chenwu/MarketAnalysis-ETL
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

删除备份版本notebook

Browse Source
This commit is contained in:
chenwu
2026-05-11 23:19:31 +08:00
parent db99577102
commit 23755b4e1e
12 changed files with 0 additions and 13599 deletions
Download Patch File Download Diff File Expand all files Collapse all files

195
AIA/02 dm_ext_aia_sales_bakup_20230327.sql
View File

@@ -1,195 +0,0 @@
-- Databricks notebook source
--字段值的初步处理
create or replace temporary view aia_rawdata_1 as
select
year,
qtr,
cast(year * 100 + ym as int) ym,
org_prov,
org_city,
ins_level,
case when ins_cd is not null and length(ins_cd) < 7 then right(concat('0000000',ins_cd),7) else ins_cd end as org_ins_cd,
atc1,
atc2,
atc3,
atc4,
atc,
mole_nm,
prod_nm,
org_prd_nm_c,
org_pk,
org_prd_str,
org_pth,
min_unit,
value,
unit,
cast(pack_desc as int) as pack_desc,
org_pk_un,
manu
from dwd.dwd_gnd_aia_rawdata
-- COMMAND ----------
--关联出AUDIT_COD、inst_code
create or replace temporary view aia_rawdata_2 as
with hospital_mapping as (
select distinct
case when length(cpa_hospital_code) < 7 then right(concat('0000000',cpa_hospital_code),7) else cpa_hospital_code end as cpa_hospital_code,
ins_cd_nl as inst_code
from dwd.dwd_gnd_hospital_not_provided
)
select upper(nvl(t2.inst_code,t1.org_ins_cd)) AUDIT_COD,
t1.*,
nvl(t2.inst_code,t1.org_ins_cd) as inst_code
from aia_rawdata_1 t1
left join hospital_mapping t2
on t1.org_ins_cd = t2.cpa_hospital_code
-- left join dm.dm_td_institution t3
-- on t2.ins_code = t3.inst_code
-- left join dm.dm_ims_td_geo t4
-- on t3.city_name_en = t4.city
-- COMMAND ----------
/*
修改人 chenwu
修改时间 20250513
修改内容 打包分子的数据，不论 打通表 和 事实表，都没有系数，需要用 事实表里的 最小制剂单位数量 作为counting_unit
*/
--关联出pack_cod、corp_cod、counting_unit
create or replace temporary view aia_rawdata_3 as
select case when length(trim(t2.iqvia_pack_code)) < 12 and trim(t2.iqvia_pack_code) REGEXP '^[0-9]' then right(concat('000000000000',trim(t2.iqvia_pack_code)),12)
else trim(t2.iqvia_pack_code)
end as pack_cod,
t3.corp_cod,
case when left(t2.iqvia_pack_code,4) = 'AZP_' then cast(t1.min_unit/t1.org_pk_un as decimal(38,10)) else t3.counting_unit end counting_unit,
-- t3.counting_unit,20250513
t1.*
from aia_rawdata_2 t1
left join (select distinct trim(iqvia_pack_code) as iqvia_pack_code,
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 from dwd.dwd_gnd_ext_aia_cpt_data) t2
--ATC1编码+ATC2编码+ATC3编码+ATC4编码+药品编码(ATC)+药品通用名+药品产品名+药品商品名+规格+剂型+给药途径+最小销售包装单位+包装规格+生产企业
on nvl(t1.ATC1,'') = nvl(t2.ATC1,'')
and nvl(t1.ATC2,'') = nvl(t2.ATC2,'')
and nvl(t1.ATC3,'') = nvl(t2.ATC3,'')
and nvl(t1.ATC4,'') = nvl(t2.ATC4,'')
and nvl(t1.ATC,'') = nvl(t2.org_prd_cd,'')
and nvl(t1.MOLE_NM,'') = nvl(t2.org_mole_nm_c,'')
and nvl(t1.PROD_NM,'') = nvl(t2.org_manu_prd_nm_c,'')
and nvl(t1.ORG_PRD_NM_C,'') = nvl(t2.org_prd_nm_c,'')
and nvl(t1.ORG_PK,'') = nvl(t2.org_pk,'')
and nvl(t1.ORG_PRD_STR,'') = nvl(t2.org_prd_str,'')
and nvl(t1.ORG_PTH,'') = nvl(t2.org_pth,'')
and nvl(t1.UNIT,'') = nvl(t2.min_pk_unit,'')
and nvl(t1.PACK_DESC,'') = nvl(t2.org_pk_unit,'')
and nvl(t1.MANU,'') = nvl(t2.org_manu_nm,'')
left join (select distinct PACK_COD,corp_cod,counting_unit from dm.dm_aia_pack_property) t3
on t2.iqvia_pack_code = t3.PACK_COD
-- COMMAND ----------
--计算出ly销量
create or replace temporary view aia_rawdata_4 as
with max_ym as (
select max(ym) as max_ym from aia_rawdata_3
)
select
YM,
org_ins_cd,
inst_code,
AUDIT_COD,
pack_cod,
corp_cod,
sum(value) as value,
sum(volume) as volume,
sum(counting_unit) as counting_unit,
sum(value_ly) as value_ly,
sum(volume_ly) as volume_ly,
sum(counting_unit_ly) as counting_unit_ly
from (
select
ym,
org_ins_cd,
inst_code,
AUDIT_COD,
pack_cod,
corp_cod,
cast(value as decimal(38,10)) as value,
cast(ORG_PK_UN as decimal(38,10)) as volume,
cast(ORG_PK_UN as decimal(38,10)) * coalesce(cast(counting_unit as decimal(38,10)),1) as counting_unit,
0 as value_ly,
0 as volume_ly,
0 as counting_unit_ly
from aia_rawdata_3
union all
select
cast(cast(ym as int) + 100 as string) as ym,
org_ins_cd,
inst_code,
AUDIT_COD,
pack_cod,
corp_cod,
0 as value,
0 as volume,
0 as counting_unit,
cast(value as decimal(38,10)) as value_ly,
cast(ORG_PK_UN as decimal(38,10)) as volume_ly,
cast(ORG_PK_UN as decimal(38,10)) * coalesce(cast(counting_unit as decimal(38,10)),1) as counting_unit_ly
from aia_rawdata_3 t1
join max_ym t2
on 1=1
where cast(cast(ym as int) + 100 as string) <= t2.max_ym
)
group by ym,
org_ins_cd,
inst_code,
AUDIT_COD,
pack_cod,
corp_cod
-- COMMAND ----------
insert overwrite table dm.dm_ext_aia_sales(
ym,
org_ins_cd,
inst_code,
AUDIT_COD,
pack_cod,
corp_cod,
value,
volume,
counting_unit,
value_ly,
volume_ly,
counting_unit_ly
)
select
ym,
org_ins_cd,
inst_code,
AUDIT_COD,
pack_cod,
corp_cod,
sum(value) as value,
sum(volume) as volume,
sum(counting_unit) as counting_unit,
sum(value_ly) as value_ly,
sum(volume_ly) as volume_ly,
sum(counting_unit_ly) as counting_unit_ly
from aia_rawdata_4
group by
ym,
org_ins_cd,
inst_code,
AUDIT_COD,
pack_cod,
corp_cod

