appsflyer_total.sh

#!/usr/bin/env bash

source ../dmp_env.sh
source ././../ga_rawdata_analysis/common/tools.sh

today=$(date -d "$ScheduleTime 1 days ago" +"%Y/%m/%d")
dt_today=$(date -d "$ScheduleTime 1 days ago" +"%Y%m%d")
dt_today_dash=$(date -d "$ScheduleTime 1 days ago" +"%Y-%m-%d")
umount_time=$(date -d "$ScheduleTime 10 days ago" +"%Y%m%d")
umount_date_path=$(date -d "$ScheduleTime 10 days ago" +"%Y/%m/%d")

expire_path=$(date +%Y/%m/%d -d "-6 day $today")
expire_date=$(date +%Y%m%d -d "-6 day $today")

echo ${today}


## INPUT_PATH="${ETL_AF_ORG_DAILY}/${today}"
INPUT_PATH="${ETL_AF_ORG_NEW_DAILY}/${dt_today}"


OUTPUT_TOTAL_PATH="${ETL_AF_ORG_TOTAL}/${today}"
UMOUNT_OUTPUT_TOTAL_PATH="${ETL_AF_ORG_TOTAL}/${umount_date_path}"


DMP_USER_INFO_OUTPUT_PATH="${ODS_DMP_USER_INFO}/${today}/appsflyer"
DMP_USER_INFO_UNMOUNT_PATH="${ODS_DMP_USER_INFO}/${expire_path}/appsflyer"


DMP_INSTALL_LIST_OUTPUT_PATH="${DM_INSTALL_LIST}_v2/${today}/appsflyer"



## check_await "${INPUT_PATH}/_SUCCESS"
## check_await "${AF_ORG_TOTAL_YES}/_SUCCESS"

## mount_partition "etl_af_org_daily" "dt='${dt_today}'" "$INPUT_PATH"
mount_partition "etl_appsflyer_audience_org" "dt='${dt_today}'" "$INPUT_PATH"

mount_partition "dm_install_list_v2" "dt='${dt_today}', business='appsflyer'" "${DMP_INSTALL_LIST_OUTPUT_PATH}"



hadoop fs -rm -r "${OUTPUT_TOTAL_PATH}/"
hadoop fs -rm -r "${DMP_USER_INFO_OUTPUT_PATH}/"
hadoop fs -rm -r "${DMP_INSTALL_LIST_OUTPUT_PATH}/"




mount_partition "etl_af_org_total" "dt='${dt_today}'" "${OUTPUT_TOTAL_PATH}"
mount_partition "ods_dmp_user_info" "dt='${dt_today}', business='appsflyer'" "${DMP_USER_INFO_OUTPUT_PATH}"
: '
HIVE_CMD=$(hive_func)

$HIVE_CMD	-v  -hivevar  dt_today ${dt_today}     -hivevar update_date ${dt_today_dash}  -f	appsflyer_total.sql
'


 spark-submit --class mobvista.dmp.datasource.appsflyer.AppsFlyerTotal \
  --conf spark.yarn.executor.memoryOverhead=2048  \
  --conf spark.network.timeout=720s \
  --conf spark.default.parallelism=10 \
  --conf spark.sql.autoBroadcastJoinThreshold=31457280 \
  --jars s3://mob-emr-test/dataplatform/DataWareHouse/offline/myjar/hive-hcatalog-core-2.3.3.jar \
  --master yarn --deploy-mode cluster --name  apps_flyer_total --executor-memory 4g --driver-memory 4g  --executor-cores 3  --num-executors 5 \
  ../${JAR}  -outputtotal ${OUTPUT_TOTAL_PATH} -dmpuserinfo ${DMP_USER_INFO_OUTPUT_PATH} \
  -coalesce 20 \
  -today ${dt_today}  -update_date ${dt_today_dash}

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


unmount_partition "etl_af_org_total" "dt='${umount_time}'" "${UMOUNT_OUTPUT_TOTAL_PATH}"
unmount_partition "ods_dmp_user_info" "dt='${expire_date}', business='appsflyer'" "${DMP_USER_INFO_UNMOUNT_PATH}"



hive_cmd  "insert overwrite table dwh.dm_install_list_v2 partition(dt='${dt_today}',business='appsflyer')
     select device_id,
     device_type,
     platform,
     package_names,
     '${dt_today_dash}' update_date
     from dwh.etl_af_org_total where dt ='${dt_today}' "



hadoop fs -test -e ${OUTPUT_TOTAL_PATH}
if [ $? -ne 0 ];then
  hadoop fs -mkdir  ${OUTPUT_TOTAL_PATH}
fi

hadoop fs -test -e ${DMP_USER_INFO_OUTPUT_PATH}
if [ $? -ne 0 ];then
  hadoop fs -mkdir  ${DMP_USER_INFO_OUTPUT_PATH}
fi
hadoop fs -test -e ${DMP_INSTALL_LIST_OUTPUT_PATH}
if [ $? -ne 0 ];then
  hadoop fs -mkdir  ${DMP_INSTALL_LIST_OUTPUT_PATH}
fi

hadoop fs -touchz ${OUTPUT_TOTAL_PATH}/_SUCCESS
hadoop fs -touchz ${DMP_USER_INFO_OUTPUT_PATH}/_SUCCESS
hadoop fs -touchz ${DMP_INSTALL_LIST_OUTPUT_PATH}/_SUCCESS