def test_harvesting_arxiv_workflow_manual_accepted(
mocked_refextract_extract_refs,
mocked_matching_match,
mocked_api_request_magpie,
mocked_api_request_beard,
mocked_download_utils,
mocked_download_arxiv,
mocked_package_download,
workflow_app,
mocked_external_services,
):
record = generate_record()
"""Test a full harvesting workflow."""
workflow_uuid, eng, obj = get_halted_workflow(app=workflow_app, record=record)
do_accept_core(app=workflow_app, workflow_id=obj.id)
eng = WorkflowEngine.from_uuid(workflow_uuid)
obj = eng.processed_objects[0]
assert obj.status == ObjectStatus.WAITING
do_robotupload_callback(app=workflow_app, workflow_id=obj.id, recids=[12345])
eng = WorkflowEngine.from_uuid(workflow_uuid)
obj = eng.processed_objects[0]
# It was accepted
assert obj.status == ObjectStatus.COMPLETED
assert obj.extra_data["approved"] is True
def test_harvesting_arxiv_workflow_manual_accepted(
mocked_refextract_extract_refs,
mocked_matching_match,
mocked_api_request_magpie,
mocked_api_request_beard,
mocked_download_utils,
mocked_download_arxiv,
mocked_package_download,
workflow_app,
mocked_external_services,
):
record = generate_record()
"""Test a full harvesting workflow."""
workflow_uuid, eng, obj = get_halted_workflow(app=workflow_app, record=record)
do_accept_core(app=workflow_app, workflow_id=obj.id)
eng = WorkflowEngine.from_uuid(workflow_uuid)
obj = eng.processed_objects[0]
assert obj.status == ObjectStatus.WAITING
do_robotupload_callback(app=workflow_app, workflow_id=obj.id, recids=[12345])
obj = workflow_object_class.get(obj.id)
assert obj.status == ObjectStatus.WAITING
do_webcoll_callback(app=workflow_app, recids=[12345])
eng = WorkflowEngine.from_uuid(workflow_uuid)
obj = eng.processed_objects[0]
# It was accepted
assert obj.status == ObjectStatus.COMPLETED
assert obj.extra_data["approved"] is True
def test_harvesting_arxiv_workflow_manual_accepted(
mocked_refextract_extract_refs,
mocked_matching_search,
mocked_api_request_magpie,
mocked_api_request_beard,
mocked_download_utils,
mocked_download_arxiv,
workflow_app,
):
record = generate_record()
"""Test a full harvesting workflow."""
with requests_mock.Mocker() as requests_mocker:
requests_mocker.register_uri(
requests_mock.ANY,
re.compile('.*(indexer|localhost).*'),
real_http=True,
)
requests_mocker.register_uri(
'POST',
re.compile('https?://localhost:1234.*', ),
text=u'[INFO]',
status_code=200,
)
workflow_uuid, eng, obj = get_halted_workflow(
app=workflow_app,
extra_config={'PRODUCTION_MODE': False},
record=record,
)
do_accept_core(
app=workflow_app,
workflow_id=obj.id,
)
eng = WorkflowEngine.from_uuid(workflow_uuid)
obj = eng.processed_objects[0]
assert obj.status == ObjectStatus.WAITING
response = do_robotupload_callback(
app=workflow_app,
workflow_id=obj.id,
recids=[12345],
)
assert response.status_code == 200
obj = workflow_object_class.get(obj.id)
assert obj.status == ObjectStatus.WAITING
response = do_webcoll_callback(app=workflow_app, recids=[12345])
assert response.status_code == 200
eng = WorkflowEngine.from_uuid(workflow_uuid)
obj = eng.processed_objects[0]
# It was accepted
assert obj.status == ObjectStatus.COMPLETED
def test_harvesting_arxiv_workflow_manual_accepted(
mocked_refextract_extract_refs,
mocked_matching_match,
mocked_api_request_magpie,
mocked_api_request_beard,
mocked_download_utils,
mocked_download_arxiv,
workflow_app,
mocked_external_services,
):
record = generate_record()
"""Test a full harvesting workflow."""
