chenwu/MarketAnalysis-ETL
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

new file: 01 dm_tf_external_sales.sql

Browse Source 
new file:   02 dm_td_external_keycompatitor.sql
	new file:   03 dm_td_external_brand_market.sql
	new file:   04 dm_td_external_calendar.sql
	new file:   05 dm_td_external_exchangerate.sql
	new file:   07 dm_td_external_packinfo.sql
	new file:   08 dm_td_external_corp.sql
	new file:   09 dm_td_external_geo_type.sql
	new file:   11 DM_TD_EXTERNAL_MARKET_NEW.sql
	new file:   12 dm_td_external_org.sql
	new file:   13 external auth.sql
	new file:   14 dm_tf_external_retail_special_bkp.sql
	new file:   AIA/01 dm_aia_pack_property.sql
	new file:   AIA/02 dm_ext_aia_sales.sql
	new file:   AIA/02 dm_ext_aia_sales_bakup_20230327.sql
	new file:   AIA/03 dm_aia_flag.sql
	new file:   AIA/04 dm_aia_provided_flag.sql
	new file:   AIA/06 DM_TD_EXT_AIA_PACK_PROPERTY.sql
	new file:   AIA/07 DM_TD_EXT_AIA_PACK2MARKET.sql
	new file:   AIA/08 DM_TD_EXT_AIA_MARKET_RATIO.sql
	new file:   AIA/09 DM_TD_EXT_AIA_MARKET_BRAND_RATIO.sql
	new file:   AIA/10 DM_TD_EXT_AIA_MARKET_PACK_MAPPING.sql
	new file:   AIA/11 DM_TD_EXT_AIA_TARGET_INST.sql
	new file:   AIA/11 dm_aia_targethp_flag.sql
	new file:   "AIA/z_03 dm_aia_flag_\345\244\207\344\273\275\347\224\250.sql"
	new file:   CHC/01 dm_chc_pack_property.sql
	new file:   CHC/02 DM_TF_EXT_CHC_SALES.sql
	new file:   CHC/03 DM_TD_EXT_CHC_PACK_PROPERTY.sql
	new file:   CHC/04 DM_TD_EXT_CHC_PACK2MARKET.sql
	new file:   CHC/05 DM_TD_EXT_CHC_MARKET_RATIO.sql
	new file:   CHC/06 DM_TD_EXT_CHC_MARKET_BRAND_RATIO.sql
	new file:   CHC/07 DM_TD_EXT_CHC_MARKET_PACK_MAPPING.sql
	new file:   CHPA/01 dwd_ims_atc_hierarchy.sql
	new file:   CHPA/01 dwd_ims_nfc_hierarchy.sql
	new file:   CHPA/01 dwd_ims_td_manufacturer_corp.sql
	new file:   CHPA/01 dwd_ims_td_pack_property.sql
	new file:   CHPA/01 dwd_update.sql
	new file:   CHPA/01_FB_BLOB_TO_DWD.sql
	new file:   CHPA/02 DWS_IMS_TD_GEO.sql
	new file:   CHPA/02 dws_ims_td_atc_cn.sql
	new file:   CHPA/02 dws_ims_td_corp_cn.sql
	new file:   CHPA/02 dws_ims_td_date.sql
	new file:   CHPA/02 dws_ims_td_manu_cn.sql
	new file:   CHPA/02 dws_ims_td_market.sql
	new file:   CHPA/02 dws_ims_td_market_ta.sql
	new file:   CHPA/02 dws_ims_td_nfc_cn.sql
	new file:   CHPA/02 dws_ims_td_prod_cn.sql
	new file:   CHPA/02 tmp_ims_td_prod_tmp.sql
	new file:   CHPA/02 tmp_ims_tf_fact_sales.sql
	new file:   CHPA/03 DM_TD_EXT_CHPA_MARKET_BRAND_RATIO.sql
	new file:   CHPA/03 DM_TD_EXT_CHPA_MARKET_PACK_MAPPING.sql
	new file:   CHPA/03 DM_TD_EXT_CHPA_MARKET_RATIO.sql
	new file:   CHPA/03 DM_TD_EXT_CHPA_PACK2MARKET.sql
	new file:   CHPA/03 DM_TD_EXT_CHPA_PACK_PROPERTY.sql
	new file:   CHPA/03 DM_TF_EXT_CHPA_SALES.sql
	new file:   CHPA/03 dm_ims_td_calendar.sql
	new file:   CHPA/03 dm_ims_td_geo.sql
	new file:   CHPA/03 dm_ims_td_market_property.sql
	new file:   CHPA/03 dm_ims_td_org.sql
	new file:   CHPA/03 dm_ims_td_org_hvh.sql
	new file:   CHPA/03 dm_ims_td_pack_property.sql
	new file:   CHPA/03 dm_ims_tf_sales.sql
	new file:   CHPA/03 dm_td_chpa_market_definition.sql
	new file:   CHPA/03 dm_td_ims_city_mapping.sql
	new file:   EC/03 ec_load_data.sql
	new file:   EC/04 DM_TD_EXT_EC_PACK_PROPERTY.sql
	new file:   EC/05 DM_TF_EXT_EC_SALES.sql
	new file:   EC/06 DM_TD_EXT_EC_PACK2MARKET.sql
	new file:   EC/07 DM_TD_EXT_EC_MARKET_RATIO.sql
	new file:   EC/08 DM_TD_EXT_EC_MARKET_BRAND_RATIO.sql
	new file:   EC/09 DM_TD_EXT_EC_MARKET_PACK_MAPPING.sql
	new file:   EC/1 (ec)blob_to_dwd.sql
	new file:   EC/2 dwd_inc_gnd_ext_ec_nationnal_pack_union_all.py
	new file:   Merged_Data/Merged_Data_Config_table_bkp.sql
	new file:   Merged_Data/Merged_Data_Config_table_bymonth.sql
	new file:   Merged_Data/dm_tf_exteranl_sales_merged_data_dtp_others_bkp.sql
	new file:   Merged_Data/dm_tf_exteranl_sales_merged_data_dtp_others_bymonth_bkp.sql
	new file:   ORG/DM_TD_EXT_AIA_ORG.sql
	new file:   ORG/DM_TD_EXT_CHC_ORG.sql
	new file:   ORG/DM_TD_EXT_CHPA_ORG.sql
	new file:   ORG/DM_TD_EXT_COUNTY_ORG.sql
	new file:   ORG/DM_TD_EXT_EC_ORG.sql
	new file:   ORG/DM_TD_EXT_RETAIL_ORG.sql
	new file:   ORG/DM_TD_EXT_THC_ORG.sql
	new file:   ORG/DM_TD_EXT_XIEHE_ORG.sql
	new file:   OTHERS/01 dm_td_report_url.sql
	new file:   OTHERS/02 dws_ext_email_warning.sql
	new file:   OTHERS/external_triggered_email.py
	new file:   Retail/01 load_tmp_data.py
	new file:   Retail/02 split_brand_data.py
	new file:   Retail/03 split_pack_data.py
	new file:   Retail/04 map_to_dws_table.py
	new file:   Retail/05 load_dtp_temp_data.py
	new file:   Retail/06 split_dtp_brand_data.py
	new file:   Retail/07 split_dtp_pack_data.py
	new file:   Retail/08 map_to_dtp_dws_table.py
