@@ -31,3 +31,5 @@ ext/word_tagger/word_tagger.so lib/word_tagger.so ext/word_tagger/word_tagger.bundle lib/word_tagger.bundle +pkg/ +doc/ .gitignore @@ -2,7 +2,11 @@ $:.unshift(File.dirname(__FILE__)) unless $:.include?(File.dirname(__FILE__)) || $:.include?(File.expand_path(File.dirname(__FILE__))) module RbTagger - + class << self + def tags_from_file( file ) + File.read(file).split("\n").map{|t| t.strip} + end + end end require 'word/tagger' lib/rbtagger.rb @@ -1,7 +1,13 @@ module Word require 'word_tagger' class Tagger < Tagger::WordTagger + def initialize( tags_file, options = {} ) + load_tags( RbTagger.tags_from_file( tags_file ) ) + set_words( options[:words] || 2 ) + end + def execute( text ) + # strip non alpha characters super( text.gsub(/[^\w]/,' ') ) end end lib/word/tagger.rb @@ -4,9 +4,7 @@ class TestWordTagger < Test::Unit::TestCase def setup if !defined?($wtagger) - $wtagger = Word::Tagger.new - $wtagger.load_tags( File.read(File.join(File.dirname(__FILE__),'fixtures','tags.txt') ).split("\n").map{|t| t.strip} ) - $wtagger.set_words( 4 ) + $wtagger = Word::Tagger.new( File.join(File.dirname(__FILE__),'fixtures','tags.txt'), :words => 4 ) end end test/test_word_tagger.rb e8cabf6eed2925cbf7288d4cd52da3a763a46469 Todd Fisher todd.fisher@gmail.com http://github.com/taf2/rb-brill-tagger/commit/32be09d11dbe475e9dc31ed7c88060fa00f30b48 32be09d11dbe475e9dc31ed7c88060fa00f30b48 2008-05-14T09:43:50-07:00 2008-05-14T09:43:50-07:00 update tests e0c7c4783f816f708ff505d2388bd4f9b822dea9 Todd Fisher todd.fisher@gmail.com