def test_data_file_not_indexed(create_alias, create_index, get_index_uuid,
get_index_hash, client, pg_driver, admin,
submitter, cgci_blgsp, monkeypatch):
"""
Test node and data file creation when neither exist and no ID is provided.
"""
monkeypatch.setitem(flask.current_app.config, 'USE_SIGNPOST', True)
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
get_index_uuid.return_value = None
get_index_hash.return_value = None
# signpost create will return a document
document = MagicMock()
document.did = '14fd1746-61bb-401a-96d2-342cfaf70000'
create_index.return_value = document
resp = submit_metadata_file(client, pg_driver, admin, submitter,
cgci_blgsp)
# index creation should be called with no args for SIGNPOST
assert create_index.call_count == 1
args, kwargs = create_index.call_args_list[0]
assert not args
assert not kwargs
# no support for aliases
assert not create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity['action'] == 'create'
def test_data_file_update_url(
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
cgci_blgsp,
):
"""
Test submitting the same data again but updating the URL field (should
get added to the indexed file in index service).
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
document = MagicMock()
document.did = "14fd1746-61bb-401a-96d2-342cfaf70000"
document.urls = [DEFAULT_URL]
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
submit_metadata_file(client, pg_driver, admin, submitter, cgci_blgsp)
# now submit again but change url
new_url = "some/new/url/location/to/add"
updated_file = copy.deepcopy(DEFAULT_METADATA_FILE)
updated_file["urls"] = new_url
resp = submit_metadata_file(client,
pg_driver,
admin,
submitter,
cgci_blgsp,
data=updated_file)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# make sure new url are in the document and patch gets called
assert document.urls == [new_url]
assert document.patch.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "update"
assert entity["id"] != document.did
def test_data_file_update_multiple_urls(create_alias, create_index,
get_index_uuid, get_index_hash, client,
pg_driver, admin, submitter,
cgci_blgsp):
"""
Test submitting the same data again but updating the URL field (should
get added to the indexed file in index service).
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
document = MagicMock()
document.did = '14fd1746-61bb-401a-96d2-342cfaf70000'
document.urls = [DEFAULT_URL]
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
submit_metadata_file(client, pg_driver, admin, submitter, cgci_blgsp)
# now submit again but change url
new_url = 'some/new/url/location/to/add'
another_new_url = 'some/other/url'
updated_file = copy.deepcopy(DEFAULT_METADATA_FILE)
# comma separated list of urls INCLUDING the url that's already there
updated_file['urls'] = DEFAULT_URL + ',' + new_url + ',' + another_new_url
resp = submit_metadata_file(client,
pg_driver,
admin,
submitter,
cgci_blgsp,
data=updated_file)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# make sure original url and new url are in the document and patch gets called
assert DEFAULT_URL in document.urls
assert new_url in document.urls
assert another_new_url in document.urls
assert document.patch.called
# make sure no duplicates were added
assert len(document.urls) == 3
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity['action'] == 'update'
def test_data_file_update_url_id_provided(create_alias, create_index,
get_index_uuid, get_index_hash,
client, pg_driver, admin, submitter,
cgci_blgsp):
"""
Test submitting the same data again (with the id provided) and updating the
URL field (should get added to the indexed file in index service).
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
document = MagicMock()
document.did = '14fd1746-61bb-401a-96d2-342cfaf70000'
document.urls = [DEFAULT_URL]
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
submit_metadata_file(client, pg_driver, admin, submitter, cgci_blgsp)
# now submit again but change url
new_url = 'some/new/url/location/to/add'
updated_file = copy.deepcopy(DEFAULT_METADATA_FILE)
updated_file['urls'] = new_url
updated_file['id'] = document.did
resp = submit_metadata_file(client,
pg_driver,
admin,
submitter,
cgci_blgsp,
data=updated_file)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# make sure original url and new url are in the document and patch gets called
assert DEFAULT_URL in document.urls
assert new_url in document.urls
assert document.patch.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity['action'] == 'update'
# make sure uuid in node is the same as the uuid from index
# FIXME this is a temporary solution so these tests will probably
# need to change in the future
assert entity['id'] == document.did
def test_data_file_already_indexed_id_provided(
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
cgci_blgsp,
):
"""
Test submitting when the file is already indexed in the index client and
an id is provided in the submission.
