def parse_transfers_from_messages(
spine_messages: Iterable[Message],
time_range: DateTimeRange) -> Iterator[Transfer]:
conversations = group_into_conversations(spine_messages)
parsed_conversations = _parse_conversations(conversations)
conversations_started_in_range = filter_conversations_by_request_started_time(
parsed_conversations, time_range)
transfers = derive_transfers(conversations_started_in_range)
return transfers
def test_group_into_conversations_sorts_messages_within_conversations():
message_one = build_message(conversation_id="abc", time=datetime(year=2020, month=6, day=6))
message_two = build_message(conversation_id="abc", time=datetime(year=2020, month=6, day=5))
messages = [message_one, message_two]
expected = [Conversation("abc", [message_two, message_one])]
actual = group_into_conversations(messages)
assert list(actual) == expected
def test_group_into_conversations_produces_correct_conversations():
message_one = build_message(conversation_id="abc")
message_two = build_message(conversation_id="xyz")
messages = [message_one, message_two]
expected = [Conversation("abc", [message_one]), Conversation("xyz", [message_two])]
actual = group_into_conversations(messages)
assert list(actual) == expected
first_message.time)
def filter_for_full_conversations(conversations, date_range):
return (c for c in conversations
if conversation_is_started_in(c, date_range))
def message_code(message, requester):
party = "R" if message.from_party_asid == requester else "S"
interaction = abbreviations[message.interaction_id]
error = str(message.error_code) if message.error_code is not None else ""
return party + ":" + interaction + "[" + error + "]"
def extract_pattern(conversation):
requester = conversation.messages[0].from_party_asid
return "-".join(
(message_code(m, requester) for m in conversation.messages))
spine_messages = read_spine_csv_gz_files(input_files)
gp2gp_conversations = group_into_conversations(spine_messages)
full_conversations = filter_for_full_conversations(gp2gp_conversations,
date_range)
counts = Counter((extract_pattern(c) for c in full_conversations))
with open("./gp2gp-patterns-jan-21.csv", 'w') as f:
for pattern, count in counts.most_common():
print(pattern, count)