rendaw/pidgoon

dragon : dragoon :: pigeon : pidgoon

Pidgoon

A Java GLR-like parser generator with tools to parse both byte streams and streams of arbitrary events (input events, tokens, tree nodes, etc). It has no lexer, parses all branches simultaneously, has no ambiguity, and can handle left recursion if you have enough memory. Actually, that it can have a lexer if you want. You can even have two lexers, and nest them.

Probably not particularly fast.

Include with maven with:

<dependency>
    <groupId>com.zarbosoft</groupId>
    <artifactId>pidgoon</artifactId>
    <version>0.0.1</version>
</dependency>

Use like:

Grammar grammar = new Grammar();
grammar.add(
    "root",
    new Operator(
        new Union().add(Grammar.stringSequence("hello"), Grammar.stringSequence("yes")),
        store -> store.pushStack("parsed")
    )
);
assertThat(new Parse().grammar(grammar).root("root").parse("hello"), equals("parsed"));

Documentation

There are two subpackages: events and bytes. events can parse streams of arbitrary data, but bytes has helper methods for dealing with byte streams including constructing nodes for character sequences and error reporting including stream position.

Grammar describes a grammar. If you don't go out of your way to modify it while parsing it is effectively immutable and thread safe.

Node is a element of the grammar.

Parse is a specific parse. For each file/stream you parse you need a new Parse object. Call parse to start the parse on your data. Parsing raises an error if the stream doesn't match the grammar. Multiple branches of your grammar can match simultaneously. parse returns whatever's on the top of the stack of the first branch that matches with exactly one value on the stack, otherwise null.

Sorry, at the moment there's no online Javadoc AFAIK.