FORMAT_CONVERSION_MAXCOMPUTE/format_convert/convert_xls.py

import os
import sys

from format_convert.convert_tree import _Document

sys.path.append(os.path.dirname(__file__) + "/../")
import logging
import traceback
from format_convert import get_memory_info
from format_convert.convert_need_interface import from_office_interface
from format_convert.convert_xlsx import xlsx2text, XlsxConvert
from format_convert.utils import judge_error_code


@get_memory_info.memory_decorator
def xls2text(path, unique_type_dir):
    logging.info("into xls2text")
    try:
        # 调用libreoffice格式转换
        file_path = from_office_interface(path, unique_type_dir, 'xlsx')
        if judge_error_code(file_path):
            return file_path

        text = xlsx2text(file_path, unique_type_dir)
        if judge_error_code(text):
            return text

        return text
    except Exception as e:
        logging.info("xls2text error!")
        print("xls2text", traceback.print_exc())
        return [-1]


class XlsConvert:
    def __init__(self, path, unique_type_dir):
        self._doc = _Document(path)
        self.path = path
        self.unique_type_dir = unique_type_dir

    def convert(self):
        # 调用office格式转换
        file_path = from_office_interface(self.path, self.unique_type_dir, 'xlsx')
        if judge_error_code(file_path):
            self._doc = file_path
            return
        _xlsx = XlsxConvert(file_path, self.unique_type_dir)
        _xlsx.convert()
        self._doc = _xlsx._doc

    def get_html(self):
        try:
            self.convert()
        except:
            traceback.print_exc()
            self._doc.error_code = [-1]
        if self._doc.error_code is not None:
            return self._doc.error_code
        print(self._doc.children)
        return self._doc.get_html()