update20260427

2026-04-27 11:04:09 +08:00
commit 3e14e78af3
173 changed files with 62579 additions and 0 deletions
--- a/dm_aia_pack_property.sql
+++ b/dm_aia_pack_property.sql
@@ -0,0 +1,355 @@
+-- Databricks notebook source
+-- CREATE OR REPLACE TABLE dm.dm_aia_pack_property (
+--   PACK_COD STRING,
+--   pack_des STRING,
+--   stgh_des STRING,
+--   pack_lch STRING,
+--   PROD_COD STRING,
+--   prod_des STRING,
+--   prod_des_c STRING,
+--   Family_Code STRING,
+--   Family_Name STRING,
+--   cmps_cod STRING,
+--   cmps_des STRING,
+--   cmps_des_c STRING,
+--   atc1_cod STRING,
+--   atc1_des STRING,
+--   atc1_des_c STRING,
+--   atc2_cod STRING,
+--   atc2_des STRING,
+--   atc2_des_c STRING,
+--   atc3_cod STRING,
+--   atc3_des STRING,
+--   atc3_des_c STRING,
+--   atc4_cod STRING,
+--   atc4_des STRING,
+--   atc4_des_c STRING,
+--   app1_cod STRING,
+--   app1_des STRING,
+--   app1_des_c STRING,
+--   app2_cod STRING,
+--   app2_des STRING,
+--   app2_des_c STRING,
+--   app3_cod STRING,
+--   app3_des STRING,
+--   app3_des_c 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_cod STRING,
+--   manu_des STRING,
+--   manu_des_c STRING,
+--   mnfl_cod STRING,
+--   mnfl_des STRING,
+--   corp_cod STRING,
+--   corp_des STRING,
+--   corp_des_c STRING,
+--   BrandType STRING,
+--   IS_AZ STRING,
+--   AZ_MAIN STRING,
+--   AZ_Related STRING,
+--   COUNTING_UNIT STRING,
+--   ETL_INSERT_DT TIMESTAMP,
+--   ETL_UPDATE_DT TIMESTAMP)
+-- USING delta
+-- -- LOCATION 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_aia_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_aia_pack_property';
+
+-- COMMAND ----------
+
+--对用户上传的表进行去重处理
+--对部分编码字段进行补0
+create or replace temporary view aia_pack_property_distinct as
+select distinct
+  iqvia_lineno,
+  case
+    when
+      length(trim(iqvia_pack_code)) < 12
+      and trim(iqvia_pack_code) REGEXP '^[0-9]'
+    then
+      right(concat('000000000000', trim(iqvia_pack_code)), 12)
+    else trim(iqvia_pack_code)
+  end as iqvia_pack_code,
+  case
+    when length(iqvia_prod_code) < 9 then right(concat('000000000', iqvia_prod_code), 9)
+    else iqvia_prod_code
+  end as iqvia_prod_code,
+  iqvia_notes,
+  IS_NEW_PRD,
+  IS_IN_CHPA,
+  ATC1,
+  ATC2,
+  ATC3,
+  ATC4,
+  org_prd_cd,
+  org_mole_nm_c,
+  org_manu_prd_nm_c,
+  org_prd_nm_c,
+  org_pk,
+  org_prd_str,
+  org_pth,
+  min_pk_unit,
+  org_pk_unit,
+  org_manu_nm,
+  app1_cod,
+  app1_des,
+  app1_des_c,
+  app2_cod,
+  app2_des,
+  app2_des_c,
+  app3_cod,
+  app3_des,
+  app3_des_c,
+  atc1_cod,
+  atc1_des,
+  atc1_des_c,
+  atc2_cod,
+  atc2_des,
+  atc2_des_c,
+  atc3_cod,
+  atc3_des,
+  atc3_des_c,
+  atc4_cod,
+  atc4_des,
+  atc4_des_c,
+  bio_desc,
+  case
+    when length(cmps_cod) < 6 then right(concat('000000', cmps_cod), 6)
+    else cmps_cod
+  end as cmps_cod,
+  cmps_des,
+  cmps_des_c,
+  corp_cod,
+  corp_des,
+  corp_des_c,
+  edl_desc,
+  eth_otc_desc,
+  gene_orig_desc,
+  gqce_desc,
+  manu_cod,
+  manu_des,
+  manu_des_c,
+  mnfl_cod,
+  mnfl_des,
+  nrdl_desc,
+  pack_des,
+  stgh_des,
+  pack_lch,
+  paed_desc,
+  prod_des,
+  prod_des_c,
+  tcm_desc,
+  vbp_desc,
+  unit,
