FORMAT_CONVERSION_MAXCOMPUTE/format_convert/convert_xlsx.py

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


@get_memory_info.memory_decorator
def xlsx2text(path, unique_type_dir):
    logging.info("into xlsx2text")
    try:
        try:
            # sheet_name=None, 即拿取所有sheet，存为dict
            df_dict = pandas.read_excel(path, header=None, keep_default_na=False, sheet_name=None)
        except Exception as e:
            logging.info("xlsx format error!")
            return [-3]

        df_list = [sheet for sheet in df_dict.values()]
        sheet_text = ""
        for df in df_list:
            text = '<table border="1">' + "\n"
            for index, row in df.iterrows():
                text = text + "<tr>"
                for r in row:
                    text = text + "<td>" + str(r) + "</td>" + "\n"
                    # print(text)
                text = text + "</tr>" + "\n"
            text = text + "</table>" + "\n"
            sheet_text += text

        return [sheet_text]
    except Exception as e:
        logging.info("xlsx2text error!")
        print("xlsx2text", traceback.print_exc())
        return [-1]


class XlsxConvert:
    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):
        # 各个包初始化
        try:
            self.df = pandas.read_excel(self.path, header=None, keep_default_na=False, sheet_name=None)
        except:
            logging.info("cannot open xlsx!")
            traceback.print_exc()
            self._doc.error_code = [-3]

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

        sheet_list = [sheet for sheet in self.df.values()]
        sheet_no = 0
        for sheet in sheet_list:
            self._page = _Page(None, sheet_no)
            self.convert_page(sheet)

            if self._doc.error_code is None and self._page.error_code is not None:
                self._doc.error_code = self._page.error_code
            self._doc.add_child(self._page)
            sheet_no += 1

    def convert_page(self, sheet):
        text = '<table border="1">' + "\n"
        for index, row in sheet.iterrows():
            text = text + "<tr>"
            for r in row:
                text = text + "<td>" + str(r) + "</td>" + "\n"
                # print(text)
            text = text + "</tr>" + "\n"
        text = text + "</table>" + "\n"

        _table = _Table(text, (0, 0, 0, 0), is_html=True)
        self._page.add_child(_table)

    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()