rov-server/utils.py

import traceback

import requests
import json
import time

from datetime import datetime
# from db_helper import HologresHelper, RedisHelper, MysqlHelper
from db_helper import RedisHelper, MysqlHelper
from config import set_config
from log import Log

config_ = set_config()
log_ = Log()


def send_msg_to_feishu(msg_text):
    """发送消息到飞书"""
    # webhook地址
    webhook = 'https://open.feishu.cn/open-apis/bot/v2/hook/8de4de35-30ed-4692-8854-7a154e89b2f2'
    # 自定义关键词key_word
    key_word = '服务报警'
    headers = {'Content-Type': 'application/json'}
    payload_message = {
        "msg_type": "text",
        "content": {
            "text": '{}: {}'.format(key_word, msg_text)
        }
    }
    response = requests.request('POST', url=webhook, headers=headers, data=json.dumps(payload_message))
    print(response.text)


def request_post(request_url, request_data, timeout):
    """
    post 请求 HTTP接口
    :param request_url: 接口URL
    :param request_data: 请求参数
    :param timeout: 超时时间，单位为秒，type-float or tuple(connect_timeout, read_timeout)
    :return: res_data json格式
    """
    try:
        headers = {"Connection": "close"}
        response = requests.post(url=request_url, json=request_data, timeout=timeout, headers=headers)
        if response.status_code == 200:
            res_data = json.loads(response.text)
            return res_data
        else:
            return None
    except Exception as e:
        log_.error('url: {}, exception: {}, traceback: {}'.format(request_url, e, traceback.format_exc()))
        return None


def get_videos_remain_view_count(app_type, videos):
    """
    获取视频在流量池中的剩余可分发数
    :param app_type: 产品标识 type-int
    :param videos: 视频信息 (视频id, 流量池标记) type-list，[{'videoId': video_id, 'flowPool': flow_pool}, ...]
    :return: data type-list，[(video_id, flow_pool, view_count), ...]
             error_flag 错误标记，True为错误
    """
    error_flag = False
    if not videos:
        return [], error_flag

    request_data = {'appType': app_type, 'videos': videos}
    result = request_post(request_url=config_.GET_REMAIN_VIEW_COUNT_URL, request_data=request_data, timeout=(0.1, 1))

    if result is None:
        error_flag = True
        return [], error_flag

    if result['code'] != 0:
        log_.info('获取视频在流量池中的剩余可分发数失败')
        error_flag = True
        return [], error_flag

    data = [(item['videoId'], item['flowPool'], item['distributeCount']) for item in result['data']]
    return data, error_flag


def get_videos_local_distribute_count(video_id, flow_pool):
    """
    获取流量池视频本地分发数
    :param video_id: video_id
    :param flow_pool: 流量池标记
    :return: current_count 本地记录的分发数
    """
    # redis_h = datetime.now().hour
    # if datetime.now().minute >= 30:
    #     redis_h += 0.5
    # key_name = config_.LOCAL_DISTRIBUTE_COUNT_PREFIX + str(redis_h)
    key_name = '{}{}.{}'.format(config_.LOCAL_DISTRIBUTE_COUNT_PREFIX, video_id, flow_pool)
    redis_helper = RedisHelper()
    # video = '{}-{}'.format(video_id, flow_pool)
    # current_count = redis_helper.get_score_with_value(key_name=key_name, value=video)
    current_count = redis_helper.get_data_from_redis(key_name=key_name)
    if current_count is not None:
        return int(current_count)
    else:
        return None


def update_video_w_h_rate(video_id, key_name):
    """
    获取横屏视频的宽高比，并存入redis中 （width/height>1)
    :param video_id: videoId type-int
    :param key_name: redis key
    :return: None
    """
    # 获取数据
    sql = "SELECT id, width, height, rotate FROM longvideo.wx_video WHERE id = {};".format(video_id)
    mysql_helper = MysqlHelper()
    data = mysql_helper.get_data(sql=sql)
    if len(data) == 0:
        return
    # 更新到redis
    width, height, rotate = int(data[0][1]), int(data[0][2]), int(data[0][3])
    if width == 0 or height == 0:
        return
    if rotate in (90, 270):
        w_h_rate = height / width
    else:
        w_h_rate = width / height
    if w_h_rate > 1:
        info_data = {int(video_id): w_h_rate}
    else:
        return
    redis_helper = RedisHelper()
    # 写入新数据
    if len(info_data) > 0:
        redis_helper.add_data_with_zset(key_name=key_name, data=info_data)


