async def test_trigger_data_importer_validate_exception(self, monkeypatch):
bot = 'test_events'
user = '******'
test_data_path = os.path.join(pytest.tmp_dir, str(datetime.utcnow()))
os.mkdir(test_data_path)
def _path(*args, **kwargs):
return test_data_path
monkeypatch.setattr(Utility, "get_latest_file", _path)
DataImporterLogProcessor.add_log(bot,
user,
files_received=REQUIREMENTS -
{"http_actions"})
await EventsTrigger.trigger_data_importer(bot, user, False, False)
logs = list(DataImporterLogProcessor.get_logs(bot))
assert len(logs) == 2
assert not logs[0].get('intents').get('data')
assert not logs[0].get('stories').get('data')
assert not logs[0].get('utterances').get('data')
assert not logs[0].get('http_actions').get('data')
assert not logs[0].get('training_examples').get('data')
assert not logs[0].get('domain').get('data')
assert not logs[0].get('config').get('data')
assert logs[0].get('exception') == 'Some training files are absent!'
assert logs[0]['is_data_uploaded']
assert logs[0]['start_timestamp']
assert logs[0]['end_timestamp']
assert logs[0]['status'] == 'Failure'
assert logs[0]['event_status'] == EVENT_STATUS.FAIL.value
async def test_trigger_data_importer_validate_file_with_errors(
self, monkeypatch):
bot = 'test_events'
user = '******'
test_data_path = os.path.join(pytest.tmp_dir, str(datetime.utcnow()))
shutil.copytree('tests/testing_data/validator/intent_name_mismatch',
test_data_path)
def _path(*args, **kwargs):
return test_data_path
monkeypatch.setattr(Utility, "get_latest_file", _path)
DataImporterLogProcessor.add_log(bot,
user,
files_received=REQUIREMENTS -
{"http_actions"})
await EventsTrigger.trigger_data_importer(bot, user, True, False)
logs = list(DataImporterLogProcessor.get_logs(bot))
assert len(logs) == 4
assert logs[0].get('intents').get('data')
assert not logs[0].get('stories').get('data')
assert not logs[0].get('utterances').get('data')
assert not logs[0].get('http_actions').get('data')
assert not logs[0].get('training_examples').get('data')
assert not logs[0].get('domain').get('data')
assert not logs[0].get('config').get('data')
assert not logs[0].get('exception')
assert logs[0]['is_data_uploaded']
assert logs[0]['start_timestamp']
assert logs[0]['end_timestamp']
assert logs[0]['status'] == 'Failure'
assert logs[0]['event_status'] == EVENT_STATUS.COMPLETED.value
def test_add_log_success(self):
bot = 'test'
user = '******'
DataImporterLogProcessor.add_log(bot,
user,
files_received=list(
REQUIREMENTS.copy()),
is_data_uploaded=False)
DataImporterLogProcessor.add_log(
bot,
user,
status='Success',
event_status=EVENT_STATUS.COMPLETED.value)
log = list(DataImporterLogProcessor.get_logs(bot))
assert not log[0].get('intents').get('data')
assert not log[0].get('stories').get('data')
assert not log[0].get('utterances').get('data')
assert not log[0].get('http_actions').get('data')
assert not log[0].get('training_examples').get('data')
assert not log[0].get('domain').get('data')
assert not log[0].get('config').get('data')
assert not log[0].get('exception')
assert not log[0]['is_data_uploaded']
assert log[0]['start_timestamp']
assert log[0].get('end_timestamp')
assert all(file in log[0]['files_received'] for file in REQUIREMENTS)
assert log[0].get('status') == 'Success'
assert log[0]['event_status'] == EVENT_STATUS.COMPLETED.value
def test_is_event_in_progress_true(self):
bot = 'test'
user = '******'
DataImporterLogProcessor.add_log(bot, user, is_data_uploaded=False)
assert DataImporterLogProcessor.is_event_in_progress(bot, False)
with pytest.raises(AppException):
DataImporterLogProcessor.is_event_in_progress(bot)
async def test_trigger_data_importer_rules_only(self, monkeypatch,
get_training_data):
bot = 'test_trigger_data_importer_rules_only'
user = '******'
test_data_path = os.path.join(pytest.tmp_dir, str(datetime.utcnow()))
data_path = os.path.join(test_data_path, 'data')
Utility.make_dirs(data_path)
shutil.copy2('tests/testing_data/validator/valid/data/rules.yml',
data_path)
nlu, story_graph, domain, config, http_actions = await get_training_data(
'tests/testing_data/validator/valid')
