def flask_predict():
try:
content = get_json(request)
if 'query' not in content:
return jsonerror('Invalid JSON object')
elif 'model_id' not in content:
return jsonerror('Model ID must be provided')
get_config(app)
model_count = len(list(app.config['MODELS'].keys())) \
if 'MODELS' in app.config \
else 0
if model_count == 0:
return jsonerror('No model trained')
model_id = content['model_id'].strip()
if model_id not in app.config['MODELS']:
return jsonerror('Model ID not found')
model_config = app.config['MODELS'][model_id]
nlu_init_model(
model_id,
model_config['CLF_MODEL_PATH'],
model_config['ENT_MODEL_PATH']
)
contexts = content['contexts'] if 'contexts' in content else None
result = nlu_predict(model_id, content['query'], contexts)
return jsonify(result)
except:
traceback.print_exc(limit=2, file=sys.stdout)
return jsonerror('Runtime exception encountered when handling request')
def flask_get_status():
try:
content = get_json(request)
if 'model_id' not in content:
return jsonerror('Model ID must be provided')
get_config(app)
model_id = content['model_id']
if model_id not in app.config['MODELS']:
return jsonerror('Model ID not found')
model_config = app.config['MODELS'][model_id]
if model_id in TRAIN_PROCESSES:
process = TRAIN_PROCESSES[model_id]
return_code = process.poll()
if return_code is None:
return jsonify({'status': 'training'})
elif return_code < 0:
return jsonify({'status': 'failed', 'error_code': return_code})
if path.exists(model_config['CLF_MODEL_PATH']):
return jsonify({'status': 'completed'})
else:
return jsonify({'status': 'not_found'})
except:
traceback.print_exc(limit=2, file=sys.stdout)
return jsonerror('Runtime exception encountered when handling request')
def flask_delete_model():
try:
content = get_json(request)
if 'model_id' not in content:
return jsonerror('Missing model_id argument in request')
else:
model_id = content['model_id']
get_config(app)
all_models = app.config['MODELS']
if model_id not in all_models:
return jsonerror('Model ID not found')
else:
model_config = all_models[model_id]
if path.exists(model_config['CLF_MODEL_PATH']):
remove(model_config['CLF_MODEL_PATH'])
if path.exists(model_config['ENT_MODEL_PATH']):
remove(model_config['ENT_MODEL_PATH'])
del all_models[model_id]
app.config['MODELS'] = all_models
save_config(app)
return jsonify({'model_id': model_id})
except:
traceback.print_exc(limit=2, file=sys.stdout)
return jsonerror('Runtime exception encountered when handling request')
def flask_visualize():
try:
content = get_json(request)
if 'items' not in content:
return jsonerror('Invalid JSON object')
get_config(app)
model_id = None
if 'model_id' in content:
model_id = content['model_id'].strip()
if model_id not in app.config['MODELS']:
return jsonerror('Model ID not found')
model_config = app.config['MODELS'][model_id]
nlu_init_model(model_id, model_config['CLF_MODEL_PATH'],
model_config['ENT_MODEL_PATH'])
n_clusters = content.get('n_clusters', None)
result = nlu_visualize(content.get('items', []),
model_id=model_id,
n_clusters=n_clusters)
return jsonify(result)
except:
traceback.print_exc(limit=2, file=sys.stdout)
return jsonerror('Runtime exception encountered when handling request')
def upload():
try:
if 'file' not in request.files:
return jsonerror('No file included')
u_file = request.files['file']
if u_file.filename == '':
return jsonerror('No selected file')
if not u_file or not allowed_file(u_file.filename):
return jsonerror('Invalid file')
request_form = request.form
callback_url = request_form.get('callback_url', '')
if 'model_id' in request_form:
prev_model_id = request_form['model_id']
if prev_model_id in TRAIN_PROCESSES:
process = TRAIN_PROCESSES[prev_model_id]
return_code = process.poll()
if return_code is None: # process is still running
process.kill() # immediately kill the process
# This deletes the model before training is complete so...
app.logger.info('Deleting previous model', prev_model_id)
# delete_model(app, prev_model_id)
model_id = str(uuid.uuid4())
filename = model_id + '_' + secure_filename(u_file.filename)
save_path = path.join(app.config['UPLOAD_FOLDER'], filename)
u_file.save(save_path)
app.logger.info('Upload complete. Beginning training model', model_id)
clf_model_path = save_path + '.cls.bin'
ent_model_path = save_path + '.ent.bin'
get_config(app)
if not app.config.get('USE_QUEUE', True):
clf_model_path, ent_model_path = nlu_train_file(
model_id, save_path, clf_model_path, ent_model_path)
else:
log_file_name = path.join(app.config['LOGS_FOLDER'],
model_id + '.log')
with open(log_file_name, 'w', encoding='utf8') as log_fp:
TRAIN_PROCESSES[model_id] = subprocess.Popen([
PYTHON_PATH, '-m', 'flask_app.nlu_train', '--model_id',
model_id, '--save_path', save_path, '--clf_model_path',
clf_model_path, '--ent_model_path', ent_model_path,
'--callback_url', callback_url
],
stdout=log_fp)
if app.config.get('MODELS', None) is None:
app.config['MODELS'] = {}
app.config['MODELS'][model_id] = {
'CLF_MODEL_PATH': clf_model_path,
'ENT_MODEL_PATH': ent_model_path,
'created': time.time()
}
save_config(app)
return jsonify({'model_id': model_id})
except:
traceback.print_exc(limit=2, file=sys.stdout)
return jsonerror('Runtime exception encountered when handling request')