Add EXTERNAL SQL scripts

EXTERNAL/ORG/DM_TD_EXT_AIA_ORG.sql

@@ -0,0 +1,461 @@
+-- Databricks notebook source
+-- CREATE OR REPLACE TABLE DM.DM_TD_EXT_AIA_ORG (
+--   REGION_CODE STRING,
+--   REGION_NAME_EN STRING,
+--   RSM_NAME STRING,
+--   RC_CODE STRING,
+--   RC_NAME_EN STRING,
+--   RC_NAME STRING,
+--   REGIONCENTER STRING,
+--   AUDIT_COD STRING,
+--   MARKET STRING,
+--   BU_CATEGORY STRING,
+--   BU_NAME_EN STRING,
+--   SUB_BU_NAME STRING,
+--   SUB_BU_CODE_AUTH STRING,
+--   BU_CODE_AUTH STRING,
+--   DATA_SOURCE STRING,
+--   REGION_RATIO FLOAT,
+--   TEAM STRING,
+--   MR_TRTY_CODE STRING,
+--   NSD_KCODE STRING,
+--   NSD_NAME STRING,
+--   RSD_KCODE STRING,
+--   RSD_NAME STRING,
+--   RSM_KCODE STRING,
+--   DSM_KCODE STRING,
+--   DSM_NAME STRING,
+--   MR_KCODE STRING,
+--   MR_NAME STRING)
+-- USING delta
+-- LOCATION 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_td_ext_aia_org';
+
+-- COMMAND ----------
+
+-- MAGIC %run ../../../Common/config
+
+-- COMMAND ----------
+
+-- MAGIC %python
+-- MAGIC spark.read.table(f'`{CDW_CATALOG}`.`dwd`.`dim_product_wide`').createOrReplaceTempView('cdw_dwd_dim_product_wide')
+
+-- COMMAND ----------
+
+-- 20260130 由于DM.dm_td_org数据需要转换，新增这个CELL。
+create or replace temporary view dm_td_org_temp
+as 
+SELECT org_key,mr_trty_tag,district_code,rsm_name,rc_name,p2.region_name,p2.rsd_kcode,p2.rsd_name
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.bu_code = 'PC12' then 'PC09' else p2.bu_code end bu_code
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.bu_report = 'EC' THEN 'Retail' ELSE p2.bu_report end bu_report
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.sub_bu_code = 'NA27' then 'NA06' else p2.sub_bu_code end sub_bu_code
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.sub_bu_name = 'EC' then 'Retail' else p2.sub_bu_name end sub_bu_name
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.sub_bu_name_en = 'EC' then 'Retail' else p2.sub_bu_name_en end sub_bu_name_en
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.bu_name = 'EC' then 'Retail' else p2.bu_name end bu_name
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.bu_name_en = 'EC' then 'Retail' else p2.bu_name_en end bu_name_en
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.nsd_kcode = 'ND_EC0001_V' then 'KSGX559' else p2.nsd_kcode end nsd_kcode
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.bu_name = 'EC'and p2.nsd_name = 'Vacant' then '李白杨' else p2.nsd_name end nsd_name
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.nsd_trty_code = 'ND_EC0001' then 'ND_RT0001' else p2.nsd_trty_code end nsd_trty_code
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.bsd_kcode = 'VP_EC0001_V' then 'VP_RT0001_V' else p2.bsd_kcode end bsd_kcode
+  ,CASE WHEN p2.yyyymm >= '202601' and p2.bsd_trty_code = 'VP_EC0001' then 'VP_RT0001' else p2.bsd_trty_code end bsd_trty_code
