修改dm_td_external_geo的region_type不为空

Reviewed-on: #2

07 dm_td_external_packinfo.sql

@@ -298,11 +298,11 @@ select * from tmp.tmp_county_dm_td_external_packinfo
 insert overwrite table tmp.tmp_thc_dm_td_external_packinfo 
 with temp as (    -- 临时表补充维度
   SELECT 
-    iqvia_pack_code PACK_COD,
+    PACK_COD PACK_COD,
     MAX(B.PACK_DES) PACK_DES,
     MAX(B.STGH_DES) STGH_DES,
     MAX(B.PACK_LCH) PACK_LCH,
-    MAX(B.iqvia_prod_code) PROD_COD,
+    MAX(B.prod_cod) PROD_COD,
     MAX(B.PROD_DES) PROD_DES,
     MAX(B.PROD_DES_C) PROD_DES_C,
     MAX(B.CMPS_COD) CMPS_COD,
@@ -358,8 +358,8 @@ with temp as (    -- 临时表补充维度
       WHEN MAX(B.GENE_ORIG_DESC) = 'TCM' THEN 4
       ELSE 5 
 END GEN_RN
-FROM  dwd.dwd_gnd_ext_thc_pack_property B 
+FROM  dm.dm_thc_pack_property B 
-  GROUP BY B.iqvia_pack_code 
+  GROUP BY B.pack_cod
 )
 select
   A.PACK_COD
@@ -435,11 +435,11 @@ with temp as (                    -- dm.dm_td_external_packinfo_temp 该表数
                                   --01首先是根据主表数据(带多个source)关联chpa的pack_porperty表,取出相关维度
                                   --02chpa中,没有的code，他的维度字段会被置为空，所以要关联我们自己处理好的维度表，来补充相关维度
   SELECT                         
-    iqvia_pack_code PACK_COD,
+    pack_cod PACK_COD,
     MAX(B.PACK_DES) PACK_DES,
     MAX(B.STGH_DES) STGH_DES,
     MAX(B.PACK_LCH) PACK_LCH,
-    MAX(B.iqvia_prod_code) PROD_COD,
+    MAX(B.prod_cod) PROD_COD,
     MAX(B.PROD_DES) PROD_DES,
     MAX(B.PROD_DES_C) PROD_DES_C,
     MAX(B.CMPS_COD) CMPS_COD,
@@ -495,8 +495,8 @@ with temp as (                    -- dm.dm_td_external_packinfo_temp 该表数
       WHEN MAX(B.GENE_ORIG_DESC) = 'TCM' THEN 4
       ELSE 5 
 END GEN_RN
-FROM  dwd.dwd_gnd_ext_chc_pack_property B 
+FROM  dm.dm_chc_pack_property B 
-GROUP BY B.iqvia_pack_code 
+GROUP BY B.pack_cod
 )
 select
   A.PACK_COD

09 dm_td_external_geo_type.sql

@@ -876,7 +876,7 @@ with original_table(
     A.AUDIT_COD,
     CASE
         WHEN A.PROVINCE_C IN (
-            'BBU_OtherProv', 'OBU_OtherProv', 'Others',
+            'Others',
             'ROC', 'Other Low Tiers', 'OTHER LOW TIERS'
             )
         OR A.PROVINCE_C IS NULL
@@ -1431,7 +1431,7 @@ ins_level
 SELECT
   A.DATA_SOURCE,
   A.AUDIT_COD,
-  A.REGION_TYPE,
+  nvl(A.REGION_TYPE,'') REGION_TYPE,
   CASE
     WHEN A.DATA_SOURCE = 'EC(Monthly)' THEN '全国'
     WHEN A.DATA_SOURCE = 'THC(Quarterly)' AND A.AUDIT_COD = 'CHT' THEN '全国'
@@ -1512,7 +1512,7 @@ PROVINCE_GROUP
 select
   DATA_SOURCE,
   AUDIT_COD,
-  null as REGION_TYPE,
+  '' as REGION_TYPE,
   PROVINCE_GROUP
 from 
 dm_td_external_geo_type_temp

11 DM_TD_EXTERNAL_MARKET_NEW.sql

@@ -162,6 +162,10 @@
 -- ('Retail(Quarterly)', 'Inhaled Extended Market', '1'),
 -- ('Retail(Quarterly)', 'BUDESONIDE Market', '1'),
 -- ('Retail(Quarterly)', 'PPI Oral ESOMEPRAZOLE Market', '1'),
+-- -- Retail藏数逻辑调整，Symbicort Market、Respules Market、Respiratory Maintenance Market省份数据要放出来，影响AI取数
+-- ('Retail(Quarterly)', 'Symbicort Market', '1'),
+-- ('Retail(Quarterly)', 'Respules Market', '1'),
+-- ('Retail(Quarterly)', 'Respiratory Maintenance', '1'),
 
