testcase-data/Ours/get_update_wrong_code.py

import json
import re
from collections import defaultdict

def get_transformed_id(problem_id):
    if "地底蔷薇" in problem_id:
        return "地底蔷薇"
    tcb_id = problem_id.replace('/', ' ')
    tcb_id = re.sub(r'^[^.]*\.', '', tcb_id)
    i = tcb_id.find('」')
    if i != -1:
        tcb_id = tcb_id[i+1:]

    tcb_id = tcb_id.strip()
    if tcb_id == "":
        print(problem_id)
        return problem_id
    return tcb_id

new_wrong_code_base = json.load(open("/home/i-luoxianzhen/data/TestCase-Gen/data/Ours/balance_v1.0.dev_with_code.json", "r", encoding="utf-8"))
tcb_bench = json.load(open("/home/i-luoxianzhen/data/TestCase-Gen/data/Ours/TestcaseBench-v6.json", "r", encoding="utf-8"))

new_wrong_code_base_dict = {}

for k, v in new_wrong_code_base.items():
    tcb_id = get_transformed_id(k)
    new_wrong_code_base_dict[tcb_id] = v['cowrong_codes']

update = 0
for item in tcb_bench:
    if item['tcb_id'] not in new_wrong_code_base_dict.keys():
        item['wrong_code'] = item['wrong_code']['wrong_code']
        continue
    item['wrong_code'] = new_wrong_code_base_dict[item['tcb_id']]
    update += 1

print(update)
json.dump(tcb_bench, open(f"/home/i-luoxianzhen/data/TestCase-Gen/data/Ours/TestcaseBench-v7.json", "w", encoding="utf-8"), indent=4, ensure_ascii=False)