FORMAT_CONVERSION_MAXCOMPUTE/format_convert/convert_rar.py

import os
import sys
sys.path.append(os.path.dirname(__file__) + "/../")
from format_convert.convert_tree import _Document, _Table, _Page, _Sentence
import logging
import traceback
from format_convert import get_memory_info
from format_convert.utils import get_platform, rename_inner_files, judge_error_code, judge_format, slash_replace


@get_memory_info.memory_decorator
def rar2text(path, unique_type_dir):
    from format_convert.convert import getText
    logging.info("into rar2text")
    try:
        rar_path = unique_type_dir
        try:
            # shell调用unrar解压
            _signal = os.system("unrar x " + path + " " + rar_path)
            print("rar2text _signal", _signal)
            # =0, 解压成功
            if _signal != 0:
                raise Exception
        except Exception as e:
            logging.info("rar format error!")
            print("rar format error!", e)
            return [-3]

        # 获取文件名
        # file_list = []
        # for root, dirs, files in os.walk(rar_path, topdown=False):
        #     for name in dirs:
        #         file_list.append(os.path.join(root, name) + os.sep)
        #     for name in files:
        #         file_list.append(os.path.join(root, name))

        if get_platform() == "Windows":
            print("============= rar file list")

        # 内部文件重命名
        file_list = rename_inner_files(rar_path)
        if judge_error_code(file_list):
            return file_list

        text = []
        for file in file_list:
            if os.path.isdir(file):
                continue

            # 无文件后缀，猜格式
            if len(file.split(".")) <= 1:
                logging.info(str(file) + " has no type! Guess type...")
                _type = judge_format(file)
                if _type is None:
                    logging.info(str(file) + "cannot guess type!")
                    sub_text = [""]
                else:
                    logging.info(str(file) + " guess type: " + _type)
                    new_file = str(file) + "." + _type
                    os.rename(file, new_file)
                    file = new_file
                    sub_text = getText(_type, file)
            # 有文件后缀，截取
            else:
                _type = file.split(".")[-1]
                sub_text = getText(_type, file)

            if judge_error_code(sub_text, code=[-3]):
                continue
            if judge_error_code(sub_text):
                return sub_text

            # print("sub text", sub_text, file, _type)
            text = text + sub_text
        return text
    except Exception as e:
        logging.info("rar2text error!")
        print("rar2text", traceback.print_exc())
        return [-1]


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

    def init_package(self):
        # 各个包初始化
        try:
            # shell调用unrar解压
            _signal = os.system("unrar x " + self.path + " " + self.rar_path)
            print("rar2text _signal", _signal)
            # =0, 解压成功
            if _signal != 0:
                raise Exception
        except:
            logging.info("cannot open rar!")
            traceback.print_exc()
            self._doc.error_code = [-3]

    def convert(self):
        from format_convert.convert import getText

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

        # 内部文件重命名
        file_list = rename_inner_files(self.rar_path)
        if judge_error_code(file_list):
            return file_list

        self._page = _Page(None, 0)
        file_no = 0
        for file in file_list:
            if os.path.isdir(file):
                continue

            bbox = (0, file_no, 0, 0)
            # 无文件后缀，猜格式
            if len(file.split(".")) <= 1:
                logging.info(str(file) + " has no type! Guess type...")
                _type = judge_format(file)
                if _type is None:
                    logging.info(str(file) + "cannot guess type!")
                    continue
                else:
                    logging.info(str(file) + " guess type: " + _type)
                    new_file = str(file) + "." + _type
                    os.rename(file, new_file)
                    file = new_file
                    sub_html = getText(_type, file)
            # 有文件后缀，截取
            else:
                _type = file.split(".")[-1]
                sub_html = getText(_type, file)

            if judge_error_code(sub_html, code=[-3]):
                continue
            if judge_error_code(sub_html):
                self._doc.error_code = sub_html
                return
            _sen = _Sentence(sub_html[0], bbox, is_html=True)
            self._page.add_child(_sen)

    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()