643
AIA/z_03 dm_aia_flag_备份用.sql
View File

@@ -1,643 +0,0 @@
-- Databricks notebook source
-- MAGIC %run ../../../Common/config
-- COMMAND ----------
-- MAGIC %python
-- MAGIC spark.read.table(f'`{CDW_CATALOG}`.`dwd`.`dim_product_wide`').createOrReplaceTempView('cdw_dwd_dim_product_wide')
-- MAGIC spark.read.table(f'`{CDW_CATALOG}`.`dwd`.`dwd_hospital_classification`').createOrReplaceTempView('cdw_dwd_dwd_hospital_classification')
-- COMMAND ----------
-----------------------------------------------------------------------------------
--修改时间：20241021
--修改人：FanXujia
--修改内容：
--只取最新年-1 ~ 最新年范围内的，称为“2年共有”。增加
-----------------------------------------------------------------------------------
--共有机构
insert overwrite table dm.dm_aia_hp_flag
with max_year as (
select
max(cast(year as int)) as max_year,
max(cast(year as int) - 1) as max_year_ly
from dwd.dwd_gnd_hospital_not_provided
)
,hp_flag as (
select
distinct concat(t1.year, '-Y') as hp_his_flag,
case when length(t1.cpa_hospital_code) < 7 then right(concat('0000000',t1.cpa_hospital_code),7) else t1.cpa_hospital_code end as cpa_hospital_code
from
dwd.dwd_gnd_hospital_not_provided t1
cross join max_year t2
where cast(t1.year as int) >= t2.max_year_ly
)
,hp_flag2 as (
SELECT
cpa_hospital_code,
CONCAT_WS(' ∩ ', SORT_ARRAY(COLLECT_LIST(hp_his_flag))) AS FLAG
FROM
hp_flag
GROUP BY
cpa_hospital_code
)
,inst_mapping as (
-- select
-- distinct right(concat('0000000', org_cd), 7) org_cd,
-- ins_cd
-- from
-- dwd.dwd_gnd_hospitalmapping
select distinct
case when length(cpa_hospital_code) < 7 then right(concat('0000000',cpa_hospital_code),7) else cpa_hospital_code end as cpa_hospital_code,
ins_cd_nl as inst_code
from dwd.dwd_gnd_hospital_not_provided
)
,ALL_INS as (
select distinct inst_code
from dm.dm_ext_aia_sales
)
,flag as (
select distinct
coalesce(t2.inst_code,t1.cpa_hospital_code) as inst_code,
t1.flag as aia_hp_flag,
'Y' as FLAG,
'AIA(Monthly)' as DATA_SOURCE
from hp_flag2 t1
left join inst_mapping t2 on t1.cpa_hospital_code = t2.cpa_hospital_code
where
length(t1.flag) > 6
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'IQVIA-CHPA(Monthly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'XH Data(Quarterly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'EC(Monthly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'Retail(Quarterly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'CHC(Quarterly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'THC(Quarterly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'IQVIA-COUNTY(Quarterly)' as DATA_SOURCE
)
select t1.inst_code,
NVL(t2.aia_hp_flag,'') as aia_hp_flag,
NVL(t2.FLAG,'N') as FLAG,
'AIA(Monthly)' as DATA_SOURCE
from ALL_INS t1
left join flag t2
on t1.inst_code = t2.inst_code
-- COMMAND ----------
-----------------------------------------------------------------------------------
--修改时间：20241021
--修改人：FanXujia
--修改内容：
--只取最新年-1 ~ 最新年范围内的，称为“2年共有”。增加
-----------------------------------------------------------------------------------
--共有机构
insert overwrite table dm.dm_aia_hp_flag
with max_year as (
select
max(cast(year as int)) as max_year,
max(cast(year as int) - 1) as max_year_ly
from dwd.dwd_gnd_hospital_not_provided
)
,hp_flag as (
select
distinct concat(t1.year, '-Y') as hp_his_flag,
case when length(t1.cpa_hospital_code) < 7 then right(concat('0000000',t1.cpa_hospital_code),7) else t1.cpa_hospital_code end as cpa_hospital_code
from
dwd.dwd_gnd_hospital_not_provided t1
cross join max_year t2
where cast(t1.year as int) >= t2.max_year_ly
)
,hp_flag2 as (
SELECT
cpa_hospital_code,
CONCAT_WS(' ∩ ', SORT_ARRAY(COLLECT_LIST(hp_his_flag))) AS FLAG
FROM
hp_flag
GROUP BY
cpa_hospital_code
)
,inst_mapping as (
-- select
-- distinct right(concat('0000000', org_cd), 7) org_cd,
-- ins_cd
-- from
-- dwd.dwd_gnd_hospitalmapping
select distinct
case when length(cpa_hospital_code) < 7 then right(concat('0000000',cpa_hospital_code),7) else cpa_hospital_code end as cpa_hospital_code,
ins_cd_nl as inst_code
from dwd.dwd_gnd_hospital_not_provided
)
,ALL_INS as (
select distinct inst_code
from dm.dm_ext_aia_sales
)
,flag as (
select distinct
coalesce(t2.inst_code,t1.cpa_hospital_code) as inst_code,
t1.flag as aia_hp_flag,
'Y' as FLAG,
'AIA(Monthly)' as DATA_SOURCE
from hp_flag2 t1
left join inst_mapping t2 on t1.cpa_hospital_code = t2.cpa_hospital_code
where
length(t1.flag) > 6
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'IQVIA-CHPA(Monthly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'XH Data(Quarterly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'EC(Monthly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'Retail(Quarterly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'CHC(Quarterly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'THC(Quarterly)' as DATA_SOURCE
union all
select '' as inst_code,'' as aia_hp_flag,'' as FLAG,'IQVIA-COUNTY(Quarterly)' as DATA_SOURCE
)
select t1.inst_code,
NVL(t2.aia_hp_flag,'') as aia_hp_flag,
NVL(t2.FLAG,'N') as FLAG,
'AIA(Monthly)' as DATA_SOURCE
from ALL_INS t1
left join flag t2
on t1.inst_code = t2.inst_code
-- COMMAND ----------
--连续取数机构
----------------------------------------------------------------------------------------------------------------------
--修改时间：20241029
--修改人：FanXujia
--修改内容：
--根据事实表最新月，往前推12个月（含最新月），按机构 + 年月粒度汇总，如果每个月销量都不等于0，这家机构称为：滚动一年有数
--根据事实表最新月，往前推24个月（含最新月），按机构 + 年月粒度汇总，如果每个月销量都不等于0，这家机构称为：滚动两年有数
--2022年没有销量为正常情况，判断时仅考虑2023年01月起销量是否等于0
----------------------------------------------------------------------------------------------------------------------
insert overwrite table dm.dm_aia_provided_flag
with max_ym as (
select max(ym) as max_ym
from dm.dm_ext_aia_sales
)
,ym_range as (
select max_ym,
case when max_ym <= '202412' then cast(right(max_ym,2) as int)
else cast(12 as int)
end ym_12m,
case when max_ym <= '202312' then '202401'
when right(max_ym,2) = '12' then concat(left(max_ym,4),'01')
else concat(cast(left(max_ym, 4) - 1 as int),right(concat('0', cast(right(max_ym, 2) + 1 as int)), 2))
end as ym_12m_before,
case when max_ym <= '202312' then cast(right(max_ym,2) as int)
when max_ym <= '202512' then cast(right(max_ym,2) as int) + 12
else cast(24 as int)
end as ym_24m,
case when max_ym <= '202512' then '202401'
when right(max_ym,2) = '12' then concat(cast(left(max_ym,4) - 1 as int),'01')
else concat(cast(left(max_ym, 4) -2 as int),right(concat('0', cast(right(max_ym, 2) + 1 as int)), 2))
end as ym_24m_before
from max_ym
)
,sales_1 as (
select
inst_code,
ym
from
dm.dm_ext_aia_sales
group by
ym,
inst_code
having
sum(value) <> 0
)
,sales_2 as (
select
t1.inst_code,
t1.ym,
t2.max_ym,
t2.ym_12m,
t2.ym_12m_before,
t2.ym_24m,
t2.ym_24m_before
from sales_1 t1
cross join ym_range t2
)
,flag_12m as (
select inst_code,
'滚动一年有数' as aia_provided_flag,
'AIA(Monthly)' as DATA_SOURCE
from sales_2
where ym <= max_ym
and ym >= ym_12m_before
group by inst_code
having count(1) = (select ym_12m from ym_range)
)
,flag_24m as (
select inst_code,
'滚动两年有数' as aia_provided_flag,
'AIA(Monthly)' as DATA_SOURCE
from sales_2
where ym <= max_ym
and ym >= ym_24m_before
group by inst_code
having count(1) = (select ym_24m from ym_range)
)
select inst_code,
aia_provided_flag,
DATA_SOURCE
from flag_12m
union all
select inst_code,
aia_provided_flag,
DATA_SOURCE
from flag_24m
union all
select '' as inst_code,'' as aia_provided_flag,'IQVIA-CHPA(Monthly)' as data_source
union all
select '' as inst_code,'' as aia_provided_flag,'XH Data(Quarterly)' as data_source
union all
select '' as inst_code,'' as aia_provided_flag,'EC(Monthly)' as data_source
union all
select '' as inst_code,'' as aia_provided_flag,'Retail(Quarterly)' as data_source
union all
select '' as inst_code,'' as aia_provided_flag,'CHC(Quarterly)' as data_source
union all
select '' as inst_code,'' as aia_provided_flag,'THC(Quarterly)' as data_source
union all
select '' as inst_code,'' as aia_provided_flag,'IQVIA-COUNTY(Quarterly)' as data_source
-- COMMAND ----------
---------------------------------------------------------------------------
--修改时间：20241108
--修改人：Fanxujia
--修改内容：