	new file:   Retail/09 dwd_inc_gnd_ext_retail_nataional.py
	new file:   Retail/10 map_to_retail_dm_table.py
	new file:   Retail/11 map_to_overview_dm_table.py
	new file:   Retail/12 dws_tf_external_retail_dtp_special.sql
	new file:   Retail/13 DM_TF_EXT_RETAIL_SALES.sql
	new file:   Retail/14 DM_TF_EXT_RETAIL_DTP_SALES.sql
	new file:   Retail/15 DM_TD_EXT_RETAIL_PACK_PROPERTY.sql
	new file:   Retail/16 DM_TD_EXT_RETAIL_DTP_PACK_PROPERTY.sql
	new file:   Retail/17 DM_TD_EXT_DTP_PACK2MARKET.sql
	new file:   Retail/17 DM_TD_EXT_RETAIL_PACK2MARKET.sql
	new file:   Retail/18 DM_TD_EXT_DTP_MARKET_RATIO.sql
	new file:   Retail/18 DM_TD_EXT_RETAIL_MARKET_RATIO.sql
	new file:   Retail/19 DM_TD_EXT_DTP_MARKET_BRAND_RATIO.sql
	new file:   Retail/19 DM_TD_EXT_RETAIL_MARKET_BRAND_RATIO.sql
	new file:   Retail/20 DM_TD_EXT_DTP_MARKET_PACK_MAPPING.sql
	new file:   Retail/20 DM_TD_EXT_RETAIL_MARKET_PACK_MAPPING.sql
	new file:   "Retail/z1 dwd_inc_gnd_ext_retail_nataional_\344\275\234\345\272\237.py"
	new file:   "Retail/z2 retail_load_data_\344\275\234\345\272\237.sql"
	new file:   "Retail/z3 retail_overview_data_\344\275\234\345\272\237.sql"
	new file:   THC/01 dm_thc_pack_property.sql
	new file:   THC/02 dm_ext_thc_sales.sql
	new file:   THC/02 dm_ext_thc_sales_bakup_20260327.sql
	new file:   THC/03 DM_TF_EXT_THC_MARKET_SALES_CHT.sql
	new file:   THC/04 dm_tf_external_sales_thc.sql
	new file:   THC/05 DM_TD_EXT_THC_PACK_PROPERTY.sql
	new file:   THC/06 DM_TD_EXT_THC_PACK2MARKET.sql
	new file:   THC/07 DM_TD_EXT_THC_MARKET_RATIO.sql
	new file:   THC/08 DM_TD_EXT_THC_MARKET_BRAND_RATIO.sql
	new file:   THC/09 DM_TD_EXT_THC_MARKET_PACK_MAPPING.sql
	new file:   UNIONALL/DM_TD_EXT_UNIONALL_MARKET_PACK_MAPPING.sql
	new file:   UNIONALL/DM_TD_EXT_UNIONALL_PACKINFO.sql
	new file:   UNIONALL/DM_TF_EXT_UNIONALL_MARKET_SALES.sql
	new file:   UNIONALL/DM_TF_EXT_UNIONALL_MARKET_SALES_v1.0.sql
	new file:   UNIONALL/DM_TF_EXT_UNIONALL_SALES.sql
	new file:   UNIONALL/DM_TF_EXT_UNIONALL_SALES_MAPPING.sql
	new file:   UNIONALL/DM_TF_EXT_UNIONALL_SALES_MAPPING_NIAD.sql
	new file:   XIEHE/01 xiehe_blob_to_dwd.py
	new file:   XIEHE/02 dm_xiehe_pack_property.sql
	new file:   XIEHE/03 dm_ext_xiehe_geo.sql
	new file:   XIEHE/04 dm_ext_xiehe_sales.sql
	new file:   XIEHE/05 dm_td_xiehe_core_dept.sql
	new file:   XIEHE/06 DM_TF_EXT_XIEHE_SALES.sql
	new file:   XIEHE/07 DM_TD_EXT_XIEHE_PACK_PROPERTY.sql
	new file:   XIEHE/08 DM_TD_EXT_XIEHE_PACK2MARKET.sql
	new file:   XIEHE/09 DM_TD_EXT_XIEHE_MARKET_RATIO.sql
	new file:   XIEHE/10 DM_TD_EXT_XIEHE_MARKET_BRAND_RATIO.sql
	new file:   XIEHE/11 DM_TD_EXT_XIEHE_MARKET_PACK_MAPPING.sql
	new file:   XIEHE/bkp_01 xiehe_blob2dwd.py
	new file:   XIEHE/bkp_02 dm_ext_xiehe_sales.sql
	new file:   XIEHE/bkp_03 dm_ext_xiehe_pack_property.sql
	new file:   county/01 tmp_ims_county_fact_sales_sum.sql
	new file:   county/02 tmp_imscounty_Result.sql
	new file:   county/03 dm_ims_td_county_geo.sql
	new file:   county/04 dws_ext_county_tf_sales.sql
	new file:   county/05 dm_ext_county_td_pack_property.sql
	new file:   county/06 dm_td_county_pack_region.sql
	new file:   county/07 dm_ext_county_tf_sales_region.sql
	new file:   county/08 DM_TD_EXT_COUNTY_PACK_PROPERTY.sql
	new file:   county/09 DM_TF_EXT_COUNTY_SALES.sql
	new file:   county/10 DM_TD_EXT_COUNTY_PACK2MARKET.sql
	new file:   county/11 DM_TD_EXT_COUNTY_MARKET_RATIO.sql
	new file:   county/12 DM_TD_EXT_COUNTY_MARKET_BRAND_RATIO.sql
	new file:   county/13 DM_TD_EXT_COUNTY_MARKET_PACK_MAPPING.sql
	new file:   for_AIA_Dashboard/01 dm_td_aia_inst_mkt.sql
	new file:   for_AIA_Dashboard/02 dm_td_aia_auth_sales.sql
	new file:   for_AIA_Dashboard/03 dm_td_aia_original_col.sql
	new file:   for_AIA_Dashboard/04 dm_td_aia_nosales_inst.sql
	new file:   for_AIA_Dashboard/05 dm_td_aia_is_eagle_flag.sql
	new file:   for_AIA_Dashboard/06 dm_td_aia_rank.sql
	new file:   for_AIA_Dashboard/07 dm_ext_aia_data_remove_flag.sql
	new file:   for_AIA_Dashboard/07 dm_td_aia_remove_special_ins_bkp.py
	new file:   for_AIA_Dashboard/08 dm_ext_aia_data_quality_flag.sql
	new file:   z 01 dm_tf_external_sales.sql
	new file:   "z 10 dm_td_external_market_pack_mapping_\344\275\234\345\272\237.sql"
	new file:   "z 11 dm_td_external_market_\344\275\234\345\272\237.sql"
	new file:   "\344\270\212\347\272\277\350\204\232\346\234\254.sql"
	new file:   "\346\225\260\346\215\256\351\252\214\350\257\201.sql"
This commit is contained in:
chenwu
2026-04-27 15:48:38 +08:00
parent d2485035c0
commit 832c7bcd69
173 changed files with 62579 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1544
EC/03 ec_load_data.sql Normal file
View File