 -- -- DTP(Quarterly) 部分
 -- ('DTP(Quarterly)', 'EGFR TKI Market', '1'),

ORG/DM_TD_EXT_COUNTY_ORG.sql

@@ -64,7 +64,7 @@ FROM DM.dm_td_org P2
 -- COMMAND ----------
 
 -- 
-insert into DM.DM_TD_EXT_RETAIL_ORG
+insert overwrite DM.DM_TD_EXT_COUNTY_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,
@@ -171,7 +171,7 @@ province_market AS (
     LEFT JOIN dwd.dwd_td_institution    t2 ON t1.inst_code   = t2.inst_code
     INNER JOIN family_2_markt            t3 ON t1.family_code = t3.prd_family_cd
 )
-
+,part1(
 -- ════════════════════════════════════════════
 -- 主查询 Part 1：正常地理覆盖行
 -- ════════════════════════════════════════════
@@ -208,6 +208,24 @@ INNER JOIN province_market t2 ON t1.mr_trty_code   = t2.trty_code
 INNER JOIN external_geo    t3 ON t2.province_name  = t3.province_c
 WHERE t2.market    IS NOT NULL
   AND t3.audit_cod IS NOT NULL
+)
+
+,all_combinations AS (
+  SELECT 
+    /*+ BROADCAST(geo) */
+    p.market,
+    P.subbu,
+    geo.audit_cod
+  FROM (SELECT DISTINCT market,subbu FROM pack_2_market) p
+  CROSS JOIN external_geo geo
+  GROUP BY     
+    p.market,
+    P.subbu,
+    geo.audit_cod
+)
+
+
+SELECT * FROM part1
 
 UNION ALL
 
@@ -220,12 +238,12 @@ SELECT DISTINCT
     'ROC'                                                               AS rc_code,
     'ROC'                                                               AS rc_name_en,
     'Other Low Tiers'                                                   AS RegionCenter,
-    'ROC'                                                               AS AUDIT_COD,
+    ac.audit_cod                                                        AS AUDIT_COD,
-    t2.MARKET,
+    ac.market,
     'All Channel'                                                       AS bu_category,
-    nvl(subbu,'BBU_County')           AS bu_name_en,
+    NVL(subbu,'ROC')           AS bu_name_en,
-    nvl(subbu,'BBU_County')           AS sub_bu_name,
+    NVL(subbu,'ROC')           AS sub_bu_name,
-    nvl(subbu,'BBU_County')           AS sub_bu_code_auth,
+    NVL(subbu,'ROC')           AS sub_bu_code_auth,
     'IQVIA-COUNTY(Quarterly)'                                           AS bu_code_auth,
     'IQVIA-COUNTY(Quarterly)'                                           AS DATA_SOURCE,
     1                                                                   AS REGION_RATIO,
@@ -242,4 +260,7 @@ SELECT DISTINCT
     ''                                                                  AS DSM_NAME,
     ''                                                                  AS MR_KCODE,
     ''                                                                  AS MR_NAME
-FROM pack_2_market t2
+FROM all_combinations ac
+LEFT ANTI JOIN part1 p
+           ON ac.market    = p.market
+          AND ac.audit_cod = p.audit_cod

ORG/DM_TD_EXT_RETAIL_ORG.sql

@@ -89,10 +89,10 @@ market_ta (
 )
 
 ,pack_market as (
-select distinct a.pack_cod,a.market,b.subbu
+select distinct a.pack_code as pack_cod,a.market,b.subbu
- from dm.dm_td_external_market_pack_mapping a
+ from dm.dm_td_ext_unionall_market_pack_mapping a
 left join market_ta b on a.MARKET = b.market
-where data_source = 'Retail(Quarterly)'
+where datasource = 'Retail(Quarterly)'
 )
 ,external_geo as (
 select distinct province_c,audit_cod,REGION_CENTER
@@ -132,7 +132,7 @@ from dm.dm_td_sd_territory_indication t1
 inner join YM_CVH t2
 on t1.yyyymm = t2.key_value
 where bu in ('Retail') --20260320 chenwu Retail数据源的架构，去掉OBU，只保留Retail
-and t1.key_hp<>3 --chenwu 20250904 排除双考影响
+and t1.key_hp=1 --chenwu 20260519 排除双考影响
 )
 ,province_market as (
 select t1.trty_code,

