Add basic support for OpenNLP.

lib/treat/config/data/core.rb

@@ -4,7 +4,7 @@
     'abw', 'doc', 'yaml', 'uea',
     'lda', 'pdf', 'ptb', 'dot',
     'ai', 'id3', 'svo', 'mlp',
-    'svm', 'srx'],
+    'svm', 'srx', 'nlp'],
 
   encodings: 
     {language_to_code: {

lib/treat/config/data/languages/english.rb

@@ -33,7 +33,7 @@
     processors: {
       parsers: [:stanford],
       segmenters: [:scalpel, :srx, :tactful, :punkt, :stanford],
-      tokenizers: [:ptb, :stanford, :punkt]
+      tokenizers: [:ptb, :stanford, :punkt, :open_nlp]
     }
   },
   stop_words:

lib/treat/config/data/libraries.rb

@@ -8,5 +8,9 @@
   stanford: {
     jar_path: nil, 
     model_path: nil
+  },
+  open_nlp: {
+    jar_path: nil,
+    model_path: nil
   }
 }

lib/treat/helpers/string.rb

@@ -54,7 +54,7 @@ def camel_case
       if @@cc_cache[o_phrase]
         return @@cc_cache[o_phrase] 
       end
-      if Treat.core.acronyms.include?(phrase)
+      if Treat.core.acronyms.include?(phrase.downcase)
         phrase = phrase.upcase
       else
         phrase.gsub!(Regex) { |a| a.upcase }

lib/treat/loaders/bind_it.rb

@@ -0,0 +1,48 @@
+class Treat::Loaders::BindIt
+
+  # Keep track of whether its loaded or not.
+  @@loaded = {}
+
+  # Load CoreNLP package for a given language.
+  def self.load(klass, name, language = nil)
+
+    return if @@loaded[klass]
+
+    language ||= Treat.core.language.default
+
+    jar_path   = Treat.libraries[name].jar_path || 
+                 Treat.paths.bin + "#{name}/"
+    model_path = Treat.libraries[name].model_path || 
+                 Treat.paths.models + "#{name}/"
+
+    if !File.directory?(jar_path)
+      raise Treat::Exception, "Looking for #{klass} " +
+      "library JAR files in #{jar_path}, but it is " +
+      "not a directory. Please set the config option " +
+      "Treat.libraries.#{name}.jar_path to a folder " +
+      "containing the appropriate JAR files."
+    end
+
+    if !File.directory?(model_path)
+      raise Treat::Exception, "Looking for #{klass} " +
+      "library model files in #{model_path}, but it " +
+      "is not a directory. Please set the config option " +
+      "Treat.libraries.#{name}.model_path to a folder " +
+      "containing the appropriate JAR files."
+    end
+
+    klass.jar_path = jar_path
+    klass.model_path = model_path
+    klass.use language
+
+    if Treat.core.verbosity.silence
+      klass.log_file = '/dev/null' 
+    end
+
+    klass.bind
+
+    @@loaded[klass] = true
+
+  end
+
+end

lib/treat/loaders/open_nlp.rb

@@ -0,0 +1,12 @@
+require 'treat/loaders/bind_it'
+
+# A helper class to load the OpenNLP package.
+class Treat::Loaders::OpenNLP < Treat::Loaders::BindIt
+
+  require 'open-nlp'
+
+  def self.load(language = nil)
+    super(OpenNLP, :open_nlp, language)
+  end
+
+end

lib/treat/loaders/stanford.rb

@@ -1,53 +1,14 @@
+require 'treat/loaders/bind_it'
+
 # A helper class to load the CoreNLP package.
-class Treat::Loaders::Stanford
+class Treat::Loaders::Stanford < Treat::Loaders::BindIt
 
-  # Keep track of whether its loaded or not.
-  @@loaded = false
+  require 'stanford-core-nlp'
 
