search_answer_ctr.py 11 KB
# -*- coding:UTF-8 -*-
# @Time  : 2020/9/4 17:07
# @File  : search_meigou_ctr.py
# @email : litao@igengmei.com
# @author : litao

import hashlib
import json

import pymysql
import xlwt, datetime
import redis
# from pyhive import hive
from maintenance.func_send_email_with_file import send_file_email
from typing import Dict, List
from elasticsearch_7 import Elasticsearch
from elasticsearch_7.helpers import scan
import sys
import time
from pyspark import SparkConf
from pyspark.sql import SparkSession, DataFrame


# from pyspark.sql.functions import lit
# import pytispark.pytispark as pti


def con_sql(sql):
    # 从数据库的表里获取数据

    db = pymysql.connect(host='172.16.40.158', port=4000, user='st_user', passwd='aqpuBLYzEV7tML5RPsN1pntUzFy',
                         db='jerry_prod')
    cursor = db.cursor()
    cursor.execute(sql)
    result = cursor.fetchall()
    db.close()
    return result


startTime = time.time()
sparkConf = SparkConf()
sparkConf.set("spark.sql.crossJoin.enabled", True)
sparkConf.set("spark.debug.maxToStringFields", "100")
sparkConf.set("spark.tispark.plan.allow_index_double_read", False)
sparkConf.set("spark.tispark.plan.allow_index_read", True)
sparkConf.set("spark.hive.mapred.supports.subdirectories", True)
sparkConf.set("spark.hadoop.mapreduce.input.fileinputformat.input.dir.recursive", True)
sparkConf.set("spark.serializer", "org.apache.spark.serializer.KryoSerializer")
sparkConf.set("mapreduce.output.fileoutputformat.compress", False)
sparkConf.set("mapreduce.map.output.compress", False)
sparkConf.set("prod.gold.jdbcuri",
              "jdbc:mysql://172.16.30.136/doris_prod?user=doris&password=o5gbA27hXHHm&rewriteBatchedStatements=true")
sparkConf.set("prod.mimas.jdbcuri",
              "jdbc:mysql://172.16.30.138/mimas_prod?user=mimas&password=GJL3UJe1Ck9ggL6aKnZCq4cRvM&rewriteBatchedStatements=true")
sparkConf.set("prod.gaia.jdbcuri",
              "jdbc:mysql://172.16.30.143/zhengxing?user=work&password=BJQaT9VzDcuPBqkd&rewriteBatchedStatements=true")
sparkConf.set("prod.tidb.jdbcuri",
              "jdbc:mysql://172.16.40.158:4000/eagle?user=st_user&password=aqpuBLYzEV7tML5RPsN1pntUzFy&rewriteBatchedStatements=true")
sparkConf.set("prod.jerry.jdbcuri",
              "jdbc:mysql://172.16.40.158:4000/jerry_prod?user=st_user&password=aqpuBLYzEV7tML5RPsN1pntUzFy&rewriteBatchedStatements=true")
sparkConf.set("prod.tispark.pd.addresses", "172.16.40.158:2379")
sparkConf.set("prod.tispark.pd.addresses", "172.16.40.170:4000")
sparkConf.set("prod.tidb.database", "jerry_prod")
sparkConf.setAppName("search_answer_ctr")

spark = (SparkSession.builder.config(conf=sparkConf).config("spark.sql.extensions", "org.apache.spark.sql.TiExtensions")
         .config("spark.tispark.pd.addresses", "172.16.40.170:2379").enableHiveSupport().getOrCreate())

spark.sql("ADD JAR hdfs:///user/hive/share/lib/udf/brickhouse-0.7.1-SNAPSHOT.jar")
spark.sql("ADD JAR hdfs:///user/hive/share/lib/udf/hive-udf-1.0-SNAPSHOT.jar")
spark.sql("CREATE TEMPORARY FUNCTION json_map AS 'brickhouse.udf.json.JsonMapUDF'")
spark.sql("CREATE TEMPORARY FUNCTION is_json AS 'com.gmei.hive.common.udf.UDFJsonFormatCheck'")
spark.sql("CREATE TEMPORARY FUNCTION arrayMerge AS 'com.gmei.hive.common.udf.UDFArryMerge'")

task_list = []
task_days = 3
for t in range(0, task_days):
    day_num = 0 - t
    now = (datetime.datetime.now() + datetime.timedelta(days=day_num))
    last_30_day_str = (now + datetime.timedelta(days=-30)).strftime("%Y%m%d")
    today_str = now.strftime("%Y%m%d")
    yesterday_str = (now + datetime.timedelta(days=-1)).strftime("%Y%m%d")
    one_week_age_str = (now + datetime.timedelta(days=-7)).strftime("%Y%m%d")

    sql_search_ctr = r"""
            SELECT
                   exp.partition_date as partition_date
                   ,active_type
                   ,device_os_type
                   ,sum(service_exp_pv) as service_exp_pv
                   ,sum(neirong_exp_pv) as neirong_exp_pv
                   ,sum(service_click_pv) as service_click_pv
                   ,sum(neirong_click_pv) as neirong_click_pv
               FROM
               (
                   SELECT t1.partition_day as partition_date,device_id
                       ,service_exp_pv,neirong_exp_pv,service_click_pv,neirong_click_pv
                   FROM
                   (--搜索结果页卡片精准曝光
                     SELECT  partition_day,
                             device_id,
                             count(CASE WHEN card_content_type='service' THEN 1 END) as service_exp_pv,
                             count(CASE WHEN card_content_type<>'service' THEN 1 END) as neirong_exp_pv
                     FROM
                     (
                       SELECT device_id,partition_day,card_content_type 
                       FROM ml.mid_ml_c_et_pe_preciseexposure_dimen_d
                       WHERE partition_day >= '{partition_day}'
                       and partition_day < '{end_date}'
                       and action in ('page_precise_exposure','home_choiceness_card_exposure')
                       and is_exposure = '1'
                       and page_code in ('search_result_diary','search_result_doctor','search_result_hospital','search_result_more'
                               ,'search_result_more_infomation','search_result_more_user','search_result_post','search_result_welfare'
                               ,'search_result_wiki','search_result_question_answer')
                       AND (card_content_type IN ('answer','qa','question') or card_type in ('answer','qa','question'))
                     )a
                     group by partition_day,card_content_type,device_id
                   )t1

