def run(train_file, label_file, flag):
examples = []
ex = Example(None, None, [])
cur_entity = Entity(0, 0, None, None)
text = ""
entities = []
with open(train_file, "r") as tr:
with open(label_file, "r") as lb:
for line in tr.readlines():
if line == "\n":
if cur_entity.to_add:
entities.append(cur_entity)
cur_label = lb.readline()
ex.text = text
ex.intent = transform_intent(cur_label.split("\n")[0])
for ent in entities:
ent.find_start_end(ex.text)
ex.entities.append(ent)
examples.append(ex)
ex = Example(None, None, [])
text = ""
entities = []
cur_entity = Entity(0, 0, None, None)
else:
word, tag = line.split("\t")
if text == "":
text = word
else:
text = text + " " + word
tag = tag.split("\n")[0]
if tag == "O":
if cur_entity.entity is not None:
entities.append(cur_entity)
cur_entity = Entity(0, 0, None, None)
elif tag.startswith("B"):
if cur_entity.entity is not None:
entities.append(cur_entity)
cur_entity = Entity(0, 0, word, tag[2:])
else:
cur_entity.entity = tag[2:]
cur_entity.value = word
cur_entity.to_add = True
elif tag.startswith("I"):
if cur_entity is not None:
if cur_entity.entity == tag[2:]:
cur_entity.value = cur_entity.value + " " + word
cur_entity.to_add = True
final_json = {
"rasa_nlu_data": {
"common_examples": [example.get_json() for example in examples],
"entity_examples": [],
"intent_examples": []
}
}
file_names = ["exact.json", "test.json"]
with open("basic_intents.json") as f:
data = json.load(f)
for el in data:
final_json["rasa_nlu_data"]["common_examples"].append(el)
with open(file_names[flag], "w") as j:
json.dump(final_json, j)