+  ,mr_kcode,dsm_trty_code,rsm_trty_code,dsm_kcode,rsd_trty_code
+  ,district_name,dsm_name,mr_name,yyyymm,mr_trty_code,REGION_CODE,rc_code,bu_category,rsm_kcode
+FROM DM.dm_td_org P2
+
+-- COMMAND ----------
+
+---------------------------------------------------------------------------------------
+--修改时间：20241105
+--修改人：Fanxujia
+--修改内容：
+--for AIA Dashboard，新增NSD、RSD、RSM、DSM、MR、TEAM等信息
+--仅AIA渠道的数据增加值，其他渠道都为空
+---------------------------------------------------------------------------------------
+---------------------------------------------------AIA------------------------------------------------------------------------------
+--AIA参考CHPA
+------所有BU能看AIA
+
+--从销量表取到全部的机构、市场组合
+with ALL_AUDIT_Market as (
+select distinct NVL(t1.AUDIT_COD,'ROC') AUDIT_COD,
+NVL(t2.market,'ALL Market') market
+from dm.dm_ext_aia_sales t1
+left join dm.dm_td_ext_aia_market_pack_mapping t2
+on t1.pack_cod = t2.PACK_CODE
+)
+,pack_market as (
+select distinct PACK_CODE AS PACK_COD,market
+from dm.dm_td_ext_aia_market_pack_mapping
+)
+
+--获取城市与大区中心的mapping
+,EXTERNAL_GEO as (
+select distinct
+concat(city_c, '市') as city_c,
+REGION_CENTER
+from DM.DM_TD_EXTERNAL_GEO_TEMP
+where DATA_SOURCE = 'AIA(Monthly)'
+)
+--CVH架构月
+,CVH_YM as (
+select key_value as yyyymm
+from dm.dm_td_date_config
+where key_code = 'pfmc_cvh'
+)
+--org表
+,org as (
+select distinct
+t1.yyyymm,
+t1.bu_category,
+t1.bu_code,
+t1.bu_name_en,
+t1.sub_bu_code,
+t1.sub_bu_name,
+t1.REGION_CODE,
+t1.region_name,
+t1.rc_code,
+t1.rc_name,
+t1.district_name as team,
+t1.mr_trty_code,
+t1.nsd_kcode,
+case when t1.nsd_name = 'Vacant' then t1.nsd_kcode else t1.nsd_name end as nsd_name,
+t1.rsd_kcode,
+case when t1.rsd_name = 'Vacant' then t1.rsd_kcode else t1.rsd_name end as rsd_name,
+t1.rsm_kcode,
+case when t1.rsm_name = 'Vacant' then t1.rsm_kcode else t1.rsm_name end as rsm_name,
+t1.dsm_kcode,
+case when t1.dsm_name = 'Vacant' then t1.dsm_kcode else t1.dsm_name end as dsm_name,
+t1.mr_kcode,
+case when t1.mr_name = 'Vacant' then t1.mr_kcode else t1.mr_name end as mr_name
+from dm_td_org_temp t1
+inner join CVH_YM t2
+on t1.yyyymm = t2.yyyymm
+)
+--territory表
+,territory as (
+select distinct
+t1.yyyymm,
+t1.trty_code,
+t1.inst_code,
+t1.family_code
+from dm.dm_td_sd_territory_indication t1
+inner join CVH_YM t2
+on t1.yyyymm = t2.yyyymm
+where t1.bu not in ('CUBE','KA')
+and ((t1.key_hp in (2) and t1.bu <> 'Retail' and t1.trty_code <>'Eagle') or t1.key_hp = 1)--chenwu 20250904 排除双考影响
+)
+--将内部的Family_code转为外部的pack_cod
+,inst_pack_previous as (
+select distinct
+t1.yyyymm,
+t1.trty_code,
+t1.inst_code,
+t2.city_name,
+t5.market,
+case when t2.area_code = 'SOUTH' then 'South'
+     when t2.area_code = 'WEST' then 'West'
+		 when t2.area_code = 'CENTRAL' then 'Central'
+		 when t2.area_code = 'NORTH' then 'North'
+		 when t2.area_code = 'EAST' then 'East'
+end as REGION_CENTER
+from territory t1
+left join dwd.dwd_td_institution t2