mongo_processor = MongoProcessor()
mongo_processor.save_domain(domain, bot, user)
mongo_processor.save_nlu(nlu, bot, user)
config["bot"] = bot
config["user"] = user
config_obj = Configs._from_son(config)
config_obj.save()
mongo_processor.save_stories(story_graph.story_steps, bot, user)
mongo_processor.save_http_action(http_actions, bot, user)
def _path(*args, **kwargs):
return test_data_path
monkeypatch.setattr(Utility, "get_latest_file", _path)
DataImporterLogProcessor.add_log(bot, user, files_received=["rules"])
await EventsTrigger.trigger_data_importer(bot, user, True, False)
logs = list(DataImporterLogProcessor.get_logs(bot))
assert len(logs) == 1
assert not logs[0].get('intents').get('data')
assert not logs[0].get('stories').get('data')
assert not logs[0].get('utterances').get('data')
assert not logs[0].get('http_actions').get('data')
assert not logs[0].get('training_examples').get('data')
assert not logs[0].get('domain').get('data')
assert not logs[0].get('config').get('data')
assert not logs[0].get('exception')
assert logs[0]['is_data_uploaded']
assert logs[0]['start_timestamp']
assert logs[0]['end_timestamp']
assert logs[0]['status'] == 'Success'
assert logs[0]['event_status'] == EVENT_STATUS.COMPLETED.value
assert len(mongo_processor.fetch_stories(bot)) == 2
assert len(list(mongo_processor.fetch_training_examples(bot))) == 7
assert len(list(mongo_processor.fetch_responses(bot))) == 2
assert len(mongo_processor.fetch_actions(bot)) == 2
assert len(mongo_processor.fetch_rule_block_names(bot)) == 3
def test_add_log(self):
bot = 'test'
user = '******'
DataImporterLogProcessor.add_log(bot, user, is_data_uploaded=False)
log = ValidationLogs.objects(bot=bot).get().to_mongo().to_dict()
assert not log.get('intents').get('data')
assert not log.get('stories').get('data')
assert not log.get('utterances').get('data')
assert not log.get('http_actions').get('data')
assert not log.get('training_examples').get('data')
assert not log.get('domain').get('data')
assert not log.get('config').get('data')
assert not log.get('exception')
assert not log['is_data_uploaded']
assert log['start_timestamp']
assert not log.get('end_timestamp')
assert not log.get('validation_status')
assert log['event_status'] == EVENT_STATUS.INITIATED.value
async def test_trigger_data_importer_validate_and_save_append(
self, monkeypatch):
bot = 'test_events'
user = '******'
test_data_path = os.path.join(pytest.tmp_dir, str(datetime.utcnow()))
shutil.copytree('tests/testing_data/validator/append', test_data_path)
def _path(*args, **kwargs):
return test_data_path
monkeypatch.setattr(Utility, "get_latest_file", _path)
DataImporterLogProcessor.add_log(bot,
user,
files_received=REQUIREMENTS -
{"http_actions", "rules"})
await EventsTrigger.trigger_data_importer(bot, user, True, False)
logs = list(DataImporterLogProcessor.get_logs(bot))
assert len(logs) == 6
assert not logs[0].get('intents').get('data')
assert not logs[0].get('stories').get('data')
assert not logs[0].get('utterances').get('data')
assert not logs[0].get('http_actions').get('data')
assert not logs[0].get('training_examples').get('data')
assert not logs[0].get('domain').get('data')
assert not logs[0].get('config').get('data')
assert not logs[0].get('exception')
assert logs[0]['is_data_uploaded']
assert logs[0]['start_timestamp']
assert logs[0]['end_timestamp']
assert logs[0]['status'] == 'Success'
assert logs[0]['event_status'] == EVENT_STATUS.COMPLETED.value
processor = MongoProcessor()
assert 'greet' in processor.fetch_intents(bot)
assert 'deny' in processor.fetch_intents(bot)
assert 'location' in processor.fetch_intents(bot)
assert 'affirm' in processor.fetch_intents(bot)
assert len(processor.fetch_stories(bot)) == 4
assert len(list(processor.fetch_training_examples(bot))) == 13
assert len(list(processor.fetch_responses(bot))) == 6
assert len(processor.fetch_actions(bot)) == 4
assert len(processor.fetch_rule_block_names(bot)) == 4
async def validate_training_data(
background_tasks: BackgroundTasks,
current_user: User = Depends(auth.get_current_user_and_bot),
):
"""
Validates bot training data.
