word_parse_of_eng/get_word_content/server.py

import json
import os
import sys
from pprint import pprint

import bottle
from bottle import redirect
from bottle import request
from bottle import response
from bottle import route, run
from bottle import template
from bottle import error

import config
from filepath2text import route_filename
from utils import get_dir_next_num

bottle.BaseRequest.MEMFILE_MAX = 8 * 1024 * 1024  # bytes

save_path = '../upload'
if not os.path.isdir(save_path):
    os.mkdir(save_path)

file_next_num = get_dir_next_num(save_path, which="files")


def start_word2html_app(kill_mathtype=True):
    if kill_mathtype:
        os.system("taskkill /f /im MathType.exe")
        os.system("taskkill /f /im WINWORD.EXE")
        os.system("taskkill /f /im ConsoleApplication1.exe")
    os.system("start {}".format(config.word2html_exe))  # start 在新窗口中打开


def set_response_header(response):
    response.set_header("Access-Control-Allow-Headers", "x-requested-with,content-type,Authorization")
    response.set_header("Access-Control-Allow-Methods", "POST,GET,OPTIONS")
    response.set_header("Access-Control-Allow-Origin", "*")


@route('/', method='GET')
@route('/', method='POST')
@route('/', method='OPTIONS')
def index():
    set_response_header(response)
    return redirect("/ocr_page")


@route('/ocr_page', method='GET')
@route('/ocr_page', method='POST')
@route('/ocr_page', method='OPTIONS')
def ocr_page():
    set_response_header(response)
    return template("index.html")


@route('/ocr', method='GET')
@route('/ocr', method='POST')
@route('/ocr', method='OPTIONS')
def do_ocr():
    global file_next_num

    set_response_header(response)
    upload = request.files.get('mydata')
    if upload:
        print("upload file: {}".format(upload.raw_filename))

        num_str = str(file_next_num)
        file_next_num += 1
        cur_save_path = save_path
        upload.raw_filename = num_str + os.path.splitext(upload.raw_filename)[1]
        upload.save(cur_save_path, overwrite=True)  # 把文件保存到save_path路径下
        filename = os.path.join(cur_save_path, upload.raw_filename)
        filename = os.path.abspath(filename)
        try:

            text = route_filename(filename)
            # pprint(text)

            if text:
                status = "OK"
                print("parse is over\n")
            else:
                status = "ERROR"
                text = "无法解析文件"

            # print([text])
            # pprint([text])
        except:
            status = "ERROR"
            text = "无法解析文件"
    else:
        status = "ERROR"
        text = "request.files.get('mydata') is None, 请检查文件名是否包含中文"

    res = {"status": status, "text": text}
    return json.dumps(res, ensure_ascii=False)


if __name__ == "__main__":
    start_word2html_app()

    if len(sys.argv) == 1:  # 默认在本地运行
        ip, port = config.LocalCfg.ip, config.LocalCfg.port
        run(host=ip, port=port, server='tornado')
    else:
        if sys.argv[1] == 'local':
            ip, port = config.LocalCfg.ip, config.LocalCfg.port
            run(host=ip, port=port, server='tornado')
        elif sys.argv[1] == 'testing':
            ip, port = config.TestingCfg.ip, config.TestingCfg.port
            run(host=ip, port=port, server='tornado')
        elif sys.argv[1] == 'production':
            ip, port = config.ProductionCfg.ip, config.ProductionCfg.port
            # run(host=ip, port=port, server='tornado', debug=True, reloader=True)
            run(host=ip, port=port, debug=True)
        else:
            print('cmd should be: python server.py local')
            print('or: python server.py testing')
            print('or: python server.py production')