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
document = MagicMock()
document.did = "14fd1746-61bb-401a-96d2-342cfaf70000"
get_index_uuid.return_value = document
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
file = copy.deepcopy(DEFAULT_METADATA_FILE)
file["id"] = document.did
resp = submit_metadata_file(client,
pg_driver,
admin,
submitter,
cgci_blgsp,
data=file)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "create"
# make sure uuid in node is the same as the uuid from index
# FIXME this is a temporary solution so these tests will probably
# need to change in the future
assert entity["id"] == document.did
def test_data_file_not_indexed_id_provided(
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
cgci_blgsp,
require_index_exists_off,
):
"""
Test node and data file creation when neither exist and an ID is provided.
That ID should be used for the node and file index creation
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
get_index_uuid.return_value = None
get_index_hash.return_value = None
file = copy.deepcopy(DEFAULT_METADATA_FILE)
file["id"] = DEFAULT_UUID
resp = submit_metadata_file(client,
pg_driver,
admin,
submitter,
cgci_blgsp,
data=file)
# index creation
assert create_index.call_count == 1
args, kwargs = create_index.call_args_list[0]
assert "did" in kwargs
did = kwargs["did"]
assert "hashes" in kwargs
assert kwargs["hashes"].get("md5") == DEFAULT_FILE_HASH
assert "urls" in kwargs
assert DEFAULT_URL in kwargs["urls"]
# alias creation
assert create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "create"
# make sure uuid in node is the same as the uuid from index
# FIXME this is a temporary solution so these tests will probably
# need to change in the future
assert did == DEFAULT_UUID
assert entity["id"] == DEFAULT_UUID
def test_can_submit_data_file_with_asterisk_json(
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
cgci_blgsp,
):
"""
Test that we can submit a file when some fields have asterisks prepended
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
document = MagicMock()
document.did = "14fd1746-61bb-401a-96d2-342cfaf70000"
get_index_uuid.return_value = document
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
file = copy.deepcopy(DEFAULT_METADATA_FILE)
file["id"] = document.did
# insert asterisks before the property names
for key in file.keys():
file["*{}".format(key)] = file.pop(key)
resp = submit_metadata_file(client,
pg_driver,
admin,
submitter,
cgci_blgsp,
data=file)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "create"
def test_link_case_insensitivity(
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
cgci_blgsp,
):
"""
Test that links are case insensitive.
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
for i in range(16):
document = MagicMock()
document.did = "14fd1746-61bb-401a-96d2-342cfaf7000" + str(i)
document.urls = [DEFAULT_URL]
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
updated_file = copy.deepcopy(DEFAULT_METADATA_FILE)
updated_file["submitter_id"] = str(i)
updated_file["experiments"]["submitter_id"] = "".join(
random.choice([k.upper(), k.lower()])
for k in updated_file["experiments"]["submitter_id"])
resp = submit_metadata_file(client,
pg_driver,
admin,
submitter,
cgci_blgsp,
data=updated_file)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "create"
def test_data_file_already_indexed(
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
cgci_blgsp,
):
"""
Test submitting when the file is already indexed in the index client and
no ID is provided. sheepdog should fall back on the hash/size of the file
to find it in indexing service.
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
document = MagicMock()
document.did = "14fd1746-61bb-401a-96d2-342cfaf70000"
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
resp = submit_metadata_file(client, pg_driver, admin, submitter,
cgci_blgsp)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "create"
path = "/v0/submission/CGCI/BLGSP/export/?format=json&ids={nid}".format(
nid=entity["id"])
r = client.get(path, headers=submitter)
data = r.json
assert len(data) == 1
assert data[0]["object_id"] == document.did
assert data[0]["id"] != document.did
def test_data_file_not_indexed(
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
cgci_blgsp,
require_index_exists_off,
):
"""
Test node and data file creation when neither exist and no ID is provided.
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
get_index_uuid.return_value = None
get_index_hash.return_value = None
resp = submit_metadata_file(client, pg_driver, admin, submitter,
cgci_blgsp)
# index creation
assert create_index.call_count == 1
_, kwargs = create_index.call_args_list[0]
assert "did" in kwargs
did = kwargs["did"]
assert "hashes" in kwargs
assert kwargs["hashes"].get("md5") == DEFAULT_FILE_HASH
assert "urls" in kwargs
assert DEFAULT_URL in kwargs["urls"]
# alias creation
assert create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "create"
path = "/v0/submission/CGCI/BLGSP/export/?format=json&ids={nid}".format(
nid=entity["id"])
r = client.get(path, headers=submitter)
data = r.json
assert len(data) == 1
assert did == None
def test_data_file_already_indexed(
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
cgci_blgsp,
monkeypatch,
):
"""
Test submitting when the file is already indexed in the index client and
no ID is provided. sheepdog should fall back on the hash/size of the file
to find it in indexing service.