--for AIA Dashboard
--增加sub_channel,dragon_flag,ka_flag,target_ins_level字段
---------------------------------------------------------------------------
---------------------------------------------------------------------------
--修改时间：20241209
--修改人：Fanxujia
--修改内容：
--增加sub channel的排序
---------------------------------------------------------------------------
--目标医院
insert overwrite table dm.dm_aia_targethp_flag --获取全部的机构 + 市场
with sales as (
select distinct t1.inst_code,t2.MARKET
from dm.dm_ext_aia_sales t1
left join dm.dm_aia_pack_property t2
on t1.pack_cod = t2.PACK_COD
)
--获取产品与市场的映射关系
,pack_market as (
select distinct
PACK_COD,
MARKET
from dm.dm_aia_pack_property
)
--辖区表，获取到KA_Flag、target_ins_level以及目标机构Flag(在辖区表里存在的就是目标机构)
,td_territory_inst as (
select distinct
inst_code,
trty_code,
family_code,
ka_flag,
TARGET_LEVEL as target_ins_level
from dm.dm_td_sd_territory_indication --eagle的非重点医院同样加为目标机构
where yyyymm in (
select
key_value
from
dm.dm_td_date_config
where
key_code = 'pfmc_cvh'
)
and bu not in ('CUBE','KA')
and key_hp<>3 --chenwu 20250904 排除双考影响
)
--汇报线表，获取到Channel，后续用于判断出sub_channel
,td_org as (
select distinct
t1.mr_trty_code,
case when t1.bu_code in ('PC14','PC16','PC18','PC19') then 'BU'
when t1.bu_code = 'SC01' and t1.sub_bu_code = 'NA29' then sub_bu_name
when t1.bu_code = 'SC01' then 'BU'
when t1.bu_code <> 'SC01' and t2.bm_name = 'BCBH' then 'BU'
when t1.bu_code <> 'SC01' then t1.bu_name
end as channel
from dm.dm_td_org t1
left join dws.dws_td_bu_active t2
on t1.bu_name = t2.bu_name
where t1.yyyymm in (
select
key_value
from
dm.dm_td_date_config
where
key_code = 'pfmc_cvh'
)
)
--Family粒度和Brand粒度的映射表
,product_wide as (
select distinct
prd_family_cd,
prd_brand_cd
from cdw_dwd_dim_product_wide
)
--内部Brand粒度与外部pack_cod的映射表
,imscode_azcode as (
select distinct
brandcode,
case when length(ims_pack_cod) < 12 and ims_pack_cod REGEXP '^[0-9]' then right(concat('00000000000',ims_pack_cod),12) else ims_pack_cod end as ims_pack_cod
from dwd.dwd_td_imscode_azcode
)
--综合上面的几个表，得到KA_flag、target_ins_level、channel、目标机构Flag
,target_hp_all as (
select
distinct
t1.inst_code,
t1.ka_flag,
t1.target_ins_level,
t2.channel,
t5.MARKET
from
td_territory_inst t1
inner join td_org t2 on t1.trty_code = t2.mr_trty_code
left join product_wide t3 on t3.prd_family_cd = t1.family_code
left join imscode_azcode t4 on t3.prd_brand_cd = t4.brandcode
left join pack_market t5 on t4.ims_pack_cod = t5.pack_cod
where t5.MARKET is not null
)
--target_hp_all里一个机构一个市场对应了多个属性，这里按优先级依次取唯一值
--ka_flag
,ka_flag1 as (
select distinct
inst_code,
MARKET,
ka_flag
from target_hp_all
)
,ka_flag2 as (
select
inst_code,
MARKET,
ka_flag,
row_number() over(partition by inst_code,market
order by case when ka_flag = 'KA' then 1
ELSE 2 end asc) as RN
from ka_flag1
)
,ka_flag_final as (
select
inst_code,
MARKET,
ka_flag
from ka_flag2
where RN = 1
)
--target_ins_level不取唯一值
--加个hardcoding，ONC Market和ALL Market设置成NA
,target_ins_level1 as (
select distinct
inst_code,
MARKET,
case when MARKET in ('AIA ALL Market','ONC Market') then 'NA'
else target_ins_level
end as target_ins_level
from target_hp_all
)
,target_ins_level_final as (
select distinct
inst_code,
MARKET,
target_ins_level
from target_ins_level1
)
--target_hp_all里一个机构一个市场对应了多个属性，这里按优先级依次取唯一值
--Channel
,Channel1 as (
select distinct
inst_code,
MARKET,
channel
from target_hp_all
)
,channel2 as (
select
inst_code,
MARKET,
channel,
row_number() over(partition by inst_code,market
order by case when channel = 'BU' then 1
when channel = 'BBU_County' then 2
when channel = 'CHC' then 3
when channel = 'Eagle' then 4
ELSE 5 end asc) as RN
from channel1
)
,channel_final as (
select
inst_code,
MARKET,
channel
from channel2
where RN = 1
)
--确定唯一值后重新放一起
,target_hp1 as (
select distinct
inst_code,
MARKET
from target_hp_all
)
,target_hp as (
select t1.inst_code,
t1.market,
t2.ka_flag,
t3.target_ins_level,
t4.channel
from target_hp1 t1
left join ka_flag_final t2
on t1.inst_code = t2.inst_code
and t1.market = t2.market
left join target_ins_level_final t3
on t1.inst_code = t3.inst_code
and t1.market = t3.market
left join channel_final t4
on t1.inst_code = t4.inst_code
and t1.market = t4.market
)
--获取Dragon_type
--判断Dragon_type是否为County
,county as (
select distinct inst_code,county_tier_desc
from dm.dm_td_institution
)
--判断另外2种
-- 20250825修改
,dragon_type as (
select src_hcc_code as inst_code,
src_prod_code as family_code,
max(is_core) is_core
from cdw_dwd_dwd_hospital_classification
where active_status = 1
and src_sales_cycle = (select key_value from dm.dm_td_date_config where key_code = 'pfmc_cvh')
group by src_hcc_code,
src_prod_code
)
--将Family_code转为pack_cod
,dragon_type_market1 as (
select distinct
t1.inst_code,
t1.is_core,
t4.market
from dragon_type t1
left join cdw_dwd_dim_product_wide t2
on t1.family_code = t2.prd_family_cd
left join dwd.dwd_td_imscode_azcode t3
on t2.prd_brand_cd = t3.brandcode
left join pack_market t4
on t3.ims_pack_cod = t4.pack_cod
where t4.market is not null
)
--按优先级取唯一值
,dragon_type_market2 as (
select distinct
inst_code,
is_core,
market,
row_number() over(partition by inst_code,market
order by case when is_core = 'CORE' then 1
when is_core = 'EMERGING' then 2
else 3 end asc) as RN
from dragon_type_market1
)
,dragon_type_market as (
select distinct
inst_code,
is_core,
market
from dragon_type_market2
where RN = 1
)
--综合前面几个表，得出dragon_flag
,dragon_flag1 as (
select distinct
t0.inst_code,
t0.market,
case when t1.county_tier_desc in ('County','County level city') then 'County'
when t2.is_core is null or length(t2.is_core) = 0 then 'Others'
else t2.is_core
end as dragon_flag
from target_hp t0
left join county t1
on t0.inst_code = t1.inst_code
left join dragon_type_market t2
on t0.inst_code = t2.inst_code
and t0.MARKET = t2.MARKET
)
--按优先级取唯一值
,dragon_flag2 as (
select inst_code,
market,
dragon_flag,
row_number() over(partition by inst_code,market
order by case when dragon_flag = 'CORE' then 1
when dragon_flag = 'EMERGING' then 2
when dragon_flag = 'County' then 3
else 4 end asc) as RN
from dragon_flag1
)
,dragon_flag as (
select inst_code,
market,
dragon_flag
from dragon_flag2
where RN = 1
)
--下面是为了把其他渠道并进来
,other_data_source as (
select explode(array('IQVIA-CHPA(Monthly)','XH Data(Quarterly)','EC(Monthly)','Retail(Quarterly)','CHC(Quarterly)','THC(Quarterly)','IQVIA-COUNTY(Quarterly)'))
)
,col as (
select '' as inst_code,
'' as MARKET,
'' as is_target_hp,
'' as sub_channel,
'' as dragon_flag,
'' as ka_flag,
'' as target_ins_level
)
,temp_1 as (
select distinct
case when t1.inst_code is null or t1.inst_code = '' then 'ROC'
else t1.inst_code
end as inst_code,
t1.MARKET,
case when t2.inst_code is null then 'N' else 'Y' end as is_target_hp,
case when t2.channel = 'BU' and geo.county_tier_desc in ('County','County level city') then 'BU_County'
when t2.channel = 'BU' and geo.county_tier_desc in ('City-3b','City-4a','City-4b') then 'BU_Emerging'
when t2.channel = 'BU' and t3.dragon_flag = 'CORE' then 'BU_Core'
when t2.channel = 'BU' then 'BU_Emerging'
when t2.channel in ('Eagle','CHC','BBU_County') then t2.channel
when t2.inst_code is null then 'NON_TARGET'
else 'NA'
end as sub_channel,
case when t3.dragon_flag is null then 'NON_TARGET'
else t3.dragon_flag
end as dragon_flag,
case when t2.ka_flag is null then 'NON_TARGET'
else t2.ka_flag
end as ka_flag,
case when t2.target_ins_level is null then 'NON_TARGET'
else t2.target_ins_level
end as target_ins_level,
'AIA(Monthly)' as DATA_SOURCE
from sales t1
left join target_hp t2
on t1.inst_code = t2.inst_code
and t1.MARKET = t2.MARKET
left join dragon_flag t3
on t1.inst_code = t3.inst_code
and t1.MARKET = t3.MARKET
left join dm.dm_td_institution inst
on t1.inst_code =inst.inst_code
left join dm.dm_td_geography geo
on inst.county_code = geo.geo_key
union all
select *
from col
cross join other_data_source
)
,sub_channel_rank as (
select distinct SUB_CHANNELSTD,SUB_CHANNEL_ORDER
from dm.dm_td_sd_channel
)
select t1.*,NVL(t2.SUB_CHANNEL_ORDER,100) as SUB_CHANNEL_ORDER
from temp_1 t1
left join sub_channel_rank t2
on t1.sub_channel = t2.SUB_CHANNELSTD