File diff suppressed because it is too large Load Diff

146
EC/04 DM_TD_EXT_EC_PACK_PROPERTY.sql Normal file
View File

@@ -0,0 +1,146 @@
-- Databricks notebook source
-- CREATE OR REPLACE TABLE DM.DM_TD_EXT_EC_PACK_PROPERTY
-- (
-- MARKET_PACK_KEY STRING,
-- PACK_CODE STRING,
-- PACK_DESC STRING,
-- STGH_DESC STRING,
-- PACK_LCH STRING,
-- FAMILY_CODE STRING,
-- FAMILY_NAME STRING,
-- PROD_CODE STRING,
-- PROD_DESC STRING,
-- PROD_DESC_C STRING,
-- CMPS_CODE STRING,
-- CMPS_DESC STRING,
-- CMPS_DESC_C STRING,
-- ATC1_CODE STRING,
-- ATC2_CODE STRING,
-- ATC3_CODE STRING,
-- ATC4_CODE STRING,
-- APP1_CODE STRING,
-- APP2_CODE STRING,
-- APP3_CODE STRING,
-- BIO_DESC STRING,
-- GENE_ORIG_DESC STRING,
-- ETH_OTC_DESC STRING,
-- NRDL_DESC STRING,
-- NRDL_ENTRY_DATE STRING,
-- EDL_DESC STRING,
-- TCM_DESC STRING,
-- PAED_DESC STRING,
-- GQCE_DESC STRING,
-- VBP_DESC_V STRING,
-- VBP_DESC STRING,
-- MANU_CODE STRING,
-- MANU_DESC STRING,
-- MANU_DESC_C STRING,
-- MNFL_CODE STRING,
-- MNFL_DESC STRING,
-- CORP_CODE STRING,
-- CORP_DESC STRING,
-- CORP_DESC_C STRING,
-- BRANDTYPE STRING,
-- MARKET STRING,
-- KEY_COMPETITOR STRING,
-- IS_AZ STRING,
-- AZ_MAIN STRING,
-- AZ_RELATED STRING,
-- ATC1_DESC STRING,
-- ATC1_DESC_C STRING,
-- ATC2_DESC STRING,
-- ATC2_DESC_C STRING,
-- ATC3_DESC STRING,
-- ATC3_DESC_C STRING,
-- ATC4_DESC STRING,
-- ATC4_DESC_C STRING,
-- APP1_DESC STRING,
-- APP1_DESC_C STRING,
-- APP2_DESC STRING,
-- APP2_DESC_C STRING,
-- APP3_DESC STRING,
-- APP3_DESC_C STRING,
-- CLASS STRING,
-- MARKET_RATIO STRING,
-- COUNTINGUNIT STRING,
-- VBP_BRAND STRING,
-- REPLENISH_FALG STRING,
-- ETL_INSERT_DT TIMESTAMP,
-- ETL_UPDATE_DT TIMESTAMP
-- )
-- USING delta
-- -- LOCATION 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_pack_property';
-- -- 上面是生产环境location，下面是测试环境location
-- LOCATION 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_pack_property';
-- COMMAND ----------
INSERT OVERWRITE TABLE DM.DM_TD_EXT_EC_PACK_PROPERTY
SELECT
MARKET_PACK_KEY,
iqvia_pack_code AS PACK_CODE,
pack_des AS PACK_DESC,
stgh_des AS STGH_DESC,
pack_lch AS PACK_LCH,
Family_Code AS FAMILY_CODE,
Family_Name AS FAMILY_NAME,
prod_cod AS PROD_CODE,
prod_des AS PROD_DESC,
prod_des_c AS PROD_DESC_C,
cmps_cod AS CMPS_CODE,
cmps_des AS CMPS_DESC,
cmps_des_c AS CMPS_DESC_C,
atc1_cod AS ATC1_CODE,
atc2_cod AS ATC2_CODE,
atc3_cod AS ATC3_CODE,
atc4_cod AS ATC4_CODE,
app1_cod AS APP1_CODE,
app2_cod AS APP2_CODE,
app3_cod AS APP3_CODE,
bio_desc AS BIO_DESC,
gene_orig_desc AS GENE_ORIG_DESC,
eth_otc_desc AS ETH_OTC_DESC,
nrdl_desc AS NRDL_DESC,
NRDL_ENTRY_DATE,
edl_desc AS EDL_DESC,
tcm_desc AS TCM_DESC,
paed_desc AS PAED_DESC,
gqce_desc AS GQCE_DESC,
NULL AS VBP_DESC_V,
vbp_desc AS VBP_DESC,
manu_cod AS MANU_CODE,
manu_des AS MANU_DESC,
manu_des_c AS MANU_DESC_C,
mnfl_cod AS MNFL_CODE,
mnfl_des AS MNFL_DESC,
corp_cod AS CORP_CODE,
corp_des AS CORP_DESC,
CORP_DES_C AS CORP_DESC_C,
BrandType AS BRANDTYPE,
market AS MARKET,
KEY_COMPETITOR,
is_az AS IS_AZ,
AZ_MAIN,
AZ_Related AS AZ_RELATED,
atc1_des AS ATC1_DESC,
atc1_des_c AS ATC1_DESC_C,
atc2_des AS ATC2_DESC,
atc2_des_c AS ATC2_DESC_C,
atc3_des AS ATC3_DESC,
atc3_des_c AS ATC3_DESC_C,
atc4_des AS ATC4_DESC,
atc4_des_c AS ATC4_DESC_C,
app1_des AS APP1_DESC,
app1_des_c AS APP1_DESC_C,
app2_des AS APP2_DESC,
app2_des_c AS APP2_DESC_C,
app3_des AS APP3_DESC,
app3_des_c AS APP3_DESC_C,
class AS CLASS,
extend_market_ratio AS MARKET_RATIO,
counting_unit AS COUNTINGUNIT,
NULL AS VBP_BRAND,
NULL AS REPLENISH_FALG,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_INSERT_DT,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_UPDATE_DT
FROM DM.DM_ZK_EC_PACK_PROPERTY;

