Link Search Menu Expand Document


Returns the first match of <pattern> within the <expression>. If the pattern does not match, returns NULL. If you want to extract all matches, use REGEXP_EXTRACT_ALL.


REGEXP_EXTRACT(<expression>, <pattern>[,'<flag>[...]',[<index>]])


Parameter Description Supported input types
<expression> The string from which to extract a substring from, based on a regular expression. TEXT
<pattern A re2 regular expression for matching with the string. TEXT
<flag> Optional. Flag that allow additional controls over the regular’s expression matching behavior. If using multiple flags, you can include them in the same single-quote block without any separator character. Firebolt supports the following RE2 flags to override default matching behavior. With - in front you can disable the flag.
* i - Specifies case-insensitive matching.
* m - Specifies multi-line mode. In this mode, ^ and $ characters in the regex match the beginning and end of line.
* s - (Enabled per default) Specifies that the . metacharacter in regex matches the newline character in addition to any character in .
* U - Specifies non-greedy mode. In this mode, the meaning of the metacharacters * and + in regex <pattern> are swapped with *? and +?, respectively. See the examples using flags below for the difference in how results are returned.
<index> Optional. Indicates which subgroup of the expression match should be returned. Default value is 0 which means the whole match is returned, independent of any number of given subgroups. An INTEGER between 0 and N where N is the number subgroups in the <pattern>.

Return Types



	REGEXP_EXTRACT('Hello Year 2023 DEF', '[A-Za-z]+');

Returns: ["Hello", "Year"]

Despite using subgroups in the regular expression, the each full match will be returned as the optional <index> argument is not set (the default value 0 is used instead).

	REGEXP_EXTRACT('Learning about #REGEX in #Firebolt 2023', '#([A-Z])[a-z]+');

Returns: ["Learning", "Firebolt"]

The regular expression contains two subgroups which allows us to set the <index> argument to something between 0 and 2. Every other value will cause an exception to be thrown. Setting <index> to 0 would cause the all full matches ["Learning", "Firebolt"] to be returned (same behavior as not setting this value, see the example above), while a 2 would return the second subgroup of each match ["earning", "irebolt"].

	REGEXP_EXTRACT('Learning about #REGEX in #Firebolt 2023', '#([A-Z])[a-z]+', '', '1');

Returns: ["L", "F"]