grammar_cop 0.1.0 → 0.1.1

data/.gitignore

@@ -1,4 +1,5 @@
 *.gem
 .bundle
 Gemfile.lock
+.DS_Store
 pkg/*

data/README

@@ -0,0 +1,22 @@
+grammar_cop is a standalone ruby extension for the link-grammar-4.7.4 C library. You don't have to install anything extra, so this is deployable on platforms like heroku, cloudfoundry. 
+Example:
+
+require 'grammar_cop'
+
+puts dir = File.expand_path('../data', File.dirname(__FILE__))
+GrammarCop::Dictionary.set_dictionary_path(dir)
+
+dict = GrammarCop::Dictionary.create("en")
+opts = GrammarCop::ParseOptions.create
+text = "Grammar is useless because there's nothing to say -- Gertrude Stein."
+sentence = GrammarCop::Sentence.create(text, dict)
+puts num = GrammarCop::Sentence.parse(sentence, opts)
+if num > 0
+  linkage = GrammarCop::Linkage.create(0, sentence, opts)
+  puts GrammarCop::Linkage.create_diagram(linkage)
+  puts "sublinkages=#{GrammarCop::Linkage.sublinkages_count(linkage)}"
+  puts "num_words=#{GrammarCop::Linkage.num_words_count(linkage)}"
+  puts "num_links=#{GrammarCop::Linkage.num_links_count(linkage)}"
+  #GrammarCop::Linkage.destroy(linkage)
+end
+GrammarCop::Sentence.destroy(sentence)

data/ext/link_grammar/link_grammar.c

@@ -17,6 +17,16 @@ void Init_link_grammar() {
 	rb_define_module_function(lg, "linkage_limit_get", linkage_limit_get, 1);
 	rb_define_module_function(lg, "short_length_set", short_length_set, 2);
 	rb_define_module_function(lg, "short_length_get", short_length_get, 1);
+	rb_define_module_function(lg, "disjunct_cost_set", disjunct_cost_set, 2);
+	rb_define_module_function(lg, "disjunct_cost_get", disjunct_cost_get, 1);
+	rb_define_module_function(lg, "min_null_count_set", min_null_count_set, 2);
+	rb_define_module_function(lg, "min_null_count_get", min_null_count_get, 1);
+	rb_define_module_function(lg, "max_null_count_set", max_null_count_set, 2);
+	rb_define_module_function(lg, "max_null_count_get", max_null_count_get, 1);	
+	rb_define_module_function(lg, "options_null_block_set", options_null_block_set, 2);
+	rb_define_module_function(lg, "options_null_block_get", options_null_block_get, 1);
+	rb_define_module_function(lg, "islands_ok_set", islands_ok_set, 2);
+	rb_define_module_function(lg, "islands_ok_get", islands_ok_get, 1);
 	rb_define_module_function(lg, "create_sentence", create_sentence, 2);
 	rb_define_module_function(lg, "parse_sentence", parse_sentence, 2);
 	rb_define_module_function(lg, "delete_sentence", delete_sentence, 1);
@@ -136,6 +146,71 @@ VALUE short_length_get(const VALUE self, VALUE opts) {
 	return INT2FIX(l);
 }
 
