update20260427

THC/02 dm_ext_thc_sales.sql

@@ -0,0 +1,354 @@
+-- Databricks notebook source
+-- CREATE OR REPLACE TABLE DM.DM_TF_EXT_THC_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_thc_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_thc_sales';
+
+-- COMMAND ----------
+
+--------------------------------------------------------------------------
+--修改时间：20241106
+--修改人：Fanxujia
+--修改内容：
+--THC原本只有全国数据，现在增加14个省份的数据。
+--但是全国数据是半年提供一次，省份数据一个季度提供一次
+--目前需求是将季度数据放入模型，而全国数据单独成一个事实表，供PBI单独展示
+--这里生成的事实表里全国和省份都包含了，在写入模型的时候用where条件筛选
+---------------------------------------------------------------------------
+
+-- COMMAND ----------
+
+--rawdata年月处理
+CREATE OR REPLACE TEMPORARY VIEW thc_rawdata1
+AS 
+select case when right(yq,2) = 'Q1' then concat(left(yq,4),'03')
+            when right(yq,2) = 'Q2' then concat(left(yq,4),'06')
+            when right(yq,2) = 'Q3' then concat(left(yq,4),'09')
+            when right(yq,2) = 'Q4' then concat(left(yq,4),'12')
+            else yq
+        end as yyyymm,
+        city_c,
+        pack_cod,
+        prod_cod,
+        value,
+        totalunit,
+        countingunit 
+from dwd.dwd_gnd_ext_thc_rawdata
+
+-- COMMAND ----------
+
+--CHPA渠道对应的公司编码以及名称
+CREATE OR REPLACE TEMPORARY VIEW chpa_corp
+AS 
+select distinct
+PACK_COD,
+CORP_COD,
+CORP_DES
+from dm.dm_ims_td_pack_property
+
+-- COMMAND ----------
+
+--从thc渠道的pack_property里取到对应的iqvia_pack_code、countingunit
+--然后corp_cod优先取CHPA渠道的
+CREATE OR REPLACE TEMPORARY VIEW thc_corp
+AS
+select t1.pack_cod,
+t1.iqvia_pack_code,
+t1.countingunit,
+coalesce(t2.CORP_COD, t1.CORP_COD) as CORP_COD ,
+coalesce(t2.CORP_DES,t1.CORP_DES ) as CORP_DES
+from dwd.dwd_gnd_dept_pack_property t1
+left join chpa_corp t2
+on t1.iqvia_pack_code = t2.PACK_COD
+
+-- COMMAND ----------
+
+--编辑thc渠道的audit_cod
+--全国数据设置成CHT
+--省份数据使用公共的地理维表里的省份编码作为audit_cod
+CREATE OR REPLACE TEMPORARY VIEW thc_geo
+AS
+select 
+t1.city_c,
+t1.city_e,
+t1.province_c,
+t1.province_e,
+case when t1.province_c = '全国' then 'CHT'
+     when t2.geo_key is null then t1.province_e
+     else t2.geo_key
+end as AUDIT_COD
+from dwd.dwd_gnd_ext_thc_geo t1
+left join dm.dm_td_geography t2
+--THC最细到省，用省关联
+on t1.province_c = t2.province_name
+and t2.geo_key = t2.province_code
+
+-- COMMAND ----------
+
+--将上面的rawdata、corp、audit信息合到一起
+CREATE OR REPLACE TEMPORARY VIEW thc_rawdata2
+AS
+select t1.yyyymm,
+t1.city_c,
+t1.pack_cod as pack_cod_old,
+t2.iqvia_pack_code as pack_cod,
+t2.CORP_COD,
+t3.AUDIT_COD,
+t1.prod_cod,
+t1.value,
+t1.totalunit,
+t1.totalunit * t2.countingunit as countingunit
+from thc_rawdata1 t1
