video-subtitle-extract/OcrUtils.py

#!/usr/bin/env python
# -*- encoding: utf-8 -*-
'''
@Author  :   liuyuqi
@Contact :   liuyuqi.gov@msn.cn
@Time    :   2020/02/17 22:13:38
@Version :   1.0
@License :   Copyright © 2017-2020 liuyuqi. All Rights Reserved.
@Desc    :   ocr 文字识别
'''

import base64
import json,os,sys,re,requests
import user_agent

class OcrUtils():
    def __init__(self):
        pass

    def ocrImg(self, imgData, ocrType):
        if ocrType == 0:
            return self.bdGeneralOcr(imgData)
        elif ocrType == 1:
            return self.bdAccurateOcr(imgData)
        elif ocrType == 1:
            return self.sogouMobileOcr(imgData)
        else:
            return self.sogouWebOcr(imgData)

    def bdGeneralOcr(self, imgData):
        return self.bdBaseOcr(imgData, "general_location")

    def sogouMobileOcr(self, imgData):
        pass

    def bdAccurateOcr(self, imgData):
        self.bdBaseOcr(imgData, "https://aip.baidubce.com/rest/2.0/ocr/v1/accurate")
        data={}
        res=requests.post(url=url,data=data,headers=user_agent.getheaders()).content.decode("utf8")

    def sogouWebOcr(self, imgData):
        url = "https://deepi.sogou.com/api/sogouService"
        referer = "https://deepi.sogou.com/?from=picsearch&tdsourcetag=s_pctim_aiomsg"
        imageData = base64.encode(imgData)


    def bdBaseOcr(self, imgData, param):
        urlArr = ["http://ai.baidu.com/tech/ocr/general", "http://ai.baidu.com/index/seccode?action=show"]

if __name__ == '__main__':
    test=OcrUtils()
    print(test.sogouWebOcr())