+VALUE disjunct_cost_set(const VALUE self, VALUE opts, volatile VALUE cost) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int c = NUM2INT(cost);
+	parse_options_set_disjunct_cost(ptr->opts, c);
+	return Qnil;
+}
+
+VALUE disjunct_cost_get(const VALUE self, VALUE opts) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int c = parse_options_get_disjunct_cost(ptr->opts);
+	return INT2FIX(c);
+}
+
+VALUE min_null_count_set(const VALUE self, VALUE opts, volatile VALUE count) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int c = NUM2INT(count);
+	parse_options_set_min_null_count(ptr->opts, c);
+	return Qnil;
+}
+
+VALUE min_null_count_get(const VALUE self, VALUE opts) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int c = parse_options_get_min_null_count(ptr->opts);
+	return INT2FIX(c);
+}
+
+VALUE max_null_count_set(const VALUE self, VALUE opts, volatile VALUE count) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int c = NUM2INT(count);
+	parse_options_set_max_null_count(ptr->opts, c);
+	return Qnil;
+}
+
+VALUE max_null_count_get(const VALUE self, VALUE opts) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int c = parse_options_get_max_null_count(ptr->opts);
+	return INT2FIX(c);
+}
+
+VALUE options_null_block_set(const VALUE self, VALUE opts, volatile VALUE n_block) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int b = NUM2INT(n_block);
+	parse_options_set_null_block(ptr->opts, b);
+	return Qnil;
+}
+
+VALUE options_null_block_get(const VALUE self, VALUE opts) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int n_block = parse_options_get_null_block(ptr->opts);
+	return INT2FIX(n_block);
+}
+
+VALUE islands_ok_set(const VALUE self, VALUE opts, volatile VALUE islands_ok) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int i = NUM2INT(islands_ok);
+	parse_options_set_islands_ok(ptr->opts, i);
+	return Qnil;
+}
+
+VALUE islands_ok_get(const VALUE self, VALUE opts) {
+	ParseOptionsPtr *ptr = retrieve_parse_options(opts);
+	int i = parse_options_get_islands_ok(ptr->opts);
+	return INT2FIX(i);
+}
+
 VALUE create_sentence(const VALUE self, volatile VALUE str, VALUE dict) {
 	char *text = StringValuePtr(str);
 	DictionaryPtr *dict_ptr = retrieve_dictionary(dict);

data/ext/link_grammar/link_grammar.h

@@ -86,6 +86,16 @@ VALUE linkage_limit_set(const VALUE self, VALUE opts, volatile VALUE limit);
 VALUE linkage_limit_get(const VALUE self, VALUE opts);
 VALUE short_length_set(const VALUE self, VALUE opts, volatile VALUE length);
 VALUE short_length_get(const VALUE self, VALUE opts);
+VALUE disjunct_cost_set(const VALUE self, VALUE opts, volatile VALUE cost);
+VALUE disjunct_cost_get(const VALUE self, VALUE opts);
+VALUE min_null_count_set(const VALUE self, VALUE opts, volatile VALUE count);
+VALUE min_null_count_get(const VALUE self, VALUE opts);
+VALUE max_null_count_set(const VALUE self, VALUE opts, volatile VALUE count);
+VALUE max_null_count_get(const VALUE self, VALUE opts);
+VALUE options_null_block_set(const VALUE self, VALUE opts, volatile VALUE n_block);
+VALUE options_null_block_get(const VALUE self, VALUE opts);
+VALUE islands_ok_set(const VALUE self, VALUE opts, volatile VALUE islands_ok);
+VALUE islands_ok_get(const VALUE self, VALUE opts);
 
 VALUE create_sentence(const VALUE self, volatile VALUE str, VALUE dict);
 VALUE parse_sentence(const VALUE self, VALUE sentence, VALUE opts);

data/lib/grammar_cop.rb

@@ -5,5 +5,6 @@ require 'grammar_cop/sentence'
 require 'grammar_cop/linkage'
 
 module GrammarCop
-  GrammarCop::Dictionary.set_dictionary_path("#{Gem.loaded_specs['grammar_cop'].full_gem_path}/data")
+  #GrammarCop::Dictionary.set_dictionary_path("#{Gem.loaded_specs['grammar_cop'].full_gem_path}/data")
+  GrammarCop::Dictionary.set_dictionary_path(File.expand_path('../data', File.dirname(__FILE__)))
 end

data/lib/grammar_cop/parse_options.rb

@@ -28,5 +28,44 @@ module GrammarCop
       short_length_get(opts)
     end
 
+    def self.set_disjunct_cost(opts, cost)
+      disjunct_cost_set(opts, cost)
+    end
+    
+    def self.get_disjunct_cost(opts)
+      disjunct_cost_get(opts)
+    end
+    
+    def self.set_min_null_count(opts, count)
+      min_null_count_set(opts, count)
+    end
+    
+    def self.get_min_null_count(opts)
+      min_null_count_get(opts)
+    end
+    
+    def self.set_max_null_count(opts, count)
+      max_null_count_set(opts, count)
+    end
+    
+    def self.get_max_null_count(opts)
+      max_null_count_get(opts)
+    end
+    
+    def self.set_null_block(opts, block)
+      options_null_block_set(opts, block)
+    end
+    
+    def self.get_null_block(opts)
+      options_null_block_get(opts)
+    end
+    
+    def self.set_islands_ok(opts, ok)
+      options_null_block_set(opts, ok)
+    end
+    
+    def self.get_islands_ok(opts)
+      islands_ok_get(opts)
+    end
   end
 end

