Postgres regular expression regexp_split_to_array

You look like you're almost there -- you just have to append the '20' to the extracted '12'/'11':

UPDATED (PostgreSQL has no boundary \b (?!)):

SELECT '20'||substring(my_field from E'(?:^|\\W)S(\\d\\d)(?!\\w)') ....

-- NOTE: Old version. However there's no '\b' in this regex flavour.
SELECT '20'||substring(my_field from E'\\bS(\\d\\d)\\b') ....

You have to double-escape backslashes. Also, the use of the brackets in the regex means that the selected substring will be '12' or '11', not including the 'S'. The '20'||xxxx just concatenates the two strings.

The (?:^|\\W) makes sure the Sxx is either at the start of the string, or preceded by a non-word character (like space, or comma). This prevents the S13 in FUT,GRAS13,S12 from being matched.

The (?!\\w) makes sure the Sxx isn't followed by another word character (ie is not part of a larger word). This prevents S13 being matched in S132,GRA,S12.

(Note - I previously had \b which means "word boundary" but PostgreSQL regex doesn't support that, so the (?:^|\W) and (?!\w) are ways to simulate that.

AFAIK, there's no word boundary in PostgreSQL's regex, \b means backspace.

I would do:

SELECT '20'||substring('FUT,GRA,S12,1055559' from E'\\WS(\\d{2})\\W')

where \W means: not a word character.