20260501

DBR更新，包含haoyi的调整
2026-05-02 06:48:24 +00:00
parent 8b73f0cc7c
commit 3a03ba1ad2
6 changed files with 724 additions and 219 deletions
--- a/UNIONALL/DM_TF_EXT_UNIONALL_MARKET_SALES.sql
+++ b/UNIONALL/DM_TF_EXT_UNIONALL_MARKET_SALES.sql
@@ -521,4 +521,147 @@ SELECT
  FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(), 'UTC+8') AS ETL_INSERT_DT,
  FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(), 'UTC+8') AS ETL_UPDATE_DT
 FROM
-  joined;
+  joined;
+
+-- COMMAND ----------
+
+INSERT into TABLE
+  DM.DM_TF_EXT_UNIONALL_MARKET_SALES
+WITH
+  t1_filtered AS (
+    SELECT
+      *
+    FROM
+      DM.DM_TF_EXT_RETAIL_NIAD_SALES
+  ),
+  mapping_cleaned AS (
+    SELECT
+      *
+    FROM
+      (
+        SELECT
+          PACK_CODE,
+          DATASOURCE,
+          MARKET,
+          KEY_COMPETITOR,
+          CLASS,
+          MARKET_RATIO,
+          Value_brand_Ratio,
+          Unit_brand_Ratio,
+          CountingUnit_brand_Ratio,
+          PDOT_brand_Ratio,
+          GREATEST                 (
+            starttime,
+            Value_brand_Ratio_start,
+            Unit_brand_start,
+            CountingUnit_brand_start,
+            PDOT_brand_start
+          ) AS eff_start,
+          LEAST (
+            endtime,
+            Value_brand_Ratio_end,
+            Unit_brand_end,
+            CountingUnit_brand_end,
+            PDOT_brand_end
+          ) AS eff_end
+        FROM
+          DM.dm_td_ext_unionall_market_pack_mapping
+      ) tmp
+    WHERE
+      eff_start <= eff_end -- 在外层子查询中过滤
+  )  ,
+  joined AS (
+    SELECT
+      /*+ BROADCAST(t2, t3) */ 
+      t1.*,
+      -- 当年                    ratio
+      t2.MARKET,
+      t2.KEY_COMPETITOR,
+      t2.CLASS,
+      t2.MARKET_RATIO AS cy_market_ratio,
+      t2.Value_brand_Ratio AS cy_value_ratio,
+      t2.Unit_brand_Ratio AS cy_unit_ratio,
+      t2.CountingUnit_brand_Ratio AS cy_cu_ratio,
+      t2.PDOT_brand_Ratio AS cy_pdot_ratio,
+      -- 去年 ratio
+      t3.MARKET_RATIO AS ly_market_ratio,
+      t3.Value_brand_Ratio AS ly_value_ratio,
+      t3.Unit_brand_Ratio AS ly_unit_ratio,
+      t3.CountingUnit_brand_Ratio AS ly_cu_ratio,
+      t3.PDOT_brand_Ratio AS ly_pdot_ratio
+    FROM
+      t1_filtered t1
+      LEFT JOIN mapping_cleaned t2 ON t1.PACK_CODE = t2.PACK_CODE
+      AND t1.DATA_SOURCE = t2.DATASOURCE
+      AND t1.YYYYMM >= t2.eff_start
+      AND t1.YYYYMM <= t2.eff_end
+      LEFT JOIN mapping_cleaned t3 ON t1.PACK_CODE = t3.PACK_CODE
+      AND t1.DATA_SOURCE = t3.DATASOURCE
+      AND (t1.YYYYMM - 100) >= t3.eff_start
+      AND (t1.YYYYMM - 100) <= t3.eff_end
+      AND t2.MARKET = t3.MARKET -- 保持原逻辑
+  )
+SELECT
+  NVL(MARKET, 'ALL Market')     AS MARKET,
+  NVL(KEY_COMPETITOR, 'OTHERS') AS KEY_COMPETITOR,
+  CASE 
+    WHEN trim(lower(CLASS)) = 'others' THEN 'OTHERS'
+    WHEN CLASS IS NULL THEN 'OTHERS'
+    ELSE CLASS 
+  END AS CLASS,
+  YYYYMM,
+  PACK_CODE,
+  CORP_CODE,
+  AUDIT_CODE,
+  PLATFORM_TYPE,
