zhangjf's picture
support downloading updated data
1d413b9
raw
history blame
1.65 kB
import os
import json
import collections
def read_qs():
qs = []
directory = "./dialogues_set"
filenames = [
'dialogues_film.json',
'dialogues_jindong.json',
'dialogues_music.json',
'dialogues_natural.json',
'dialogues_taobao.json',
'dialogues_travel_kd.json'
]
for filename in filenames:
with open(f"{directory}/{filename}", "r", encoding="utf-8") as f:
for idx,line in enumerate(f):
idx2query = json.loads(line)
query = idx2query[str(idx)]
qs.append(query)
print(f"read {len(qs)} queries from files")
return qs
def read_qas():
qas = []
directory = "./dialogues_set"
for filename in os.listdir(directory):
if filename.endswith(".json") and "qas" in filename:
with open(f"{directory}/{filename}", "r", encoding="utf-8") as f:
for qa in json.loads(f.read()):
qas.append(qa)
print(f"read {len(qas)} query-answers from files")
return qas
def merge(qs, qas):
q_to_as = collections.defaultdict(lambda:[])
for qa in qas:
q_to_as[qa["q"]].append(qa["a"])
qas = []
for q in qs:
if len(q_to_as[q])==0:
continue
a = q_to_as[q].pop()
qas.append({"q":q, "a":a})
print(f"merge {len(qas)} query-answers from files")
return qas
if __name__ == "__main__":
qs = read_qs()
qas = read_qas()
qas = merge(qs, qas)
with open("./dialogues_set/qas.json", "w", encoding="utf-8") as f:
f.write(json.dumps(qas, ensure_ascii=False, indent=2))