#!/usr/bin/env bash

source ../../dmp_env.sh

dt_slash_today=$(date -d "$ScheduleTime 1 days ago" +"%Y/%m/%d")
check_await "${TMP_EGGPLANTS_OUTPUT_PATH}/${dt_slash_today}"

hadoop fs -test -e "${ODS_OTHER_DEVICE_DAILY}/${dt_slash_today}"
if [ $? -ne 0 ];then
  hadoop fs -mkdir -p "${ODS_OTHER_DEVICE_DAILY}/${dt_slash_today}"
fi

OUTPUT_PATH01="${TMP_COM_REYUN_PRACTICALTOOL_PATH}/${dt_slash_today}/01"
OUTPUT_PATH02="${TMP_COM_REYUN_PRACTICALTOOL_PATH}/${dt_slash_today}/02"
hadoop fs -rm -r "${OUTPUT_PATH01}"
hadoop fs -rm -r "${OUTPUT_PATH02}"

spark-submit --class mobvista.dmp.datasource.dm.ComReyunPracticaltool \
 --conf spark.yarn.executor.memoryOverhead=2048  \
 --conf spark.default.parallelism=2000 \
 --conf spark.sql.shuffle.partitions=2000 \
 --conf spark.driver.maxResultSize=4g \
 --conf spark.network.timeout=720s \
 --deploy-mode cluster  --executor-memory 8g --driver-memory 6g  --executor-cores 5  --num-executors 10 \
 ../../${JAR}   -output01 ${OUTPUT_PATH01}  -output02 ${OUTPUT_PATH02}   -coalesce 400


if [[ $? -ne 0 ]]; then
    exit 255
fi

hadoop distcp -m20  "${OUTPUT_PATH01}/*" "${TMP_EGGPLANTS_OUTPUT_PATH}/${dt_slash_today}/"
hadoop distcp -m20  "${OUTPUT_PATH02}/*" "${ODS_OTHER_DEVICE_DAILY}/${dt_slash_today}/"


: '
任务说明
运营侧有需要针对热云提供给我们的标签，拉取对应标签的设备进行伪包名入库的需求
运行以下sql
select
device_id
from dwh.device_tag_weight_event_all_weekly
where dt='20210922'
and id_type in('oaid','imei')
and tag_code = '0414'  ---需要更改tag_code
group by device_id

tag_code和人群包的命名关系如下
一共有5个，需要限制不同的tag_code
0414-com.reyun_practicaltool
041404-com.reyun_clean
040507-com.reyun_smalltool
041403-com.reyun_wifi
041406-com.reyun_security
'