def test_compute_data(init_drone: SimpleBibDrone):
"""
test whether data is extracted as expected
:param init_drone: un-connected simpleBibDrone instance
:return:
None
"""
list_record_id = init_drone.read(init_drone.path)
data = init_drone.process_item(list_record_id[0])
assert "citations" in data
assert "text" in data
def test_process_item(init_drone: SimpleBibDrone):
"""
Test whether data is expaneded correctly and whether meta data is added
:param init_drone: un-connected simpleBibDrone instance
:return:
None
"""
list_record_id = init_drone.read(init_drone.path)
data = init_drone.process_item(list_record_id[0])
assert "citations" in data
assert "text" in data
assert "record_key" in data
assert "last_updated" in data
assert "documents" in data
assert "state_hash" in data
def test_record_id(init_drone: SimpleBibDrone):
"""
Test validity of RecordIdentifier
:param init_drone: un-connected simpleBibDrone instance
:return:
None
"""
list_record_id = init_drone.read(init_drone.path)
record0 = list_record_id[0]
assert record0.parent_directory == init_drone.path
assert record0.last_updated < datetime.now()
assert len(record0.documents) > 0
# state hash does not change when the file is not changed
assert record0.compute_state_hash() == record0.state_hash
def test_read(init_drone: SimpleBibDrone):
"""
Test whether read function is correct
:param init_drone: un-connected simpleBibDrone instance
:return:
None
"""
list_record_id = init_drone.read(init_drone.path)
assert len(list_record_id) == 7
state_hashes = [r.state_hash for r in list_record_id]
assert len(state_hashes) == len(list_record_id) # all record_id has hash
assert len((set(state_hashes))) == len(state_hashes) # all unique hashes
num_docs = sum([len(r.documents) for r in list_record_id])
assert num_docs == 12
def test_compute_record_identifier_key(init_drone: SimpleBibDrone):
list_record_id = init_drone.read(init_drone.path)
record0 = list_record_id[0]
doc0 = record0.documents[0]
assert record0.record_key == init_drone.compute_record_identifier_key(doc0)