def provider_dereference(self, record):
"""
Check the URL that the DOI dereferences to, by taking advantage of the fact that
DOI lookups use HTTP 303 to redirect you to the resource. Append to the record['provider']['url']
list with the string which describes the provider (ideally a URI)
"""
# check that we can actually work on this record
# - must have an indentifier
# - must be a doi
# - must have a canonical form
if not "identifier" in record:
return
if not "type" in record["identifier"]:
return
if record["identifier"]["type"] != "doi":
return
if not "canonical" in record["identifier"]:
return
# first construct a dereferenceable doi (prefix it with dx.doi.org)
canon = record['identifier']['canonical']
loc = self.dereference(canon)
# either way we are going to copy the doi into the provider object
recordmanager.record_provider_doi(record, canon)
if loc is None:
return
# if we find something, record it
recordmanager.record_provider_url(record, loc)
def test_01_record_provider_url(self):
record = {}
recordmanager.record_provider_url(record, "http://hello")
assert "provider" in record
assert "url" in record["provider"]
assert len(record["provider"]["url"]) == 1
assert record["provider"]["url"][0] == "http://hello"
def provider_resolver(self, record):
"""
Take a pubmed id (if that is the type) and obtain a reference to the base
URL of the resource that it links to and append it to the record['provider']['url'] list
"""
# check that we can actually work on this record
# - must have an indentifier
# - must be a pmid
# - must have a canonical form
if not "identifier" in record:
return
if not "type" in record["identifier"]:
return
if record["identifier"]["type"] != "pmid":
return
if not "canonical" in record["identifier"]:
return
# see if we can resolve a doi for the item
canon = record['identifier']['canonical']
doi, loc = self._resolve_doi(canon)
if loc is not None:
# if we find something, record it
recordmanager.record_provider_url(record, loc)
recordmanager.record_provider_doi(record, doi)
return
# if we get to here, the DOI lookup failed, so we need to scrape the NCBI site for possible urls
urls = self._scrape_urls(canon)
if urls is not None and len(urls) > 0:
# if we find something, record it
recordmanager.record_provider_urls(record, urls)
