FORMAT_CONVERSION_MAXCOMPUTE/format_convert/convert_txt.py

import os
import sys
sys.path.append(os.path.dirname(__file__) + "/../")
from format_convert.convert_tree import _Document, _Page, _Sentence
import logging
import traceback
import chardet
from format_convert import get_memory_info


@get_memory_info.memory_decorator
def txt2text(path):
    logging.info("into txt2text")
    try:
        # 判断字符编码
        with open(path, "rb") as ff:
            data = ff.read()
        encode = chardet.detect(data).get("encoding")
        print("txt2text judge code is", encode)

        try:
            if encode is None:
                logging.info("txt2text cannot judge file code!")
                return [-3]
            with open(path, "r", encoding=encode) as ff:
                txt_text = ff.read()
            return [txt_text]
        except:
            logging.info("txt2text cannot open file with code " + encode)
            return [-3]
    except Exception as e:
        print("txt2text", traceback.print_exc())
        logging.info("txt2text error!")
        return [-1]


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

    def init_package(self):
        try:
            # 判断字符编码
            with open(self.path, "rb") as ff:
                data = ff.read()
            encode = chardet.detect(data).get("encoding")
            print("txt2text judge code is", encode)
            if encode is None:
                logging.info("txt2text cannot judge file code!")
                raise Exception
            with open(self.path, "r", encoding=encode) as ff:
                self.txt_text = ff.read()
        except:
            logging.info("cannot open txt!")
            traceback.print_exc()
            self._doc.error_code = [-3]

    def convert(self):
        self.init_package()
        if self._doc.error_code is not None:
            return

        self._page = _Page(None, 0)
        _sen = _Sentence(self.txt_text, (0, 0, 0, 0))
        self._page.add_child(_sen)
        self._doc.add_child(self._page)

    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
        return self._doc.get_html()