An Alternative to the Use of Patterns in String Processing

Abstract

~v SNOBOL4 is best known for its string processing facilities, which are based on patterns as da~: objects. Despite the demonstrated success of patterns, there are many shortcomings associated with, their use. The concept of patterns in SNOBOL4 is examined and problem areas are discussed. A~. alternative method for high-level string processing is described. This method, implemented in thg programming language Icon, employs generators, which are capable of producing alternative value~ Generators, coupled with a goal-driven method of expression evaluation, provide the string processingfacilities of SNOBOL4 without the disadvantages associated with patterns. Comparisons betwee~ SNOBOL4 and Icon are included and the broader implications of the new approach are discussed. ;~

The Evaluation of Expressions in Icon

“…SNOBOIA is closer to Icon, in that it has expressions that may produce no result. Within the framework of its pattern-matching language [5], it also has expressions that may produce more than one result, although these results are not accessible in the context of ordinary expression evaluation.…”

“…(The term "nondeterministic" is sometimes used in this sense [3].) This mode of evaluation is a general feature of Icon and is not limited to a specific part of the language, as it is to pattern matching in SNOBOIA [5].…”

The implementation of generators and goal‐directed evaluation in icon

“…The result sequence for this control structure consists of the results for exprl followed by the results for expr2. For example, (1 to 5) I (7 to 10) has the result sequence 11, 2, 3,4, 5, 7, 8,9, 10).…”

Generators in Icon

“…A complete discussion of the issues involved in this effort is given in [12]. Briefly stated, the major issue is that, in most languages, string manipulation facilities constitute a separate sublanguage, and interaction between the sublanguage and the language proper is awkward at best.…”

“…This limited "scope" of generators avoids the inefficiency of uncontrolled backtracking and gives the programmer more explicit control in the search for alternatives. This latter capability is especially important in light of the observation that, in many cases, backtracking is simply not needed [12,25]. This paper describes the use of generators in Icon, concentrating on their use in applications other than string processing, which is treated in [10].…”