Parse the text file of the item list from geechs and output TSV

geechs_project_list_parser.py

import sys
import pathlib

def parse_project_data(file_path):
    with open(file_path, 'r', encoding='utf-8') as f:
        content = f.read()
        parts = content.split('＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝＝')
    projects = []
    for i in range(0, len(parts)-2, 2):
        projects.append(parts[i] + parts[i+1])


    project_data = []
    current_key = None
    for project in projects:
        lines = project.strip().split('\n')
        project_dict = {}
        for line in lines:
            line = line.strip()
            if line == '':
                continue
            elif '＞' in line:
                project_dict["会社名"] = line.strip('＜＞')
                current_key = "会社URL"
            elif line.startswith('【') and '】' in line:
                current_key, value = line.split('】', 1)
                current_key = current_key.strip('【')
                project_dict[current_key] = value.strip()
            else:
                if current_key is not None:
                    project_dict[current_key] = project_dict.get(current_key, '') + line.strip() + '\n'

        project_data.append(project_dict)

    return project_data


def quote(value):
    return '"' + value.strip(" \t\n") + '"'

keys = [
    '会社名',
    '会社URL',
    '案件名称/案件概要',
    '作業内容',
    '開発環境',
    '作業場所',
    '必須経験',
    '歓迎経験',
    '単金',
    '時間幅',
    '契約期間',
    '商談回数',
    '備考',
]
def tsv_header():
    return '\t'.join(keys)

def convert_to_tsv(data):
    tsv_output = ''
    for project in data:
        row = [ quote(project.get(key, '')) for key in keys ]
        tsv_row = '\t'.join(row)
        tsv_output += tsv_row + '\n'

    return tsv_output


def main():
    filenames = sys.argv[1:]

    print(tsv_header())
    for filename in filenames:

        filename = str(pathlib.Path(filename).resolve())
        project_data = parse_project_data(filename)
        tsv_output = convert_to_tsv(project_data)

        # TSV形式のテキストを出力します
        print(tsv_output)

if __name__ == '__main__':
    main()