SUMMARY Regular expressions are used in many applications to specify patterns because any regular expression can be compiled into a very efficient one-pass pattern matcher called a finite automaton. Finding matches is useful, but even more useful is parse extraction, which describes in detail how a pattern matches some input. After matching an address, for example, parse extraction makes it easy to find out the Zip code part of the address. We present an elegant, efficient algorithm for extracting a parse after matching with a finite automaton. In addition, we extend the regular expression language to include new operators for matching arbitrary left context and single character right context. The extended language can be matched as efficiently as the usual regular expression language, but is more expressive. Finally, we suggest how to apply the matching algorithms to match regular expressions containing arbitrary right context and single character left context. In effect, this allows one to specify patterns that seem to require an unlimited amount of look-ahead to match.

Extending regular expressions with context operators and parse extraction

The word paradigm, is used in computer science to talk about a family of notations, that share a common way for describing program implementations. Since each paradigm is well suited for solving only a range of problems, ideally a large system should be subdivided into components, each of which should be implemented in the most appropriate paradigm. Multiparadigm programming, allows the programmer to implement a system, in a number of different paradigms. The use of multiparadigm programming techniques, could lower implementation costs, and result in more reliable and efficient applications. The difficulties that arise with multiparadigm programming can be separated into the areas of application development in multiple paradigms, design and implementation of multiparadigm environment, and generators for creating such environments. We propose the use of object-oriented design techniques as a method for encapsulating programming paradigms within multiparadigm applications, and for abstracting common characteristics across paradigms. These techniques can be used to design practical multiparadigm environment generators, and implement multiparadigm programming environments supporting a wide variety of paradigms. In order to demonstrate the validity of our approach, we describe the design and implementation of three prototypes: one in each problem area. The integrator is a multiparadigm application dealing with the numeric and symbolic evaluation of integrals. The symbolic evaluation is based on the backtracking resolution mechanism offered by the logic programming paradigm, and the numeric evaluation, on the infinite streams implemented in the functional paradigm. Additionally, lexical analysis of the input expressions is described using regular expressions, and the expression grammar is described using a BNF syntax. Finally, expression simplification uses a term rewrite system, and graphing of functions is done by directly interacting with Unix tools. Integrator is implemented in the blueprint multiparadigm programming environment, a prototype implementation of the six paradigms based on our object-oriented approach. Many different implementation techniques have been applied in order to demonstrate the wide applicability of our approach. Some of the paradigms are implemented as compilers (using existing implementations where possible), and others are implemented as interpreters. Finally, the implementation of blueprint is based on the MPSS multiparadigm environment generator. This, allows the description of paradigms as object classes using paradigm description files. Additionally, it provides a multiparadigm link editor, and support for incorporating existing compilers into multiparadigm programming environments.

/pdf/programming-paradigms-as-object-classes-a-structuring-lfhlatm3vl.pdf

Programming Paradigms as Object Classes: A Structuring Mechanism for Multiparadigm Programming

Regular expressions are used in many applications to specify patterns because they can be compiled into very efficient one-pass pattern matchers; at the same time, they can specify a wide range of patterns of practical interest. The thesis extends regular expressions in 4 ways. First, we give a simple, efficient algorithm for matching an extended regular expression language. The language includes a number of new operators for expressing patterns, including operators that specify the context of a match, and operators that can require arbitrary lookahead. The new operators make it feasible to specify many more patterns than practical before, but the new matching algorithm retains the efficient of one-pass pattern matching. Second, we show how to extract a parse from a successful match, which provides significantly more information from a match than possible before. Third, we describe practical optimizations for efficiently implementing the matching and parse extraction algorithms. Fourth, we demonstrate how the new pattern matching capabilities can be integrated with an existing programming language, by describing a program generator for pattern matching applications. The program generator is called TLex. We describe the results of its use in several real-life applications.

Extending regular expressions

The InterLingua information processing system is proposed as a practical approach to providing analysis and access to large sets of multilanguage publications pertinent to a given industry and a particular world region. An existing natural language text analysis system (NLTAS) provides the base technology. The NLTAS facilitates construction of knowledge bases of the ideas that occur in verbatim text. The system is based on characterizing expressions of ideas as patterns of words. The NLTAS is language independent for all languages that have a finite set of characters. These knowledge bases can then be used to (i) partition large volumes of text into sets of text on narrower topics, (ii) generate intelligent indices to large sets of multilanguage documents without the high cost of verbatim translation, (iii) provide the basis for machine-aided verbatim translation services, and (iv) serve as the statistical basis for the tracking of trends in the given industry. >

InterLingua: a multilanguage business information center

On string pattern matching: a quantitative analysis and a proposal

This paper presents an extension of Pascal with string pattern matching. Pattern definitions are built using six basic operations: alternation, concatenation, immediate value assignment, intersection, difference and complement. The last three have not been previously implemented and they increase the expressive power beyond context-free languages. The pattern matching actions are augmented with three options: trace, prefix and suffix. Comparisons with a SNOBOL4 implementation are also presented. This experiment demonstrates that Pascal with pattern matching is a useful tool for string processing applications.

Ken-Chih Liu

Papers

A string pattern matching extension to Pascal and some comparisons with SNOBOL4