workflow_uuid, eng, obj = get_halted_workflow(
app=workflow_app,
record=record,
)
do_accept_core(
app=workflow_app,
workflow_id=obj.id,
)
eng = WorkflowEngine.from_uuid(workflow_uuid)
obj = eng.processed_objects[0]
assert obj.status == ObjectStatus.WAITING
response = do_robotupload_callback(
app=workflow_app,
workflow_id=obj.id,
recids=[12345],
)
assert response.status_code == 200
obj = workflow_object_class.get(obj.id)
assert obj.status == ObjectStatus.WAITING
response = do_webcoll_callback(app=workflow_app, recids=[12345])
assert response.status_code == 200
eng = WorkflowEngine.from_uuid(workflow_uuid)
obj = eng.processed_objects[0]
# It was accepted
assert obj.status == ObjectStatus.COMPLETED
def test_edit_article_workflow_sending_to_hep(workflow_app,
mocked_external_services):
app_client = workflow_app.test_client()
user = User.query.filter_by(email='*****@*****.**').one()
login_user_via_session(app_client, user=user)
record = {
'$schema':
'http://localhost:5000/schemas/records/hep.json',
'arxiv_eprints': [{
'categories': ['nucl-th'],
'value': '1802.03287'
}],
'control_number':
123,
'document_type': ['article'],
'titles': [{
'title':
'Resource Pooling in Large-Scale Content Delivery Systems'
}],
'self': {
'$ref': 'http://localhost:5000/schemas/records/hep.json'
},
'_collections': ['Literature']
}
factory = TestRecordMetadata.create_from_kwargs(json=record)
with patch.dict(
workflow_app.config, {
'FEATURE_FLAG_ENABLE_REST_RECORD_MANAGEMENT': True,
'INSPIREHEP_URL': "http://web:8000"
}):
with requests_mock.Mocker() as requests_mocker:
requests_mocker.register_uri(
'PUT',
'{url}/literature/{cn}'.format(
url=workflow_app.config.get("INSPIREHEP_URL"),
cn=record['control_number'],
),
headers={'content-type': 'application/json'},
status_code=200,
json={
"metadata": {
'control_number': record['control_number'],
},
'id_': 1
})
eng_uuid = start('edit_article', data=factory.record_metadata.json)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING
assert obj.extra_data['callback_url']
user_id = user.get_id()
obj.id_user = user_id
# simulate changes in the editor and save
new_title = 'Somebody edited this fancy title'
obj.data['titles'][0]['title'] = new_title
payload = {
'id': obj.id,
'metadata': obj.data,
'_extra_data': obj.extra_data
}
app_client.put(obj.extra_data['callback_url'],
data=json.dumps(payload),
content_type='application/json')
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING # waiting for robot_upload
assert obj.data['titles'][0]['title'] == new_title
do_robotupload_callback(
app=workflow_app,
workflow_id=obj.id,
recids=[obj.data['control_number']],
)
record = get_db_record('lit', 123)
assert record['titles'][0]['title'] == new_title
# assert record edit transaction is by user who created the edit workflow
revision = record.revisions[record.revision_id]
transaction_id = revision.model.transaction_id
Transaction = transaction_class(RecordMetadata)
transaction = Transaction.query.filter(
Transaction.id == transaction_id).one()
assert transaction.user_id == int(user_id)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.COMPLETED
pending_records = WorkflowsPendingRecord.query.filter_by(
workflow_id=obj.id).all()
assert not pending_records
def test_edit_article_workflow_deleting(workflow_app,
mocked_external_services):
app_client = workflow_app.test_client()
user = User.query.filter_by(email='*****@*****.**').one()
login_user_via_session(app_client, user=user)
record = {
'$schema':
'http://localhost:5000/schemas/records/hep.json',
'arxiv_eprints': [{
'categories': ['nucl-th'],
'value': '1802.03287'
}],
'control_number':
123,
'document_type': ['article'],
'titles': [{
'title':
'Resource Pooling in Large-Scale Content Delivery Systems'
}],
'self': {
'$ref': 'http://localhost:5000/schemas/records/hep.json'
},
'_collections': ['Literature']
}
factory = TestRecordMetadata.create_from_kwargs(json=record)
eng_uuid = start('edit_article', data=factory.record_metadata.json)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
obj.id_user = user.get_id()
assert obj.status == ObjectStatus.WAITING
assert obj.extra_data['callback_url']
record = get_db_record('lit', 123)
search = LiteratureSearch()
search.get_source(record.id)
# simulate changes in the editor and save
obj.data['deleted'] = True
payload = {
'id': obj.id,
'metadata': obj.data,
'_extra_data': obj.extra_data
}
app_client.put(obj.extra_data['callback_url'],
data=json.dumps(payload),