NOTE: signpostclient will create another file in the index service regardless
of whether or not the file exists. signpostclient does not
have capabilities of searching for files based on hash/size
"""
monkeypatch.setitem(flask.current_app.config, "USE_SIGNPOST", True)
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
# signpostclient cannot find by hash/size
get_index_hash.return_value = None
get_index_uuid.return_value = None
# signpost create will return a document
document = MagicMock()
document.did = "14fd1746-61bb-401a-96d2-342cfaf70000"
create_index.return_value = document
resp = submit_metadata_file(client, pg_driver, admin, submitter, cgci_blgsp)
# index creation should be called with no args for SIGNPOST
assert create_index.call_count == 1
args, kwargs = create_index.call_args_list[0]
assert not args
assert not kwargs
# no support for aliases
assert not create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "create"
def test_data_file_already_indexed(create_alias, create_index, get_index_uuid,
get_index_hash, client, pg_driver, admin,
submitter, cgci_blgsp):
"""
Test submitting when the file is already indexed in the index client and
no ID is provided. sheepdog should fall back on the hash/size of the file
to find it in indexing service.
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
document = MagicMock()
document.did = '14fd1746-61bb-401a-96d2-342cfaf70000'
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
resp = submit_metadata_file(client, pg_driver, admin, submitter,
cgci_blgsp)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity['action'] == 'create'
# make sure uuid in node is the same as the uuid from index
# FIXME this is a temporary solution so these tests will probably
# need to change in the future
assert entity['id'] == document.did
def test_data_file_not_indexed(
create_alias, create_index, get_index_uuid, get_index_hash,
client, pg_driver, admin, submitter, cgci_blgsp,
require_index_exists_off):
"""
Test node and data file creation when neither exist and no ID is provided.
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
get_index_uuid.return_value = None
get_index_hash.return_value = None
resp = submit_metadata_file(client, pg_driver, admin, submitter, cgci_blgsp)
# index creation
assert create_index.call_count == 1
_, kwargs = create_index.call_args_list[0]
assert 'did' in kwargs
did = kwargs['did']
assert 'hashes' in kwargs
assert kwargs['hashes'].get('md5') == DEFAULT_FILE_HASH
assert 'urls' in kwargs
assert DEFAULT_URL in kwargs['urls']
# alias creation
assert create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity['action'] == 'create'
# make sure uuid in node is the same as the uuid from index
# FIXME this is a temporary solution so these tests will probably
# need to change in the future
assert entity['id'] == did
def test_data_file_already_indexed_object_id_provided_hash_match(
update_acl_uploader_indexd,
create_alias,
create_index,
get_index_uuid,
get_index_hash,
client,
pg_driver,
admin,
submitter,
submitter_name,
cgci_blgsp,
):
"""
Test submitting when the file is already indexed in the index client,
an id is provided in the submission, and the size and hash match those
of the indexed file. The empty acl means the file was just uploaded.
The submission should succeed but no new index should be created. The acl
and uploader field should have been updated as part of the data upload
flow.
"""
submit_first_experiment(client, pg_driver, admin, submitter, cgci_blgsp)
file = copy.deepcopy(DEFAULT_METADATA_FILE)
# provide the object_id of an existing indexed file
file["object_id"] = "14fd1746-61bb-401a-96d2-342cfaf70000"
document = MagicMock()
document.did = file["object_id"]
document.size = file["file_size"]
document.hashes = {"md5": file["md5sum"], "other": "abc123"}
document.acl = []
document.uploader = submitter_name
get_index_uuid.return_value = document
get_index_hash.return_value = document
# only return the correct document by uuid IF the uuid provided is
# the one from above
def get_index_by_uuid(uuid):
if uuid == document.did:
return document
else:
return None
get_index_uuid.side_effect = get_index_by_uuid
resp = submit_metadata_file(client,
pg_driver,
admin,
submitter,
cgci_blgsp,
data=file)
# no index or alias creation
assert not create_index.called
assert not create_alias.called
# response
assert_positive_response(resp)
entity = assert_single_entity_from_response(resp)
assert entity["action"] == "create"
path = "/v0/submission/CGCI/BLGSP/export/?format=json&ids={nid}".format(
nid=entity["id"])
r = client.get(path, headers=submitter)
data = r.json
assert len(data) == 1
assert data[0]["object_id"] == file["object_id"]
# check that the acl and uploader fields have been updated in indexd
assert update_acl_uploader_indexd.called