1544
EC/03 ec_load_data_backup.sql
View File

File diff suppressed because it is too large Load Diff

228
Merged_Data/Merged_Data_Config_table_bkp.sql
View File

@@ -1,228 +0,0 @@
-- Databricks notebook source
create or replace temporary view brand_province_config -- brand配置表
as
select upper(`brand`) as brand, `province` as province, datasource as data_source from dwd.dwd_gnd_merge_data_brand_province
-- COMMAND ----------
create or replace temporary view market_province_config --market 配置表
as
select market, province, datasource as data_source from dwd.dwd_gnd_merge_data_brand_market
-- COMMAND ----------
insert overwrite dws.dws_external_merge_data_config_special
-- LOCATION 'abfss://master@azcdatalakeprd.dfs.core.chinacloudapi.cn/DWS/dws_external_merge_data_config_special' --prd
-- LOCATION 'abfss://master@retaildlstoragetest.dfs.core.chinacloudapi.cn/DWS/dws_external_merge_data_config_special' --test
with pack_and_brand ( --brand 转 pack
select
upper(brandnamee) brand_name
,ims_pack_cod PACK_COD --取所有code带出market
from dwd.dwd_td_imscode_azcode
where ims_pack_cod is not null and startswith(ims_pack_cod,'0')
and upper(brandnamee) in (select brand from brand_province_config group by 1 )--取出在配置表中的pack
group by 1,2
)
,pack_to_market (
select
a.PACK_COD
,MARKET
from dm.dm_td_external_market_pack_mapping a
inner join pack_and_brand b on a.PACK_COD = b.pack_cod --取出配置表中的brand对应的 market
where DATA_SOURCE in ( 'AIA(Monthly)')--只取 AIA中的市场，chpa的市场太多
and lower(a.MARKET) not like '%all market%'
and market not in (
'RE Extended Market'
,'RE Market'
,'RE Market (Global)'
,'Pulmicort VBP Benchmark Market'
,'Respules Market'
)
group by 1,2
)
,market_to_pack ( -- 获取上述market中所有的pack
select
market
,pack_cod
from dm.dm_td_external_market_pack_mapping a
where DATA_SOURCE in ( 'AIA(Monthly)','IQVIA-CHPA(Monthly)')--AIA和chpa的市场
and market in (select market from pack_to_market group by 1)
group by 1,2
)
,province_to_audit (
select
PROVINCE_C
,AUDIT_COD
,DATA_SOURCE
,CASE WHEN DATA_SOURCE = 'AIA(Monthly)' THEN 'AIA'
WHEN DATA_SOURCE = 'IQVIA-CHPA(Monthly)' THEN 'CHPA'
ELSE NULL
END AS SOURCENAME
from dm.dm_td_external_geo
where DATA_SOURCE in ( 'AIA(Monthly)','IQVIA-CHPA(Monthly)')
group by 1,2,3
)
,not_selected_datasource (--因为pack2market2pack 会导致部分大市场包含所有的pack，故用配置表再删除一些不需要配置数据源的
select
a.ims_pack_cod PACK_COD --取所有code
,b.province
from dwd.dwd_td_imscode_azcode a
inner join (
select
UPPER(brand) as brand
,province
from brand_province_config
where data_source = '/'
group by 1,2
) b
on upper(a.brandnamee) = upper(b.brand )
where a.ims_pack_cod is not null and startswith(a.ims_pack_cod,'0')
group by 1,2
union
select
pack_cod,
'天津'
from market_to_pack
where market= 'EGFR TKI Market'
group by 1
)
select DATA_SOURCE,PACK_COD,AUDIT_COD
from not_selected_datasource a
left join province_to_audit b on a.province = b.province_c
group by 1,2,3
-- COMMAND ----------
INSERT OVERWRITE dws.dws_external_merge_data_config
-- create or replace table dws.dws_external_merge_data_config
/*
1通过brand找到pack，通过pack找到市场以市场下的所有pack
2通过province找到对应的省份或者机构，从而获取到audit_code
*/
----pack
with pack_and_brand ( --brand 转 pack
select
upper(brandnamee) brand_name
,ims_pack_cod PACK_COD --取所有code带出market
from dwd.dwd_td_imscode_azcode
where ims_pack_cod is not null and startswith(ims_pack_cod,'0')
and upper(brandnamee) in (select brand from brand_province_config group by 1 )--取出在配置表中的pack
group by 1,2
)
,pack_to_market (
select
a.PACK_COD
,MARKET
from dm.dm_td_external_market_pack_mapping a
inner join pack_and_brand b on a.PACK_COD = b.pack_cod --取出配置表中的brand对应的 market
where DATA_SOURCE in ( 'AIA(Monthly)')--只取 AIA中的市场，chpa的市场太多
and lower(a.MARKET) not like '%all market%'
and market not in (
'RE Extended Market'
,'RE Market'
,'RE Market (Global)'
,'Pulmicort VBP Benchmark Market'
,'Respules Market'
)
group by 1,2
)
,market_to_pack ( -- 获取上述market中所有的pack
select
market
,pack_cod
from dm.dm_td_external_market_pack_mapping a
where DATA_SOURCE in ( 'AIA(Monthly)','IQVIA-CHPA(Monthly)')--AIA和chpa的市场
and market in (select market from pack_to_market group by 1)
group by 1,2
)
,province_to_audit (
select
PROVINCE_C
,AUDIT_COD
,DATA_SOURCE
,CASE WHEN DATA_SOURCE = 'AIA(Monthly)' THEN 'AIA'
WHEN DATA_SOURCE = 'IQVIA-CHPA(Monthly)' THEN 'CHPA'
ELSE NULL
END AS SOURCENAME
from dm.dm_td_external_geo
where DATA_SOURCE in ( 'AIA(Monthly)','IQVIA-CHPA(Monthly)')
group by 1,2,3
)
,pack_result as (
select
D.pack_cod
,E.audit_cod
,A.province
,A.data_source as SOURCENAME
,E.data_source
from brand_province_config A
inner join pack_and_brand B on A.brand = B.brand_name
inner join pack_to_market C on B.pack_cod = C.pack_cod
inner join market_to_pack D on C.market = D.market
inner join province_to_audit E on A.province = E.province_c AND A.data_source = E.SOURCENAME
where A.data_source <> '/'
group by 1,2,3,4,5
)
----market
,config_market (
select
market
from market_province_config
group by 1
)
,market_to_pack_dtp (
select
PACK_COD,MARKET
from dm.dm_td_external_market_pack_mapping
where DATA_SOURCE = 'DTP(Quarterly)' and PACK_COD in (
select
PACK_COD
from dm.dm_tf_external_sales
where DTP_FLAG =1
group by 1)
)
,final_market_to_pack (--获取需要转换market成pack的部分
select PACK_COD,a.MARKET from market_to_pack a inner join config_market b on a.market = b.market
union
select PACK_COD,a.MARKET from market_to_pack_dtp a inner join config_market b on a.market = b.market
)
----market
,market_result as (
select
B.pack_cod
,C.AUDIT_COD
,A.province
,A.data_source as SOURCENAME
,C.DATA_SOURCE
from market_province_config a
left join final_market_to_pack B on A.market = B.market
left join province_to_audit C on A.province = C.province_c AND A.data_source = C.SOURCENAME
where a.data_source <> '/' AND C.AUDIT_COD IS not NULL
group by 1,2,3,4,5)
,need_to_be_changed( --算出不在pack 中的pack code，替换成 market中的pack code 的配置信息
select
pack_cod,AUDIT_COD,DATA_SOURCE,sourcename
from market_result
where pack_cod in (select pack_cod from pack_result group by 1)
)
--最终写入表
select
pack_cod,AUDIT_COD,DATA_SOURCE,sourcename, 'pack' CONFIG_DATASOURCE
from pack_result
union ALL
select
pack_cod,AUDIT_COD,DATA_SOURCE,sourcename, 'market' CONFIG_DATASOURCE
from pack_result
where pack_cod not in (select pack_cod from need_to_be_changed group by 1)
union all
select
pack_cod,AUDIT_COD,DATA_SOURCE,sourcename, 'market' CONFIG_DATASOURCE
from need_to_be_changed

