def delete(cls, record, account, public_id=None):
"""
Record a "delete" request, with the associated (minimal) record data, requested by the given account
Delete requests need only come with enough record data to identify the public record
:param record: The mimimal record data for identifying the record
:param account: the account to carry out the request on behalf of
:param public_id: the public id of a public record for which this is a delete request
:return: the Request object that was created
"""
if record is None:
raise RequestAPIException("You can't call 'delete' with a NoneType record argument")
if account is None:
raise RequestAPIException("You can't call 'delete' with a NoneType account argument")
req = Request()
req.record = record
req.owner = account.id
req.action = "delete"
if public_id is not None:
req.public_id = public_id
req.save()
return req
def test_05_request_dao(self):
# Check the DAO methods on the Request object
dao = Request()
source = RequestFixtureFactory.example()
req = Request(source)
req.owner = "test1"
req.action = "update"
req.public_id = "abcdefg"
req.save(blocking=True)
req2 = dao.pull(req.id)
assert req2 is not None
# check successful queries for identifiers
res = dao.find_by_identifier("doi", "10.1234/me", "test1")
assert len(res) == 1
res = dao.find_by_identifier("pmcid", "PMC1234", "test1")
assert len(res) == 1
res = dao.find_by_identifier("pmid", "87654321", "test1")
assert len(res) == 1
res = dao.find_by_identifier("url", "http://example.com/whatever", "test1")
assert len(res) == 1
# check unsuccessful ones
res = dao.find_by_identifier("doi", "10.1234/you", "test1")
assert len(res) == 0
res = dao.find_by_identifier("pmcid", "PMC5678", "test1")
assert len(res) == 0
res = dao.find_by_identifier("pmid", "123456789", "test1")
assert len(res) == 0
res = dao.find_by_identifier("url", "http://example.com/this", "test1")
assert len(res) == 0
# and check using the wrong owner
res = dao.find_by_identifier("doi", "10.1234/me", "test2")
assert len(res) == 0
res = dao.find_by_identifier("pmcid", "PMC1234", "test2")
assert len(res) == 0
res = dao.find_by_identifier("pmid", "87654321", "test2")
assert len(res) == 0
res = dao.find_by_identifier("url", "http://example.com/whatever", "test2")
assert len(res) == 0
def test_03_pull_request(self):
# Pull a Request through the ApiRequest object
acc = MonitorUKAccount()
acc.id = "abcdefghij"
acc.save(blocking=True)
# first make a request which contains that doi
req_source = RequestFixtureFactory.example()
req = Request(req_source)
req.save(blocking=True)
# you can't pull a request object, so just show that that's true...
# pull by doi should fail
result = ApiRequest.pull("10.1234/me", account=acc)
assert result is None
# pull by request id should fail
result = ApiRequest.pull(req.id, account=acc)
assert result is None
def test_10_find_request(self):
# Find a Request through a number of routes
source = RequestFixtureFactory.example()
req = Request(source)
req.save(blocking=True)
time.sleep(2)
source = RequestFixtureFactory.example()
req1 = Request(source)
req1.save(blocking=True)
# document to form the basis of the queries
source2 = RequestFixtureFactory.example()
# create sources with one of each kind of identifier, then look them up using the
# find_request_by_identifier method
result = RequestApi.find_request_by_identifier("doi", "10.1234/me", "abcdefghij")
assert result is not None
assert result.created_date == req1.created_date
result = RequestApi.find_request_by_identifier("pmid", "87654321", "abcdefghij")
assert result is not None
assert result.created_date == req1.created_date
result = RequestApi.find_request_by_identifier("pmcid", "PMC1234", "abcdefghij")
assert result is not None
assert result.created_date == req1.created_date
result = RequestApi.find_request_by_identifier("url", "http://example.com/whatever", "abcdefghij")
assert result is not None
assert result.created_date == req1.created_date
# finally, ensure that you don't get a match when you shouldn't
result = RequestApi.find_request_by_identifier("doi", "10.1234/another", "abcdefghij")
assert result is None
result = RequestApi.find_request_by_identifier("doi", "10.1234/me", "test")
assert result is None
def update(cls, record, account, public_id=None):
"""
Record an "update" request, with the associated record data, requested by the given account
:param record: The raw dict data (e.g. coming from the web API) which makes up the body of the request
:param account: the account to carry out the request on behalf of
:param public_id: The public id of a public record for which this is an update
:return: the Request object that was created
"""
if record is None:
raise RequestAPIException("You can't call 'update' with a NoneType record argument")
if account is None:
raise RequestAPIException("You can't call 'update' with a NoneType account argument")
req = Request()
req.record = record
req.owner = account.id
req.action = "update"
if public_id is not None:
req.public_id = public_id
req.save()
return req
def test_13_request_iterator(self):
# Check we can iterate successfully over lists of Requests
sources = RequestFixtureFactory.request_per_day("2001-01", 10)
for s in sources:
req = Request(s)
req.save()
time.sleep(2)
dao = Request()
gen = dao.list_all_since("2001-01-01T00:00:00Z", page_size=5) # set the page size small, to ensure the iterator has to work
results = [x for x in gen]
assert len(results) == 10
dates = [r.created_date for r in results]
comp = deepcopy(dates)
comp.sort() # this puts the dates in ascending order (i.e. oldest first)