class FilterVideos(object):
    """视频过滤"""
    def __init__(self, app_type, video_ids, mid='', uid=''):
        """
        初始化
        :param app_type: 产品标识 type-int
        :param video_ids: 需过滤的视频列表 type-list
        :param mid: mid type-string
        :param uid: uid type-string
        """
        self.app_type = app_type
        self.mid = mid
        self.uid = uid
        self.video_ids = video_ids

    def filter_video_status_h(self, video_ids, rule_key, ab_code, province_code, key_flag=''):
        """召回小时级更新的视频状态过滤"""
        # 根据Redis缓存中的数据过滤
        redis_helper = RedisHelper()
        # 获取不符合推荐状态的视频
        if ab_code in [code for _, code in config_.AB_CODE['region_rank_by_h'].items()]:
            if key_flag == 'region_24h':
                key_prefix = f"{config_.REGION_H_VIDEO_FILER_24H}{province_code}."
            else:
                key_prefix = f"{config_.REGION_H_VIDEO_FILER}{province_code}."
        elif ab_code in [code for _, code in config_.AB_CODE['rank_by_24h'].items()]:
            key_prefix = config_.H_VIDEO_FILER_24H
        else:
            key_prefix = config_.H_VIDEO_FILER
        filter_videos_list = redis_helper.get_data_from_set(key_name=f"{key_prefix}{rule_key}")
        if not filter_videos_list:
            return video_ids
        filter_videos = [int(video) for video in filter_videos_list]
        filtered_videos = [video_id for video_id in video_ids if video_id not in filter_videos]
        return filtered_videos

    def filter_videos_h(self, rule_key, ab_code, province_code, key_flag=''):
        """召回小时级更新的视频过滤"""
        # 预曝光过滤
        st_pre = time.time()
        filtered_pre_result = self.filter_video_previewed(self.video_ids)
        et_pre = time.time()
        log_.info('filter by previewed: app_type = {}, mid = {}, uid = {}, result = {}, execute time = {}ms'.format(
            self.app_type, self.mid, self.uid, filtered_pre_result, (et_pre - st_pre) * 1000))
        if not filtered_pre_result:
            return None

        # 视频状态过滤
        st_status = time.time()
        filtered_status_result = self.filter_video_status_h(video_ids=filtered_pre_result, rule_key=rule_key,
                                                            ab_code=ab_code, province_code=province_code,
                                                            key_flag=key_flag)
        et_status = time.time()
        log_.info('filter by video status: result = {}, execute time = {}ms'.format(
            filtered_status_result, (et_status - st_status) * 1000))
        if not filtered_status_result:
            return None

        # 视频已曝光过滤
        st_viewed = time.time()
        filtered_viewed_result = self.filter_video_viewed(video_ids=filtered_status_result)
        et_viewed = time.time()
        log_.info('filter by viewed: app_type = {}, mid = {}, uid = {}, result = {}, execute time = {}ms'.format(
            self.app_type, self.mid, self.uid, filtered_viewed_result, (et_viewed - st_viewed) * 1000))
        if not filtered_viewed_result:
            return None
        else:
            return [int(video_id) for video_id in filtered_viewed_result]

    def filter_videos(self):
        """视频过滤"""
        # 预曝光过滤
        st_pre = time.time()
        filtered_pre_result = self.filter_video_previewed(self.video_ids)
        et_pre = time.time()
        log_.info('filter by previewed: app_type = {}, mid = {}, uid = {}, result = {}, execute time = {}ms'.format(
            self.app_type, self.mid, self.uid, filtered_pre_result, (et_pre - st_pre) * 1000))
        if not filtered_pre_result:
            return None

        # 视频状态过滤采用离线定时过滤方案
        # 视频状态过滤
        # st_status = time.time()
        # filtered_status_result = self.filter_video_status(video_ids=filtered_pre_result)
        # et_status = time.time()
        # log_.info('filter by video status: result = {}, execute time = {}ms'.format(
        #     filtered_status_result, (et_status - st_status) * 1000))
        # if not filtered_status_result:
        #     return None

