/docs/help"
document_uris_from_highwire_pdf.return_value = [
{"uri": "uri_1"},
{"uri": "uri_2"},
{"uri": "uri_3"},
]
document_uris = document_claims.document_uris_from_data(
document_data=document_data, claimant=claimant
)
document_uris_from_highwire_pdf.assert_called_once_with(
document_data["highwire"], claimant
)
for document_uri in document_uris_from_highwire_pdf.return_value:
assert document_uri in document_uris