jsonl文件格式示例:
{'ab8145f6d1ee0d7a7a1a2730d50105c0645c21e663767754ef5eba85cf41ddc1': 'Stadacona.'}
{'3a72ecafa9cc526ec3a44a6230aaa7e243e319d2212015b207582ae9c65f9705': 'Stadacona.'}
需要转化为如下格式json文件:
格式1:
{
"ab8145f6d1ee0d7a7a1a2730d50105c0645c21e663767754ef5eba85cf41ddc1": "Stadacona.",
"3a72ecafa9cc526ec3a44a6230aaa7e243e319d2212015b207582ae9c65f9705": "Stadacona."
}
格式2:
[
{
"ab8145f6d1ee0d7a7a1a2730d50105c0645c21e663767754ef5eba85cf41ddc1": "Stadacona."
},
{
"3a72ecafa9cc526ec3a44a6230aaa7e243e319d2212015b207582ae9c65f9705": "Stadacona."
}
]
转化为格式1的代码:(需要注意的是,读取文件和写入文件的encoding需要指定为一致,不然会导致gbk和utf混用,导致评测和微调的时候出现字符错误)
import json
def jsonl_to_json(jsonl_file, json_file):
with open(jsonl_file, 'r', encoding='utf-8') as f:
jsonl_data = f.readlines()
# print(jsonl_data)
# print(jsonl_data[0])
# print(jsonl_data[0][1:-2])
# print(jsonl_data[14])
with open(json_file, 'w', encoding='utf-8') as f:
for line in jsonl_data:
line = line[1:-2]
f.write(' ' + line + ',' + '\n')
f.close()
jsonl_file = 'C:/Users/Desktop/1.jsonl'
json_file = 'C:/Users/Desktop/test.json'
jsonl_to_json(jsonl_file, json_file)
如果json文件的anwer有多个答案,如:
{'afec84fa0ca103df7b548b9d6a9c33d916ee6464f6585b836bb076e4cdeb8d0b': 'Great Lakes, Pacific Ocean, Atlantic Ocean.'}
{'f492297aef364879271558f8f334d261b4c513357f52bd3c89265d30865eb03f': 'Candy cane, Sprint car racing, ice cream cone, Thanksgiving.'}
那么需要用如下代码转换:
import json
def jsonl_to_json(jsonl_file, json_file):
converted_items = {}
with open(jsonl_file, 'r', encoding='utf-8') as f:
for line in f:
try:
item = eval(line)
except json.decoder.JSONDecodeError:
print(line)
return
pid = list(item.keys())[0]
print(pid)
answer = item[pid]
answers = answer.replace(".", "").split(", ")
converted_items[pid] = answers
with open(json_file, 'w', encoding='utf-8') as f:
json.dump(converted_items, f)
jsonl_file = 'C:/Users/Desktop/epoch_13_KoRC_outputs.jsonl'
json_file = 'C:/Users/Desktop/iid.json'
jsonl_to_json(jsonl_file, json_file)
转化为格式2的代码:(注意需要将jsonl的’转化为”)
import json
def jsonl_to_json(jsonl_file, json_file):
json_data = []
with open(jsonl_file, 'r', encoding='utf-8') as f:
for line in f:
try:
obj = json.loads(line)
json_data.append(obj)
except json.JSONDecodeError:
print(f"Ignoring invalid JSON object: {line}")
with open(json_file, 'w', encoding='utf-8') as f:
json.dump(json_data, f, indent=4)
jsonl_file = 'C:/Users/Desktop/1.jsonl'
json_file = 'C:/Users/Desktop/test.json'
jsonl_to_json(jsonl_file, json_file)