FORMAT_CONVERSION_MAXCOMPUTE/format_convert/convert_ofd_test.py

import base64
import os
import re
import sys
import time
os.environ["CUDA_VISIBLE_DEVICES"] = "-1"
sys.path.append(os.path.abspath(os.path.dirname(__file__)) + "/../")

from format_convert.utils import judge_error_code, get_logger, log, register_all_fonts
# register_all_fonts("/usr/share/fonts/")

from format_convert.easyofd.easyofd.ofd import OFD
from format_convert.convert_tree import _Document, _Sentence, _Page
import logging
import traceback
from format_convert.convert_pdf import PDFConvert


class OfdConvert:
    def __init__(self, path, unique_type_dir):
        self._doc = _Document(path)
        self.path = path
        self.unique_type_dir = unique_type_dir
        self.ofd = OFD()  # 初始化OFD 工具类

    def convert(self):
        start_time = time.time()
        file_prefix = os.path.splitext(os.path.split(self.path)[1])[0]

        with open(self.path, "rb") as f:
            ofd_b64 = str(base64.b64encode(f.read()), "utf-8")

        self.ofd.read(ofd_b64, save_xml=False, xml_name=f"{file_prefix}_xml",
                      save_dir=self.unique_type_dir)  # 读取ofdb64
        # print("ofd.data", ofd.data) # ofd.data 为程序解析结果
        pdf_bytes = self.ofd.to_pdf()  # 转pdf

        self.ofd.del_data()

        file_name = re.split('[/\\\]', self.path)[-1]
        new_path = self.unique_type_dir + file_name[:-4] + '.pdf'

        with open(new_path, "wb") as f:
            f.write(pdf_bytes)
        log('odf to pdf path ' + new_path + ' cost: ' + str(time.time()-start_time))

        # 用pdf提取
        self._pdf = PDFConvert(new_path, self.unique_type_dir, need_page_no=None)
        # _pdf.convert()
        # self._doc = _pdf._doc

    def get_html(self):
        try:
            self.convert()
        except:
            traceback.print_exc()
            self._doc.error_code = [-1]

        # 直接返回doc处理的html
        if self._doc.error_code is not None:
            return self._doc.error_code
        else:
            return self._pdf.get_html()


if __name__ == '__main__':
    _p = "C:/Users/Administrator/Downloads/0c71fe77-f052-414d-8189-3e8cb4f2a607.ofd"
    _p = '../1750381792388.ofd'
    # _p = "C:/Users/Administrator/Desktop/test_wps/error2.wps"
    save_dir = "/data/fangjiasheng/format_conversion_maxcompute/format_convert/temp" + '/'
    c = OfdConvert(_p, save_dir)
    _html = c.get_html()
    print(_html)