VideoToArticles/task/baidu_search_task.py

"""
@author: luojunhui
"""
import json
import time

import requests
from uuid import uuid4

from applications.config import db_config
from applications.functions import whisper
from applications.ai import kimi_ai
from spider.baijiahao_article import bjh_article, bjh_url_list


class BaiduGenerateTask(object):
    """
    视频匹配文章流程
    流程
    1. 拿视频id，标题等信息匹配账号
    """

    def __init__(self, mysql_client):
        """
        :param mysql_client mysql服务池
        """
        self.mysql_client = mysql_client

    async def whisper_task(self):
        """
        执行定时任务，把库里面的视频转文本
        :return:
        """
        select_sql = f"""SELECT video_id FROM {db_config} WHERE status_code = 0 ORDER BY id ASC limit 1;"""
        video_list = await self.mysql_client.select(select_sql)

        async def whisper_and_update(video_id, mysql_client):
            """
            whisper处理视频并且把信息更新到mysql表中
            :param video_id:
            :param mysql_client:
            :return:
            """
            try:
                w_response = whisper(video_id)
            except:
                w_response = {"text": "whisper failed"}
            print(w_response)
            text = w_response['text'].replace("'", "")
            update_sql = f"""
            UPDATE {db_config}
            SET 
                video_text = %s,
                status_code = %s
            WHERE video_id = %s;
            """
            print(update_sql)
            await mysql_client.async_insert(sql=update_sql, params=(text, 1, video_id))

        for vid in video_list:
            await whisper_and_update(video_id=vid[0], mysql_client=self.mysql_client)

    async def materials_task(self):
        """
        获取task的材料
        :return:
        """
        select_sql = f"""SELECT task_id, video_title, video_text FROM {db_config} WHERE status_code = 1 ORDER BY id ASC limit 1;"""
        task_list = await self.mysql_client.select(select_sql)

        async def baidu_search(task_tuple, mysql_client):
            """
            :param task_tuple:
            :param mysql_client:
            """
            task_id, title, text = task_tuple
            url_list = bjh_url_list(title)
            L = []
            for url in url_list:
                a_text, img_list = bjh_article(url.split("&")[0])
                obj = {
                    "text": a_text,
                    "img_list": img_list
                }
                L.append(obj)
            update_sql = f"""
            UPDATE {db_config}
            SET materials = %s, status_code = %s
            WHERE task_id = %s;
            """
            print(update_sql)
            await mysql_client.async_insert(sql=update_sql, params=(L, 2, task_id))

        for task in task_list:
            await baidu_search(task, self.mysql_client)