173
EC/05 DM_TF_EXT_EC_SALES.sql Normal file
View File

@@ -0,0 +1,173 @@
-- Databricks notebook source
-- CREATE OR REPLACE TABLE DM.DM_TF_EXT_EC_SALES (
-- YYYYMM STRING,
-- PACK_CODE STRING,
-- CORP_CODE STRING,
-- AUDIT_CODE STRING,
-- PLATFORM_TYPE STRING,
-- STORE_NAME STRING,
-- STORE_TYPE STRING,
-- REGION_TYPE STRING,
-- PACK_FLAG INT,
-- PROD_FLAG INT,
-- DTP_FLAG INT,
-- SALES_UNIT_CAL DECIMAL(38,10),
-- SALES_UNIT_CAL_LY DECIMAL(38,10),
-- SALES_VALUE_CAL DECIMAL(38,10),
-- SALES_VALUE_CAL_LY DECIMAL(38,10),
-- CONUTING_UNIT DECIMAL(38,10),
-- CONUTING_UNIT_LY DECIMAL(38,10),
-- DATA_SOURCE STRING,
-- INST_CODE STRING COMMENT '内部机构编码',
-- CMPS_FLAG STRING COMMENT '分子式标签',
-- DEPT_NAME STRING COMMENT '科室名称',
-- PRESCRIPTION DECIMAL(38,10) COMMENT '处方张数',
-- PRESCRIPTION_LY DECIMAL(38,10) COMMENT '去年同期处方张数',
-- NEW_CODE STRING COMMENT '主数据关联CODE',
-- AREA STRING COMMENT '城市',
-- H_LEVEL STRING COMMENT '医院类型',
-- REIMBURSE STRING COMMENT '报销情况',
-- REIMBURSE_TYPE STRING COMMENT '报销类型',
-- PRESCRIPTION_SOURCE STRING COMMENT '处方来源',
-- ETL_INSERT_DT TIMESTAMP,
-- ETL_UPDATE_DT TIMESTAMP)
-- USING delta
-- -- LOCATION 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_tf_ext_ec_sales';
-- -- 上面是生产环境location，下面是测试环境location
-- LOCATION 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_tf_ext_ec_sales';
-- COMMAND ----------
CREATE OR REPLACE TEMPORARY VIEW EC_SALES
AS
SELECT
A.YYYYMM,
A.iqvia_pack_code AS PACK_CODE,
A.AUDIT_COD AS AUDIT_CODE,
'EC(Monthly)' DATA_SOURCE,
-------------------------------------
SUM(sales_unit) SALES_UNIT_CAL,
SUM(sales_unit_ly) SALES_UNIT_CAL_LY,
SUM(sales_value) SALES_VALUE_CAL,
SUM(sales_value_LY) SALES_VALUE_CAL_LY,
SUM(counting_unit) CONUTING_UNIT,
SUM(counting_unit_LY) CONUTING_UNIT_LY,
null as prescription,
null as prescription_ly,
-------------------------------------
--Retail藏数逻辑标签------------------
1 PACK_FLAG,
1 PROD_FLAG,
0 DTP_FLAG,
null as cmps_flag,
-------------------------------------
--EC数据标签--------------------------
A.PLATFORM AS PLATFORM_TYPE,
A.STORE_NAME,
A.STORE_TYPE,
-------------------------------------
--COUNTY数据标签----------------------
'' REGION_TYPE,
-------------------------------------
--AIA数据标签-------------------------
null inst_code, --可以移除，报告没有使用
-------------------------------------
--XIE HE 数据标签---------------------
'' as dept_name,
'' as new_code,
'' as area,
'' as h_level,
'' as reimburse,
'' as reimburse_type,
'' as prescription_source,
-------------------------------------
'' as CORP_CODE --可以移除，报告没有使用
-------------------------------------
FROM
dm.dm_zk_ec_sales A
where
A.YYYYMM >= '202001'
GROUP BY
A.YYYYMM,
A.iqvia_pack_code,
A.AUDIT_COD,
A.PLATFORM,
A.STORE_NAME,
A.STORE_TYPE,
A.PACK_FLAG,
A.PROD_FLAG
-- COMMAND ----------
INSERT OVERWRITE TABLE DM.DM_TF_EXT_EC_SALES
(
YYYYMM,
PACK_CODE,
AUDIT_CODE,
DATA_SOURCE,
SALES_UNIT_CAL,
SALES_UNIT_CAL_LY,
SALES_VALUE_CAL,
SALES_VALUE_CAL_LY,
CONUTING_UNIT,
CONUTING_UNIT_LY,
PRESCRIPTION,
PRESCRIPTION_LY,
PACK_FLAG,
PROD_FLAG,
DTP_FLAG,
CMPS_FLAG,
PLATFORM_TYPE,
STORE_NAME,
STORE_TYPE,
REGION_TYPE,
INST_CODE,
DEPT_NAME,
NEW_CODE,
AREA,
H_LEVEL,
REIMBURSE,
REIMBURSE_TYPE,
PRESCRIPTION_SOURCE,
CORP_CODE,
ETL_INSERT_DT,
ETL_UPDATE_DT
)
SELECT
YYYYMM,
NVL (
NULLIF(PACK_CODE, ''),
CONCAT ('PACK_CODE_', DATA_SOURCE)
) AS PACK_CODE,
NVL (NULLIF(AUDIT_CODE, ''), 'ROC') AS AUDIT_CODE,
DATA_SOURCE,
SALES_UNIT_CAL,
SALES_UNIT_CAL_LY,
SALES_VALUE_CAL,
SALES_VALUE_CAL_LY,
CONUTING_UNIT,
CONUTING_UNIT_LY,
PRESCRIPTION,
PRESCRIPTION_LY,
PACK_FLAG,
PROD_FLAG,
DTP_FLAG,
CMPS_FLAG,
PLATFORM_TYPE,
STORE_NAME,
STORE_TYPE,
REGION_TYPE,
INST_CODE,
DEPT_NAME,
NEW_CODE,
AREA,
H_LEVEL,
REIMBURSE,
REIMBURSE_TYPE,
PRESCRIPTION_SOURCE,
CORP_CODE,
FROM_UTC_TIMESTAMP (CURRENT_TIMESTAMP(), 'UTC+8') AS ETL_INSERT_DT,
FROM_UTC_TIMESTAMP (CURRENT_TIMESTAMP(), 'UTC+8') AS ETL_UPDATE_DT
FROM
EC_SALES
;