Retail/13 DM_TF_EXT_RETAIL_SALES.sql

@@ -212,58 +212,132 @@ FROM
 -- 功能: 月份NIAD不进行季度汇总, 新建DM.DM_TF_EXT_RETAIL_NIAD_SALES表做关联
 -- ****************************************************
 WITH NIAD_DWD_TABLE AS (
-  select
+  SELECT
     a.month,
-    b.iqvia_pack_code as pack_code,
+    b.iqvia_pack_code AS pack_code,
     'ROC' AS zk_region,
-    a.sales_unit,
+    SUM(a.sales_unit) AS sales_unit,
-    a.sales_value,
+    SUM(a.sales_value) AS sales_value,
-    a.sales_unit * (b.counting_unit/ coalesce(b.unit,1)) as counting_unit
+    SUM(a.sales_unit * (b.counting_unit / COALESCE(b.unit, 1))) AS counting_unit
-  from
+  FROM
     dwd.dwd_gnd_ext_retail_nataional_niad a
-      left join dwd.dwd_gnd_ext_retail_pack_property b
+      LEFT JOIN dwd.dwd_gnd_ext_retail_pack_property b
-        on case
+        ON CASE
-          when length(zk_product_id) < 7 then right(concat('0000000', zk_product_id), 7)
+          WHEN LENGTH(zk_product_id) < 7 THEN RIGHT(CONCAT('0000000', zk_product_id), 7)
-          else zk_product_id
+          ELSE zk_product_id
-        end = b.product_id
+        END = b.product_id
-  where
+  WHERE
     a.zk_region = '全国'
+  GROUP BY
+    a.month, b.iqvia_pack_code
+),
+UNION_TABLE AS (
+  -- 当期
+  SELECT
+    cur.month AS YYYYMM,
+    cur.pack_code AS PACK_CODE,
+    '' AS CORP_CODE,
+    cur.zk_region AS AUDIT_CODE,
+    '' AS PLATFORM_TYPE,
+    '' AS STORE_NAME,
+    '' AS STORE_TYPE,
+    '' AS REGION_TYPE,
+    '' AS PACK_FLAG,
+    '' AS PROD_FLAG,
+    '' AS DTP_FLAG,
+    CAST(REPLACE(cur.sales_unit, ',', '') AS DECIMAL(38, 10)) AS SALES_UNIT_CAL,
+    0 AS SALES_UNIT_CAL_LY,
+    CAST(REPLACE(cur.sales_value, ',', '') AS DECIMAL(38, 10)) AS SALES_VALUE_CAL,
+    0 AS SALES_VALUE_CAL_LY,
+    CAST(cur.counting_unit AS DECIMAL(38, 10)) AS CONUTING_UNIT,
+    0 AS CONUTING_UNIT_LY,
+    'Retail(Quarterly)' AS DATA_SOURCE,
+    '' AS INST_CODE,
+    '' AS CMPS_FLAG,
+    '' AS DEPT_NAME,
+    '' AS PRESCRIPTION,
+    '' AS PRESCRIPTION_LY,
+    '' AS NEW_CODE,
+    '' AS AREA,
+    '' AS H_LEVEL,
+    '' AS REIMBURSE,
+    '' AS REIMBURSE_TYPE,
+    '' AS PRESCRIPTION_SOURCE
+  FROM NIAD_DWD_TABLE cur
+
+  UNION ALL
+
+  -- 同期（去年）
+  SELECT
+    CAST(CAST(cur.month AS INT) + 100 AS STRING) AS YYYYMM,
+    cur.pack_code AS PACK_CODE,
+    '' AS CORP_CODE,
+    cur.zk_region AS AUDIT_CODE,
+    '' AS PLATFORM_TYPE,
+    '' AS STORE_NAME,
+    '' AS STORE_TYPE,
+    '' AS REGION_TYPE,
+    '' AS PACK_FLAG,
+    '' AS PROD_FLAG,
+    '' AS DTP_FLAG,
+    0 AS SALES_UNIT_CAL,
+    CAST(REPLACE(cur.sales_unit, ',', '') AS DECIMAL(38, 10)) AS SALES_UNIT_CAL_LY,
+    0 AS SALES_VALUE_CAL,
+    CAST(REPLACE(cur.sales_value, ',', '') AS DECIMAL(38, 10)) AS SALES_VALUE_CAL_LY,
+    0 AS CONUTING_UNIT,
+    CAST(cur.counting_unit AS DECIMAL(38, 10)) AS CONUTING_UNIT_LY,
