def query(cls, query_string, page=1, page_size=25):
quoted = quote(query_string, safe="/")
qpage = quote(str(page))
qsize = quote(str(page_size))
if qsize is None or qpage is None or quoted is None:
raise EuropePMCException(None, "unable to url escape the string")
url = app.config.get("EPMC_REST_API") + "search/query=" + query_string
url += "&resulttype=core&format=json&page=" + qpage + "&pageSize=" + qsize
app.logger.debug("Requesting EPMC metadata from " + url)
resp = http.get(url)
if resp is None:
raise EuropePMCException(
message="could not get a response from EPMC")
if resp.status_code != 200:
raise EuropePMCException(resp)
try:
j = resp.json()
except:
raise EuropePMCException(
message="could not decode JSON from EPMC response")
results = [
models.EPMCMetadata(r)
for r in j.get("resultList", {}).get("result", [])
]
return results
def test_06_epmc_compliance_data(self):
record = models.Record()
msg = workflow.WorkflowMessage(record=record)
data = json.loads(open(EPMC_MD, "r").read())
epmc_md = epmcmod.EPMCMetadata(data)
workflow.extract_metadata(msg, epmc_md)
assert record.in_epmc is True
assert record.is_oa is False
assert len(record.issn) == 1
assert "1471-2121" in record.issn
def test_05_populate_identifiers(self):
record = models.Record()
msg = workflow.WorkflowMessage(record=record)
data = json.loads(open(EPMC_MD, "r").read())
epmc_md = epmcmod.EPMCMetadata(data)
workflow.populate_identifiers(msg, epmc_md)
assert record.pmcid == "PMC4219345"
assert record.pmid == "24279897"
assert record.doi == "10.1186/1471-2121-14-52"
record.pmcid = "PMC000000"
record.pmid = "0000000"
del record.doi
workflow.populate_identifiers(msg, epmc_md)
assert record.pmcid == "PMC000000"
assert record.pmid == "0000000"
assert record.doi == "10.1186/1471-2121-14-52"
def field_search(cls, field, value, fuzzy=False, page=1):
wrap = "\"" if not fuzzy else ""
quoted = quote(value, safe="/")
qpage = quote(str(page))
if quoted is None or qpage is None:
raise EuropePMCException(None, "unable to url escape the string")
url = app.config.get("EPMC_REST_API") + "search/query=" + field + ":" + wrap + quoted + wrap
url += "&resultType=core&format=json&page=" + qpage
app.logger.debug("Requesting EPMC metadata from " + url)
resp = http.get(url)
if resp is None:
raise EuropePMCException(message="could not get a response from EPMC")
if resp.status_code != 200:
raise EuropePMCException(resp)
try:
j = resp.json()
except:
raise EuropePMCException(message="could not decode JSON from EPMC response")
results = [models.EPMCMetadata(r) for r in j.get("resultList", {}).get("result", [])]
return results
def mock_get_md(*args, **kwargs):
md = epmcmod.EPMCMetadata(json.loads(open(EPMC_MD, "r").read()))
return md, 1.0