208
EC/06 DM_TD_EXT_EC_PACK2MARKET.sql Normal file
View File

@@ -0,0 +1,208 @@
-- Databricks notebook source
-- CREATE OR REPLACE TABLE DM.DM_TD_EXT_EC_PACK2MARKET (
-- MARKET STRING,
-- PACK_CODE STRING,
-- PACK_DESC STRING,
-- STGH_DESC STRING,
-- PACK_LCH STRING,
-- PROD_CODE STRING,
-- CMPS_CODE STRING,
-- CMPS_DESC STRING,
-- ATC1_CODE STRING,
-- ATC2_CODE STRING,
-- ATC3_CODE STRING,
-- ATC4_CODE STRING,
-- APP1_CODE STRING,
-- APP2_CODE STRING,
-- APP3_CODE STRING,
-- BIO_DESC STRING,
-- GENE_ORIG_DESC STRING,
-- ETH_OTC_DESC STRING,
-- NRDL_DESC STRING,
-- NRDL_ENTRY_DATE STRING,
-- EDL_DESC STRING,
-- TCM_DESC STRING,
-- PAED_DESC STRING,
-- GQCE_DESC STRING,
-- VBP_DESC STRING,
-- MANU_CODE STRING,
-- MANU_DESC STRING,
-- MNFL_CODE STRING,
-- MNFL_DESC STRING,
-- CORP_CODE STRING,
-- CORP_DESC STRING,
-- BRANDTYPE STRING,
-- BU STRING,
-- STARTTIME STRING,
-- ENDTIME STRING,
-- MARKET_RATIO STRING,
-- ETL_INSERT_DT TIMESTAMP,
-- ETL_UPDATE_DT TIMESTAMP)
-- USING delta
-- -- LOCATION 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_pack2market';
-- -- 上面是生产环境location，下面是测试环境location
-- LOCATION 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_pack2market';
-- COMMAND ----------
-- 使用 EXCEPT（集合差操作）简化逻辑
INSERT OVERWRITE TABLE DM.DM_TD_EXT_EC_PACK2MARKET
-- 包含规则
SELECT DISTINCT
T2.MARKET,
T1.PACK_CODE,
T1.PACK_DESC,
T1.STGH_DESC,
T1.PACK_LCH,
T1.PROD_CODE,
T1.CMPS_CODE,
T1.CMPS_DESC,
T1.ATC1_CODE,
T1.ATC2_CODE,
T1.ATC3_CODE,
T1.ATC4_CODE,
T1.APP1_CODE,
T1.APP2_CODE,
T1.APP3_CODE,
T1.BIO_DESC,
T1.GENE_ORIG_DESC AS GENE_ORIG,
T1.ETH_OTC_DESC AS ETH_OTC_D,
T1.NRDL_DESC,
T1.NRDL_ENTRY_DATE AS NRDL_ENTR,
T1.EDL_DESC,
T1.TCM_DESC,
T1.PAED_DESC,
T1.GQCE_DESC,
T1.VBP_DESC,
T1.MANU_CODE,
T1.MANU_DESC,
T1.MNFL_CODE,
T1.MNFL_DESC,
T1.CORP_CODE,
T1.CORP_DESC,
T1.BRANDTYPE,
T2.BU,
T2.STARTTIME,
T2.ENDTIME,
COALESCE(T2.EXTEND_MARKET_RATIO, '1') AS MARKET_RATIO,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_INSERT_DT,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_UPDATE_DT
FROM DM.DM_TD_EXT_EC_PACK_PROPERTY AS T1
LEFT JOIN (
SELECT
MARKET,
BU,
STARTTIME,
ENDTIME,
EXTEND_MARKET_RATIO,
ATC1_CODE,
ATC2_CODE,
ATC3_CODE,
ATC4_CODE,
NFC1_CODE,
NFC2_CODE,
NFC3_CODE,
CORPORATION_CODE,
MANUFACTURER_CODE,
PRODUCT_CODE,
PACK_CODE,
STRENGTH,
MOLECULE_CODE
FROM DWD.DWD_GND_IMS_TBLMARKET_BYMONTH
WHERE EXTEND_MARKET IS NULL AND (NOT_IN_FLAG IS NULL OR NOT_IN_FLAG = '1')
) AS T2
ON (T2.ATC1_CODE IS NULL OR T1.ATC1_CODE = T2.ATC1_CODE)
AND (T2.ATC2_CODE IS NULL OR T1.ATC2_CODE = T2.ATC2_CODE)
AND (T2.ATC3_CODE IS NULL OR T1.ATC3_CODE = T2.ATC3_CODE)
AND (T2.ATC4_CODE IS NULL OR T1.ATC4_CODE = T2.ATC4_CODE)
AND (T2.NFC1_CODE IS NULL OR T1.APP1_CODE = T2.NFC1_CODE)
AND (T2.NFC2_CODE IS NULL OR T1.APP2_CODE = T2.NFC2_CODE)
AND (T2.NFC3_CODE IS NULL OR T1.APP3_CODE = T2.NFC3_CODE)
AND (T2.CORPORATION_CODE IS NULL OR T1.CORP_CODE = T2.CORPORATION_CODE)
AND (T2.MANUFACTURER_CODE IS NULL OR T1.MANU_CODE = T2.MANUFACTURER_CODE)
AND (T2.PRODUCT_CODE IS NULL OR T1.PROD_CODE = T2.PRODUCT_CODE)
AND (T2.PACK_CODE IS NULL OR T1.PACK_CODE = T2.PACK_CODE)
AND (T2.STRENGTH IS NULL OR T1.STGH_DESC = T2.STRENGTH)
AND (T2.MOLECULE_CODE IS NULL OR T1.CMPS_CODE = T2.MOLECULE_CODE)
WHERE T2.MARKET IS NOT NULL
EXCEPT
-- 排除规则
SELECT DISTINCT
T2.MARKET,
T1.PACK_CODE,
T1.PACK_DESC,
T1.STGH_DESC,
T1.PACK_LCH,
T1.PROD_CODE,
T1.CMPS_CODE,
T1.CMPS_DESC,
T1.ATC1_CODE,
T1.ATC2_CODE,
T1.ATC3_CODE,
T1.ATC4_CODE,
T1.APP1_CODE,
T1.APP2_CODE,
T1.APP3_CODE,
T1.BIO_DESC,
T1.GENE_ORIG_DESC AS GENE_ORIG,
T1.ETH_OTC_DESC AS ETH_OTC_D,
T1.NRDL_DESC,
T1.NRDL_ENTRY_DATE AS NRDL_ENTR,
T1.EDL_DESC,
T1.TCM_DESC,
T1.PAED_DESC,
T1.GQCE_DESC,
T1.VBP_DESC,
T1.MANU_CODE,
T1.MANU_DESC,
T1.MNFL_CODE,
T1.MNFL_DESC,
T1.CORP_CODE,
T1.CORP_DESC,
T1.BRANDTYPE,
T2.BU,
T2.STARTTIME,
T2.ENDTIME,
COALESCE(T2.EXTEND_MARKET_RATIO, '1') AS MARKET_RATIO,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_INSERT_DT,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_UPDATE_DT
FROM DM.DM_TD_EXT_EC_PACK_PROPERTY AS T1
LEFT JOIN (
SELECT
MARKET,
BU,
STARTTIME,
ENDTIME,
EXTEND_MARKET_RATIO,
ATC1_CODE,
ATC2_CODE,
ATC3_CODE,
ATC4_CODE,
NFC1_CODE,
NFC2_CODE,
NFC3_CODE,
CORPORATION_CODE,
MANUFACTURER_CODE,
PRODUCT_CODE,
PACK_CODE,
STRENGTH,
MOLECULE_CODE
FROM DWD.DWD_GND_IMS_TBLMARKET_BYMONTH
WHERE EXTEND_MARKET IS NULL AND NOT_IN_FLAG = '0'
) AS T2
ON (T2.ATC1_CODE IS NULL OR T1.ATC1_CODE = T2.ATC1_CODE)
AND (T2.ATC2_CODE IS NULL OR T1.ATC2_CODE = T2.ATC2_CODE)
AND (T2.ATC3_CODE IS NULL OR T1.ATC3_CODE = T2.ATC3_CODE)
AND (T2.ATC4_CODE IS NULL OR T1.ATC4_CODE = T2.ATC4_CODE)
AND (T2.NFC1_CODE IS NULL OR T1.APP1_CODE = T2.NFC1_CODE)
AND (T2.NFC2_CODE IS NULL OR T1.APP2_CODE = T2.NFC2_CODE)
AND (T2.NFC3_CODE IS NULL OR T1.APP3_CODE = T2.NFC3_CODE)
AND (T2.CORPORATION_CODE IS NULL OR T1.CORP_CODE = T2.CORPORATION_CODE)
AND (T2.MANUFACTURER_CODE IS NULL OR T1.MANU_CODE = T2.MANUFACTURER_CODE)
AND (T2.PRODUCT_CODE IS NULL OR T1.PROD_CODE = T2.PRODUCT_CODE)
AND (T2.PACK_CODE IS NULL OR T1.PACK_CODE = T2.PACK_CODE)
AND (T2.STRENGTH IS NULL OR T1.STGH_DESC = T2.STRENGTH)
AND (T2.MOLECULE_CODE IS NULL OR T1.CMPS_CODE = T2.MOLECULE_CODE)
WHERE T2.MARKET IS NOT NULL;