+    'Retail(Quarterly)' AS DATA_SOURCE,
+    '' AS INST_CODE,
+    '' AS CMPS_FLAG,
+    '' AS DEPT_NAME,
+    '' AS PRESCRIPTION,
+    '' AS PRESCRIPTION_LY,
+    '' AS NEW_CODE,
+    '' AS AREA,
+    '' AS H_LEVEL,
+    '' AS REIMBURSE,
+    '' AS REIMBURSE_TYPE,
+    '' AS PRESCRIPTION_SOURCE
+  FROM NIAD_DWD_TABLE cur
+  WHERE CAST(cur.month AS INT) + 100 <= (SELECT CAST(MAX(month) AS INT) FROM NIAD_DWD_TABLE)
 )
 INSERT OVERWRITE TABLE DM.DM_TF_EXT_RETAIL_NIAD_SALES
 SELECT
-  cur.month AS YYYYMM,
+  YYYYMM,
-  cur.pack_code AS PACK_CODE,
+  PACK_CODE,
-  '' AS CORP_CODE,
+  CORP_CODE,
-  cur.zk_region AS AUDIT_CODE,
+  AUDIT_CODE,
-  '' AS PLATFORM_TYPE,
+  PLATFORM_TYPE,
-  '' AS STORE_NAME,
+  STORE_NAME,
-  '' AS STORE_TYPE,
+  STORE_TYPE,
-  '' AS REGION_TYPE,
+  REGION_TYPE,
-  '' AS PACK_FLAG,
+  PACK_FLAG,
-  '' AS PROD_FLAG,
+  PROD_FLAG,
-  '' AS DTP_FLAG,
+  DTP_FLAG,
-  CAST(REPLACE(cur.sales_unit, ',', '') AS DECIMAL(38, 10)) AS SALES_UNIT_CAL,
+  SUM(SALES_UNIT_CAL)     AS SALES_UNIT_CAL,
-  IFNULL(CAST(REPLACE(ly.sales_unit, ',', '') AS DECIMAL(38, 10)), 0) AS SALES_UNIT_CAL_LY,
+  SUM(SALES_UNIT_CAL_LY)  AS SALES_UNIT_CAL_LY,
-  CAST(REPLACE(cur.sales_value, ',', '') AS DECIMAL(38, 10)) AS SALES_VALUE_CAL,
+  SUM(SALES_VALUE_CAL)    AS SALES_VALUE_CAL,
-  IFNULL(CAST(REPLACE(ly.sales_value, ',', '') AS DECIMAL(38, 10)), 0) AS SALES_VALUE_CAL_LY,
+  SUM(SALES_VALUE_CAL_LY) AS SALES_VALUE_CAL_LY,
-  CAST(cur.counting_unit AS DECIMAL(38, 10)) AS CONUTING_UNIT,
+  SUM(CONUTING_UNIT)      AS CONUTING_UNIT,
-  IFNULL(CAST(ly.counting_unit AS DECIMAL(38, 10)), 0) AS CONUTING_UNIT_LY,
+  SUM(CONUTING_UNIT_LY)   AS CONUTING_UNIT_LY,
-  'Retail(Quarterly)' AS DATA_SOURCE,
+  DATA_SOURCE,
-  '' AS INST_CODE,
+  INST_CODE,
-  '' AS CMPS_FLAG,
+  CMPS_FLAG,
-  '' AS DEPT_NAME,
+  DEPT_NAME,
-  '' AS PRESCRIPTION,
+  PRESCRIPTION,
-  '' AS PRESCRIPTION_LY,
+  PRESCRIPTION_LY,
-  '' AS NEW_CODE,
+  NEW_CODE,
-  '' AS AREA,
+  AREA,
-  '' AS H_LEVEL,
+  H_LEVEL,
-  '' AS REIMBURSE,
+  REIMBURSE,
-  '' AS REIMBURSE_TYPE,
+  REIMBURSE_TYPE,
-  '' AS PRESCRIPTION_SOURCE,
+  PRESCRIPTION_SOURCE,
   FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(), 'UTC+8') AS ETL_INSERT_DT,
   FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(), 'UTC+8') AS ETL_UPDATE_DT
-FROM (select * from NIAD_DWD_TABLE ) cur
+FROM UNION_TABLE
-LEFT JOIN (select * from NIAD_DWD_TABLE ) ly
+GROUP BY
-  ON CAST(CAST(ly.month AS INT) + 100 AS STRING) = cur.month
+  YYYYMM, PACK_CODE, CORP_CODE, AUDIT_CODE, PLATFORM_TYPE,
-  AND ly.pack_code = cur.pack_code
+  STORE_NAME, STORE_TYPE, REGION_TYPE, PACK_FLAG, PROD_FLAG,
-  AND ly.zk_region = cur.zk_region
+  DTP_FLAG, DATA_SOURCE, INST_CODE, CMPS_FLAG, DEPT_NAME,
+  PRESCRIPTION, PRESCRIPTION_LY, NEW_CODE, AREA, H_LEVEL,
+  REIMBURSE, REIMBURSE_TYPE, PRESCRIPTION_SOURCE