"""
DataImporterLogProcessor.is_limit_exceeded(current_user.get_bot())
DataImporterLogProcessor.is_event_in_progress(current_user.get_bot())
Utility.make_dirs(
os.path.join("training_data", current_user.get_bot(),
str(datetime.utcnow())))
DataImporterLogProcessor.add_log(current_user.get_bot(),
current_user.get_user(),
is_data_uploaded=False)
background_tasks.add_task(EventsTrigger.trigger_data_importer,
current_user.get_bot(), current_user.get_user(),
False, False)
return {"message": "Event triggered! Check logs."}
def test_update_log(self):
bot = 'test'
user = '******'
DataImporterLogProcessor.add_log(bot, user, is_data_uploaded=False)
log = next(DataImporterLogProcessor.get_logs(bot))
assert not log.get('intents').get('data')
assert not log.get('stories').get('data')
assert not log.get('utterances').get('data')
assert not log.get('http_actions').get('data')
assert not log.get('training_examples').get('data')
assert not log.get('domain').get('data')
assert not log.get('config').get('data')
assert not log.get('exception')
assert not log['is_data_uploaded']
assert log['start_timestamp']
assert not log.get('end_timestamp')
assert not log.get('validation_status')
assert log['event_status'] == EVENT_STATUS.INITIATED.value
count = COMPONENT_COUNT.copy()
count['http_action'] = 6
count['domain']['intents'] = 12
summary = {
'intents': ['Intent not added to domain'],
'config': ['Invalid component']
}
DataImporterLogProcessor.update_summary(bot, user, count, summary,
'Failed', 'Completed')
log = next(DataImporterLogProcessor.get_logs(bot))
assert log.get('intents').get('data') == ['Intent not added to domain']
assert not log.get('stories').get('data')
assert not log.get('utterances').get('data')
assert not log.get('http_actions').get('data')
assert not log.get('training_examples').get('data')
assert not log.get('domain').get('data')
assert log.get('config').get('data') == ['Invalid component']
assert not log.get('exception')
assert not log['is_data_uploaded']
assert log['start_timestamp']
assert log.get('end_timestamp')
assert log.get('status') == 'Failed'
assert log['event_status'] == EVENT_STATUS.COMPLETED.value
async def test_trigger_data_importer_import_with_intent_issues(
self, monkeypatch):
bot = 'test_trigger_data_importer_import_with_intent_issues'
user = '******'
test_data_path = os.path.join(pytest.tmp_dir, str(datetime.utcnow()))
shutil.copytree('tests/testing_data/validator/intent_name_mismatch',
test_data_path)
def _path(*args, **kwargs):
return test_data_path
monkeypatch.setattr(Utility, "get_latest_file", _path)
BotSettings(ignore_utterances=True, bot=bot, user=user).save()
DataImporterLogProcessor.add_log(
bot, user, files_received=['nlu', 'stories', 'domain', 'config'])
await EventsTrigger.trigger_data_importer(bot, user, True, True)
logs = list(DataImporterLogProcessor.get_logs(bot))
assert len(logs) == 1
assert logs[0].get('intents').get('data')
assert not logs[0].get('stories').get('data')
assert not logs[0].get('utterances').get('data')
assert not logs[0].get('http_actions').get('data')
assert not logs[0].get('training_examples').get('data')
assert not logs[0].get('domain').get('data')
assert not logs[0].get('config').get('data')
assert not logs[0].get('exception')
assert logs[0]['is_data_uploaded']
assert logs[0]['start_timestamp']
assert logs[0]['end_timestamp']
assert logs[0]['status'] == 'Failure'
assert logs[0]['event_status'] == EVENT_STATUS.COMPLETED.value
mongo_processor = MongoProcessor()
assert len(mongo_processor.fetch_stories(bot)) == 0
assert len(list(mongo_processor.fetch_training_examples(bot))) == 0
assert len(list(mongo_processor.fetch_responses(bot))) == 0
assert len(mongo_processor.fetch_actions(bot)) == 0
assert len(mongo_processor.fetch_rule_block_names(bot)) == 0
def test_add_log_exception(self):
bot = 'test'
user = '******'
DataImporterLogProcessor.add_log(bot,
user,
exception='Validation failed',
status='Failure',
event_status=EVENT_STATUS.FAIL.value)
log = ValidationLogs.objects(bot=bot).get().to_mongo().to_dict()
assert not log.get('intents').get('data')
assert not log.get('stories').get('data')
assert not log.get('utterances').get('data')
assert not log.get('http_actions').get('data')
