ANTLR v4

This is a fork of ANTLR, available at https://github.com/antlr/antlr4. These modifications to ANTLR are available under the same license as ANTLR, or the BSD 3-clause license in LICENSE.txt.

This fork adds an option -scala-extractor-objects, which adds extractor object for each production of each rule. This means that you can apply pattern matching to context objects. Take for example this grammar:

expr : expr '+' expr
     | number
     ;

Now we can match and destructure on the productions of the rule expr as such:

def func(tree: ExprContext): Unit = tree match {
  case Expr0(left, "+", right) => ???
  case Expr1(num) => ???
}

There are two caveats to this addition to ANTLR. It forces all rule productions to have alternative names (keeping pre-existing ones), so our grammar above is rewritten to something equivalent to:

expr : expr '+' expr # expr1
     | number # expr2
     ;

Furthermore, this approach builds on the target for Java by adding the extractors to that target. As we can't alter the generated context classes after the fact, the unapply methods are defined in a similarly named class, but not the same class. The file with the extractor objects is called GrammarNameParserPatterns.scala. The extractors are named the same as the alternatives, so as with the example above the context <rule.upper>Context can be destructured with <rule.upper>0 and <rule.upper>1.

Finally it is important to note that currently we can't destructure arbitrary eBNF patterns in rule productions, but you will be warned of this fact.

Compiling

How to use this fork, for example, to compile VerCors' Java grammar:

1. Run mvn install -DskipTests=true -f pom.xml. This will compile the fork into a jar.
2. Use the jar to compile the grammar:

[user@pc ~/antlr4/tool/target]
$ java -jar antlr4-4.8-2-SNAPSHOT-complete.jar -scala-extractor-objects -lib ~/vercors/parsers/lib/antlr4/  ~/vercors/parsers/src/main/antlr4/JavaParser.g4