To match a substring with the string, use the percentage sign wildcard (%expression). For example, SELECT ... WHERE Description SIMILAR TO 'car' matches only car, not sportscar. However, SELECT ... WHERE Description SIMILAR TO '%car' matches car, sportscar, and any string that ends with car.
When matching against only a sub-character class, you must include the outer square brackets, and the square brackets for
the sub-character class. For example, expression SIMILAR TO '[[:digit:]]'). For more on sub-character class matching, see Regular expressions: Special sub-character classes.
Comparisons are performed character-by-character, unlike the equivalence (=) operator and other operators where the comparison
is done string-by-string. For example, when a comparison is done in a UCA collation (CHAR or NCHAR with the collation set
to UCA), 'Æ'='AE' is true, but 'Æ' SIMILAR TO 'AE' is false.
For a character-by-character comparison to match, each single character in the expression being searched must match a single
character or a wildcard in the SIMILAR TO pattern.
SIMILAR TO use the collation to determine character equivalence and evaluate character class ranges. For example, if the database
is case- and accent-insensitive, matches are case- and accent-insensitive. Ranges are also evaluated using the collation sort
SIMILAR TO search conditions can be used to compare CHAR and NCHAR strings. In this case, character set conversion is performed
so that the comparison is done using a common data type. Then, a character-by-character comparison is performed. See Comparisons between CHAR and NCHAR.
You can specify expression or pattern as an NCHAR string literal by prefixing the quoted value with N (for example, expression SIMILAR TO N'pattern'). You can also use the CAST function to cast the pattern to CHAR or NCHAR (for example, expression SIMILAR TO CAST(pattern AS datatype).