assert not log.get('training_examples').get('data')
assert not log.get('domain').get('data')
assert not log.get('config').get('data')
assert not log.get('files_received')
assert log.get('exception') == 'Validation failed'
assert not log['is_data_uploaded']
assert log['start_timestamp']
assert log.get('end_timestamp')
assert log.get('status') == 'Failure'
assert log['event_status'] == EVENT_STATUS.FAIL.value
async def test_trigger_data_importer_validate_existing_data(
self, monkeypatch, get_training_data):
bot = 'test_trigger_data_importer_domain_only'
user = '******'
test_data_path = os.path.join(pytest.tmp_dir, str(datetime.utcnow()))
Utility.make_dirs(test_data_path)
def _path(*args, **kwargs):
return test_data_path
monkeypatch.setattr(Utility, "get_latest_file", _path)
DataImporterLogProcessor.add_log(bot, user)
await EventsTrigger.trigger_data_importer(bot, user, True, False)
logs = list(DataImporterLogProcessor.get_logs(bot))
assert len(logs) == 2
assert not logs[0].get('intents').get('data')
assert not logs[0].get('stories').get('data')
assert not logs[0].get('utterances').get('data')
assert not logs[0].get('http_actions').get('data')
assert not logs[0].get('training_examples').get('data')
assert not logs[0].get('domain').get('data')
assert not logs[0].get('config').get('data')
assert not logs[0].get('exception')
assert logs[0]['is_data_uploaded']
assert logs[0]['start_timestamp']
assert logs[0]['end_timestamp']
assert logs[0]['status'] == 'Success'
assert logs[0]['event_status'] == EVENT_STATUS.COMPLETED.value
mongo_processor = MongoProcessor()
assert len(mongo_processor.fetch_stories(bot)) == 2
assert len(list(mongo_processor.fetch_training_examples(bot))) == 7
assert len(list(mongo_processor.fetch_responses(bot))) == 2
assert len(mongo_processor.fetch_actions(bot)) == 2
assert len(mongo_processor.fetch_rule_block_names(bot)) == 3
async def trigger_data_importer(bot: Text,
user: Text,
save_data: bool,
overwrite: bool = True):
"""
Triggers data importer event which validates and imports data into kairon.
@param bot: bot id.
@param user: kairon username.
@param save_data: Flag to import data into kairon. If set to false, then only validation is run.
Otherwise, both validation and import is done.
@param overwrite: Overwrite existing data(if set to true) or append (if set to false).
@return:
"""
validation_status = 'Failure'
path = None
try:
if Utility.get_event_url("DATA_IMPORTER"):
import_flag = '--import-data' if save_data else ''
overwrite_flag = '--overwrite' if overwrite else ''
env_var = {
'BOT': bot,
'USER': user,
"IMPORT_DATA": import_flag,
"OVERWRITE": overwrite_flag
}
event_request = Utility.build_event_request(env_var)
Utility.http_request(
"POST", Utility.environment['model']['data_importer'].get(
'event_url'), None, user, event_request)
DataImporterLogProcessor.add_log(
bot, user, event_status=EVENT_STATUS.TASKSPAWNED.value)
else:
path = Utility.get_latest_file(
os.path.join('training_data', bot))
files_received = DataImporterLogProcessor.get_files_received_for_latest_event(
bot)
DataImporterLogProcessor.add_log(
bot, user, event_status=EVENT_STATUS.PARSE.value)
data_importer = DataImporter(path, bot, user, files_received,
save_data, overwrite)
DataImporterLogProcessor.add_log(
bot, user, event_status=EVENT_STATUS.VALIDATING.value)
summary, component_count = await data_importer.validate()
is_data_valid = all(
[not summary[key] for key in summary.keys()])
validation_status = 'Success' if is_data_valid else 'Failure'
DataImporterLogProcessor.update_summary(
bot,
user,
component_count,
summary,
status=validation_status,
event_status=EVENT_STATUS.SAVE.value)
if is_data_valid:
data_importer.import_data()
DataImporterLogProcessor.add_log(
bot, user, event_status=EVENT_STATUS.COMPLETED.value)
except exceptions.ConnectionError as e:
logger.error(str(e))
DataImporterLogProcessor.add_log(
bot,
user,
exception='Failed to trigger the event.',
status=validation_status,
event_status=EVENT_STATUS.FAIL.value)
except Exception as e:
logger.error(str(e))
DataImporterLogProcessor.add_log(
bot,
user,
exception=str(e),
status=validation_status,
event_status=EVENT_STATUS.FAIL.value)
if path:
Utility.delete_directory(path)