+on t1.inst_code = t2.inst_code
+left join cdw_dwd_dim_product_wide t3
+on t1.family_code = t3.prd_family_cd
+left join dwd.dwd_td_imscode_azcode t4
+on t3.prd_brand_cd = t4.brandcode
+left join pack_market t5
+on t4.ims_pack_cod = t5.PACK_COD
+)
+/*
+20251207 Chenwu 把可定的架构复制到立普妥上,直接替换市场 'Atorvastatin Market' = 'Rosuvastatin Market'
+*/
+,inst_pack as (
+  select 
+    yyyymm,
+    trty_code,
+    inst_code,
+    city_name,
+    market,
+    REGION_CENTER
+  from inst_pack_previous
+  union all
+  select
+    yyyymm,
+    trty_code,
+    inst_code,
+    city_name,
+    'Atorvastatin Market' market,
+    REGION_CENTER
+  from inst_pack_previous 
+  where market = 'Rosuvastatin Market'
+)
+
+---------------------------------------------------------------------------------------
+--修改时间：20250220
+--修改人：Zhu chenwu
+--修改内容：
+--原本存在部分Eagel的机构和市场没有挂上架构，针对于没有挂上的部分，用dws.dws_td_eg_cityterritory中的city 和family的关系对应到市场挂上架构。
+----------------------------------------------m-----------------------------------------
+,eagle_trty as 
+(---eagle的city 和 family 对应的 trtycode取数据时，如果city和family对应的trtycode 有多个，以p12，p11,p11为准，取第一个。
+  select SP_TRTY_CD,CITY_CD,FAMILY_CD from (
+  select *
+  ,row_number() over (partition by CITY_CD,FAMILY_CD order by PROJECT_CD desc) as num
+  from (
+    select SP_TRTY_CD,CITY_CD,FAMILY_CD,PROJECT_CD  from  dws.dws_td_eg_cityterritory t1 
+    inner join CVH_YM t2
+    on t1.YM = t2.yyyymm
+    where SP_TRTY_CD not in (select distinct trty_code from dm.dm_td_sd_territory_indication where bu = 'Eagle')
+    and PROJECT_CD in (select distinct project_code from DWS.dws_td_eg_project_instype WHERE ins_type='医院') 
+    and cast(nvl(SALES_RATE,0) as decimal(20,6))<>'0.000000' 
+    group by  1,2,3,4
+  )
+) where num = 1
+)
+,family_mapping as (--family和brand的mapping
+  select prd_family_cd,prd_brand_cd from cdw_dwd_dim_product_wide
+  group by 1,2
+)
+,packcode_mapping as (--pack 和 brand的mapping
+  select brandcode,ims_pack_cod from dwd.dwd_td_imscode_azcode 
+  group by 1,2
+)
+,geo_mapping as (--city的中文
+  select city_code,city_name from dm.dm_td_geography
+  group by 1,2
+)
+,eagle_mapping (--eagle表中city和family所对应的 岗位编码
+select SP_TRTY_CD,city_name,market from (
+	select 
+	t1.SP_TRTY_CD
+	-- ,t1.FAMILY_CD
+	-- ,t3.prd_brand_cd
+	,t4.ims_pack_cod
+	,t5.city_name
+	from eagle_trty t1
+	left join family_mapping t3
+	on t1.FAMILY_CD = t3.prd_family_cd --转换family 成 brand
+	left join packcode_mapping t4
+	on t3.prd_brand_cd = t4.brandcode
+	left join geo_mapping t5
+	on t1.CITY_CD = t5.city_code) A
+inner join pack_market B on A.ims_pack_cod = B.PACK_COD
+group by 1,2,3
+)
+
+,audit_geo as (--audit的城市信息
+select AUDIT_COD,max(CITY_C) as city_c from DM.DM_TD_EXTERNAL_GEO 
+where DATA_SOURCE = 'AIA(Monthly)'
+group by 1)
+
+,eagle_trty_with_city as ( --eagle中 挂上 部分机构归不到eagle的，按eagle的城市+品牌挂上
+select
+t0.AUDIT_COD,t0.market
+,t10.city_c