30
EC/07 DM_TD_EXT_EC_MARKET_RATIO.sql Normal file
View File

@@ -0,0 +1,30 @@
-- Databricks notebook source
-- CREATE OR REPLACE TABLE DM.DM_TD_EXT_EC_MARKET_RATIO (
-- MARKET STRING,
-- PACK_CODE STRING,
-- CMPS_CODE STRING,
-- STARTTIME STRING,
-- ENDTIME STRING,
-- MARKET_RATIO STRING,
-- ETL_INSERT_DT TIMESTAMP,
-- ETL_UPDATE_DT TIMESTAMP)
-- USING delta
-- -- LOCATION 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_market_ratio';
-- -- 上面是生产环境location，下面是测试环境location
-- LOCATION 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_market_ratio';
-- COMMAND ----------
INSERT OVERWRITE TABLE DM.DM_TD_EXT_EC_MARKET_RATIO
SELECT
DISTINCT
MARKET,
PACK_CODE,
CMPS_CODE,
STARTTIME,
ENDTIME,
MARKET_RATIO,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_INSERT_DT,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_UPDATE_DT
FROM DM.DM_TD_EXT_EC_PACK2MARKET
WHERE UPPER(MARKET) NOT LIKE '%ALL%'

163
EC/08 DM_TD_EXT_EC_MARKET_BRAND_RATIO.sql Normal file
View File

@@ -0,0 +1,163 @@
-- Databricks notebook source
-- CREATE OR REPLACE TABLE DM.DM_TD_EXT_EC_MARKET_BRAND_RATIO (
-- MARKET STRING,
-- PACK_CODE STRING,
-- VALUE_BRAND_RATIO FLOAT,
-- VALUE_BRAND_RATIO_START STRING,
-- VALUE_BRAND_RATIO_END STRING,
-- UNIT_BRAND_RATIO FLOAT,
-- UNIT_BRAND_START STRING,
-- UNIT_BRAND_END STRING,
-- COUNTINGUNIT_BRAND_RATIO FLOAT,
-- COUNTINGUNIT_BRAND_START STRING,
-- COUNTINGUNIT_BRAND_END STRING,
-- PDOT_BRAND_RATIO FLOAT,
-- PDOT_BRAND_START STRING,
-- PDOT_BRAND_END STRING,
-- ETL_INSERT_DT TIMESTAMP,
-- ETL_UPDATE_DT TIMESTAMP)
-- USING delta
-- -- LOCATION 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_market_brand_ratio';
-- -- 上面是生产环境location，下面是测试环境location
-- LOCATION 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_market_brand_ratio';
-- COMMAND ----------
INSERT OVERWRITE TABLE DM.DM_TD_EXT_EC_MARKET_BRAND_RATIO
SELECT
DISTINCT
T1.MARKET,
T1.PACK_CODE,
CASE
WHEN T2.RATIO IS NOT NULL THEN T2.RATIO
ELSE NVL(T5.RATIO, 1) END AS VALUE_BRAND_RATIO,
CASE
WHEN T2.RATIO IS NOT NULL THEN T2.STARTTIME
ELSE NVL(T5.STARTTIME, 200001) END AS VALUE_BRAND_RATIO_START,
CASE
WHEN T2.RATIO IS NOT NULL THEN T2.ENDTIME
ELSE NVL(T5.ENDTIME, 299912) END AS VALUE_BRAND_RATIO_END,
CASE
WHEN T3.RATIO IS NOT NULL THEN T3.RATIO
ELSE NVL(T6.RATIO, 1) END AS UNIT_BRAND_RATIO,
CASE
WHEN T3.RATIO IS NOT NULL THEN T3.STARTTIME
ELSE NVL(T6.STARTTIME, 200001) END AS UNIT_BRAND_START,
CASE
WHEN T3.RATIO IS NOT NULL THEN T3.ENDTIME
ELSE NVL(T6.ENDTIME, 299912) END AS UNIT_BRAND_END,
1 AS COUNTINGUNIT_BRAND_RATIO,
200001 AS COUNTINGUNIT_BRAND_START,
299912 AS COUNTINGUNIT_BRAND_END,
CASE
WHEN T4.RATIO IS NOT NULL THEN T4.RATIO
ELSE NVL(T7.RATIO, 1) END AS PDOT_BRAND_RATIO,
CASE
WHEN T4.RATIO IS NOT NULL THEN T4.STARTTIME
ELSE NVL(T7.STARTTIME, 200001) END AS PDOT_BRAND_START,
CASE
WHEN T4.RATIO IS NOT NULL THEN T4.ENDTIME
ELSE NVL(T7.ENDTIME, 299912) END AS PDOT_BRAND_END,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_INSERT_DT,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_UPDATE_DT
FROM DM.DM_TD_EXT_EC_MARKET_RATIO T1
LEFT JOIN ( --VALUE--PACK
SELECT
DISTINCT
MARKET,
CMPS_COD,
CMPS_DES,
PACK_COD,
PACK_DES,
LEVEL,
TYPE,
CAST(RATIO AS FLOAT) AS RATIO,
CAST(STARTTIME AS INT) STARTTIME,
CAST(ENDTIME AS INT) ENDTIME
FROM DWD.DWD_GND_IMS_TBLBRANDRATIO_BYMONTH
WHERE UPPER(TYPE) = 'VALUE' AND UPPER(LEVEL) = 'PACK'
) T2 ON T1.MARKET = T2.MARKET AND T1.PACK_CODE = T2.PACK_COD
LEFT JOIN ( --UNIT--PACK
SELECT
DISTINCT
MARKET,
CMPS_COD,
CMPS_DES,
PACK_COD,
PACK_DES,
LEVEL,
TYPE,
CAST(RATIO AS FLOAT) AS RATIO,
CAST(STARTTIME AS INT) STARTTIME,
CAST(ENDTIME AS INT) ENDTIME
FROM DWD.DWD_GND_IMS_TBLBRANDRATIO_BYMONTH
WHERE UPPER(TYPE) = 'UNIT' AND UPPER(LEVEL) = 'PACK'
) T3 ON T1.MARKET = T3.MARKET AND T1.PACK_CODE = T3.PACK_COD
LEFT JOIN ( --COUNTINGUNIT--PACK
SELECT
DISTINCT
MARKET,
CMPS_COD,
CMPS_DES,
PACK_COD,
PACK_DES,
LEVEL,
TYPE,
CAST(RATIO AS FLOAT) AS RATIO,
CAST(STARTTIME AS INT) STARTTIME,
CAST(ENDTIME AS INT) ENDTIME
FROM DWD.DWD_GND_IMS_TBLBRANDRATIO_BYMONTH
WHERE UPPER(TYPE) = 'COUNTINGUNIT' AND UPPER(LEVEL) = 'PACK'
) T4 ON T1.MARKET = T4.MARKET AND T1.PACK_CODE = T4.PACK_COD
LEFT JOIN ( --VALUE--MOLECULE
SELECT
DISTINCT
MARKET,
CMPS_COD,
CMPS_DES,
PACK_COD,
PACK_DES,
LEVEL,
TYPE,
CAST(RATIO AS FLOAT) AS RATIO,
CAST(STARTTIME AS INT) STARTTIME,
CAST(ENDTIME AS INT) ENDTIME
FROM DWD.DWD_GND_IMS_TBLBRANDRATIO_BYMONTH
WHERE UPPER(TYPE) = 'VALUE' AND UPPER(LEVEL) = 'MOLECULE'
) T5 ON T1.MARKET = T5.MARKET AND T1.CMPS_CODE = T5.CMPS_COD
LEFT JOIN ( --UNIT--MOLECULE
SELECT
DISTINCT
MARKET,
CMPS_COD,
CMPS_DES,
PACK_COD,
PACK_DES,
LEVEL,
TYPE,
CAST(RATIO AS FLOAT) AS RATIO,
CAST(STARTTIME AS INT) STARTTIME,
CAST(ENDTIME AS INT) ENDTIME
FROM DWD.DWD_GND_IMS_TBLBRANDRATIO_BYMONTH
WHERE UPPER(TYPE) = 'UNIT' AND UPPER(LEVEL) = 'MOLECULE'
) T6 ON T1.MARKET = T6.MARKET AND T1.CMPS_CODE = T6.CMPS_COD
LEFT JOIN ( --COUNTINGUNIT--MOLECULE
SELECT
DISTINCT
MARKET,
CMPS_COD,
CMPS_DES,
PACK_COD,
PACK_DES,
LEVEL,
TYPE,
CAST(RATIO AS FLOAT) AS RATIO,
CAST(STARTTIME AS INT) STARTTIME,
CAST(ENDTIME AS INT) ENDTIME
FROM DWD.DWD_GND_IMS_TBLBRANDRATIO_BYMONTH
WHERE UPPER(TYPE) = 'COUNTINGUNIT' AND UPPER(LEVEL) = 'MOLECULE'
) T7 ON T1.MARKET = T7.MARKET
AND T1.CMPS_CODE = T7.CMPS_COD
-- COMMAND ----------

