雪花中的 SUBSTRING_INDEX()

Question

What is the exact duplicate function of MySQL SUBSTRING_INDEX() in Snowflake??

I found SPLIT_PART() in Snowflake but this is not the exact same of SUBSTRING_INDEX() .

Eg SUBSTRING_INDEX("www.abc.com", ".", 2); returns www.abc

all the left side substring after 2nd delimiter '.'

but

SPLIT_PART("www.abc.com", ".", 2); return abc

it splits 1st then only returns the split part of a string.

How can I use SUBSTRING_INDEX() in the same way as MySQL in Snowflake

Answer 1

Similar effect could be achieved using ARRAY operations:

SELECT s.c, ARRAY_TO_STRING(ARRAY_SLICE(STRTOK_TO_ARRAY(s.c, '.'), 0, 2), '.')
FROM (VALUES ('www.abc.com')) AS s(c);

---

How does it works?

• STRTOK_TO_ARRAY - make an array from string
• ARRAY_SLICE - take the parts from 0 to n
• ARRAY_TO_STRING - convert array back to string using '.' as delimeter

In steps:

SELECT 
  s.c,
  STRTOK_TO_ARRAY(s.c, '.')   AS arr,
  ARRAY_SLICE(arr, 0, 2)      AS slice,
  ARRAY_TO_STRING(slice, '.') AS result
FROM (VALUES ('www.abc.com')) AS s(c);

Answer 2

You may use REGEXP_SUBSTR here:

SELECT REGEXP_SUBSTR('www.abc.com', '^[^.]+\.[^.]+');

Here is a demo showing that the regex pattern works as expected.

Answer 3

The substring_index function in MySQL returns the entire string if the substring isn't found or if the supplied occurrence is greater than the maximum occurrence. Assuming you want to preserve that behavior and that you'd also find it helpful to be able to extract non-contiguous parts of string, consider this approach.

with cte as (select 'www.abc.com' as txt)


select a.txt, listagg(b.value,'.') within group (order by b.index)
from cte a, lateral split_to_table(a.txt, '.') b
where b.index <=2 --you can also do for e.g. b.index in (1,3) to get 'www.com'
group by a.txt;