+  case
+    when
+      counting_unit is null
+      and org_pk_unit is null
+    then
+      1
+    when counting_unit is null then cast(org_pk_unit as decimal(30, 10))
+    else cast(counting_unit / coalesce(unit, 1) as decimal(30, 10))
+  end as counting_unit,
+  dosage_unit
+from
+  dwd.dwd_gnd_ext_aia_cpt_data;
+
+-- COMMAND ----------
+
+-- create or replace table dm.dm_aia_pack_property
+insert overwrite table dm.dm_aia_pack_property
+WITH dedup_l AS (
+    SELECT *,
+           ROW_NUMBER() OVER (PARTITION BY iqvia_pack_code ORDER BY iqvia_pack_code) AS rn
+    FROM aia_pack_property_distinct
+    QUALIFY rn = 1
+),
+dedup_r AS (
+    SELECT *,
+           ROW_NUMBER() OVER (PARTITION BY PACK_COD ORDER BY PACK_COD) AS rn
+    FROM tmp.tmp_ext_chpa_pack_propetry
+    QUALIFY rn = 1
+)
+SELECT
+    l.iqvia_pack_code AS PACK_COD,
+
+    -- Pack / Product 描述
+    COALESCE(r.PACK_DES,        l.pack_des)          AS pack_des,
+    COALESCE(r.STGH_DES,        l.stgh_des)          AS stgh_des,
+    COALESCE(r.PACK_LCH,        l.pack_lch)          AS pack_lch,
+    COALESCE(r.PROD_COD,        l.iqvia_prod_code)   AS PROD_COD,
+    COALESCE(r.PROD_DES,        l.prod_des)          AS prod_des,
+    COALESCE(r.PROD_DES_C,      l.prod_des_c)        AS prod_des_c,
+
+    -- 右表独有：Family
+    r.Family_Code,
+    r.Family_Name,
+
+    -- 成分
+    COALESCE(r.CMPS_COD,        l.cmps_cod)          AS cmps_cod,
+    COALESCE(r.CMPS_DES,        l.cmps_des)          AS cmps_des,
+    COALESCE(r.CMPS_DES_C,      l.cmps_des_c)        AS cmps_des_c,
+
+    -- ATC
+    COALESCE(r.ATC1_COD,        l.atc1_cod)          AS atc1_cod,
+    COALESCE(r.atc1_des,        l.atc1_des)          AS atc1_des,
+    COALESCE(r.atc1_des_c,      l.atc1_des_c)        AS atc1_des_c,
+    COALESCE(r.ATC2_COD,        l.atc2_cod)          AS atc2_cod,
+    COALESCE(r.atc2_des,        l.atc2_des)          AS atc2_des,
+    COALESCE(r.atc2_des_c,      l.atc2_des_c)        AS atc2_des_c,
+    COALESCE(r.ATC3_COD,        l.atc3_cod)          AS atc3_cod,
+    COALESCE(r.atc3_des,        l.atc3_des)          AS atc3_des,
+    COALESCE(r.atc3_des_c,      l.atc3_des_c)        AS atc3_des_c,
+    COALESCE(r.ATC4_COD,        l.atc4_cod)          AS atc4_cod,
+    COALESCE(r.atc4_des,        l.atc4_des)          AS atc4_des,
+    COALESCE(r.atc4_des_c,      l.atc4_des_c)        AS atc4_des_c,
+
+    -- APP
+    COALESCE(r.APP1_COD,        l.app1_cod)          AS app1_cod,
+    COALESCE(r.app1_des,        l.app1_des)          AS app1_des,
+    COALESCE(r.app1_des_c,      l.app1_des_c)        AS app1_des_c,
+    COALESCE(r.APP2_COD,        l.app2_cod)          AS app2_cod,
+    COALESCE(r.app2_des,        l.app2_des)          AS app2_des,
+    COALESCE(r.app2_des_c,      l.app2_des_c)        AS app2_des_c,
+    COALESCE(r.APP3_COD,        l.app3_cod)          AS app3_cod,
+    COALESCE(r.app3_des,        l.app3_des)          AS app3_des,
+    COALESCE(r.app3_des_c,      l.app3_des_c)        AS app3_des_c,
+
+    -- 标志位
+    COALESCE(r.BIO_DESC,        l.bio_desc)          AS bio_desc,
+    COALESCE(r.GENE_ORIG_DESC,  l.gene_orig_desc)    AS gene_orig_desc,
