dict 0.2.2 → 0.2.3

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    dict (0.2.1)
+    dict (0.2.2)
       nokogiri (~> 1.5.5)
       slop (~> 3.3.2)
 

data/lib/dict/dict.rb

@@ -29,7 +29,7 @@ module Dict
     def get_single_dictionary_translations(word, dictionary)
       case dictionary
       when 'wiktionary'
-        Wiktionary.new(word).translate
+        Wiktionary.new(word).translate.translations
       when 'dictpl'
         Dictpl.new(word).translate
       else Dictionary.message

data/lib/dict/dictionary.rb

@@ -1,6 +1,7 @@
 # -*- coding: utf-8 -*-
 
 require 'open-uri'
+require_relative 'result'
 
 module Dict
   class Dictionary
@@ -11,6 +12,7 @@ module Dict
       @translations = []
       @examples = []
       @word = word
+      @result = Dict::Result.new(@word)
     end
 
     # returns hash with structure as showed below

data/lib/dict/result.rb

@@ -0,0 +1,35 @@
+module Dict
+  class Result
+    attr_reader :term, :translations, :examples
+    
+    def initialize(term)
+      @term = term
+      @translations = {}
+      @examples = {}
+    end
+    
+    def add_translation(term, translation)
+      add_result(@translations, term, translation)
+    end
+    
+    def add_example(term, example)
+      add_result(@examples, term, example)
+    end
+    
+    def each_translation
+      @translations.each_pair do |term,translation|
+        yield term, translation
+      end
+    end
+    
+    private
+    def add_result(hash, key, value)
+      if hash.has_key?(key)
+        hash[key].push(value)
+      else
+        hash.merge!({ key => [value] })
+      end
+      self
+    end
+  end
+end

data/lib/dict/version.rb

@@ -1,3 +1,3 @@
 module Dict
-  VERSION = "0.2.2"
+  VERSION = "0.2.3"
 end

data/lib/dict/wiktionary.rb

@@ -1,28 +1,62 @@
 # -*- coding: utf-8 -*-
 
 require 'nokogiri'
-require 'dict/dictionary'
+require_relative 'dictionary'
 
 WIKI_URL = 'http://en.wiktionary.org/wiki/'
 
 module Dict
   class Wiktionary < Dictionary
-    # returns hash with structure as showed below
-    # { 'TRANSLATION' => ['EXAMPLE', ...], ... }
-    def translate
-      context_words = []
-      get_html(uri(WIKI_URL, @word)).css('p + ol li a').each do |node|
-        get_html(uri(WIKI_URL, node.text)).css('p + ol > li dl dd').each do |example|
-          context_words << node.text << example.text
-        end
+    # Return a Dict::Result object
+	  def translate
+      url_pl = "http://en.wiktionary.org/w/index.php?title=#{@word}&action=edit"
+      url_en = "http://pl.wiktionary.org/w/index.php?title=#{@word}&action=edit"
+      
+      content = get_html(url_pl).css('textarea#wpTextbox1').first
+      if polish?(get_html(url_pl).css('textarea#wpTextbox1').first)
+        get_translations(get_html(url_pl).css('textarea#wpTextbox1').first)
+      else
+        get_translations(get_html(url_en).css('textarea#wpTextbox1').first.content, false)
       end
-      make_hash_results(context_words)
-    end
-
+      
+      @result
+    end 
+    
     def get_html(url)
-      Nokogiri::HTML(open(url))
+      Nokogiri::HTML(open(URI.encode(url)))
     rescue OpenURI::HTTPError
       raise Dictionary::ConnectError
     end
+    
+    private
+    def polish?(content)
+       return true if /==Polish==/i.match(content)
+       false
+    end
+
+    def get_translations(content, polish = true)
+      if polish
+        translations = /Noun[^\{]+\{\{(?:head\|pl|pl\-noun)[^#]+#\s*\[\[([^\n]+)/.match(content)
+        translations = (translations && translations[1].gsub(/\[|\]/,'').split(', ')) || []
+        translations.each { |item| @result.add_translation(@result.term, item) }
+      else
+        translations_block = /angielski(?:.|\n)+\{\{znaczenia\}\}(.|\n)+(?:\{\{odmiana){1,}/.match(content)
+        return unless translations_block.instance_of?(MatchData)
+        translations_block = translations_block[0].gsub(/odmiana(.|\n)+$/,'')
+        translations = translations_block.scan(/:\s*\(\d\.?\d?\)\s*([^\n]+)/)
+        translations.map! do |translation|
+          translation[0].gsub(/\[|\]|\{\{[^\}]+\}\}|'/,'').strip
+        end
+        translations.delete_if do |item|
+          item.empty?
+        end
+        translations ||= []
+        translations.each { |item| @result.add_translation(@result.term, item) }
+      end
+    end
+    
+    def get_examples(content, polish = true)
+      # todo
+    end
   end
 end