content_type='application/json')
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING # waiting for robot_upload
assert obj.data['deleted'] is True
do_robotupload_callback(
app=workflow_app,
workflow_id=obj.id,
recids=[obj.data['control_number']],
)
record = get_db_record('lit', 123)
assert record['deleted'] is True
with pytest.raises(NotFoundError):
search.get_source(record.id)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.COMPLETED
pending_records = WorkflowsPendingRecord.query.filter_by(
workflow_id=obj.id).all()
assert not pending_records
def test_edit_article_workflow(workflow_app, mocked_external_services):
app_client = workflow_app.test_client()
user = User.query.filter_by(email='*****@*****.**').one()
login_user_via_session(app_client, user=user)
record = {
'$schema':
'http://localhost:5000/schemas/records/hep.json',
'arxiv_eprints': [{
'categories': ['nucl-th'],
'value': '1802.03287'
}],
'control_number':
123,
'document_type': ['article'],
'titles': [{
'title':
'Resource Pooling in Large-Scale Content Delivery Systems'
}],
'self': {
'$ref': 'http://localhost:5000/schemas/records/hep.json'
},
'_collections': ['Literature']
}
factory = TestRecordMetadata.create_from_kwargs(json=record)
eng_uuid = start('edit_article', data=factory.record_metadata.json)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING
assert obj.extra_data['callback_url']
user_id = user.get_id()
obj.id_user = user_id
# simulate changes in the editor and save
new_title = 'Somebody edited this fancy title'
obj.data['titles'][0]['title'] = new_title
payload = {
'id': obj.id,
'metadata': obj.data,
'_extra_data': obj.extra_data
}
app_client.put(obj.extra_data['callback_url'],
data=json.dumps(payload),
content_type='application/json')
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING # waiting for robot_upload
assert obj.data['titles'][0]['title'] == new_title
do_robotupload_callback(
app=workflow_app,
workflow_id=obj.id,
recids=[obj.data['control_number']],
)
record = get_db_record('lit', 123)
assert record['titles'][0]['title'] == new_title
# assert record edit transaction is by user who created the edit workflow
revision = record.revisions[record.revision_id]
transaction_id = revision.model.transaction_id
Transaction = transaction_class(RecordMetadata)
transaction = Transaction.query.filter(
Transaction.id == transaction_id).one()
assert transaction.user_id == int(user_id)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.COMPLETED
pending_records = WorkflowsPendingRecord.query.filter_by(
workflow_id=obj.id).all()
assert not pending_records
def test_edit_article_workflow_deleting(workflow_app, mocked_external_services):
app_client = workflow_app.test_client()
user = User.query.filter_by(email='*****@*****.**').one()
login_user_via_session(app_client, user=user)
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'arxiv_eprints': [
{
'categories': [
'nucl-th'
],
'value': '1802.03287'
}
],
'control_number': 123,
'document_type': ['article'],
'titles': [{'title': 'Resource Pooling in Large-Scale Content Delivery Systems'}],
'self': {'$ref': 'http://localhost:5000/schemas/records/hep.json'},
'_collections': ['Literature']
}
factory = TestRecordMetadata.create_from_kwargs(json=record)
eng_uuid = start('edit_article', data=factory.record_metadata.json)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
obj.id_user = user.get_id()
assert obj.status == ObjectStatus.WAITING
assert obj.extra_data['callback_url']
record = get_db_record('lit', 123)
search = LiteratureSearch()
search.get_source(record.id)
# simulate changes in the editor and save
obj.data['deleted'] = True
payload = {
'id': obj.id,
'metadata': obj.data,
'_extra_data': obj.extra_data
}
app_client.put(
obj.extra_data['callback_url'],
data=json.dumps(payload),
content_type='application/json'
)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING # waiting for robot_upload
assert obj.data['deleted'] is True
do_robotupload_callback(
app=workflow_app,
workflow_id=obj.id,
recids=[obj.data['control_number']],
)
record = get_db_record('lit', 123)
assert record['deleted'] is True
with pytest.raises(NotFoundError):
search.get_source(record.id)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.COMPLETED
pending_records = WorkflowsPendingRecord.query.filter_by(workflow_id=obj.id).all()
assert not pending_records
def test_edit_article_workflow(workflow_app, mocked_external_services):
app_client = workflow_app.test_client()