62
EC/09 DM_TD_EXT_EC_MARKET_PACK_MAPPING.sql Normal file
View File

@@ -0,0 +1,62 @@
-- Databricks notebook source
-- CREATE OR REPLACE TABLE DM.DM_TD_EXT_EC_MARKET_PACK_MAPPING
-- (
-- PACK_CODE STRING,
-- MARKET STRING,
-- CLASS STRING,
-- KEY_COMPETITOR STRING,
-- MARKET_RATIO STRING,
-- STARTTIME STRING,
-- ENDTIME STRING,
-- VALUE_BRAND_RATIO STRING,
-- VALUE_BRAND_RATIO_START STRING,
-- VALUE_BRAND_RATIO_END STRING,
-- UNIT_BRAND_RATIO STRING,
-- UNIT_BRAND_START STRING,
-- UNIT_BRAND_END STRING,
-- COUNTINGUNIT_BRAND_RATIO STRING,
-- COUNTINGUNIT_BRAND_START STRING,
-- COUNTINGUNIT_BRAND_END STRING,
-- PDOT_BRAND_RATIO STRING,
-- PDOT_BRAND_START STRING,
-- PDOT_BRAND_END STRING,
-- ETL_INSERT_DT TIMESTAMP,
-- ETL_UPDATE_DT TIMESTAMP
-- )
-- USING delta
-- -- LOCATION 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_market_pack_mapping';
-- -- 上面是生产环境location，下面是测试环境location
-- LOCATION 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_td_ext_ec_market_pack_mapping';
-- COMMAND ----------
INSERT OVERWRITE TABLE DM.DM_TD_EXT_EC_MARKET_PACK_MAPPING
SELECT DISTINCT
T1.PACK_CODE,
T1.MARKET,
T1.CLASS,
T1.KEY_COMPETITOR,
T2.MARKET_RATIO,
T2.STARTTIME,
T2.ENDTIME,
T3.VALUE_BRAND_RATIO,
T3.VALUE_BRAND_RATIO_START,
T3.VALUE_BRAND_RATIO_END,
T3.UNIT_BRAND_RATIO,
T3.UNIT_BRAND_START,
T3.UNIT_BRAND_END,
T3.COUNTINGUNIT_BRAND_RATIO,
T3.COUNTINGUNIT_BRAND_START,
T3.COUNTINGUNIT_BRAND_END,
T3.PDOT_BRAND_RATIO,
T3.PDOT_BRAND_START,
T3.PDOT_BRAND_END,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_INSERT_DT,
FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(),'UTC+8') AS ETL_UPDATE_DT
FROM
DM.DM_TD_EXT_EC_PACK_PROPERTY T1
LEFT JOIN DM.DM_TD_EXT_EC_MARKET_RATIO T2
ON T1.MARKET = T2.MARKET AND T1.PACK_CODE = T2.PACK_CODE
LEFT JOIN DM.DM_TD_EXT_EC_MARKET_BRAND_RATIO T3
ON T1.MARKET = T3.MARKET AND T1.PACK_CODE = T3.PACK_CODE
WHERE UPPER(T1.MARKET) NOT LIKE "%ALL%"