data/spec/dict/lib_dict_cli_runner_spec.rb

@@ -37,7 +37,7 @@ describe "get_translations" do
     stub_const("ARGV", ["słowik"])
     runner = Dict::CLI::Runner.new
     opts = runner.parse_parameters
-    runner.get_translations(opts, "słowik").should == {"wiktionary"=>{}, "dictpl"=>{"słowik"=>["nightingale"], "słowik białobrewy; Luscinia indicus; Tarsiger indicus (gatunek ptaka)"=>["white-browed bush-robin"], "słowik białosterny; Luscinia pectoralis (gatunek ptaka)"=>["Himalayan rubythroat", "white-tailed rubythroat"], "słowik chiński; pekińczyk żółty; Leiothrix lutea"=>["Pekin robin", "red-billed leiothrix"], "słowik chiński; pekińczyk żółty; pekińczyk koralodzioby; Leiothrix lutea"=>["Peking robin"], "słowik czarnogardły; Luscinia obscura"=>["black-throated blue robin"], "słowik himalajski; Luscinia brunnea (gatunek ptaka)"=>["Indian blue chat", "Indian blue robin"], "słowik modry; Luscinia cyane"=>["Siberian blue robin"], "słowik obrożny; Luscinia johnstoniae; Tarsiger johnstoniae (gatunek ptaka)"=>["collared bush-robin"]}}
+    runner.get_translations(opts, "słowik").should == {"wiktionary"=>{"słowik"=>["nightingale"]}, "dictpl"=>{"słowik"=>["nightingale"], "słowik białobrewy; Luscinia indicus; Tarsiger indicus (gatunek ptaka)"=>["white-browed bush-robin"], "słowik białosterny; Luscinia pectoralis (gatunek ptaka)"=>["Himalayan rubythroat", "white-tailed rubythroat"], "słowik chiński; pekińczyk żółty; Leiothrix lutea"=>["Pekin robin", "red-billed leiothrix"], "słowik chiński; pekińczyk żółty; pekińczyk koralodzioby; Leiothrix lutea"=>["Peking robin"], "słowik czarnogardły; Luscinia obscura"=>["black-throated blue robin"], "słowik himalajski; Luscinia brunnea (gatunek ptaka)"=>["Indian blue chat", "Indian blue robin"], "słowik modry; Luscinia cyane"=>["Siberian blue robin"], "słowik obrożny; Luscinia johnstoniae; Tarsiger johnstoniae (gatunek ptaka)"=>["collared bush-robin"]}}
   end
 
   it "should return results from selected dictionary for word 'słowik'" do

data/spec/dict/lib_dict_spec.rb

@@ -4,24 +4,21 @@ require 'dict/dict'
 
 describe Dict do
 
-  it "should get single translation from dictionary with two arguments given" do
-    expect{
-      Dict.get_single_dictionary_translations('samochód', 'dictpl')
-    }.to_not raise_error
-  end
-
-  it "should return hash with translations from all dictionaries" do
-    Dict.get_all_dictionaries_translations('samochód').should be_a(Hash)
-  end
-
   it "should return array of available services which is not empty" do
     arr = Dict.available_dictionaries
-		arr.should be_a(Array)
-		arr.size.should_not == 0
+    arr.should be_a(Array)
+    arr.size.should_not == 0
   end
 
-	it "should return array of available services which contains wiktionary and dictpl" do
-		Dict.available_dictionaries.should == ['wiktionary', 'dictpl']
-	end
-
+  it "should return array of available services, which contains wiktionary and dictpl" do        
+    Dict.available_dictionaries.should == ['wiktionary', 'dictpl']
+  end
+  
+ 
+  
+  it "should return whatever Dictpl returns embedded in a hash" do
+    dictpl = stub(:translate => 'DICTPL_RESULTS')
+    Dict::Dictpl.should_receive(:new).with('WORD').and_return(dictpl)
+    Dict.get_single_dictionary_translations('WORD', 'dictpl').should == 'DICTPL_RESULTS'
+  end
 end

data/spec/dict/lib_wiktionary_spec.rb

@@ -9,26 +9,22 @@ describe Dict::Wiktionary do
   end
 
   it "should return an two element array of translations of word samochód containing [\"car\",\"automobile\"]" do
-    w = Dict::Wiktionary.new('samochód')
-    w.translate
-    w.translations.should == ["car", "automobile"]
+    w = Dict::Wiktionary.new('samochód').translate
+    w.translations.should == {"samochód"=>["car", "automobile"]}
   end
 
-  it "should return array with translations" do
-    w = Dict::Wiktionary.new('samochód')
-    w.translate
-    w.translations.should be_a(Array)
+  it "should return a hash with translations" do
+    w = Dict::Wiktionary.new('samochód').translate
+    w.translations.should be_a(Hash)
   end
 
-  it "should return array with examples of translated words" do
-    w = Dict::Wiktionary.new('samochód')
-    w.translate
-    w.examples.should be_a(Array)
+  it "should return a Resut object" do
+    w = Dict::Wiktionary.new('samochód').translate
+    w.should be_a(Dict::Result)
   end
-
-  it "should return a hash from array of paired values" do
-    w = Dict::Wiktionary.new('samochód')
-    w.make_hash_results(w.translate).should be_a(Hash)
+  
+  it "should return translation from english to polish for word 'field'" do
+    result = Dict::Wiktionary.new('field').translate.translations
+    result.should eq({"field"=>["pole", "pole (magnetyczne, elektryczne, sił, itp.)", "pole (skalarne, wektorowe, itp.)", "ciało (liczb rzeczywistych, zespolonych, itp.)", "wystawić (drużynę)", "odpowiadać (na pytania)", "polowy", "polny"]})
   end
-
 end