Efficient String Matching and Easy Bottom-Up Parsing

The paper consists of the two parts. In the first one we compare three string matching algorithms: Dömölki algorithm, also known as SHIFT OR algorithm, O(n×m), Table driven O((n +m) × lgm) and Aho Corasick O(m + n). Table driven algorithm pass trough the same states as Dömölki one but have compact states encoding. The table driven algorithm can be also considered as Aho Corasick algorithm with eliminated ǫ-transitions. We advocate that the table driven algorithm is the best solution for matching multiple patterns of reasonable size. The second part of the paper deal with bottom-up syntax analysis. We have shown that the backward deterministic syntax analysis can be implemented via extension of a string matching automaton by a stack; or two stacks, if we want to go beyond context free grammars. The implementation of a parser this type is as easy as writing a recursive descent parser; we need to supply only the transition table, which can be easily derived from the grammar. Finally, we discuss some compiler engineering details.