regular expression in hive for a specific string - regex

Question

I have a column in hive table which is a address column and i want to split that into 2.
There are 2 scenarios to take care of.
Example:
Scenario 1:
Input column value:
ABC DEF123 AD
Output column values:
Column 1 should have ABC DEF
Column 2 should have 123 AD
Another example can be like below.
MICHAEL POSTON875 HYDERABAD
In this case separation should be based on a number which is part of a string value, if a string is having number in it then both should separate
Scenario 2:
Input value: ABC DEFPO BOX 5232
Output:
Column 1:- ABC DEF
Column 2:- PO BOX 5232
Another example can be like below.
Hyderabad jhillsPO BOX 522002
In this case separation should be based on PO BOX
Both the data is in same column and i would like to update the data into target based on the string format..like a case statement not sure about the approach.
NOTE:- The string length can be varied as this is address column.
Can some one please help me to provide a hive query and pyspark for the same?

Answer 1

Using CASE expression you can check which template does it match and using regexp_replace insert some delimiter, then split by the same delimiter.
Demo (Hive):
with mytable as (
select stack(4,
'ABC DEF123 AD',
'MICHAEL POSTON875 HYDERABAD',
'ABC DEFPO BOX 5232',
'Hyderabad jhillsPO BOX 522002'
) as str
) --Use your table instead of this
select columns[0] as col1, columns[1] as col2
from
(
select split(case when (str rlike 'PO BOX') then regexp_replace(str, 'PO BOX','|||PO BOX')
when (str rlike '[a-zA-Z ]+\\d+') then regexp_replace(str,'([a-zA-Z ]+)(\\d+.*)', '$1|||$2')
--add more cases and ELSE part
end,'\\|{3}') columns
from mytable
)s
Result:
col1 col2
ABC DEF 123 AD
MICHAEL POSTON 875 HYDERABAD
ABC DEF PO BOX 5232
Hyderabad jhills PO BOX 522002