A generic parser for strings and trees

Abstract

In this paper, we propose a two fold generic parser. First, it simulates the behavior of multiple parsing automata. Second, it parses strings drawn from either a context free grammar, a regular tree grammar, or from both. The proposed parser is based on an approach that defines an extended version of an automaton, called positionparsing automaton (PPA) using concepts from LR and regular tree automata, combined with a newly introduced concept, called state instantiation and transition cloning. It is constructed as a direct mapping from a grammar, represented in an expanded list format. However, PPA is a non-deterministic automaton with a generic bottom-up parsing behavior. Hence, it is efficiently transformed into a reduced one (RBA). The proposed parser is then constructed to simulate the run of the RBA automaton on input strings derived from a respective grammar. Without loss of generality, the proposed parser is used within the framework of pattern matching and code generation. Comparisons with similar and well-known approaches, such as LR and RI, have shown that our parsing algorithm is conceptually simpler and requires less space and states.

A Generic Tool for Teaching Compilers

“…For our further discussions, we assume the following definitions based on the ones given by Aho et al (2007) and by Jabri (2012). These definitions are used to specify the subsequent phases of a compiler as follows:…”

“…In this section, we propose a construction approach for the reduced automaton RAGA based on extension of the subset construction algorithms given in (Aho et al, 2007;Jabri, 2012). Such extension is a twofold.…”

“…The constructed automaton in (Jabri, 2012) has multiple transitions and respective reductions. In contrast, according to the proposed approach, the sets of follow symbols for the nonterminals are used to determine their respective reduction in a deterministic way, similar to the one adopted for SLR parsers (Aho et al, 2007 and to the translation actions (as indicated by the translation schemes respective to SPA and RPA) to be performed by RAGA (G) during its run on an input alphabet, generated from the grammar G. Thus, Algorithm 2 computes Q RGA using an ε -closure function (Aho et al, 2007).This function closes the initial states and the final states respective to the different grammar symbol types.…”

“…Our parsing approach is based on position parsing automata (PPA), as proposed in (Jabri, 2009(Jabri, , 2012. However, it extends PPA to act as a scanner/parser/translator and uses different construction approach that handles recursion by reduced stack activities in a way similar to the generalized bottom up parsers and the reduction-incorporated parsers, as proposed in (Ayock, 2001;Johnstone & Scott, 2007;Scott & Johnstone, 2005) respectively.…”

“…However, it extends PPA to act as a scanner/parser/translator and uses different construction approach that handles recursion by reduced stack activities in a way similar to the generalized bottom up parsers and the reduction-incorporated parsers, as proposed in (Ayock, 2001;Johnstone & Scott, 2007;Scott & Johnstone, 2005) respectively. In addition, it handles nondeterminism in a way similar to SLR parsing as proposed in (Jabri, 2012).…”