+left join thc_corp t2
+on t1.pack_cod = t2.pack_cod
+left join thc_geo t3
+on t1.city_c = t3.city_c
+
+-- COMMAND ----------
+
+--最大的月份，计算ly时会需要
+CREATE OR REPLACE TEMPORARY VIEW max_thc_rawdata
+AS
+select max(cast(yyyymm as int)) as max_ym
+from thc_rawdata2
+
+-- COMMAND ----------
+
+--计算ly
+CREATE OR REPLACE TEMPORARY VIEW thc_rawdata3
+AS
+select yyyymm,
+AUDIT_COD,
+pack_cod,
+CORP_COD,
+value,
+totalunit,
+countingunit,
+0 as value_ly,
+0 as totalunit_ly,
+0 as countingunit_ly
+from thc_rawdata2
+union all
+select cast(t1.yyyymm as int) + 100 as yyyymm,
+t1.AUDIT_COD,
+t1.pack_cod,
+t1.CORP_COD,
+0 as value,
+0 as totalunit,
+0 as countingunit,
+t1.value as value_ly,
+t1.totalunit as totalunit_ly,
+t1.countingunit as countingunit_ly
+from thc_rawdata2 t1
+cross join max_thc_rawdata t2
+where cast(t1.yyyymm as int) + 100 <= t2.max_ym
+
+-- COMMAND ----------
+
+--按年月、audit_cod、pack_cod、Corp_cod汇总
+CREATE OR REPLACE TEMPORARY VIEW thc_rawdata4
+AS
+select yyyymm,
+AUDIT_COD,
+pack_cod,
+CORP_COD,
+sum(value) as SALES_VALUE_CAL,
+sum(totalunit) as SALES_UNIT_CAL,
+sum(countingunit) as COUNTING_UNIT,
+sum(value_ly) as SALES_VALUE_CAL_LY,
+sum(totalunit_ly) as SALES_UNIT_CAL_LY,
+sum(countingunit_ly) as COUNTING_UNIT_LY
+from thc_rawdata3
+group by yyyymm,
+AUDIT_COD,
+pack_cod,
+CORP_COD
+
+-- COMMAND ----------
+
+CREATE OR REPLACE TEMPORARY VIEW THC_SALES
+AS
+SELECT
+    A.YYYYMM,
+    A.PACK_COD AS PACK_CODE,
+    A.AUDIT_COD AS AUDIT_CODE,
+    'THC(Quarterly)' DATA_SOURCE,
+    -------------------------------------
+    SALES_UNIT_CAL,
+    SALES_UNIT_CAL_LY,
+    SALES_VALUE_CAL,
+    SALES_VALUE_CAL_LY,
+    COUNTING_UNIT,
+    COUNTING_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--------------------------
+    '' PLATFORM_TYPE,
+    '' STORE_NAME,
+    '' 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
+  thc_rawdata4 A
+-- 条件后移到 unionall_sales
+-- WHERE
+--   AUDIT_COD = 'CHT'
+
+-- COMMAND ----------
+
+INSERT OVERWRITE TABLE DM.DM_TF_EXT_THC_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,
+  COUNTING_UNIT,
+  COUNTING_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 
+  THC_SALES
+;
+
+-- COMMAND ----------
+
+insert overwrite table dm.dm_ext_thc_sales
+(
+yyyymm,
+AUDIT_COD,
+pack_cod,
+CORP_COD,
+SALES_VALUE_CAL,
+SALES_UNIT_CAL,
+COUNTING_UNIT,
+SALES_VALUE_CAL_LY,
+SALES_UNIT_CAL_LY,
+COUNTING_UNIT_LY
+)
+select yyyymm,
+AUDIT_COD,
+pack_cod,
+CORP_COD,
+SALES_VALUE_CAL,
+SALES_UNIT_CAL,
+COUNTING_UNIT,
+SALES_VALUE_CAL_LY,
+SALES_UNIT_CAL_LY,
+COUNTING_UNIT_LY
+from thc_rawdata4
+--取2018年之后的数
+where yyyymm >= '201801'