+,eagle_mapping.SP_TRTY_CD from ALL_AUDIT_Market t0
+left join audit_geo t10 on t0.AUDIT_COD = t10.AUDIT_COD
+left join eagle_mapping on eagle_mapping.city_name = t10.city_c and eagle_mapping.market = t0.market 
+where eagle_mapping.SP_TRTY_CD is not null
+group by 1,2,3,4)
+
+insert overwrite dm.DM_TD_EXT_AIA_ORG (
+    REGION_CODE,
+    region_name_en,
+    rc_code,
+    rc_name_en,
+    RegionCenter,
+    AUDIT_COD,
+    MARKET,
+    bu_category,
+    bu_name_en,
+    sub_bu_name,
+    sub_bu_code_auth,
+    bu_code_auth,
+    DATA_SOURCE,
+    REGION_RATIO,
+    RSM_Name,
+    RC_Name,
+TEAM,
+MR_TRTY_CODE,
+NSD_KCODE,
+NSD_NAME,
+RSD_KCODE,
+RSD_NAME,
+RSM_KCODE,
+DSM_KCODE,
+DSM_NAME,
+MR_KCODE,
+MR_NAME
+  )
+
+SELECT distinct
+  COALESCE(t2.REGION_CODE,t6.REGION_CODE,'ROC') as REGION_CODE,
+  COALESCE(t2.region_name,t6.region_name,'ROC') as region_name_en,
+  COALESCE(t2.rc_code,t6.rc_code,'ROC') as rc_code,
+  COALESCE(t2.rc_name,t6.rc_name,'ROC') as rc_name_en,
+  COALESCE(t1.REGION_CENTER,t4.REGION_CENTER,t7.REGION_CENTER,'') as RegionCenter,
+  t0.AUDIT_COD,
+  t0.MARKET,
+  COALESCE(t2.bu_category,t6.bu_category,'') as bu_category,
+  COALESCE(t2.bu_name_en,t6.bu_name_en,'') as bu_name_en,
+  COALESCE(t2.sub_bu_name,t6.sub_bu_name,'') as sub_bu_name,
+  COALESCE(t2.sub_bu_code,t6.sub_bu_code,'') as sub_bu_code_auth,
+  concat(COALESCE(t2.bu_code,t6.bu_code,''),'AIA(Monthly)') as bu_code_auth,
+  'AIA(Monthly)' as DATA_SOURCE,
+  1 REGION_RATIO,
+  COALESCE(t2.rsm_name,t6.rsm_name,'Vacant') as rsm_name,
+  COALESCE(t2.rsd_name,t6.rsd_name,'Vacant') as RC_Name,
+  COALESCE(t2.team,t6.team,'ROC') as team,
+  COALESCE(t2.mr_trty_code,t6.mr_trty_code,'Vacant') as mr_trty_code,
+  COALESCE(t2.nsd_kcode,t6.nsd_kcode,'Vacant') as nsd_kcode,
+  COALESCE(t2.nsd_name,t6.nsd_name,'Vacant') as nsd_name,
+  COALESCE(t2.rsd_kcode,t6.rsd_kcode,'Vacant') as rsd_kcode,
+  COALESCE(t2.rsd_name,t6.rsd_name,'Vacant') as rsd_name,
+  COALESCE(t2.rsm_kcode,t6.rsm_kcode,'Vacant') as rsm_kcode,
+  COALESCE(t2.dsm_kcode,t6.dsm_kcode,'Vacant') as dsm_kcode,
+  COALESCE(t2.dsm_name,t6.dsm_name,'Vacant') as dsm_name,
+  COALESCE(t2.mr_kcode,t6.mr_kcode,'Vacant') as mr_kcode,
+  COALESCE(t2.mr_name,t6.mr_name,'Vacant') as mr_name
+-- ,t5.SP_TRTY_CD
+from ALL_AUDIT_Market t0
+left join inst_pack t1
+on t0.AUDIT_COD = t1.inst_code
+and t0.market = t1.market
+left join org t2
+on t1.trty_code = t2.mr_trty_code
+and t1.yyyymm = t2.yyyymm
+left join EXTERNAL_GEO t4
+on t1.city_name = t4.city_c
+
+left join eagle_trty_with_city t5
+on t0.AUDIT_COD = t5.AUDIT_COD
+and t0.market = t5.market and t1.inst_code is null
+left join org t6
+on t5.SP_TRTY_CD = t6.mr_trty_code
+left join EXTERNAL_GEO t7
+on t5.city_c = t7.city_c
+
+-- COMMAND ----------
+
+-- DBTITLE 1,add
+--修改时间：20240808
+--修改人：FanXuJia
+--修改背景：
+--事实表与维度表来自于不同手工表，目前存在一部分audit_cod在事实表有销量，但是在维度表没有这个audit_cod，导致销量缺失。
+--因此需要从事实表补全这部分audit_cod