124
Merged_Data/dm_tf_exteranl_sales_merged_data_dtp_others_bkp.sql
View File

@@ -1,124 +0,0 @@
-- Databricks notebook source
-- create or replace table dm.dm_tf_exteranl_sales_merged_data_dtp_others
insert overwrite table dm.dm_tf_exteranl_sales_merged_data_dtp_others
with all_egfr_pack as (
select * from DM.dm_td_external_market_pack_mapping
where DATA_SOURCE = 'DTP(Quarterly)' and MARKET = 'EGFR TKI Market'
)
,4brand as (
select * from all_egfr_pack
where PROD_DES_C in ('阿美乐','泰瑞沙','凯美纳','艾弗沙')
)
,otherbrand as (
select PACK_COD from all_egfr_pack
where PACK_COD not in ( select PACK_COD from `4brand` group by 1)
group by 1
)
-- -- 直取表 cu - dtp4brand 的 cu = otherbrand 的cu
,dirct_cu as (
select
yyyymm
,replace(audit_key,'DTP(Quarterly)','') audit_cod
,cast(pdot_counting_unit as decimal ) pdot_counting_unit
,cast(pdot_counting_unit_ly as decimal ) pdot_counting_unit_ly
from dm.dm_tf_external_retail_special a
where yyyymm >=202401
)
--先 4个brand 的 cu 和 value
,fact_sales_dtp_4brand(
select
YYYYMM,AUDIT_COD
,sum(cast( CONUTING_UNIT*PDOT_MARKET_RATIO as decimal)) CONUTING_UNIT
,sum(cast( CONUTING_UNIT_LY*PDOT_MARKET_RATIO as decimal)) CONUTING_UNIT_LY
from dm.dm_tf_external_sales t1
inner join 4brand t2 on t1.PACK_COD = t2.PACK_COD
where t1.DATA_SOURCE = 'DTP(Quarterly)'
and DTP_FLAG = 1
and YYYYMM >=202401
group by
YYYYMM,AUDIT_COD )
-- 其他产品 的 cu 和 value
,fact_sales_dtp_other(
select
YYYYMM
,"DTPOTHERS" PACK_COD
,"DTPOTHERS" CORP_COD
,AUDIT_COD
,'' as PLATFORM_TYPE
,'' as STORE_NAME
,'' as STORE_TYPE
,'' as REGION_TYPE
,1 as PACK_FLAG
,2 as PROD_FLAG
,1 as DTP_FLAG
,sum(SALES_UNIT_CAL) SALES_UNIT_CAL
,sum(SALES_UNIT_CAL_LY) SALES_UNIT_CAL_LY
,sum(SALES_VALUE_CAL) SALES_VALUE_CAL
,sum(SALES_VALUE_CAL_LY) SALES_VALUE_CAL_LY
,sum(CONUTING_UNIT) CONUTING_UNIT
,sum(CONUTING_UNIT_LY) CONUTING_UNIT_LY
,'DTP(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 dm.dm_tf_external_sales
where DTP_FLAG = 1
and PACK_COD in (select PACK_COD from otherbrand) --其他部分
and DATA_SOURCE = 'DTP(Quarterly)'
and YYYYMM >=202401
group by
YYYYMM,AUDIT_COD)
select
T0.YYYYMM
,t0.PACK_COD
,t0.CORP_COD
,t0.AUDIT_COD
,t0.PLATFORM_TYPE
,t0.STORE_NAME
,t0.STORE_TYPE
,t0.REGION_TYPE
,t0.PACK_FLAG
,t0.PROD_FLAG
,t0.DTP_FLAG
,t0.SALES_UNIT_CAL
,t0.SALES_UNIT_CAL_LY
,t0.SALES_VALUE_CAL
,t0.SALES_VALUE_CAL_LY
,t1.CONUTING_UNIT
,t1.CONUTING_UNIT_LY
,t0.DATA_SOURCE
,t0.inst_code
,t0.cmps_flag
,t0.dept_name
,t0.prescription
,t0.prescription_ly
,t0.new_code
,t0.area
,t0.h_level
,t0.reimburse
,t0.reimburse_type
,t0.prescription_source
from fact_sales_dtp_other t0
left join (
select
t1.yyyymm
,t1.audit_cod
,t1.pdot_counting_unit - t2.conuting_unit as CONUTING_UNIT
,t1.pdot_counting_unit_ly - t2.conuting_unit_ly as CONUTING_UNIT_LY
from dirct_cu t1
left join fact_sales_dtp_4brand t2 on t1.yyyymm = t2.yyyymm and t1.audit_cod = t2.audit_cod
) t1
on t0.audit_cod = t1.audit_cod and t0.yyyymm = t1.yyyymm
-- select * from fact_sales_dtp_4brand

209
Retail/04 map_to_dws_table_backup.py
View File

@@ -1,209 +0,0 @@
# Databricks notebook source
############################################################START##############################################################
### STEP-1: insert splited pack data into tmp final table: tmp_retail_final_sales
# COMMAND ----------
# MAGIC %sql
# MAGIC -------------------------------------------------------------------------------------
# MAGIC -- STEP-1: insert splited pack data into tmp final table
# MAGIC -- insert into tmp_retail_final_sales
# MAGIC -------------------------------------------------------------------------------------
# MAGIC
# MAGIC with tmp_pack as (
# MAGIC select
# MAGIC ------------------------------------------------------
# MAGIC -- 有月度数据使用月度数据，无月度数据用季度数据去转
# MAGIC nvl(
# MAGIC a.month,
# MAGIC CONCAT(
# MAGIC SUBSTRING(a.quarter, 1, 4), -- 提取年份（前4位）
# MAGIC CASE
# MAGIC WHEN SUBSTRING(a.quarter, 6, 1) = '1' THEN '03' -- Q1 → 03月
# MAGIC WHEN SUBSTRING(a.quarter, 6, 1) = '2' THEN '06' -- Q2 → 06月
# MAGIC WHEN SUBSTRING(a.quarter, 6, 1) = '3' THEN '09' -- Q3 → 09月
# MAGIC WHEN SUBSTRING(a.quarter, 6, 1) = '4' THEN '12' -- Q4 → 12月
# MAGIC END
# MAGIC )
# MAGIC ) as YYYYMM,
# MAGIC ------------------------------------------------------
# MAGIC a.pack_code as iqvia_pack_code,
# MAGIC a.product_id as zk_product_id,
# MAGIC case when a.product_desc <> 'others' then a.product_desc else null end as prod_des_c,
# MAGIC case when a.product_desc <> 'others' then a.product_desc else concat('Others_', a.molecule_desc) end as PROD_MAPPING,
# MAGIC a.zk_regin as province_city,
# MAGIC a.level_market as market,
# MAGIC a.sales_value,
# MAGIC a.sales_unit,
# MAGIC ------------------------------------------------------
# MAGIC -- counting_unit取值逻辑：
# MAGIC -- 不能直接取原始pack文件表中的值，改为取pack_property表中counting_unit / unit的值
# MAGIC a.sales_unit * (b.counting_unit/ coalesce(b.unit,1)) as counting_unit,
# MAGIC ------------------------------------------------------
# MAGIC case when data_flag = 0 then 1 else 2 end as pack_flag,
# MAGIC case when brand_flag = 1 then 1 else 2 end as brand_flag
# MAGIC from tmp.tmp_retail_pack_rawdata a
# MAGIC left join dwd.dwd_gnd_ext_retail_pack_property b
# MAGIC on a.product_id = b.product_id
# MAGIC ), tmp_has_roc as (
# MAGIC select
# MAGIC product_id,
# MAGIC quarter,
# MAGIC ------------------------------------------------------
# MAGIC -- 有月度数据使用月度数据，无月度数据用季度数据去转
# MAGIC nvl(
# MAGIC month,
# MAGIC CONCAT(
# MAGIC SUBSTRING(quarter, 1, 4), -- 提取年份（前4位）
# MAGIC CASE
# MAGIC WHEN SUBSTRING(quarter, 6, 1) = '1' THEN '03' -- Q1 → 03月
# MAGIC WHEN SUBSTRING(quarter, 6, 1) = '2' THEN '06' -- Q2 → 06月
# MAGIC WHEN SUBSTRING(quarter, 6, 1) = '3' THEN '09' -- Q3 → 09月
# MAGIC WHEN SUBSTRING(quarter, 6, 1) = '4' THEN '12' -- Q4 → 12月
# MAGIC END
# MAGIC )
# MAGIC ) as month,
# MAGIC ------------------------------------------------------
# MAGIC pack_code
# MAGIC from tmp.tmp_retail_pack_rawdata
# MAGIC where zk_regin = 'ROC'
# MAGIC ), tmp_pack_this_year_with_roc as (
# MAGIC select
# MAGIC *
# MAGIC from tmp_pack a
# MAGIC where exists(
# MAGIC select * from tmp_has_roc b
# MAGIC where a.YYYYMM = b.month
# MAGIC and a.iqvia_pack_code = b.pack_code
# MAGIC and a.zk_product_id = b.product_id
# MAGIC ) and a.province_city <> '全国'
# MAGIC ), tmp_pack_next_year_with_roc as (
# MAGIC select
# MAGIC cast(YYYYMM + 100 as int) as YYYYMM,
# MAGIC iqvia_pack_code,
# MAGIC zk_product_id,
# MAGIC prod_des_c,
# MAGIC PROD_MAPPING,
# MAGIC province_city,
# MAGIC market,
# MAGIC sales_value as sales_value_ly,
# MAGIC sales_unit as sales_unit_ly,
# MAGIC counting_unit as counting_unit_ly,
# MAGIC pack_flag,
# MAGIC brand_flag
# MAGIC from tmp_pack a
# MAGIC where YYYYMM + 100 <= (select max(YYYYMM) from tmp_pack)
# MAGIC and exists(
# MAGIC select * from tmp_has_roc b
# MAGIC where a.YYYYMM = b.month
# MAGIC and a.iqvia_pack_code = b.pack_code
# MAGIC and a.zk_product_id = b.product_id
# MAGIC ) and a.province_city <> '全国'
# MAGIC
# MAGIC ), tmp_pack_this_year_without_roc as (
# MAGIC select
# MAGIC *
# MAGIC from tmp_pack a
# MAGIC where not exists(
# MAGIC select * from tmp_has_roc b
# MAGIC where a.YYYYMM = b.month
# MAGIC and a.iqvia_pack_code = b.pack_code
# MAGIC and a.zk_product_id = b.product_id
# MAGIC )
# MAGIC ), tmp_pack_next_year_without_roc as (
# MAGIC select
# MAGIC cast(YYYYMM + 100 as int) as YYYYMM,
# MAGIC iqvia_pack_code,
# MAGIC zk_product_id,
# MAGIC prod_des_c,
# MAGIC PROD_MAPPING,
# MAGIC province_city,
# MAGIC market,
# MAGIC sales_value as sales_value_ly,
# MAGIC sales_unit as sales_unit_ly,
# MAGIC counting_unit as counting_unit_ly,
# MAGIC pack_flag,
# MAGIC brand_flag
# MAGIC from tmp_pack a
# MAGIC where YYYYMM + 100 <= (select max(YYYYMM) from tmp_pack)
# MAGIC and not exists(
# MAGIC select * from tmp_has_roc b
# MAGIC where a.YYYYMM = b.month
# MAGIC and a.iqvia_pack_code = b.pack_code
# MAGIC and a.zk_product_id = b.product_id
# MAGIC )
# MAGIC ), tmp_final_sales as (
# MAGIC select
# MAGIC ifnull(a.yyyymm, b.yyyymm) as yyyymm,
# MAGIC ifnull(a.iqvia_pack_code, b.iqvia_pack_code) as iqvia_pack_code,
# MAGIC ifnull(a.zk_product_id, b.zk_product_id) as zk_product_id,
# MAGIC ifnull(a.prod_des_c, b.prod_des_c) as prod_des_c,
# MAGIC ifnull(a.PROD_MAPPING, b.PROD_MAPPING) as PROD_MAPPING,
# MAGIC ifnull(a.province_city, b.province_city) as province_city,
# MAGIC ifnull(a.market, b.market) as market,
# MAGIC ifnull(a.sales_value, 0) as sales_value,
# MAGIC ifnull(a.sales_unit, 0) as sales_unit,
# MAGIC ifnull(a.counting_unit, 0) as counting_unit,
# MAGIC ifnull(a.pack_flag, b.pack_flag) as pack_flag,
# MAGIC ifnull(a.brand_flag,b.brand_flag ) as brand_flag,
# MAGIC ifnull(b.sales_value_ly, 0) as sales_value_ly,
# MAGIC ifnull(b.sales_unit_ly, 0) as sales_unit_ly,
# MAGIC ifnull(b.counting_unit_ly, 0) as counting_unit_ly
# MAGIC from tmp_pack_this_year_with_roc a
# MAGIC full outer join tmp_pack_next_year_with_roc b
# MAGIC on a.YYYYMM = b.YYYYMM
# MAGIC and a.iqvia_pack_code = b.iqvia_pack_code
# MAGIC and a.zk_product_id = b.zk_product_id
# MAGIC and a.province_city = b.province_city
# MAGIC
# MAGIC union all
# MAGIC
# MAGIC select
# MAGIC ifnull(c.yyyymm, d.yyyymm) as yyyymm,
# MAGIC ifnull(c.iqvia_pack_code, d.iqvia_pack_code) as iqvia_pack_code,
# MAGIC ifnull(c.zk_product_id, d.zk_product_id) as zk_product_id,
# MAGIC ifnull(c.prod_des_c, d.prod_des_c) as prod_des_c,
# MAGIC ifnull(c.PROD_MAPPING, d.PROD_MAPPING) as PROD_MAPPING,
# MAGIC 'ROC' as province_city,
# MAGIC ifnull(c.market, d.market) as market,
# MAGIC ifnull(c.sales_value, 0) as sales_value,
# MAGIC ifnull(c.sales_unit, 0) as sales_unit,
# MAGIC ifnull(c.counting_unit, 0) as counting_unit,
# MAGIC --ifnull(c.pack_flag, d.pack_flag) as pack_flag,
# MAGIC 2 as pack_flag, -- 此类没有拆分比例，且pack只有全国的数，pack_flag固定为2
# MAGIC ifnull(c.brand_flag,d.brand_flag ) as brand_flag,
# MAGIC ifnull(d.sales_value_ly, 0) as sales_value_ly,
# MAGIC ifnull(d.sales_unit_ly, 0) as sales_unit_ly,
# MAGIC ifnull(d.counting_unit_ly, 0) as counting_unit_ly
# MAGIC from tmp_pack_this_year_without_roc c
# MAGIC full outer join tmp_pack_next_year_without_roc d
# MAGIC on c.YYYYMM = d.YYYYMM
# MAGIC and c.iqvia_pack_code = d.iqvia_pack_code
# MAGIC and c.zk_product_id = d.zk_product_id
# MAGIC and c.province_city = d.province_city
# MAGIC )
# MAGIC
# MAGIC insert overwrite table tmp.tmp_retail_final_sales
# MAGIC
# MAGIC select
# MAGIC yyyymm,
# MAGIC iqvia_pack_code,
# MAGIC zk_product_id,
# MAGIC prod_des_c,
# MAGIC PROD_MAPPING,
# MAGIC province_city,
# MAGIC market,
# MAGIC sales_value,
# MAGIC sales_value_ly,
# MAGIC sales_unit,
# MAGIC sales_unit_ly,
# MAGIC counting_unit,
# MAGIC counting_unit_ly,
# MAGIC pack_flag,
# MAGIC brand_flag
# MAGIC from tmp_final_sales
# MAGIC order by yyyymm
# COMMAND ----------
############################################################END################################################################