                   LEFT JOIN
                   (--搜索结果页卡片点击
                     SELECT cl_id,partition_date
                         ,sum(CASE WHEN card_content_type='service' THEN click_pv END) as service_click_pv
                         ,sum(CASE WHEN card_content_type='neirong' THEN click_pv END) as neirong_click_pv
                     FROM
                     (
                         SELECT partition_date,cl_id,'service' as card_content_type,count(1) as click_pv
                         FROM online.bl_hdfs_maidian_updates
                         WHERE partition_date >= '{partition_day}'
                             AND partition_date < '{end_date}'
                         AND ((action in ('search_result_click_recommend_item','search_result_welfare_click_item')
                             AND page_name in ('search_result_more','search_result_welfare'))
                          or (action = 'goto_welfare_detail' AND params ['from'] = 'search_result_welfare_recommend')
                          or (action = 'on_click_card' AND params['card_content_type'] in ('service') AND page_name in ('search_result_more','search_result_welfare')))
                         GROUP BY partition_date,cl_id,'service' 

                         UNION ALL
                         SELECT partition_date,cl_id,'neirong' as card_content_type,count(1) as click_pv
                         FROM online.bl_hdfs_maidian_updates
                         WHERE partition_date >= '{partition_day}'
                           AND partition_date < '{end_date}'
                           AND ((action in ('on_click_topic_card','on_click_diary_card','search_result_click_infomation_item')
                               AND page_name in ('search_result_more','search_result_diary','search_result_post'))
                               or (action = 'on_click_card' AND params['card_content_type'] in ('answer','diary') AND page_name in ('search_result_more','search_result_diary','search_result_question_answer')))
                         GROUP BY partition_date,cl_id,'neirong'
                     )t2
                     GROUP BY cl_id,partition_date
                   )t2
                     ON t1.partition_day=t2.partition_date AND t1.device_id=t2.cl_id
               )exp

               JOIN
               (
                   SELECT partition_date,device_id,t2.active_type,t2.channel,t2.device_os_type
                   FROM
                   (
                       SELECT
                       partition_date,m.device_id
                       ,array(device_os_type ,'合计') as device_os_type
                       ,array(case WHEN active_type = '4'  THEN '老活' 
                             WHEN active_type  in ('1','2')  then '新增' END ,'合计') as active_type 
                       ,array(CASE WHEN is_ai_channel = 'true' THEN 'AI'  ELSE '其他' END , '合计') as channel
                       FROM online.ml_device_day_active_status m
                           LEFT JOIN 
                           (SELECT code,is_ai_channel,partition_day
                            FROM DIM.DIM_AI_CHANNEL_ZP_NEW
                            WHERE partition_day>= '{partition_day}' AND partition_day < '{end_date}'  ) tmp
                           ON  m.partition_date=tmp.partition_day AND first_channel_source_type=code
                       where partition_date >= '{partition_day}'  
                       AND partition_date < '{end_date}' 
                       AND active_type in ('1','2','4')
                   ) mas
                   LATERAL VIEW explode(mas.channel) t2 AS channel
                   LATERAL VIEW explode(mas.device_os_type) t2 AS device_os_type
                   LATERAL VIEW explode(mas.active_type) t2 AS active_type
               )dev_channel
                 on dev_channel.device_id = exp.device_id
                   AND dev_channel.partition_date = exp.partition_date
               GROUP BY exp.partition_date,active_type,device_os_type
           """.format(partition_day=yesterday_str, end_date=today_str)

    print(sql_search_ctr)
    search_ctr_df = spark.sql(sql_search_ctr)
    # spam_pv_df.createOrReplaceTempView("dev_view")
    search_ctr_df.show(1)
    sql_res = search_ctr_df.collect()

    print("-------------------------------")

    for res in sql_res:
        print(res)
        device_os_type = res.device_os_type
        active_type = res.active_type
        partition_date = yesterday_str
        pid = hashlib.md5((partition_date + device_os_type + active_type).encode("utf8")).hexdigest()
        click_num = res.neirong_click_pv
        exposure = res.neirong_exp_pv
        try:
            search_ctr = round(click_num / exposure, 5)
        except:
            search_ctr = 0
        instert_sql = """replace into search_answer_ctr(
               partition_date,device_os_type,active_type,pid,click_num,exposure,search_ctr) VALUES('{partition_date}','{device_os_type}','{active_type}','{pid}',{click_num},{exposure},{search_ctr});""".format(
            partition_date=partition_date, device_os_type=device_os_type, active_type=active_type, pid=pid,
            click_num=click_num,
            exposure=exposure, search_ctr=search_ctr
        )
        print(instert_sql)
        # cursor.execute("set names 'UTF8'")
        db = pymysql.connect(host='172.16.40.158', port=4000, user='st_user', passwd='aqpuBLYzEV7tML5RPsN1pntUzFy',
                             db='jerry_prod')
        cursor = db.cursor()
        res = cursor.execute(instert_sql)
        db.commit()
        print(res)