testcase-data/add_experience/wrong_str_distribute.py

import json
import os
import numpy as np
def load_all_wrong_code_subset(name="tcb", prefix_dir=None, testcase_alg = ""):

    ds = json.load(open("/home/luoxianzhen/yang/data/Ours/TestcaseBench-v28.json", "r", encoding="utf-8"))
    tcb_id_transform = {}
    for item in ds:
        tcb_id_transform[item['wrong_code'][0]['problem']] = item['tcb_id']
    all_wrong_code = json.load(open("/home/luoxianzhen/yang/data/Ours/all_wrong_code/data/all_wrong_code_subset.json", "r", encoding="utf-8"))

    w_distributions = {}

    for item in all_wrong_code:
        matrix = []
        for idx, c in enumerate(item["all_wrng_code"]):
            output_str = c["output_str"]
            binary = [1 if ch == 'W' else 0 for ch in output_str]
            matrix.append(binary)
        col_sum = np.sum(matrix, axis=0)
        w_distributions[item['name']] = ",".join(map(str, col_sum))
    json.dump(w_distributions, open("/home/luoxianzhen/yang/data/add_experience/wrong_code_distribute.json", "w", encoding="utf-8"), indent=4, ensure_ascii=False)

    return w_distributions


if __name__ == "__main__":

    data = load_all_wrong_code_subset(name="tcb", prefix_dir=f"/home/luoxianzhen/yang/save_tests_gpt-4o-type/lcb/", testcase_alg="algo")