367
Retail/z1 dwd_inc_gnd_ext_retail_nataional_作废.py
View File

@@ -1,367 +0,0 @@
# Databricks notebook source
# MAGIC %md
# MAGIC ### 原本逻辑
# COMMAND ----------
#当更新pack 或品牌 事实数据时需要运行此代码，否则无需运行。
# COMMAND ----------
# MAGIC %sql
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_nataional_oap' where file_name ='pack-CV-抗血栓2通用名-全国.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_htn' where file_name ='pack-CV-高血压-化学药-全国.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_atomizer' where file_name ='pack-雾化器-全国&县域数据.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_anti_asthma_copd' where file_name ='pack-RE-慢阻肺-全国.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_zk_brand' where file_name ='Brand-品牌数据报表.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_statin_xzk' where file_name ='pack-CV-他汀类+血脂康-全国.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_nataional_rd' where file_name ='pack-RD-肾科-全国.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_aagsa_ppi_oral' where file_name ='pack-GI-慢性胃炎胃溃疡-全国.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_nataional_niad' where file_name ='pack-DM-口服降糖化学药.xlsx';
# MAGIC update dwd.dwd_gnd_ext_retail_corresponding_relationship set table_name ='dwd.dwd_gnd_ext_retail_metoprolol_tartrat' where file_name ='pack-CV-酒石酸美托洛尔.xlsx';
# MAGIC
# COMMAND ----------
# pack数据自动接入 整合
#获取配置表信息(表名、brand_flag
df = spark.sql("""
SELECT DISTINCT table_name tab ,file_name brand_flag FROM dwd.dwd_gnd_ext_retail_corresponding_relationship
where type_name ='PACK'
""").collect()
def get_union_pack_data(df):
#初始化结果集
union_query = None
# niad_pdot_unit需特殊赋值对应表名：tmp.tmp_inc_gnd_ext_retail_nataional_niad
# niad_pdot_unit_flag = 'tmp.tmp_inc_gnd_ext_retail_nataional_niad'
for table in df:
# 选择当前表名
T = str(table.tab)
# 获取对应brand表维度对应得 market 名称
brand_flag = str(table.brand_flag)
sql = f"""
select
cast(t1.month as int) AS YYYYMM
,cast(left(t1.quarter, 4) as int) AS year
,right(t1.quarter, 2) AS quarter
,t1.quarter AS yq
,t1.zk_product_id
,t1.zk_region
,t1.zk_rx_otc
,t1.zk_medicine_type
,t1.zk_medicine_tier1
,t1.zk_medicine_tier2
,t1.zk_medicine_tier3
,t1.zk_medicine_tier4
,t1.zk_common_name
,t1.zk_dosage_form
,t1.zk_user_type
,t1.zk_category_name
,t1.zk_product_name
,t1.zk_brand_name
,t1.zk_manu_des
,t1.zk_corp_des
,t1.zk_pack_des
,t1.price
,CAST(replace(t1.sales_unit,',','') AS decimal(30,10)) as sales_unit
,CAST(replace(t1.sales_value,',','') AS decimal(30,10)) as sales_value
,CAST(replace(t1.digital_spread_rate,',','') AS decimal(30,10)) as digital_spread_rate
,CAST(replace(t1.weighted_spread_rate,',','') AS decimal(30,10)) as weighted_spread_rate
,CAST(replace(t1.counting_unit,',','') AS decimal(30,10)) as counting_unit
,'{brand_flag}' as brand_flag
,from_utc_timestamp(current_timestamp(),'UTC+8') AS etl_insert_dt
,from_utc_timestamp(current_timestamp(),'UTC+8') AS etl_update_dt
from {T} t1
left join dws.dws_ext_retail_td_prod t2
on t1.zk_product_id = t2.zk_product_id
where month is not null
"""
# 读取数据
current_query = spark.sql(sql)
#union 数据
if union_query ==None:
union_query=current_query
else:
union_query = union_query.union(current_query)
#返回数据集 / 写入表也行？？？
return union_query
pack_result = get_union_pack_data(df)
pack_result.write.mode("overwrite").saveAsTable("dwd.dwd_inc_gnd_ext_retail_nataional_pack_union_all")
# COMMAND ----------
# brand+ 省份数据自动接入
#获取配置表信息(表名、brand_flag
dfband = spark.sql("""
SELECT DISTINCT table_name tab ,file_name brand_flag FROM dwd.dwd_gnd_ext_retail_corresponding_relationship
where type_name ='BRAND'
""").collect()
def get_union_brand_data(df):
#数据为空
if df == None:
return None
#初始化结果集
union_query = None
for table in df:
# 选择当前表名
T = str(table.tab)
# 获取对应brand表维度对应得 market 名称
pack_flag = str(table.brand_flag)
sql = f"""
select
cast(left(quarter, 4)*100 + right(quarter,1)*3 as int ) AS YYYYMM
,cast(left(quarter, 4) as int ) AS year
,right(quarter, 2) AS quarter
,quarter AS yq
,type AS brand_cat_type
,case when ta = 'NIAD' then 'DM' else ta end AS TA
,market AS market
,zk_brand_category AS zk_brand_category
,zk_common_name AS zk_common_name
,zk_manu_des AS zk_manu_des
,rc_name_en AS rc_name_en
,province_city AS province_city
,ytd AS ytd
,cast(sales_value * 1000000 as decimal(30,10)) AS sales_val
,cast(sales_volume * 1000000 as decimal(30,10)) AS sales_vol
,cast(price as decimal(30,10)) as price
,cast(num_dist_rate as decimal(30,10)) as num_dist_rate
,cast(weig_dist_rate as decimal(30,10)) as weig_dist_rate
,cast(value_share as decimal(30,10)) as val_share
,cast(volume_share as decimal(30,10)) as vol_share
,replace(key_brand_ytd,'-','') as key_brand_ytd
,cast(replace(key_brand_rank_ytd,'-','0') as int) as key_brand_rank_ytd
,replace(top_brand_ytd,'-','') as top_brand_ytd
,cast(replace(top_brand_ms_ytd,'-','0') as decimal(30,10)) as top_brand_ms_ytd
,cast(replace(top_brand_inc_ms_ytd,'-','0') as decimal(30,10)) as top_brand_inc_ms_ytd
,cast(replace(top_brand_gr_ytd,'-','0') as decimal(30,10)) as top_brand_gr_ytd
,replace(key_brand_qtd,'-','') as key_brand_qtd
,cast(replace(key_brand_rank_qtd,'-','0') as int) as key_brand_rank_qtd
,replace(top_brand_qtd,'-','') as top_brand_qtd
,cast(replace(top_brand_ms_qtd,'-','0') as decimal(30,10)) as top_brand_ms_qtd
,cast(replace(top_brand_inc_ms_qtd,'-','0') as decimal(30,10)) as top_brand_inc_ms_qtd
,cast(replace(top_brand_gr_qtd,'-','0') as decimal(30,10)) as top_brand_gr_qtd
,ranked_by as ranked_by
,'{pack_flag}' as pack_flag
,from_utc_timestamp(current_timestamp(),'UTC+8') as etl_insert_dt
,from_utc_timestamp(current_timestamp(),'UTC+8') as etl_update_dt
from {T}
"""
# 读取数据
current_query = spark.sql(sql)
#union 数据
if union_query == None:
union_query = current_query
else:
union_query = union_query.union(current_query)
#返回数据集 / 写入表也行？？？
return union_query
brand_result = get_union_brand_data(dfband)
brand_result.write.mode("overwrite").saveAsTable("dwd.dwd_inc_gnd_ext_retail_nataional_brand_union_all")