        # 视频已曝光过滤
        st_viewed = time.time()
        filtered_viewed_result = self.filter_video_viewed(video_ids=filtered_pre_result)
        et_viewed = time.time()
        log_.info('filter by viewed: app_type = {}, mid = {}, uid = {}, result = {}, execute time = {}ms'.format(
            self.app_type, self.mid, self.uid, filtered_viewed_result, (et_viewed - st_viewed) * 1000))
        if not filtered_viewed_result:
            return None
        else:
            return [int(video_id) for video_id in filtered_viewed_result]

    def filter_video_previewed(self, video_ids):
        """
        预曝光过滤
        :param video_ids: 需过滤的视频列表 type-list
        :return: filtered_videos  过滤后的列表  type-list
        """
        if not self.mid:
            # mid为空时，不做预曝光过滤
            return video_ids
        # 根据Redis缓存中的数据过滤
        redis_helper = RedisHelper()
        # key拼接
        key_name = config_.PREVIEW_KEY_PREFIX + '{}.{}'.format(self.app_type, self.mid)
        pe_videos_list = redis_helper.get_data_from_set(key_name)
        # log_.info('****app_type = {}, mid = {}, uid = {}, pe_videos_list = {}'.format(
        #     self.app_type, self.mid, self.uid, pe_videos_list))
        # log_.info('****app_type = {}, mid = {}, uid = {}, video_ids = {}'.format(
        #     self.app_type, self.mid, self.uid, video_ids))
        if not pe_videos_list:
            return video_ids
        pe_videos = [int(video) for video in pe_videos_list]
        filtered_videos = [video_id for video_id in video_ids if video_id not in pe_videos]
        return filtered_videos

    def filter_video_status(self, video_ids):
        """
        对视频状态进行过滤
        :param video_ids: 视频id列表 type-list
        :return: filtered_videos
        """
        if len(video_ids) == 1:
            sql = "set hg_experimental_enable_shard_pruning=off; " \
                  "SELECT video_id " \
                  "FROM {} " \
                  "WHERE audit_status = 5 " \
                  "AND applet_rec_status IN (1, -6) " \
                  "AND open_status = 1 " \
                  "AND payment_status = 0 " \
                  "AND encryption_status != 5 " \
                  "AND transcoding_status = 3 " \
                  "AND video_id IN ({});".format(config_.VIDEO_STATUS, video_ids[0])
        else:
            sql = "set hg_experimental_enable_shard_pruning=off; " \
                  "SELECT video_id " \
                  "FROM {} " \
                  "WHERE audit_status = 5 " \
                  "AND applet_rec_status IN (1, -6) " \
                  "AND open_status = 1 " \
                  "AND payment_status = 0 " \
                  "AND encryption_status != 5 " \
                  "AND transcoding_status = 3 " \
                  "AND video_id IN {};".format(config_.VIDEO_STATUS, tuple(video_ids))

        hologres_helper = HologresHelper()
        data = hologres_helper.get_data(sql=sql)
        filtered_videos = [int(temp[0]) for temp in data]
        return filtered_videos

    def filter_video_viewed(self, video_ids, types=(1, 6,)):
        """
        调用后端接口过滤用户已观看视频
        :param video_ids: 视频id列表 type-list
        :param types: 过滤参数 type-tuple, 默认(1, )  1-已观看 2-视频状态 3-是否进入老年人社区 4-话题状态 5-推荐状态 6-白名单过滤
        :return: filtered_videos
        """
        # 调用http接口
        request_data = {"appType": self.app_type,
                        "mid": self.mid,
                        "uid": self.uid,
                        "types": list(types),
                        "videoIds": video_ids}
        result = request_post(request_url=config_.VIDEO_FILTER_URL, request_data=request_data, timeout=2.5)

        if result is None:
            log_.info('过滤失败，types: {}'.format(types))
            return video_ids

        if result['code'] != 0:
            log_.info('过滤失败，types: {}'.format(types))
            return video_ids

        filtered_videos = result['data']
        return filtered_videos


if __name__ == '__main__':
    # filter_ = FilterVideos(app_type=1, mid='22', uid='www', video_ids=[1, 2, 3, 55])
    # filter_.filter_videos()
    # filter_.filter_video_status(video_ids=[1, 3, 5])

    # videos = [{'videoId': 9034659, 'flowPool': '3#11#3#1637824188547'}, {'videoId': 9035052, 'flowPool': '3#11#3#1637824172827'}]
    # res = get_videos_remain_view_count(4, videos)
    # print(res)

    # text = '测试 @李倩'
    # send_msg_to_feishu(text)

    update_video_w_h_rate(video_id=113, key_name='')