String pattern matching is one of the oldest computational subdomains, used in diverse applications such as bioinformatics and natural language processing. Most mainstream languages provide only simple facilities for string processing culminating in regular expressions. Regular expressions are concise but lack power and exibility, and usually comprise a disjoint sublanguage within a more mainstream language.
Conicting design goals for string processing languages include: (1) concise expression of string patterns such as is provided by regular expressions, (2) expressive power and exibility sufcient to handle complex and dynamic pattern contexts that cannot be matched by regular and contextfree languages, and (3) peaceful co-existence and integration with ordinary (non-pattern) computational facilities. This paper describes the integration of regular expressions and SNOBOL patterns into the string scanning control structure in Unicon, a successor of the SNOBOL and Icon languages. Regular expressions compile into and mix with SNOBOL-style patterns. Unicon's matching operator is extended to execute a SNOBOL pattern match anchored to a scanning position; a new operator performs an unanchored match. Pattern matches can use string scanning expressions within patterns, supporting full bidirectional integration of the matching facilities.