report_app/app/tools/doc2docx.py

import subprocess
import os
import logging

# 获取日志记录器
logger = logging.getLogger(__name__)


def doc2docx(input_path, output_dir=None):
    logger.info("开始将doc转成docx")

    # 如果未指定输出目录，则使用输入文件所在目录
    if output_dir is None:
        output_dir = os.path.dirname(input_path)

    # 构建输出文件路径
    output_file = os.path.join(
        output_dir, os.path.splitext(os.path.basename(input_path))[0] + ".docx"
    )

    # LibreOffice 命令行工具
    libreoffice_cmd = "soffice"

    # 构建转换命令
    command = [
        libreoffice_cmd,
        "--headless",  # 无界面模式
        "--convert-to",
        "docx",  # 转换为 docx
        "--outdir",
        output_dir,  # 输出目录
        input_path,  # 输入文件
    ]

    # 执行命令
    try:
        subprocess.run(command, check=True)
        print(f"转换成功: {output_file}")
        return output_file
    except subprocess.CalledProcessError as e:
        print(f"转换失败: {e}")
        return None


if __name__ == "__main__":
    input_path = "./南方电网公司“抢修、投诉、舆情”三工单联动监测日报（3月6日）.doc"  # 替换为你的 .doc 文件路径
    doc2docx(input_path)