diff --git a/data_process/process/content_extract.py b/data_process/process/content_extract.py index cd05e29..b39d019 100644 --- a/data_process/process/content_extract.py +++ b/data_process/process/content_extract.py @@ -1,6 +1,7 @@ import os import glob import json +from typing import List, Dict valid_keys = [ "Core_Fear_Source", "Pain_Threshold", "Time_Window_Pressure", "Helplessness_Index", @@ -77,6 +78,25 @@ def extract_json_data(json_files: list, threshold: int = 10) -> dict: raise ValueError(f"Invalid key {key} in {json_file}") return filt_json_data(data, threshold) +def load_data_from_dict(data_dict: List[dict]): + """ + 不进行阈值过滤,直接加载数据 + """ + data = {} + for idx, item in enumerate(data_dict): + data[idx] = {} + for key, value in item.items(): + if key in valid_keys: + data[idx][key] = value.get("value", None) + elif key == "Follow_up_Priority": + continue + else: + match_key = try_match_error_key(key) + if match_key: + data[idx][match_key] = value.get("value", None) + else: + print(f"Warning: Invalid key {key} in data dict, skipped.") + return data if __name__=="__main__": deal_folder = "deal"