nsq_convert2txt/file_processing/file_doc.py

"""
.doc解析
"""
import subprocess, os
from loguru import logger
from file_processing.file_docx import read_docx
from docs.config import error_number


def read_doc(file_path):
    out_dir = os.path.dirname(file_path)
    try:
        args = 'soffice --headless --convert-to docx %s --outdir %s' % (file_path, out_dir)
        output = subprocess.check_output(args, shell=True)
        logger.debug(str(output))
    except subprocess.CalledProcessError as e:
        logger.error('doc文件转换出错')
        logger.error(e)
        return '', error_number["解析错误"]
    file_name = os.path.basename(file_path).split('.')[0] + '.docx'
    logger.debug('doc文件转docx后文件路径>>>' + os.path.join(out_dir, file_name))
    docx_path = os.path.join(out_dir, file_name)
    return read_docx(docx_path)