UNIONALL/DM_TF_EXT_UNIONALL_SALES_MAPPING.sql

@@ -80,10 +80,10 @@ SELECT
   T0.STORE_TYPE,
   T0.REGION_TYPE,
   T0.DATA_SOURCE,
-  CASE WHEN P.MARKET IS NOT NULL AND T0.PACK_FLAG = 1 THEN 1 
+  CASE WHEN P.MARKET IS NOT NULL THEN 1 
   WHEN P.MARKET IS NOT NULL THEN 2
   ELSE 0 END PACK_FLAG,
-  CASE WHEN P.MARKET IS NOT NULL AND T0.PROD_FLAG  = 1 THEN 1
+  CASE WHEN P.MARKET IS NOT NULL THEN 1
   WHEN P.MARKET IS NOT NULL THEN 2
   WHEN T0.MARKET = 'Metoprolol Succinate Market' AND PACK.PROD_DES = 'BETALOC ZOK' THEN 1
   WHEN T0.MARKET = 'Atorvastatin Market' AND PACK.PROD_DES = 'LIPITOR' THEN 1

UNIONALL/DM_TF_EXT_UNIONALL_SALES_MAPPING_NIAD.sql

@@ -143,16 +143,23 @@ niad_pack_cods AS (
 -- 来源：Retail pack粒度数据（EC行join后为NULL，属正常）
 -- 关联键：pack_code = PACK_COD, geo_key = AUDIT_COD, YYYYMM = YYYYMM
 kpi_pivot AS (
-  SELECT
+ select
-    pack_code,
+    a.month yyyymm,
-    YYYYMM,
+    b.iqvia_pack_code as pack_code,
-    geo_key,
+    'ROC' AS geo_key,
-    MAX(CASE WHEN KPI_name = 'digital_spread_rate'  THEN kpi_val END) AS DIGITAL_SPREAD_RATE,
+    max(a.digital_spread_rate) as DIGITAL_SPREAD_RATE,
-    MAX(CASE WHEN KPI_name = 'weighted_spread_rate' THEN kpi_val END) AS WEIGHTED_SPREAD_RATE
+    max(a.weighted_spread_rate) as WEIGHTED_SPREAD_RATE
-  FROM dm.dm_zk_retail_sales_kpi
+  from
-  WHERE KPI_name IN ( 'digital_spread_rate', 'weighted_spread_rate')
+    dwd.dwd_gnd_ext_retail_nataional_niad a
-    AND ta = 'DM'
+      left join dwd.dwd_gnd_ext_retail_pack_property b
-  GROUP BY pack_code, YYYYMM, geo_key
+        on case
+          when length(zk_product_id) < 7 then right(concat('0000000', zk_product_id), 7)
+          else zk_product_id
+        end = b.product_id
+  where
+    a.zk_region = '全国'
+  group by 
+  a.month,b.iqvia_pack_code
 )
 
 -- 主查询：提取所有关联市场的完整销售数据

XIEHE/03 dm_ext_xiehe_geo.sql

@@ -12,7 +12,7 @@ with geo_xiehe as (-- dwd层表
   group by fact_province,fact_city,geo_city,nvl(geo_province,'')
 )
 
--- 特殊区域：'黑吉辽','云贵川青','湖北','甘肃','其他地区'
+-- 特殊区域：'黑吉辽','云贵川','湖北','其他地区'
 select 
   -- a.*,
   concat('xiehe_',nvl(b.AUDIT_COD,concat(a.geo_province,a.fact_province,a.fact_city))) as AUDIT_COD

county/02 dm_td_ext_county_geo_mapping_flag.sql

@@ -14,10 +14,10 @@
 --     bbu_bu_county_flag string
 -- )
 -- USING delta
--- -- location 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_td_ext_county_geo_mapping_flag'
+-- location 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DM/dm_td_ext_county_geo_mapping_flag'
--- location 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_td_ext_county_geo_mapping_flag'
+-- -- location 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DM/dm_td_ext_county_geo_mapping_flag'
 
 -- COMMAND ----------
 
 insert overwrite dm.dm_td_ext_county_geo_mapping_flag
-select province,bbu_bu_county_flag from dwd.dwd_gnd_county_geo_mapping
+select province,bbu_bu_county  from dwd.dwd_gnd_county_geo