Language reference
SPLIT_PART (one field of a delimited string)
Syntax
SPLIT_PART(<string>, <delimiter>, <n>)
π SPLIT_PART(email, '@', 2) → domain (Users)Description
SPLIT_PART cuts a string at each occurrence of a delimiter and gives one of the pieces: SPLIT_PART('a,b,c', ',', 2) is 'b'. A negative number counts from the end, so -1 is the last piece.
Technical Description
SPLIT_PART(s: STRING, delimiter: STRING, n: NUMBER) → STRING. Fields count from 1, or from the end for a negative n; past the last field the result is ''. An empty delimiter leaves one field, the whole string. A field number of zero is an error. The delimiter is matched exactly, case included. A NULL in any argument returns NULL. PURE, DETERMINISTIC.
Examples
The domain of an address:
π SPLIT_PART(email, '@', 2) → domain (Users)PROJECT SPLIT_PART(email, '@', 2) -> domain (Users)The last segment of a path:
π SPLIT_PART(code, '/', -1) → leaf (Products)PROJECT SPLIT_PART(code, '/', -1) -> leaf (Products)Pushdown
SQL: SPLIT_PART(…) on PostgreSQL, which counts from the end for a negative n from version 14; in the engine elsewhere.