Beginnings of a JVM native ctags library?

Declarations.rb

require 'yaml'

JAVA_YAML=<<YAML
- regex:    "class\\s+(\\w+)"
  capture:  1
  type:     id
- regex:    "interface\\s+(\\w+)"
  capture:  1
  type:     id
- regex:    "(public|private).*\\s+(\\w+)\s*\\("
  capture:  2
  type:     id
YAML

RUBY_YAML=<<YAML
- regex:    "^[^#]*(class|module)\\s+(\\w+)"
  capture:  2
  type:     id
- regex:    "^[^#]*def (self\\.)(\\w+)"
  capture:  2
  type:     id
- regex:    "^[^#]*attr(_reader|_accessor|_writer)(.*)$"
  capture:  2
  type:     id-list
- regex:    "^[^#]*alias\s+:(\\w+)"
  capture:  1
  type:     id
- regex:    "^[^#]*alias_method\s+:(\\w+)"
  capture:  1
  type:     id
YAML

class Declarations
  DEFINITIONS = {
    /\.rb$/   => YAML.load(RUBY_YAML),
    /\.java$/ => YAML.load(JAVA_YAML)
  }
  
  def decls_for_file(path)
    DEFINITIONS.each do |fn_re, decls|
      if path =~ fn_re
        return decls
      end
    end
    nil
  end
  
  def match_in_file(path)
    tags = []
    begin
      file = File.read(path)
      if decls = decls_for_file(path)
        decls.each do |decl| 
          file.each_line do |line| 
            if md = line.match(Regexp.new(decl["regex"]))
              capture = md[decl["capture"]]
              case decl["type"]
              when "id"
                tags << [capture, path, md[0]]
              when "id-list"
                tags += capture.scan(/\w+/).map {|id| [id, path, md[0]] }
              end
            end
          end
        end
      end
      tags
    rescue
      []
    end
  end
  
  def generate(files)
    tags = []
    files.each do |path|
      tags += match_in_file(path)
    end
    tags.sort!
    File.open("tags", "w") do |tags_file|
      tags.each do |id, path, declaration|
        tags_file.puts "#{id}\t#{path}\t#{declaration}"
      end
    end
  end
end

files = Dir[ARGV[0]]
puts "#{files.length} files"

5.times do 
  s = Time.now
  decl = Declarations.new
  decl.generate(files)
  puts "took #{Time.now - s}s"
end

#
# Timings for running against all Ruby and Java in the Redcar repository:
#
# ~/Redcar/redcar (master) 
# $ ruby ~/Desktop/declarations.rb "**/*.{rb,java}"
# 917 files
# took 4.466576s
# took 4.694825s
# took 4.458402s
# took 4.87733s
# took 4.939915s
# 
# ^C~/Redcar/redcar (master) 
# $ jruby ~/Desktop/declarations.rb "**/*.{rb,java}"
# 917 files
# took 1.802s
# took 1.13s
# took 0.986s
# took 0.88s
# took 0.879s
#
#
# Timings for running against the entire Songkick.com Rails project:
#
#~/Redcar/redcar (master) 
# $ ruby ~/Desktop/declarations.rb "/Users/danlucraft/Songkick/skweb/**/*.rb"
# 5330 files
# took 44.585956s
# took 31.532368s
# took 31.402265s
# took 31.702252s
# took 31.957006s
# ~/Redcar/redcar (master) 
# $ jruby ~/Desktop/declarations.rb "/Users/danlucraft/Songkick/skweb/**/*.rb"
# 5330 files
# took 7.06s
# took 5.801s
# took 5.183s
# took 5.217s
# took 5.208s

isn't this more of a native ruby ctags library? :)

The language is kind of irrelevant as long as it runs on the jvm :) More important is whether this approach will work.