opener-property-tagger 3.0.5 → 3.0.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: db4ce8acc0c1dd88b2042f9a936012ea9a3c1448
-  data.tar.gz: 4ed777c8b215fb98dd3687689cb1c58d97787dd0
+  metadata.gz: c3d36e45bbff187579cbb8f2275d53261ed13030
+  data.tar.gz: 2850b27e9876336083dde1af17083658be60ef1c
 SHA512:
-  metadata.gz: 5cd57abe32b4d88f8ed10618c9e8466c1628faebda89ad1ceeaa956c571fdc00a6069cea44a1a1570df2ebe0136fc6e34c142d1850972a1bc3e2f926c640d13d
-  data.tar.gz: e64b80100aac3dbfc3229e75692ef9ab16cbd362ca878e36f2ef5b492c456843fcb535c5f26105f6e334594a433c7ff9b050758068779803373aab76acac5e96
+  metadata.gz: f95a22d030ff6dc1685ee117ddae2f0d104fce5b187e7bb2062b96dd98ba035dd55c6b9f21266a63e968ff42d9e0869ce762786c7a2d7f5b53b6cc63d5aff4a2
+  data.tar.gz: be9bfed3bb90e26a111e03973a3ec543780e886be08d7102dad057e81917e1323c599198f81d79da58257d85e5adafaf507454a0fa6e30105fc8210698c94e17

data/lib/opener/property_tagger.rb

@@ -1,12 +1,14 @@
 require 'open3'
 require 'slop'
 require 'oga'
+require 'monitor'
 
 require 'rexml/document'
 require 'rexml/formatters/pretty'
 
 require_relative 'property_tagger/version'
 require_relative 'property_tagger/cli'
+require_relative 'property_tagger/aspects_cache'
 require_relative 'property_tagger/processor'
 
 module Opener
@@ -52,24 +54,15 @@ module Opener
     end
 
     ##
-    # Processes the input and returns an Array containing the output of STDOUT,
-    # STDERR and an object containing process information.
+    # Processes the input KAF document.
     #
-    # @param [String] input The text of which to detect the language.
-    # @return [Array]
+    # @param [String] input
+    # @return [String]
     #
     def run(input)
-      output = process(input)
-
-      return output
-    end
-
-    protected
-
-    def process(input)
-      processor = Processor.new(input, path, !options[:no_time])
+      timestamp = !options[:no_time]
 
-      return processor.process
+      return Processor.new(input, path, timestamp, options[:pretty]).process
     end
   end # PolarityTagger
 end # Opener

data/lib/opener/property_tagger/aspects_cache.rb

@@ -0,0 +1,49 @@
+module Opener
+  class PropertyTagger
+    ##
+    # Thread-safe cache for storing the contents of aspect files.
+    #
+    class AspectsCache
+      include MonitorMixin
+
+      def initialize
+        super
+
+        @cache = {}
+      end
+
+      ##
+      # Returns the aspects for the given file path. If the aspects don't exist
+      # they are first loaded into the cache.
+      #
+      # @param [String] path
+      #
+      def [](path)
+        synchronize do
+          @cache[path] = load_aspects(path) unless @cache.key?(path)
+        end
+
+        return @cache[path]
+      end
+
+      alias_method :get, :[]
+
+      ##
+      # Loads the aspects of the given path.
+      #
+      # @param [String] path
+      #
+      def load_aspects(path)
+        mapping = Hash.new { |hash, key| hash[key] = [] }
+
+        File.foreach(path) do |line|
+          lemma, pos, aspect = line.chomp.split("\t")
+
+          mapping[lemma.to_sym] << aspect
+        end
+
+        return mapping
+      end
+    end # AspectsCache
+  end # PropertyTagger
+end # Opener

data/lib/opener/property_tagger/cli.rb

@@ -56,10 +56,13 @@ Examples:
 
           on :'no-time', 'Disables adding of timestamps'
 
+          on :ugly, 'Disables pretty formatting of XML (faster)'
+
           run do |opts, args|
             tagger = PropertyTagger.new(
               :args    => args,
-              :no_time => opts[:'no-time']
+              :no_time => opts[:'no-time'],
+              :pretty  => !opts[:ugly]
             )
 
             input  = STDIN.tty? ? nil : STDIN.read

data/lib/opener/property_tagger/processor.rb

@@ -4,13 +4,27 @@ module Opener
     # Class that applies property tagging to a given input KAF file.
     #
     class Processor
-      attr_accessor :document, :aspects_path, :language, :aspects, :terms,
-        :timestamp
+      attr_accessor :document, :aspects_path, :timestamp, :pretty
 