data/lib/grammar_cop/sentence.rb

@@ -25,7 +25,8 @@ module GrammarCop
       end
       
       #however, sometimes the split results in all incomplete clause so check one more time in those cases
-      if count == 0
+      #but make sure it's not a really long sentence which will cause link-grammar to choke with combinatorial explosion
+      if text.size < 250 and count == 0
         s = create_sentence(text, dictionary)
         count = parse_sentence(s, options)
         delete_sentence(s)

data/lib/grammar_cop/version.rb

@@ -1,3 +1,3 @@
 module GrammarCop
-  VERSION = "0.1.0"
+  VERSION = "0.1.1"
 end

data/test/grammar_cop_test.rb

@@ -4,11 +4,21 @@ puts dir = File.expand_path('../data', File.dirname(__FILE__))
 GrammarCop::Dictionary.set_dictionary_path(dir)
 
 dict = GrammarCop::Dictionary.create("en")
+
+#create and set parse options
 opts = GrammarCop::ParseOptions.create
 GrammarCop::ParseOptions.set_short_length(opts, 7)
 puts "short_length=#{GrammarCop::ParseOptions.get_short_length(opts)}"
 GrammarCop::ParseOptions.set_linkage_limit(opts, 10)
 puts "linkage limit=#{GrammarCop::ParseOptions.get_linkage_limit(opts)}"
+GrammarCop::ParseOptions.set_disjunct_cost(opts, 1000)
+puts "disjunct cost=#{GrammarCop::ParseOptions.get_disjunct_cost(opts)}"
+
+puts "min null count=#{GrammarCop::ParseOptions.get_min_null_count(opts)}"
+puts "max null count=#{GrammarCop::ParseOptions.get_max_null_count(opts)}"
+puts "null block=#{GrammarCop::ParseOptions.get_null_block(opts)}"
+puts "islands ok=#{GrammarCop::ParseOptions.get_islands_ok(opts)}"
+
 text = "Grammar is useless because there's nothing to say -- Gertrude Stein."
 sentence = GrammarCop::Sentence.create(text, dict)
 puts num = GrammarCop::Sentence.parse(sentence, opts)

metadata

@@ -2,7 +2,7 @@
 name: grammar_cop
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors: 
 - Quan Nguyen
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-05-08 00:00:00 Z
+date: 2011-05-13 00:00:00 Z
 dependencies: 
 - !ruby/object:Gem::Dependency 
   name: rake-compiler
@@ -36,8 +36,8 @@ files:
 - .DS_Store
 - .gitignore
 - Gemfile
+- README
 - Rakefile
-- data/.DS_Store
 - data/Makefile
 - data/Makefile.am
 - data/Makefile.in
@@ -64,7 +64,6 @@ files:
 - data/en/Makefile.in
 - data/en/README
 - data/en/tiny.dict
-- data/en/words/.DS_Store
 - data/en/words/Makefile
 - data/en/words/Makefile.am
 - data/en/words/Makefile.in
@@ -148,7 +147,6 @@ files:
 - data/lt/Makefile.am
 - data/lt/Makefile.in
 - ext/.DS_Store
-- ext/link_grammar/.DS_Store
 - ext/link_grammar/extconf.rb
 - ext/link_grammar/link-grammar/.DS_Store
 - ext/link_grammar/link-grammar/.deps/analyze-linkage.Plo
@@ -366,7 +364,6 @@ files:
 - ext/link_grammar/link_grammar.c
 - ext/link_grammar/link_grammar.h
 - grammar_cop.gemspec
-- lib/.DS_Store
 - lib/grammar_cop.rb
 - lib/grammar_cop/.DS_Store
 - lib/grammar_cop/dictionary.rb
@@ -374,7 +371,6 @@ files:
 - lib/grammar_cop/parse_options.rb
 - lib/grammar_cop/sentence.rb
 - lib/grammar_cop/version.rb
-- test/.DS_Store
 - test/grammar_cop_test.rb
 homepage: https://github.com/mquan
 licenses: []