# the point of this comparison is to show that the results came out in the right order.
# that is, oldest first
assert dates == comp
def test_11_process_requests_exception(self):
# What happens when the process_reuests method fails for a variety of reasons
sources = RequestFixtureFactory.request_per_day("2001-01", 9)
dois = ["10.1234/first", "10.1234/second", "10.1234/third"]
# we're going to construct a series of requests for each doi
# starting with a create, then an update, followed by a delete
# (not that it matters, as we're going to pump them through a mock)
for i in range(len(sources)):
s = sources[i]
doi_idx = i % 3 # iterate over the dois 3 times
doi = dois[doi_idx]
s["record"]["dc:identifier"] = [{"type": "doi", "id": doi}]
if i < 3:
s["record"]["dc:title"] = "Create"
req = Request(s)
req.action = "update"
req.save()
elif i < 6:
s["record"]["dc:title"] = "Update"
req = Request(s)
req.action = "update"
req.save()
else:
s["record"]["dc:title"] = "Delete"
req = Request(s)
req.action = "delete"
req.save()
time.sleep(2)
# set up the mocks
PublicApi.publish = publish_mock
PublicApi.remove = delete_mock
# now run the process job back to the first day
with self.assertRaises(TestException):
WorkflowApi.process_requests()
# we know this died during the 6th update request being processed,
# so just check that the workflow state reflects that
wfs_dao = WorkflowState()
wfs = wfs_dao.pull("requests")
assert wfs.last_request == "2001-01-05T00:00:00Z"
assert len(wfs.already_processed) == 1
def test_11_process_requests_cycle(self):
# Run through the process of processing a Request into a PublicAPC
source = RequestFixtureFactory.example()
if "id" in source:
del source["id"]
pub_dao = PublicAPC()
wfs_dao = WorkflowState()
# first make a record for the first time
first = deepcopy(source)
del first["record"]["dc:title"]
req = Request(first)
req.owner = "test"
req.action = "update"
req.save(blocking=True)
# run the job
WorkflowApi.process_requests()
time.sleep(2)
# first check that a public record was made
pubs = pub_dao.find_by_doi("10.1234/me")
assert len(pubs) == 1
assert pubs[0].record.get("dc:title") is None
# check that the workflow state was created
wfs = wfs_dao.pull("requests")
assert wfs is not None
assert wfs.last_request == req.created_date
assert wfs.already_processed == [req.id]
# now run an update with a different date
second = deepcopy(source)
second["record"]["dc:title"] = "Update"
second["created_date"] = "2002-01-01T00:00:00Z"
req2 = Request(second)
req2.owner = "test"
req2.action = "update"
req2.save(blocking=True)
# run the job again
WorkflowApi.process_requests()
time.sleep(2)
# check the public record was updated
pubs = pub_dao.find_by_doi("10.1234/me")
assert len(pubs) == 1
assert pubs[0].record.get("dc:title") == "Update"
# check that the workflow state was updated
wfs = wfs_dao.pull("requests")
assert wfs is not None
assert wfs.last_request == req2.created_date
assert wfs.already_processed == [req2.id]
# now run an update with the same date, to observe the difference in the workflow state
third = deepcopy(source)
third["record"]["dc:title"] = "Update 2"
third["created_date"] = "2002-01-01T00:00:00Z"
req3 = Request(third)
req3.owner = "test"
req3.action = "update"
req3.save(blocking=True)
# run the job again
WorkflowApi.process_requests()
time.sleep(2)
# check the public record was updated
pubs = pub_dao.find_by_doi("10.1234/me")
assert len(pubs) == 1
assert (
pubs[0].record.get("dc:title") == "Update 2"
) # should have been updated, as there are only apc contributions from one source
# check that the workflow state was updated
wfs = wfs_dao.pull("requests")
assert wfs is not None
assert wfs.last_request == req3.created_date
assert wfs.already_processed == [req2.id, req3.id] # processed records should have been appended
# finally issue a delete request
fourth = deepcopy(source)
fourth["created_date"] = "2003-01-01T00:00:00Z"
req4 = Request(fourth)
req4.owner = "test"
req4.action = "delete"
req4.save(blocking=True)
# run the job again
WorkflowApi.process_requests()
time.sleep(2)
# check the public record was updated
pubs = pub_dao.find_by_doi("10.1234/me")
assert len(pubs) == 0
# check that the workflow state was updated
wfs = wfs_dao.pull("requests")
assert wfs is not None
assert wfs.last_request == req4.created_date
assert wfs.already_processed == [req4.id] # processed records should have been appended