# COMMAND ----------
# MAGIC %sql
# MAGIC ----------------------多达一、天一宁、others、氨氯地平阿托伐他汀钙 数据在 高血压和他汀血脂康里面重复，但该数据在品牌报表里面没有对应的值，会导致后续拆分到pack + 省份时得到的结果不一致，因此需要将高血压或者他汀血脂康数据 根据pack+ 全国数据按省份数平均分配，并汇总到品牌维度，写入品牌数据报表，为后续pack +全国拆分到pack + 省份 提供数据基础
# MAGIC insert overwrite table dwd.dwd_inc_gnd_ext_retail_nataional_brand_union_all
# MAGIC with data_pack as (
# MAGIC ----------底表获取基础数据----pack 汇总到品牌 --并按省份数量平均
# MAGIC select brand_flag,a.YYYYMM,nvl(b.prod_des_c,'OTHERS' ) prod_des_c ,sum(a.sales_value)/sum(num) sales_value,sum(a.sales_unit)/sum(num) sales_unit ,sum(a.sales_value) sales ,sum(a.sales_unit) saleu
# MAGIC from dwd.dwd_inc_gnd_ext_retail_nataional_pack_union_all a
# MAGIC left join dwd.dwd_gnd_ext_retail_pack_property b on a.zk_product_id = b.product_id
# MAGIC cross join (select count(distinct zk_region) num from dwd.dwd_inc_gnd_ext_retail_nataional_pack_union_all where zk_region<>'全国')
# MAGIC where a.zk_common_name ='氨氯地平阿托伐他汀钙' and a.brand_flag in ('pack-CV-他汀类+血脂康-全国.xlsx' ,'pack-CV-高血压-化学药-全国.xlsx','pack-CV-抗血栓2通用名-全国.xlsx')
# MAGIC group by 1,2,3
# MAGIC ),city as (
# MAGIC ----获取省份及年月头表
# MAGIC select distinct
# MAGIC YYYYMM
# MAGIC ,year
# MAGIC ,quarter
# MAGIC ,yq
# MAGIC ,province_city
# MAGIC from dwd.dwd_inc_gnd_ext_retail_nataional_brand_union_all
# MAGIC where province_city !='全国' and ranked_by ='volume'
# MAGIC )
# MAGIC , split as (
# MAGIC ------分别补充全国数据、品牌数据------
# MAGIC select
# MAGIC city.YYYYMM
# MAGIC ,year
# MAGIC ,quarter
# MAGIC ,yq
# MAGIC ,'品牌' brand_cat_type
# MAGIC ,'CV' TA
# MAGIC ,brand_flag market
# MAGIC ,prod_des_c
# MAGIC ,''
# MAGIC ,''
# MAGIC ,''
# MAGIC ,province_city
# MAGIC ,''
# MAGIC ,sales_value
# MAGIC ,sales_unit
# MAGIC ,0,0,0,0,0,'' ,0 ,'' ,0,0,0,'',0,'',0,0,0,'volume' ,''
# MAGIC from data_pack left join city on city.YYYYMM=data_pack.YYYYMM
# MAGIC union all
# MAGIC select
# MAGIC city.YYYYMM
# MAGIC ,year
# MAGIC ,quarter
# MAGIC ,yq
# MAGIC ,'品牌' brand_cat_type
# MAGIC ,'CV' TA
# MAGIC ,brand_flag market
# MAGIC ,prod_des_c
# MAGIC ,''
# MAGIC ,''
# MAGIC ,''
# MAGIC ,'全国' province_city
# MAGIC ,''
# MAGIC ,sales sales_value
# MAGIC ,saleu sales_unit
# MAGIC ,0,0,0,0,0,'' ,0 ,'' ,0,0,0,'',0,'',0,0,0,'volume' ,''
# MAGIC from data_pack left join (select distinct year,quarter,yyyymm,yq from city ) city on city.YYYYMM=data_pack.YYYYMM
# MAGIC
# MAGIC )
# MAGIC select * from dwd.dwd_inc_gnd_ext_retail_nataional_brand_union_all
# MAGIC union all
# MAGIC select *
# MAGIC ,from_utc_timestamp(current_timestamp(),'UTC+8') AS etl_insert_dt
# MAGIC ,from_utc_timestamp(current_timestamp(),'UTC+8') AS etl_update_dt
# MAGIC from split
# COMMAND ----------
# MAGIC %sql
# MAGIC --补位
# MAGIC with tmp as (
# MAGIC select
# MAGIC YYYYMM,
# MAGIC year,
# MAGIC quarter,
# MAGIC yq,
# MAGIC case when length(zk_product_id) < 7 then right(concat('0000000',zk_product_id),7) else zk_product_id end as zk_product_id,
# MAGIC zk_region,
# MAGIC zk_rx_otc,
# MAGIC zk_medicine_type,
# MAGIC zk_medicine_tier1,
# MAGIC zk_medicine_tier2,
# MAGIC zk_medicine_tier3,
# MAGIC zk_medicine_tier4,
# MAGIC zk_common_name,
# MAGIC zk_dosage_form,
# MAGIC zk_user_type,
# MAGIC zk_category_name,
# MAGIC zk_product_name,
# MAGIC zk_brand_name,
# MAGIC zk_manu_des,
# MAGIC zk_corp_des,
# MAGIC zk_pack_des,
# MAGIC price,
# MAGIC sales_unit,
# MAGIC sales_value,
# MAGIC digital_spread_rate,
# MAGIC weighted_spread_rate,
# MAGIC counting_unit,
# MAGIC brand_flag,
# MAGIC etl_insert_dt,
# MAGIC etl_update_dt
# MAGIC from dwd.dwd_inc_gnd_ext_retail_nataional_pack_union_all
# MAGIC -- pack-CV-酒石酸美托洛尔.xlsx的数据没有了，所以排除 酒石酸美托洛尔
# MAGIC where brand_flag<>'pack-CV-酒石酸美托洛尔.xlsx'
# MAGIC )
# MAGIC
# MAGIC insert overwrite dwd.dwd_inc_gnd_ext_retail_nataional_pack_union_all
# MAGIC select *
# MAGIC from tmp;
# COMMAND ----------
# MAGIC %md
# MAGIC ### 新逻辑
# MAGIC - 修改brand数据，先拆分成月维度的数据
# COMMAND ----------
# MAGIC %sql
# MAGIC /*
# MAGIC 修改时间：20250311
# MAGIC 修改人：chenwu
# MAGIC 修改内容：brand来数频率为 季度来数， 但是 pack 为 月度来数据，需要用季度的数据/3得到月度的
# MAGIC */
# MAGIC insert overwrite table dwd.dwd_inc_gnd_ext_retail_nataional_brand_union_all
# MAGIC with quarterly_table as (
# MAGIC select
# MAGIC *
# MAGIC from dwd.dwd_inc_gnd_ext_retail_nataional_brand_union_all
# MAGIC where market not in ('NIAD','Inhaled Extended Market','布地奈德雾化溶液')
# MAGIC -- 范围内只能是 季度来数据的，如果有月度来数据的需要排除掉
# MAGIC )
# MAGIC
# MAGIC ,month_table as (--转化成月度数据
# MAGIC SELECT
# MAGIC SUBSTR(q.yq, 1, 4)*100 + -- 提取年份
# MAGIC LPAD(m.month_num, 2, '0') -- 补零月份
# MAGIC AS YYYYMM -- 月份首日
# MAGIC ,`year`
# MAGIC ,`quarter`
# MAGIC ,yq
# MAGIC ,brand_cat_type
# MAGIC ,TA
# MAGIC ,market
# MAGIC ,zk_brand_category
# MAGIC ,zk_common_name
# MAGIC ,zk_manu_des
# MAGIC ,rc_name_en
# MAGIC ,province_city
# MAGIC ,ytd
# MAGIC ,sales_val /3 --除3
# MAGIC ,sales_vol /3 --除3
# MAGIC ,price
# MAGIC ,num_dist_rate
# MAGIC ,weig_dist_rate
# MAGIC ,val_share
# MAGIC ,vol_share
# MAGIC ,key_brand_ytd
# MAGIC ,key_brand_rank_ytd
# MAGIC ,top_brand_ytd
# MAGIC ,top_brand_ms_ytd
# MAGIC ,top_brand_inc_ms_ytd
# MAGIC ,top_brand_gr_ytd
# MAGIC ,key_brand_qtd
# MAGIC ,key_brand_rank_qtd
# MAGIC ,top_brand_qtd
# MAGIC ,top_brand_ms_qtd
# MAGIC ,top_brand_inc_ms_qtd
# MAGIC ,top_brand_gr_qtd
# MAGIC ,ranked_by
# MAGIC ,pack_flag
# MAGIC ,etl_insert_dt
# MAGIC ,etl_update_dt
# MAGIC FROM
# MAGIC quarterly_table q
# MAGIC LATERAL VIEW EXPLODE( -- 为每季度生成三个月
# MAGIC CASE
# MAGIC WHEN RIGHT(q.yq, 2) = 'Q1' THEN ARRAY(1, 2, 3)
# MAGIC WHEN RIGHT(q.yq, 2) = 'Q2' THEN ARRAY(4, 5, 6)
# MAGIC WHEN RIGHT(q.yq, 2) = 'Q3' THEN ARRAY(7, 8, 9)
# MAGIC WHEN RIGHT(q.yq, 2) = 'Q4' THEN ARRAY(10, 11, 12)
# MAGIC END
# MAGIC ) m AS month_num
# MAGIC )
# MAGIC
# MAGIC ,other_not_quarterly_table (
# MAGIC select
# MAGIC *
# MAGIC from dwd.dwd_inc_gnd_ext_retail_nataional_brand_union_all
# MAGIC where market in ('NIAD','Inhaled Extended Market','布地奈德雾化溶液')
# MAGIC -- 范围内只能是 月度来数据的
# MAGIC )
# MAGIC
# MAGIC select * from month_table
# MAGIC union all
# MAGIC select * from other_not_quarterly_table