+  STORE_NAME,
+  STORE_TYPE,
+  REGION_TYPE,
+  DATA_SOURCE,
+  PACK_FLAG,
+  PROD_FLAG,
+  DTP_FLAG,
+  CMPS_FLAG,
+  NEW_CODE,
+  INST_CODE,
+  NULL                          AS AIA_HP_FLAG,
+  DEPT_NAME,
+  H_LEVEL,
+  REIMBURSE,
+  REIMBURSE_TYPE,
+  PRESCRIPTION_SOURCE           AS PRESCRIPTION_TYPE,
+  PRESCRIPTION,
+  PRESCRIPTION_LY,
+  --                            预计算公用因子，避免重复计算
+  NVL(NVL(cy_market_ratio, 1) * cy_value_ratio, 1) AS VA,
+  NVL(NVL(cy_market_ratio, 1) * cy_unit_ratio, 1) AS UT,
+  NVL(NVL(cy_market_ratio, 1) * cy_cu_ratio, 1) AS CU,
+  NVL(NVL(cy_market_ratio, 1) * cy_pdot_ratio, 1) AS PT,
+  CAST(
+    SALES_UNIT_CAL * NVL(cy_unit_ratio, 1) * NVL(cy_market_ratio, 1) AS DECIMAL(35, 6)
+  ) AS SALES_UNIT_CAL,
+  CAST(
+    SALES_UNIT_CAL_LY * NVL(ly_unit_ratio, 1) * NVL(ly_market_ratio, 1) AS DECIMAL(35, 6)
+  ) AS SALES_UNIT_CAL_LY,
+  CAST(
+    SALES_VALUE_CAL * NVL(cy_value_ratio, 1) * NVL(cy_market_ratio, 1) AS DECIMAL(35, 6)
+  ) AS SALES_VALUE_CAL,
+  CAST(
+    SALES_VALUE_CAL_LY * NVL(ly_value_ratio, 1) * NVL(ly_market_ratio, 1) AS DECIMAL(35, 6)
+  ) AS SALES_VALUE_CAL_LY,
+  CAST(
+    CONUTING_UNIT * NVL(cy_cu_ratio, 1) * NVL(cy_market_ratio, 1) AS DECIMAL(35, 6)
+  ) AS CONUTING_UNIT,
+  CAST(
+    CONUTING_UNIT_LY * NVL(ly_cu_ratio, 1) * NVL(ly_market_ratio, 1) AS DECIMAL(35, 6)
+  ) AS CONUTING_UNIT_LY,
+  CAST(
+    CONUTING_UNIT * NVL(cy_pdot_ratio, 1) * NVL(cy_market_ratio, 1) AS DECIMAL(35, 6)
+  ) AS PDOT,
+  CAST(
+    CONUTING_UNIT_LY * NVL(ly_pdot_ratio, 1) * NVL(ly_market_ratio, 1) AS DECIMAL(35, 6)
+  ) AS PDOT_LY,
+  FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(), 'UTC+8') AS ETL_INSERT_DT,
+  FROM_UTC_TIMESTAMP(CURRENT_TIMESTAMP(), 'UTC+8') AS ETL_UPDATE_DT
+FROM
+  joined;
--- a/UNIONALL/DM_TF_EXT_UNIONALL_SALES.sql
+++ b/UNIONALL/DM_TF_EXT_UNIONALL_SALES.sql
@@ -346,4 +346,4 @@ SELECT
  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_TF_EXT_XIEHE_SALES
-;
+;
--- a/UNIONALL/DM_TF_EXT_UNIONALL_SALES_MAPPING.sql
+++ b/UNIONALL/DM_TF_EXT_UNIONALL_SALES_MAPPING.sql
@@ -473,7 +473,7 @@ SELECT
  END AS CONUTING_UNIT,
  CASE
    WHEN A.CONUTING_UNIT_LY = 0 THEN NULL
-    https://adb-1064580724995799.3.databricks.azure.cn/editor/notebooks/3456519074298505?o=1064580724995799$0ELSE CAST(A.CONUTING_UNIT_LY AS DECIMAL(35, 10))
+    ELSE CAST(A.CONUTING_UNIT_LY AS DECIMAL(35, 10))
  END AS CONUTING_UNIT_LY,
  CASE
    WHEN A.PDOT = 0 THEN NULL
--- a/UNIONALL/DM_TF_EXT_UNIONALL_SALES_MAPPING_NIAD.sql
+++ b/UNIONALL/DM_TF_EXT_UNIONALL_SALES_MAPPING_NIAD.sql
@@ -41,7 +41,7 @@
 -- DBTITLE 1,NIAD单独页面
 -- ****************************************************
 -- 功能说明：提取NIAD（非胰岛素类抗糖尿病药物）市场的销售数据并创建映射表
-- 数据源：EC(Monthly)月度电商数据 + Retail(Quarterly)季度零售数据
+-- 数据源：EC(Monthly)月度电商数据 
 -- 创建日期：2026-04-02
 -- 修改日期：2026-04-02
 -- ****************************************************
@@ -51,11 +51,91 @@ WITH

 -- CTE 1：获取NIAD市场的产品编码集合
 -- 注：直接下推双重过滤条件，减少扫描数据量
+-- 20260430拆分