103
EC/1 (ec)blob_to_dwd.sql Normal file
View File

@@ -0,0 +1,103 @@
-- Databricks notebook source
-- MAGIC %python
-- MAGIC import datetime
-- MAGIC from pyspark.sql.functions import current_timestamp, expr, date_format
-- MAGIC
-- MAGIC # 计算当前日期
-- MAGIC current_date_utc = datetime.datetime.utcnow()
-- MAGIC offset = datetime.timedelta(hours=8)
-- MAGIC current_date = current_date_utc + offset
-- MAGIC
-- MAGIC today_path = "{:04d}/{:02d}/{:02d}/".format(
-- MAGIC current_date.year,
-- MAGIC current_date.month,
-- MAGIC current_date.day,
-- MAGIC )
-- MAGIC # 基础路径
-- MAGIC # 测试环境
-- MAGIC base_path_0 = f"abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/ODS/GND/UserUpload/"
-- MAGIC # 生产环境
-- MAGIC # base_path_0 = f"abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/ODS/GND/UserUpload/"
-- MAGIC base_path = base_path_0 + today_path
-- MAGIC print(base_path)
-- MAGIC
-- MAGIC # 检查基础路径是否存在
-- MAGIC def check_path_exists(path):
-- MAGIC try:
-- MAGIC dbutils.fs.ls(path)
-- MAGIC return True
-- MAGIC except Exception as e:
-- MAGIC return False
-- MAGIC
-- MAGIC if check_path_exists(base_path):
-- MAGIC # 列出所有批次路径
-- MAGIC batch_paths = dbutils.fs.ls(base_path)
-- MAGIC
-- MAGIC # 从已存在的配置表中读取数据
-- MAGIC config_df = spark.table("dwd.dwd_gnd_ec_config_table")
-- MAGIC
-- MAGIC # 逐批处理
-- MAGIC for batch in batch_paths:
-- MAGIC current_batch_number = int(batch.name.strip('/'))
-- MAGIC # print(f"Checking batch {current_batch_number} at {batch.path}")
-- MAGIC
-- MAGIC files_in_batch = dbutils.fs.ls(batch.path)
-- MAGIC
-- MAGIC print("该批次中的文件:")
-- MAGIC for file in files_in_batch:
-- MAGIC print(file.name)
-- MAGIC
-- MAGIC for row in config_df.collect():
-- MAGIC file_name = row['file_name'].strip().lower()
-- MAGIC table_name = row['table_name']
-- MAGIC
-- MAGIC # 检查文件是否匹配
-- MAGIC matching_files = [f for f in files_in_batch if f.name.strip().lower() == file_name]
-- MAGIC
-- MAGIC for match in matching_files:
-- MAGIC csv_file_path = batch.path + match.name
-- MAGIC print(f"找到匹配的文件: {csv_file_path}")
-- MAGIC
-- MAGIC # 读取 CSV 文件
-- MAGIC df = spark.read.format("csv").option("header", "true").option("charset", "GBK").load(csv_file_path)
-- MAGIC
-- MAGIC # 对列进行重命名（假设所有文件的列名相同）
-- MAGIC df = df.withColumnRenamed('时间（月度）', 'time')
-- MAGIC df = df.withColumnRenamed('平台', 'platform')
-- MAGIC df = df.withColumnRenamed('店铺名称', 'store_name')
-- MAGIC df = df.withColumnRenamed('店铺类型', 'store_type')
-- MAGIC df = df.withColumnRenamed('产品ID', 'product_id')
-- MAGIC df = df.withColumnRenamed('品牌', 'brand_name')
-- MAGIC df = df.withColumnRenamed('品名', 'category_name')
-- MAGIC df = df.withColumnRenamed('商品名', 'product_name')
-- MAGIC df = df.withColumnRenamed('通用名', 'common_name')
-- MAGIC df = df.withColumnRenamed('厂家', 'factory')
-- MAGIC df = df.withColumnRenamed('集团权益', 'group_interest')
-- MAGIC df = df.withColumnRenamed('规格', 'specification')
-- MAGIC df = df.withColumnRenamed('单件包装盒数', 'pcs_per_box')
-- MAGIC df = df.withColumnRenamed('剂型', 'dosage')
-- MAGIC df = df.withColumnRenamed('细分一', 'des1')
-- MAGIC df = df.withColumnRenamed('细分二', 'des2')
-- MAGIC df = df.withColumnRenamed('细分三', 'des3')
-- MAGIC df = df.withColumnRenamed('细分四', 'des4')
-- MAGIC df = df.withColumnRenamed('销售额', 'sales_amount')
-- MAGIC df = df.withColumnRenamed('成交件数', 'sold_pcs')
-- MAGIC df = df.withColumnRenamed('平均单价（元/件）', 'aup_pices')
-- MAGIC df = df.withColumnRenamed('销售量（盒）', 'sales_qty')
-- MAGIC df = df.withColumnRenamed('平均单价（元/盒）', 'aup_box')
-- MAGIC df = df.withColumnRenamed('PROD_COD', 'PROD_COD')
-- MAGIC df = df.withColumnRenamed('PACK_COD', 'PACK_COD')
-- MAGIC df = df.withColumnRenamed('APP3_COD', 'APP3_COD')
-- MAGIC df = df.withColumnRenamed('ATC4_COD', 'ATC4_COD')
-- MAGIC
-- MAGIC # 添加 'etl_insert_dt' 列，包含当前时间并加上 8 小时的时差
-- MAGIC df = df.withColumn('etl_insert_dt', date_format(expr("current_timestamp() + INTERVAL 8 HOURS"), 'yyyy-MM-dd HH:mm:ss'))
-- MAGIC
-- MAGIC # 将数据保存到目标表
-- MAGIC df.write.mode("overwrite").saveAsTable(table_name)
-- MAGIC
-- MAGIC print(f"数据已写入表 {table_name}")
-- MAGIC else:
-- MAGIC print("未找到批次或路径不存在。")
-- MAGIC
-- MAGIC

74
EC/2 dwd_inc_gnd_ext_ec_nationnal_pack_union_all.py Normal file
View File

@@ -0,0 +1,74 @@
# Databricks notebook source
from pyspark.sql.functions import lit
from pyspark.sql.utils import AnalysisException
# pack数据自动接入 整合
#获取配置表信息(表名、brand_flag
df = spark.sql("""
SELECT DISTINCT table_name tab ,file_name brand_flag FROM dwd.dwd_gnd_ec_config_table
where type_name ='PACK'
""").collect()
# print(df)
def get_union_pack_data(df):
#初始化结果集
union_query = None
# niad_pdot_unit需特殊赋值对应表名：tmp.tmp_inc_gnd_ext_retail_nataional_niad
# niad_pdot_unit_flag = 'tmp.tmp_inc_gnd_ext_retail_nataional_niad'
for table in df:
# 选择当前表名
table_name = table['tab']
if not table_name:
print("由于table_name为空，故跳过查询")
continue
# 获取对应brand表维度对应得 market 名称
# brand_flag = str(table.brand_flag)
sql = f"""
select
t1.time,
t1.platform,
t1.store_name,
t1.store_type,
t1.product_id,
t1.brand_name,
t1.category_name,
t1.product_name,
t1.common_name,
t1.factory,
t1.group_interest,
t1.specification,
t1.pcs_per_box,
t1.dosage,
t1.des1,
t1.des2,
t1.des3,
t1.des4,
t1.sales_amount,
t1.sold_pcs,
t1.aup_pices,
t1.sales_qty,
t1.aup_box,
t1.PROD_COD,
t1.PACK_COD,
t1.APP3_COD,
t1.ATC4_COD,
t1.etl_insert_dt
from {table_name} t1
"""
# 尝试读取数据，只有在表名非空时执行
try:
current_query = spark.sql(sql)
except AnalysisException as e:
print(f"Error querying table {table_name}: {e}")
continue # 如果查询发生错误，跳过当前循环迭代
#union 数据
if union_query ==None:
union_query=current_query
else:
union_query = union_query.union(current_query)
#返回数据集
return union_query
pack_result = get_union_pack_data(df)
if pack_result is not None:
pack_result.write.mode("overwrite").saveAsTable("dwd.dwd_inc_gnd_ext_ec_nationnal_pack_union_all")
else:
print("没有数据写入这张表里")