user = User.query.filter_by(email='*****@*****.**').one()
login_user_via_session(app_client, user=user)
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'arxiv_eprints': [
{
'categories': [
'nucl-th'
],
'value': '1802.03287'
}
],
'control_number': 123,
'document_type': ['article'],
'titles': [{'title': 'Resource Pooling in Large-Scale Content Delivery Systems'}],
'self': {'$ref': 'http://localhost:5000/schemas/records/hep.json'},
'_collections': ['Literature']
}
factory = TestRecordMetadata.create_from_kwargs(json=record)
eng_uuid = start('edit_article', data=factory.record_metadata.json)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING
assert obj.extra_data['callback_url']
user_id = user.get_id()
obj.id_user = user_id
# simulate changes in the editor and save
new_title = 'Somebody edited this fancy title'
obj.data['titles'][0]['title'] = new_title
payload = {
'id': obj.id,
'metadata': obj.data,
'_extra_data': obj.extra_data
}
app_client.put(
obj.extra_data['callback_url'],
data=json.dumps(payload),
content_type='application/json'
)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING # waiting for robot_upload
assert obj.data['titles'][0]['title'] == new_title
do_robotupload_callback(
app=workflow_app,
workflow_id=obj.id,
recids=[obj.data['control_number']],
)
record = get_db_record('lit', 123)
assert record['titles'][0]['title'] == new_title
# assert record edit transaction is by user who created the edit workflow
revision = record.revisions[record.revision_id]
transaction_id = revision.model.transaction_id
Transaction = transaction_class(RecordMetadata)
transaction = Transaction.query.filter(Transaction.id == transaction_id).one()
assert transaction.user_id == int(user_id)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.COMPLETED
pending_records = WorkflowsPendingRecord.query.filter_by(workflow_id=obj.id).all()
assert not pending_records
def test_edit_article_workflow_sending_to_hep(workflow_app, mocked_external_services):
app_client = workflow_app.test_client()
user = User.query.filter_by(email='*****@*****.**').one()
login_user_via_session(app_client, user=user)
record = {
'$schema': 'http://localhost:5000/schemas/records/hep.json',
'arxiv_eprints': [
{
'categories': [
'nucl-th'
],
'value': '1802.03287'
}
],
'control_number': 123,
'document_type': ['article'],
'titles': [{'title': 'Resource Pooling in Large-Scale Content Delivery Systems'}],
'self': {'$ref': 'http://localhost:5000/schemas/records/hep.json'},
'_collections': ['Literature']
}
factory = TestRecordMetadata.create_from_kwargs(json=record)
with patch.dict(workflow_app.config, {
'FEATURE_FLAG_ENABLE_REST_RECORD_MANAGEMENT': True,
'INSPIREHEP_URL': "http://web:8000"
}):
with requests_mock.Mocker() as requests_mocker:
requests_mocker.register_uri(
'PUT', '{url}/literature/{cn}'.format(
url=workflow_app.config.get("INSPIREHEP_URL"),
cn=record['control_number'],
),
headers={'content-type': 'application/json'},
status_code=200,
json={
"metadata": {
'control_number': record['control_number'],
},
'id_': 1
}
)
eng_uuid = start('edit_article', data=factory.record_metadata.json)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING
assert obj.extra_data['callback_url']
user_id = user.get_id()
obj.id_user = user_id
# simulate changes in the editor and save
new_title = 'Somebody edited this fancy title'
obj.data['titles'][0]['title'] = new_title
payload = {
'id': obj.id,
'metadata': obj.data,
'_extra_data': obj.extra_data
}
app_client.put(
obj.extra_data['callback_url'],
data=json.dumps(payload),
content_type='application/json'
)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.WAITING # waiting for robot_upload
assert obj.data['titles'][0]['title'] == new_title
do_robotupload_callback(
app=workflow_app,
workflow_id=obj.id,
recids=[obj.data['control_number']],
)
record = get_db_record('lit', 123)
assert record['titles'][0]['title'] == new_title
# assert record edit transaction is by user who created the edit workflow
revision = record.revisions[record.revision_id]
transaction_id = revision.model.transaction_id
Transaction = transaction_class(RecordMetadata)
transaction = Transaction.query.filter(Transaction.id == transaction_id).one()
assert transaction.user_id == int(user_id)
obj = WorkflowEngine.from_uuid(eng_uuid).objects[0]
assert obj.status == ObjectStatus.COMPLETED
pending_records = WorkflowsPendingRecord.query.filter_by(workflow_id=obj.id).all()
assert not pending_records
