There have been a hand full of releases since I have posted an announcement
here. Since then, I have implemented a simple ERB-based template output mode,
in lieu of the StringTemplate-based template mode provided by ANTLR's Java and
Python targets. I have also fixed several bugs, made a number of code tweaks,
and authored a significant amount of usage and API documentation. Scroll on
down for a catalog of significant changes since version 1.3.
Gem installation: gem install antlr3
Homepage: http://antlr.ohboyohboyohboy.org
Rubygems.org: http://rubygems.org/gems/antlr3
GitHub: http://github.com/ohboyohboyohboy/antlr3
== About
Fully-featured ANTLR 3 parser generation for Ruby.
ANTLR (ANother Tool for Language Recognition) is a tool that is used to
generate code for performing a variety of language recognition tasks: lexing,
parsing, abstract syntax tree construction and manipulation, tree structure
recognition, and input translation. The tool operates similarly to other
parser generators, taking in a grammar specification written in the special
ANTLR metalanguage and producing source code that implements the recognition
functionality.
While the tool itself is implemented in Java, it has an extensible design that
allows for code generation in other programming languages. To implement an
ANTLR language target, a developer may supply a set of templates written in
the StringTemplate (http://www.stringtemplate.org) language.
ANTLR is currently distributed with a fairly limited Ruby target
implementation. While it does provide implementation for basic lexer and
parser classes, the target does not provide any implementation for abstract
syntax tree construction, tree parser class generation, input translation, or
a number of the other ANTLR features that give the program an edge over
traditional code generators.
This gem packages together a complete implementation of the majority of
features ANTLR provides for other language targets, such as Java and Python.
It contains:
* A customized version of the latest ANTLR program, bundling all necessary
java code and templates for producing fully featured language recognition
in ruby code
* a ruby run-time library that collects classes used throughout the code that
ANTLR generates
* a wrapper script, `antlr4ruby', which executes the ANTLR command line tool
after ensuring the ruby-tailored ANTLR jar is in java's class path
== Features
* powerful LALR lexer, parser, and tree-parser class generation
* rapid development with ANTLR features like LL(*) prediction,
backtracking w/ memoization, and predicates
* ANTLR run-time library code that aims to be more intuitive and stylized for
the conventions ruby developers are accustomed to
* integrated Abstract Syntax Tree (AST) construction, rewriting, and parsing
* an ERB-based template output mode to simplify source translation and
extraction
* built-in test driver code to quickly try out grammar output
* debug and profile modes to help develop and optimize grammars
* extra utilities such as DOT-based graph generation and tree building
wizards to help with the development process
== History
=== 1.7.2 / 05-16-10
* Minor Enhancements
- added new #hold convenience method to ANTLR3::CommonTokenStream
* Bug Fixes
- corrected a typo in ANTLR3::Error#FailedPredicate which caused a
NoMethodError to raise instead of FailedPredicate when semantic
predicates failed
* Documentation
- added more content to the ruby antlr3 guide
- set up an official project website at
http://antlr.ohboyohboyohboy.org
=== 1.7.0 / 04-20-10
* Minor Enhancements
- added new #each_on_channel and #walk convenience methods to
ANTLR3::CommonTokenStream
* Bug Fixes
- discovered and fixed a rare but serious bug in the way the output code
evaluates
syntactic predicates
* Documentation
- began developing a more thorough usage guide for the package
- the guide is currently available on GitHub at
http://ohboyohboyohboy.github.com/antlr3
=== 1.6.3 / 02-12-10
* 4 Minor Enhancements
- added --help and --version options to antlr4ruby
- documented the antlr4ruby source and added license information
- prettied up the source code in all of the library files a little
- added a hoe-based rakefile with the gem distribution of the source code
=== 1.6.0 / 01-27-10
* 4 Minor Enhancements to run-time library
- CommonTokenStream and subclasses can now also accept another
CommonTokenStream
object, serving sort of as a copy constructor. This provides a reasonable
way
to cast a CommonTokenStream into a TokenRewriteStream and vice-versa
- CommonTreeNodeStream can take an options hash at the end of any of its
various
signatures to provide values for attributes like token_stream
- If a parser is created with a CharacterStream object (i.e., StringStream),
it
will also try to cast it by checking for an associated lexer (like it
currently
does when created with String or IO objects).
- Template-output mode parsers have a new class method "load_templates( path
)"
which will attempt to load the template group file at +path+ and, if
successful,
set it to the parser's default template library. (See the CMinus example
set for
an example of how this can be used)
* 2 bug fixes / 1 change to templates
- backtracking was broken in situations where recognition failed when
a semantic predicate was encountered as the template code did not
throw BacktrackingFailed when it was in backtracking mode instead of
PredicateFailed
- rule memoization also had a major bug in that rules that were memoized
after failure did not throw BacktrackingFailed when it was needed
- I changed `attribute scope' structures (not rule return scopes, but
the groups of attributes you can assign to rules with the `scope'
keyword),
from being implemented as named constants in the class body to class
variables belonging to the recognizer. As class variables do not need
to follow naming conventions, this provides more freedom in naming without
all of the ugly name mangling. These are generally used privately by
parsers,
so it shouldn't be too drastic of a change as to break end user's code.
=== 1.5.0 / 01-25-10
* 2 minor enhancements
- Added BaseTree#walk, which iterates in a top-down fashion across all notes
in a tree.
- Added BaseTree#prune, which stops decent into the current node during
BaseTree#walk.
* 1 optimization
- changed `closure loops' ( uses of the `*' or `+' modifiers ) from using
Kernel#loop to `while true ... end' style loops, which should be slightly
faster
* new test code
- test/functional/template-output/template-output.rb serves as a basic
functional test implementation for the new template output mode
=== 1.4.0 / 01-17-10
* 1 major new feature
- finished basic implementation of a template output mode.
- lib/antlr3/template.rb defines a number of classes for
ERB-based template output translation
- templates/ST.stg defines the Ruby target templates to
generate template output mode recognizers
* new test code
- added template unit tests at test/unit/test-template.rb
- test/functional/template-output/template-output.rb serves as a basic
functional test implementation for the new template output mode
Best Regards,
Kyle Yetter