I need to translate this query from Oracle to Postgres:
select
case when (not regexp_like(upper(location),'^[0-9]{1,5} [A-Z]') or not regexp_like(upper(location),'^[0-9]{1,5}.* [a-z]') or regexp_like(upper(location),'^[0-9]{1,5}')) and regexp_like(upper(modelname),'CEDMI') then '74'
when (not regexp_like(upper(location),'^[0-9]{1,5} [A-Z]') or not regexp_like(upper(location),'^[0-9]{1,5}.* [a-z]') or regexp_like(upper(location),'^[0-9]{1,5}')) and regexp_like(upper(modelname),'CEDNA') then '768'
when (not regexp_like(upper(location),'^[0-9]{1,5} [A-Z]') or not regexp_like(upper(location),'^[0-9]{1,5}.* [a-z]') or regexp_like(upper(location),'^[0-9]{1,5}')) and regexp_like(upper(modelname),'CEDRM') then '598'
when (not regexp_like(upper(modelname),'CEDMI') or not regexp_like(upper(modelname),'CEDNA') or not regexp_like(upper(modelname),'CEDRM')) and (regexp_like(upper(location),'^[A-Z]') or location is null) then null
else location
end as LOCATION
from stagingarea.ENEL_TLC_T_DEVICE
can someone help me?
thanks in advance for your attention and support
CodePudding user response:
In postgres the operateur ~
performs a regex comparison as thus replaces the Oracle function regexp_like()
. Your query therefore becomes.
I would like to alert your attention that the test not upper(location) ~ '^[0-9]{1,5}.* [a-z]'
will always be true because the application of upper()
makes it impossible to match [a-z]
.
create table ENEL_TLC_T_DEVICE( location varchar(100), modelname varchar(100) );
✓
insert into ENEL_TLC_T_DEVICE (location, modelname) values ('12A','CEDMI'), ('12A','CEDRM'), ('12A','CEDNA'), ('12AA','CEDMI'), ('12AA','CEDRM'), ('12AA','CEDNA'), (null,null);
7 rows affected
select location, modelname, case when (not upper(location) ~ '^[0-9]{1,5} [A-Z]' or not upper(location) ~ '^[0-9]{1,5}.* [a-z]' or upper(location) ~ '^[0-9]{1,5}') and upper(modelname) ~ 'CEDMI' then '74' when (not upper(location) ~ '^[0-9]{1,5} [A-Z]' or not upper(location) ~ '^[0-9]{1,5}.* [a-z]' or upper(location) ~ '^[0-9]{1,5}') and upper(modelname) ~ 'CEDNA' then '768' when (not upper(location) ~ '^[0-9]{1,5} [A-Z]' or not upper(location) ~ '^[0-9]{1,5}.* [a-z]' or upper(location) ~ '^[0-9]{1,5}') and upper(modelname) ~ 'CEDRM' then '598' when (not upper(modelname) ~ 'CEDMI' or not upper(modelname) ~ 'CEDNA' or not upper(modelname) ~ 'CEDRM') and ( upper(location) ~ '^[A-Z]' or location is null) then null else location end as LOCATION from ENEL_TLC_T_DEVICE;
location | modelname | location :------- | :-------- | :------- 12A | CEDMI | 74 12A | CEDRM | 598 12A | CEDNA | 768 12AA | CEDMI | 74 12AA | CEDRM | 598 12AA | CEDNA | 768 null | null | null
db<>fiddle here