def manual_import(args):
if args.dry_run:
MDB.set_storage_mode("runtime")
files = []
for file in args.files:
if file.is_file():
files.append(file)
elif file.is_dir():
for sub_file in list(file.iterdir()):
if sub_file.is_file():
files.append(sub_file)
for file in files:
logger.info("reading \"{}\" now...".format(file.as_posix()))
transcripts = list()
if file.suffix.lower() == ".json":
logger.info("reading json based transcript file now...")
file_content = read_transcripts_json_file(file)
else:
logger.info("reading xml based transcript file now...")
file_content = [read_transcript_xml_file(file)]
logger.info("extracting communication model now...".format(file.as_posix()))
for metadata, inter_candidates in file_content:
transcript = Transcript.from_interactions(
metadata=metadata,
interactions=extract_communication_model(
candidates=inter_candidates,
add_debug_objects=args.add_debug_objects))
# insert into DB
if not args.dry_run:
transcript_dict = transcript.dict(exclude_none=True, exclude_unset=True)
logger.info(f"writing transcript with '{len(transcript_dict['interactions'])}' interactions into db.")
database.update_one("session", {"session_id": transcript.session_no}, transcript_dict)
transcripts.append(transcript)
# notify sentiment group
if args.notify and transcript:
utils.notify_sentiment_analysis_group([str(transcript.session_no)])
cm = CommunicationModel(transcripts=transcripts)
if args.dry_run:
out_file: Path = file.with_suffix(".converted.json")
logger.info("writing transcripts into {}.".format(out_file.absolute().as_posix()))
with open(out_file, "w", encoding="utf-8") as o:
o.write(cm.json(exclude_none=True, indent=4, ensure_ascii=False))
with open(out_file.parent / "mdb.json", "w", encoding="utf-8") as o:
safe_json_dump(MDB._mdb_runtime_storage, o)
import unittest
from datetime import datetime
from cme.domain import InteractionCandidate, MDB, Faction
from cme.extraction import extract_communication_model
MDB.set_storage_mode("runtime")
def _build_candidate(comment: str) -> InteractionCandidate:
return InteractionCandidate(speaker=MDB.find_and_add_in_storage(
forename="Likey",
surname="McUnittest",
memberships=[(datetime.min, None, Faction.NONE)]),
paragraph="Unittest",
comment=comment)
class TestExtraction(unittest.TestCase):
def test_extract_sample1(self):
comment = "(Beifall bei der FDP sowie bei Abgeordneten der CDU/CSU, der SPD und des BÜNDNISSES 90/DIE GRÜNEN – Dr. Eberhardt Alexander Gauland [AfD]: Ha, ha, ha!)"
cm = extract_communication_model([_build_candidate(comment)])
interaction_0 = cm[0]
interaction_1 = cm[1]
interaction_2 = cm[2]
interaction_3 = cm[3]
interaction_4 = cm[4]
self.assertEqual(interaction_0.sender, Faction.FDP)
self.assertEqual(
interaction_0.message,