+    COALESCE(r.ETH_OTC_DESC,    l.eth_otc_desc)      AS eth_otc_desc,
+    COALESCE(r.NRDL_DESC,       l.nrdl_desc)         AS nrdl_desc,
+    r.NRDL_Entry_Date,
+    COALESCE(r.EDL_DESC,        l.edl_desc)          AS edl_desc,
+    COALESCE(r.TCM_DESC,        l.tcm_desc)          AS tcm_desc,
+    COALESCE(r.PAED_DESC,       l.paed_desc)         AS paed_desc,
+    COALESCE(r.GQCE_DESC,       l.gqce_desc)         AS gqce_desc,
+    COALESCE(r.VBP_DESC,        l.vbp_desc)          AS vbp_desc,
+
+    -- 厂商
+    COALESCE(r.MANU_COD,        l.manu_cod)          AS manu_cod,
+    COALESCE(r.MANU_DES,        l.manu_des)          AS manu_des,
+    COALESCE(r.MANU_DES_C,      l.manu_des_c)        AS manu_des_c,
+    COALESCE(r.MNFL_COD,        l.mnfl_cod)          AS mnfl_cod,
+    COALESCE(r.MNFL_DES,        l.mnfl_des)          AS mnfl_des,
+
+    -- 公司
+    COALESCE(r.CORP_COD,        l.corp_cod)          AS corp_cod,
+    COALESCE(r.CORP_DES,        l.corp_des)          AS corp_des,
+    COALESCE(r.CORP_DES_C,      l.corp_des_c)        AS corp_des_c,
+
+    -- 右表独有：品牌 / AZ 标志
+    r.BrandType,
+    r.IS_AZ,
+    r.AZ_MAIN,
+    r.AZ_Related,
+    -- 左表独有
+    l.COUNTING_UNIT,
+    from_utc_timestamp(current_timestamp(),'UTC+8') as ETL_INSERT_DT,
+    from_utc_timestamp(current_timestamp(),'UTC+8') as ETL_UPDATE_DT
+FROM dedup_l l
+LEFT JOIN dedup_r r
+    ON l.iqvia_pack_code = r.PACK_COD
+
+-- COMMAND ----------
+
+-- -----------------------------------------------------------
+-- --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
+-- --由于AIA涉及后面的Flag逻辑、NEW AIA Dashboard，因此不能直接在最后的DM表进行限制，需要在这里限制。
+-- -----------------------------------------------------------
+-- with hidden_market as (
+-- select distinct 
+-- upper(hidden_market) as hidden_market
+-- from dwd.dwd_gnd_ims_hidden_market
+-- where upper(data_source) = 'AIA'
+-- )
+-- ,tmp as (
+-- select 
+-- MARKET_PACK_KEY,
+-- PACK_COD,
+-- PACK_DES,
+-- STGH_DES,
+-- PACK_LCH,
+-- Family_Code,
+-- Family_Name,
+-- PROD_COD,
+-- PROD_DES,
+-- PROD_DES_C,
+-- CMPS_COD,
+-- CMPS_DES,
+-- CMPS_DES_C,
+-- ATC1_COD,
+-- ATC2_COD,
+-- ATC3_COD,
+-- ATC4_COD,
+-- APP1_COD,
+-- APP2_COD,
+-- APP3_COD,
+-- BIO_DESC,
+-- GENE_ORIG_DESC,
+-- ETH_OTC_DESC,
+-- NRDL_DESC,
+-- NRDL_Entry_Date,
+-- EDL_DESC,
+-- TCM_DESC,
+-- PAED_DESC,
+-- GQCE_DESC,
+-- VBP_DESC,
+-- MANU_COD,
+-- MANU_DES,
+-- MANU_DES_C,
+-- MNFL_COD,
+-- MNFL_DES,
+-- CORP_COD,
+-- CORP_DES,
+-- CORP_DES_C,
+-- BrandType,
+-- t1.MARKET,
+-- t1.MARKET_CHPA,
+-- KEY_COMPETITOR,
+-- IS_AZ,
+-- AZ_MAIN,
+-- AZ_Related,
+-- atc1_des,
+-- atc1_des_c,
+-- atc2_des,
+-- atc2_des_c,
+-- atc3_des,
+-- atc3_des_c,
+-- atc4_des,
+-- atc4_des_c,
+-- app1_des,
+-- app1_des_c,
+-- app2_des,
+-- app2_des_c,
+-- app3_des,
+-- app3_des_c,
+-- Class,
+-- ETL_INSERT_DT,
+-- ETL_UPDATE_DT,
+-- market_ratio,
+-- counting_unit
+-- from dm.dm_aia_pack_property t1
+-- left join hidden_market t3
+-- on upper(t1.MARKET) = t3.hidden_market
+-- where t3.hidden_market is null
+-- union
+-- select *
+-- from dm.dm_aia_pack_property
+-- where MARKET = 'AIA ALL Market'
+-- )
+
+-- insert overwrite dm.dm_aia_pack_property
+-- select * from tmp