RedPen の Json 形式の出力ファイルをCSVに変換するスクリプト

redpen_json_to_csv.py

#!/usr/bin/env python

RESULT_JSON_FILE = "./check_results/redpen_results.json"
RESULT_CSV_FILE = "./check_results/redpen_results.csv"


def flatten_json(y):
    results = []
    document = y[0].get("document")
    errors = y[0].get("errors")

    for error in errors:
        elems = [document, error.get("sentence"), error.get("validator")]
        end_position = error.get("endPosition")
        if end_position is None:
            elems.append("")
            elems.append("")
        else:
            elems.append(end_position.get("offset", ""))
            elems.append(end_position.get("lineNum", ""))
        elems.append(error.get("DoubledWord"))
        elems.append(error.get("level"))
        elems.append(error.get("sentenceStartColumnNum"))
        elems.append(error.get("message"))
        start_position = error.get("startPosition")
        if start_position is None:
            elems.append("")
            elems.append("")
        else:
            elems.append(start_position.get("offset"))
            elems.append(start_position.get("lineNum"))
        results.append(elems)
    return results


def get_headers():
    elems = ["document", "sentence", "validator", "endPosition.offset", "endPosition.lineNum", "DoubledWord", "level",
             "sentenceStartColumnNum", "message", "startPosition.offset", "startPosition.lineNum"]
    return elems


def main():
    import json
    from io import StringIO
    
    flat_results = [get_headers()]
    
    # 一行ずつ読み込む、ファイルをオープンする
    with open(RESULT_JSON_FILE, "r") as check_results:
       for line in check_results:
           io = StringIO(line)
           json_dict = json.load(io)
           flat_json = flatten_json(json_dict)
           flat_results.extend(flat_json)

    import csv
    with open(RESULT_CSV_FILE, "w+") as results_csv:
        csv_writer = csv.writer(results_csv, delimiter=',')
        csv_writer.writerows(flat_results)

if __name__ == '__main__':
    main()

redpen_results.csv