-  # Load CoreNLP package for a given language.
   def self.load(language = nil)
-
-    return if @@loaded
-
-    language ||= Treat.core.language.default
-
-    jar_path   = Treat.libraries.stanford.jar_path || 
-                 Treat.paths.bin + 'stanford/'
-    model_path = Treat.libraries.stanford.model_path || 
-                 Treat.paths.models + 'stanford/'
-
-    if !File.directory?(jar_path)
-      raise Treat::Exception, "Looking for Stanford " +
-      "CoreNLP JAR files in #{jar_path}, but it is " +
-      "not a directory. Please set the config option " +
-      "Treat.libraries.stanford.jar_path to a folder " +
-      "containing the Stanford JAR files."
-    end
-
-    if !File.directory?(model_path)
-      raise Treat::Exception, "Looking for Stanford " +
-      "CoreNLP model files in #{model_path}, but it " +
-      "is not a directory. Please set the config option " +
-      "Treat.libraries.stanford.model_path to a folder " +
-      "containing the Stanford JAR files."
-    end
-
-    require 'stanford-core-nlp'
-
-    StanfordCoreNLP.jar_path = jar_path
-    StanfordCoreNLP.model_path = model_path
-    StanfordCoreNLP.use(language)
-
-    if Treat.core.verbosity.silence
-      StanfordCoreNLP.log_file = '/dev/null' 
-    end
-
-    StanfordCoreNLP.bind
-
-    @@loaded = true
-
+    super(StanfordCoreNLP, :stanford, language)
   end
-  
+
   def self.find_model(name, language)
     language = language.intern
     model_file = StanfordCoreNLP::Config::Models[name][language]
@@ -57,4 +18,4 @@ def self.find_model(name, language)
     File.join(model_path, model_dir, model_file)
   end
 
-end
+end

lib/treat/proxies/proxy.rb

@@ -11,14 +11,15 @@ module Proxy
     def method_missing(sym, *args, &block)
       if [:do, :apply].include?(sym) || 
         Treat::Workers.lookup(sym)
-          to_entity.send(sym, *args)
+        to_entity.send(sym, *args)
       else
         super(sym, *args, &block)
       end
     end
+
     # Create an unknown type of entity by default.
     def to_entity(builder = nil)
-      Treat::Entities::Unknown(self.to_s)
+      Treat::Entities::Unknown.new(self.to_s)
     end
   end
 

lib/treat/workers/groupable.rb

@@ -15,7 +15,7 @@ def const_missing(const)
       require file
       if not self.const_defined?(const)
         raise Treat::Exception,
-        "File #{file} does not define " +
+        "File #{file}.rb does not define " +
         "#{self}::#{const}."
       end
       const_get(const)

lib/treat/workers/processors/tokenizers/maxent.rb → lib/treat/workers/processors/tokenizers/open_nlp.rb

@@ -1,8 +1,10 @@
 # Maximum entropy tokenization supplied by OpenNLP.
-class Treat::Workers::Processors::Tokenizers::Maxent
+class Treat::Workers::Processors::Tokenizers::OpenNlp
 
   require 'open-nlp'
-  OpenNLP.load
+  Treat::Loaders::OpenNLP.load
+
+  @@tokenizers = {}
 
   # Maximum entropy tokenization.
   def self.tokenize(entity, options = {})
@@ -20,8 +22,7 @@ def self.tokenize(entity, options = {})
     tokens = tokenizer.tokenize(str).to_a
 
     tokens.each do |token|
-      entity << Treat::Entities
-      ::Token.from_string(chunk)
+      entity << Treat::Entities::Token.from_string(token)
     end
 
   end

spec/helper.rb

@@ -7,17 +7,21 @@ module Treat::Specs
   require 'rspec'
 
   # Some configuration options for devel.
-=begin
+
   Treat.databases.mongo.db = 'treat_test'
   Treat.libraries.stanford.model_path =
   '/ruby/stanford-core-nlp-minimal/models/'
   Treat.libraries.stanford.jar_path =
   '/ruby/stanford-core-nlp-minimal/bin/'
+  Treat.libraries.open_nlp.jar_path = 
+  '/ruby/open-nlp-english/bin/'
+  Treat.libraries.open_nlp.model_path =
+  '/ruby/open-nlp-english/models/'
   Treat.libraries.punkt.model_path =
   '/ruby/punkt/models/'
   Treat.libraries.reuters.model_path =
   '/ruby/reuters/models/'
-=end
+
   # Mimic the ./lib structure.
   module Entities; end
   module Workers; end