 niad_pack_cods AS (
  SELECT DISTINCT PACK_COD
  FROM DM.DM_TF_EXT_UNIONALL_MARKET_SALES
  WHERE MARKET      = 'NIAD (Non-Insulin Anti-Diabetic) Market'
-    AND DATA_SOURCE IN ('EC(Monthly)', 'Retail(Quarterly)')
+    AND DATA_SOURCE ='EC(Monthly)'
+)
+
+-- 主查询：提取所有关联市场的完整销售数据
+-- BROADCAST 小表 niad_markets，避免大表 shuffle
+SELECT 
+  -- 业务维度
+  t.MARKET,
+  t.KEY_COMPETITOR,
+  t.CLASS,
+  t.YYYYMM,
+
+  -- 产品标识
+  t.PACK_COD,
+  t.AUDIT_COD,
+
+  -- 数据来源
+  t.DATA_SOURCE,
+
+  -- 调整系数
+  t.VA,  -- Value Adjustment
+  t.UT,  -- Unit Adjustment
+  t.CU,  -- Counting Unit Adjustment
+  t.PT,  -- PDOT Adjustment
+
+  -- 当期销售指标
+  t.SALES_UNIT_CAL,
+  t.SALES_VALUE_CAL,
+  t.CONUTING_UNIT,   -- 注：字段名原始拼写保持不变
+  t.PDOT,
+
+  -- 去年同期指标
+  t.SALES_UNIT_CAL_LY,
+  t.SALES_VALUE_CAL_LY,
+  t.CONUTING_UNIT_LY,
+  t.PDOT_LY,
+
+  -- 处方指标
+  t.PRESCRIPTION,
+  t.PRESCRIPTION_LY,
+
+  -- 衍生组合键
+  CONCAT(t.PACK_COD,   t.DATA_SOURCE) AS PACK_SOURCE,
+  CONCAT(t.AUDIT_COD,  t.DATA_SOURCE) AS AUDIT_SOURCE,
+  CONCAT(t.MARKET,  t.DATA_SOURCE) AS MARKET_SOURCE,
+  CONCAT(t.MARKET,t.CLASS,  t.DATA_SOURCE) AS CLASS_SOURCE,
+
+  -- 时间维度：优化为 TO_DATE 直接解析 yyyyMM 格式，避免字符串拼接
+  TO_DATE(t.YYYYMM, 'yyyyMM') AS DATE_KEY,
+  PLATFORM_TYPE,
+  STORE_NAME,
+  STORE_TYPE,
+
+  -- 零售铺货率（仅 Retail 行有值，EC 行为 NULL）
+  NULL AS DIGITAL_SPREAD_RATE,
+  NULL AS WEIGHTED_SPREAD_RATE
+  
+FROM DM.DM_TF_EXT_UNIONALL_MARKET_SALES t
+LEFT SEMI JOIN niad_pack_cods m ON t.PACK_COD = m.PACK_COD
+WHERE t.DATA_SOURCE IN ('EC(Monthly)')
+
+-- COMMAND ----------
+
+-- ****************************************************
+-- 功能说明：提取NIAD（非胰岛素类抗糖尿病药物）市场的销售数据并创建映射表
+-- 数据源：Retail(Quarterly)季度零售数据
+-- 创建日期：2026-04-02
+-- 修改日期：2026-04-02
+-- ****************************************************
+INSERT into DM.DM_TF_EXT_UNIONALL_SALES_MAPPING_NIAD
+
+WITH
+
+-- CTE 1：获取NIAD市场的产品编码集合
+-- 注：直接下推双重过滤条件，减少扫描数据量
+niad_pack_cods AS (
+  SELECT DISTINCT PACK_COD
+  FROM DM.DM_TF_EXT_UNIONALL_MARKET_SALES
+  WHERE MARKET      = 'NIAD (Non-Insulin Anti-Diabetic) Market'
+    AND DATA_SOURCE IN ('Retail(Quarterly)')
 ),

 -- CTE 2：从KPI宽转长表中横向取出三个零售铺货率/价格指标
@@ -66,11 +146,10 @@ kpi_pivot AS (
    pack_code,
    YYYYMM,
    geo_key,
-    MAX(CASE WHEN KPI_name = 'pack_price'           THEN kpi_val END) AS PRICE,
    MAX(CASE WHEN KPI_name = 'digital_spread_rate'  THEN kpi_val END) AS DIGITAL_SPREAD_RATE,
    MAX(CASE WHEN KPI_name = 'weighted_spread_rate' THEN kpi_val END) AS WEIGHTED_SPREAD_RATE
  FROM dm.dm_zk_retail_sales_kpi
-  WHERE KPI_name IN ('pack_price', 'digital_spread_rate', 'weighted_spread_rate')
+  WHERE KPI_name IN ( 'digital_spread_rate', 'weighted_spread_rate')
    AND ta = 'DM'
  GROUP BY pack_code, YYYYMM, geo_key
 )
@@ -137,4 +216,4 @@ LEFT JOIN kpi_pivot kpi
      AND kpi.YYYYMM    = t.YYYYMM
      AND kpi.geo_key   = t.AUDIT_COD
      AND t.DATA_SOURCE = 'Retail(Quarterly)'
-WHERE t.DATA_SOURCE IN ('EC(Monthly)', 'Retail(Quarterly)')
+WHERE t.DATA_SOURCE IN ('Retail(Quarterly)')