-      def initialize(file, aspects_path, timestamp = true)
+      ##
+      # Global cache used for storing loaded aspects.
+      #
+      # @return [Opener::PropertyTagger::AspectsCache.new]
+      #
+      ASPECTS_CACHE = AspectsCache.new
+
+      ##
+      # @param [String|IO] file The KAF file/input to process.
+      # @param [String] aspects_path Path to the aspects.
+      # @param [TrueClass|FalseClass] timestamp Add timestamps to the KAF.
+      # @param [TrueClass|FalseClass] pretty Enable pretty formatting, disabled
+      #  by default due to the performance overhead.
+      #
+      def initialize(file, aspects_path, timestamp = true, pretty = false)
         @document     = Oga.parse_xml(file)
         @aspects_path = aspects_path
         @timestamp    = timestamp
+        @pretty       = pretty
 
         raise 'Error parsing input. Input is required to be KAF' unless is_kaf?
       end
@@ -20,64 +34,52 @@ module Opener
       # @return [String]
       #
       def process
-        @language = get_language
-        @aspects  = load_aspects
-        @terms    = get_terms
-
         existing_aspects = extract_aspects
 
         add_features_layer
         add_properties_layer
 
-        index = 1
+        existing_aspects.each_with_index do |(key, value), index|
+          index += 1
 
-        existing_aspects.each_pair do |key,value|
           add_property(key, value, index)
-          index += 1
         end
 
         add_linguistic_processor
 
-        return pretty_print(document)
+        return pretty ? pretty_print(document) : document.to_xml
       end
 
       ##
-      # Loads the aspects from the txt file
       # @return [Hash]
       #
-      def load_aspects
-        aspects_hash = {}
-
-        File.foreach(aspects_file) do |line|
-          lemma, pos, aspect = line.gsub("\n", "").split("\t")
-
-          aspects_hash[lemma.to_sym] = [] unless aspects_hash[lemma.to_sym]
-          aspects_hash[lemma.to_sym] << aspect
-        end
-
-        return aspects_hash
+      def aspects
+        return ASPECTS_CACHE[aspects_file]
       end
 
       ##
       # Get the language of the input file.
+      #
       # @return [String]
       #
-      def get_language
-        document.at_xpath('KAF').get('xml:lang')
+      def language
+        return @language ||= document.at_xpath('KAF').get('xml:lang')
       end
 
       ##
       # Get the terms from the input file
       # @return [Hash]
       #
-      def get_terms
-        terms_hash = {}
+      def terms
+        unless @terms
+          @terms = {}
 
-        document.xpath('KAF/terms/term').each do |term|
-          terms_hash[term.get('tid').to_sym] = term.get('lemma')
+          document.xpath('KAF/terms/term').each do |term|
+            @terms[term.get('tid').to_sym] = term.get('lemma')
+          end
         end
 
-        return terms_hash
+        return @terms
       end
 
       ##
@@ -93,7 +95,7 @@ module Opener
         # lemmas) belong to a property.
         max_ngram = 2
 
-        uniq_aspects = {}
+        uniq_aspects = Hash.new { |hash, key| hash[key] = [] }
 
         while current_token < terms.count
           (0..max_ngram).each do |tam_ngram|
@@ -107,7 +109,6 @@ module Opener
                 properties.uniq.each do |property|
                   next if !property or property.strip.empty?
 
-                  uniq_aspects[property.to_sym] = [] unless uniq_aspects[property.to_sym]
                   uniq_aspects[property.to_sym] << [ids,ngram]
                 end
               end
@@ -223,8 +224,12 @@ module Opener
         return !!document.at_xpath('KAF')
       end
 
+      ##
+      # @return [String]
+      #
       def aspects_file
-        return File.expand_path("#{aspects_path}/#{language}.txt", __FILE__)
+        return @aspects_file ||=
+          File.expand_path("#{aspects_path}/#{language}.txt", __FILE__)
       end
     end # Processor
   end # PropertyTagger

data/lib/opener/property_tagger/version.rb

@@ -1,5 +1,5 @@
 module Opener
   class PropertyTagger
-    VERSION = '3.0.5'
+    VERSION = '3.0.6'
   end # PropertyTagger
 end # Opener

data/opener-property-tagger.gemspec

@@ -33,4 +33,5 @@ Gem::Specification.new do |gem|
   gem.add_development_dependency 'rspec', '~> 3.0'
   gem.add_development_dependency 'cucumber'
   gem.add_development_dependency 'rake'
+  gem.add_development_dependency 'benchmark-ips', '~> 2.0'
 end

data/task/test.rake

@@ -2,5 +2,6 @@ desc 'Runs the tests'
 task :test => [:lexicons] do
   ENV['RESOURCE_PATH'] = File.expand_path('../../tmp/lexicons/hotel', __FILE__)
 
-  sh('cucumber features')
+  sh 'cucumber features'
+  sh 'rspec spec --order random'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: opener-property-tagger
 version: !ruby/object:Gem::Version
-  version: 3.0.5
+  version: 3.0.6
 platform: ruby
 authors:
 - development@olery.com
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-01-26 00:00:00.000000000 Z
+date: 2015-02-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: opener-daemons
@@ -108,6 +108,20 @@ dependencies:
     - - ">="
       - !ruby/object:Gem::Version
         version: '0'
+- !ruby/object:Gem::Dependency
+  name: benchmark-ips
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '2.0'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '2.0'
 description: Property tagger for hotels in Dutch and English.
 email: 
 executables:
@@ -125,6 +139,7 @@ files:
 - config.ru
 - exec/property-tagger.rb
 - lib/opener/property_tagger.rb
+- lib/opener/property_tagger/aspects_cache.rb
 - lib/opener/property_tagger/cli.rb
 - lib/opener/property_tagger/processor.rb
 - lib/opener/property_tagger/public/markdown.css