pdf2markdown/fix_requirements.py

import re


def simplify_editable_git_path(line: str) -> str:
    if not line.strip().startswith("-e"):
        return line

    # 匹配 subdirectory 参数
    match = re.search(r"subdirectory=([^\s&#]+)", line)
    if match:
        sub_path = match.group(1)
        if sub_path.startswith("third_party/"):
            return f"-e {sub_path}\n"
    return line

def fix_requirements():
    with open('requirements.txt', "r") as infile:
        lines = infile.readlines()

    fixed_lines = [simplify_editable_git_path(line) for line in lines]

    with open('requirements.txt', "w") as outfile:
        outfile.writelines(fixed_lines)

    print(f"✅ 已更新 requirements.txt：简化了包含 third_party 的 git 路径。")

if __name__ == "__main__":
    fix_requirements()