6276
Retail/z2 retail_load_data_作废.sql
View File

File diff suppressed because it is too large Load Diff

1201
Retail/z3 retail_overview_data_作废.sql
View File

File diff suppressed because it is too large Load Diff

1187
z 01 dm_tf_external_sales.sql
View File

File diff suppressed because it is too large Load Diff

1439
z 10 dm_td_external_market_pack_mapping_作废.sql
View File

File diff suppressed because it is too large Load Diff

186
z 11 dm_td_external_market_作废.sql
View File

@@ -1,186 +0,0 @@
-- Databricks notebook source
INSERT OVERWRITE dm.dm_td_external_market (
MARKET, DATA_SOURCE, TA, AZ_RELATED, MARKET_DESC, TA_RN, PRD_FLAG
)
SELECT
MARKET,
DATA_SOURCE,
MAX(TA) TA,
MAX(AZ_RELATED) AZ_RELATED,
MAX(MARKET_DESC) MARKET_DESC,
MAX(TA_RN) TA_RN,
case
when
DATA_SOURCE = 'Retail(Quarterly)'
and MARKET in (
'Antacid anti GI swelling agent+PPI Oral',
'HTN Market',
'NIAD (Non-Insulin Anti-Diabetic) Market',
'Respules Market',
'Statin+XZK Market',
'BUDESONIDE Market',
'Rosuvastatin Market',
'Atorvastatin Market',
'Metoprolol Succinate Market',
'PPI Oral ESOMEPRAZOLE Market',
'PPI Oral OMEPRAZOLE Market',
--'Inhaled Extended Market by Brand',
'Inhaled Extended Market',
'Non-Omeprazole',
'Metoprolol Tartrate Market',
'Pediatric Cough Market',
'Anti-HER2 Market',
'Anti-HER2 mBC Market',
'Fasenra Market',
'Severe Asthma'
)
then
'1'
when
data_source = 'DTP(Quarterly)'
and market in (
'EGFR TKI Market',
'Anti-HER2 Market',
'Anti-HER2 mBC Market',
'Fasenra/Tezspire Related Market',
'Severe Asthma',
'Tezspire CRSwNP Market',
'CRSwNP Market',
'PAM Market'
)
then
'1'
else '0'
end as PRD_FLAG
FROM
DM.dm_td_external_market_pack_mapping_temp
GROUP BY
MARKET,
DATA_SOURCE
-- COMMAND ----------
refresh table dm.DM_TD_EXTERNAL_MARKET
-- COMMAND ----------
insert overwrite dm.dm_td_external_market_ta(MARKET,DATA_SOURCE,TA,TA_RN,MARKET_DESC)
SELECT MARKET,DATA_SOURCE, AZ_RELATED TA, TA_RN,MARKET_DESC
FROM DM.DM_TD_EXTERNAL_MARKET
WHERE MARKET NOT LIKE '%ALL Market%' and MARKET <>'Retail All Market'
UNION all
SELECT MARKET,DATA_SOURCE, 'CVRM' TA,0 TA_RN,MARKET_DESC
FROM DM.DM_TD_EXTERNAL_MARKET
WHERE MARKET = 'CVRM Market'
-- COMMAND ----------
-----------------------------------------------------------------------------------
--修改时间：20240822
--修改人：FanXujia
--修改内容：
--Retail、EC渠道，新增NIAD这个TA。数据与DM相同
-----------------------------------------------------------------------------------
insert into dm.dm_td_external_market_ta
(
MARKET,
DATA_SOURCE,
TA,
TA_RN,
MARKET_DESC
)
select
market,
DATA_SOURCE,
'NIAD' as TA,
7 as TA_RN,
MARKET_DESC
from dm.dm_td_external_market_ta
where DATA_SOURCE = 'Retail(Quarterly)'
and ta = 'DM'
union all
select
market,
DATA_SOURCE,
'NIAD' as TA,
7 as TA_RN,
MARKET_DESC
from dm.dm_td_external_market_ta
where DATA_SOURCE = 'EC(Monthly)'
and ta = 'DM'
--20250701 chenwu 新增RARE = ONCO的Soliris Related Market
union all
select
market,
DATA_SOURCE,
'RARE' as TA,
8 as TA_RN,
MARKET_DESC
from dm.dm_td_external_market_ta
where DATA_SOURCE = 'Retail(Quarterly)'
and ta = 'ONCO'
and MARKET = 'Soliris Related Market'
-- COMMAND ----------
-- log 20250226 auth:庄伟
-- 对market_ta表中市场进行补全，并依据手工文件维护的 market 与 ta 映射，匹配出完整市场对应的TA_NAME
insert overwrite table dm.dm_td_external_market_ta
(
MARKET,
DATA_SOURCE,
TA,
TA_MAP,
TA_RN,
MARKET_DESC
)
-----找出在market维度表中不存在的market列表
with temp_ma_ext_market
as
(
select
distinct
MARKET
,DATA_SOURCE
from dm.dm_td_external_market
where MARKET NOT IN (
select distinct MARKET
from dm.dm_td_external_market_ta
)
)
------补充完整的market列表，并通过market匹配手工文件维护的market对应的TA
select
distinct
dim_mkt_ta.MARKET,
DATA_SOURCE,
dim_mkt_ta.TA,
nvl(mkt_property.ta,mkt_property_key.TA) TA_MAP,
dim_mkt_ta.TA_RN,
dim_mkt_ta.MARKET_DESC
from
(
select MARKET,DATA_SOURCE,TA,TA_RN,MARKET_DESC
from dm.dm_td_external_market_ta
)dim_mkt_ta
left join dm.dm_ims_td_market_property mkt_property
on dim_mkt_ta.market = mkt_property.market
left join dm.dm_ims_td_market_property mkt_property_key
on dim_mkt_ta.MARKET = mkt_property_key.MARKET_KEY
union all
select
ext_market.MARKET,
ext_market.DATA_SOURCE,
null TA,
mkt_property.TA TA_MAP,
'99' as TA_RN,
null MARKET_DESC
from temp_ma_ext_market ext_market
left join dm.dm_ims_td_market_property mkt_property
on ext_market. market =mkt_property.MARKET
where nvl(mkt_property.TA,'')<>'';