+with sales as (
+select distinct PACK_COD,AUDIT_COD,DATA_SOURCE from dm.dm_tf_external_sales
+union
+select distinct PACK_COD,AUDIT_COD,DATA_SOURCE from dm.dm_tf_external_sales_thc
+)
+,market_pack as (
+select distinct PACK_COD,MARKET,DATA_SOURCE from dm.dm_td_external_market_pack_mapping
+)
+,org as (
+select distinct AUDIT_COD,MARKET,DATA_SOURCE from dm.DM_TD_EXT_AIA_ORG
+)
+,tmp_org as (
+select distinct a.AUDIT_COD,b.MARKET,a.DATA_SOURCE
+from sales a 
+left join market_pack b on a.PACK_COD = b.PACK_COD and a.DATA_SOURCE = b.DATA_SOURCE
+left join org c on a.DATA_SOURCE = c.DATA_SOURCE and a.AUDIT_COD = c.AUDIT_COD and b.MARKET = c.MARKET
+where C.AUDIT_COD IS NULL
+)
+insert into table dm.DM_TD_EXT_AIA_ORG
+(
+REGION_CODE,
+region_name_en,
+RSM_Name,
+rc_code,
+rc_name_en,
+RC_Name,
+RegionCenter,
+AUDIT_COD,
+MARKET,
+bu_category,
+bu_name_en,
+sub_bu_name,
+sub_bu_code_auth,
+bu_code_auth,
+DATA_SOURCE,
+REGION_RATIO,
+TEAM,
+MR_TRTY_CODE,
+NSD_KCODE,
+NSD_NAME,
+RSD_KCODE,
+RSD_NAME,
+RSM_KCODE,
+DSM_KCODE,
+DSM_NAME,
+MR_KCODE,
+MR_NAME
+)
+select
+'ROC' as REGION_CODE,
+'ROC' as region_name_en,
+'Vacant' as RSM_Name,
+'ROC' as rc_code,
+'ROC' as rc_name_en,
+'Vacant' as RC_Name,
+null as RegionCenter,
+AUDIT_COD,
+MARKET,
+null as bu_category,
+null as bu_name_en,
+null as sub_bu_name,
+null as sub_bu_code_auth,
+DATA_SOURCE as bu_code_auth,
+DATA_SOURCE,
+1 as REGION_RATIO,
+'ROC' as TEAM,
+'Vacant' as MR_TRTY_CODE,
+'Vacant' as NSD_KCODE,
+'Vacant' as NSD_NAME,
+'Vacant' as RSD_KCODE,
+'Vacant' as RSD_NAME,
+'Vacant' as RSM_KCODE,
+'Vacant' as DSM_KCODE,
+'Vacant' as DSM_NAME,
+'Vacant' as MR_KCODE,
+'Vacant' as MR_NAME
+from tmp_org
+where DATA_SOURCE = 'AIA(Monthly)'
+
+-- COMMAND ----------
+
+-- DBTITLE 1,hidden_market
+-----------------------------------------------------------
+--修改时间：20241118
+--修改人：Fanxujia
+--修改内容：
+--使用手工文件排除掉每个渠道的不需要显示的市场
+-----------------------------------------------------------
+with hidden_market as (
+select distinct 
+case when upper(data_source) = 'CHPA'    then 'IQVIA-CHPA(Monthly)'
+     when upper(data_source) = '科室市场' then 'XH Data(Quarterly)'
+     when upper(data_source) = 'EC'      then 'EC(Monthly)'
+     when upper(data_source) = 'RETAIL'  then 'Retail(Quarterly)'
+     when upper(data_source) = 'CHC'     then 'CHC(Quarterly)'
+     when upper(data_source) = 'AIA'     then 'AIA(Monthly)'
+     when upper(data_source) = 'THC'     then 'THC(Quarterly)'
+     when upper(data_source) = 'COUNTY'  then 'IQVIA-COUNTY(Quarterly)'
+     else data_source
+end as data_source,
+upper(hidden_market) as hidden_market
+from dwd.dwd_gnd_ims_hidden_market
+union all 
+select
+'AIA(Monthly)' as data_source,
+'AIA ALL Market'
+)
+
+
+MERGE into dm.DM_TD_EXT_AIA_ORG as t1
+USING hidden_market t2
+on t1.data_source = t2.data_source
+and upper(t1.MARKET) = upper(t2.hidden_market)
+WHEN MATCHED THEN
+DELETE