analects 0.2.1 → 0.3.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 89bd80acac983054fa373c4f54039b1a89b155e6
-  data.tar.gz: dc1003d42fb5ef4c0f935326f498c6e5aeabc55f
+  metadata.gz: c11afafa3e6efdbad98b84a73d3a190e2dc7fb52
+  data.tar.gz: 74b50dbb99b0fbb20182e682d1a5b044930e089a
 SHA512:
-  metadata.gz: 11e1845f8585ed02645d0b39da4b1e7e7a40e1243d3bd06acd5f220a382d5b6e12201b7ae33b7f9d7583a9ddabe16025695412485c3b01a2878127f80b04a6a0
-  data.tar.gz: a52fdb67b1be3af05b93962ca42517a4f0b97652fc23de24e3a4ed20ecd6b8255a4093bf81b335ea317581990538d721465f2375a2b9c883392ce91c8ef94b97
+  metadata.gz: bf0ef56017cdf8a29b8931ca7b6e838c0c0255b7e9cfa9f878e23a531bbb43c176f4ffc1840308c0a82427e2c00b764e625226d20b912b7eb4731a8ba8d4f834
+  data.tar.gz: ad3972c46c16426698709ecf2a1cf86246b7221333398ec5d8a2d9b51e9b3dd0532e66754c173bb3e04c05cb288bf475e20229dc844b5a8e0991d00101cceccf

data/.travis.yml

@@ -1,8 +1,8 @@
 language: ruby
 rvm:
   - 1.9.3
-  - 2.0.0
-  - 2.1.0
+  - 2.0
+  - 2.1.1
   - jruby-19mode
   - rbx
   - ruby-head
@@ -11,4 +11,3 @@ matrix:
   allow_failures:
     - rvm: ruby-head
     - rvm: jruby-head
-script: bundle exec rake ci:travis

data/Gemfile

@@ -5,13 +5,4 @@ group :development, :test do
   eval_gemfile 'Gemfile.devtools'
 end
 
-if ENV['LOCAL']
-  source = ->(s) { { path: ENV['HOME'] + "/github/#{s}" } }
-else
-  source = ->(s) { { github: "plexus/#{s}" } }
-end
-
-gem 'rmmseg'   , source.('rmmseg')
-gem 'ting'     , source.('ting')
-
 gemspec

data/Gemfile.lock

@@ -1,15 +1,3 @@
-GIT
-  remote: git://github.com/plexus/rmmseg.git
-  revision: 73fb91682ddf955e6e72722cc3f2adbecaf7508d
-  specs:
-    rmmseg (0.1.6)
-
-GIT
-  remote: git://github.com/plexus/ting.git
-  revision: 63491759a6b40c8dddc7aadcf16fce833ea107ba
-  specs:
-    ting (0.9.0)
-
 GIT
   remote: https://github.com/rom-rb/devtools.git
   revision: 62d3fa16941e5b61e1e8bb520a1b40d6179aebf5
@@ -19,7 +7,7 @@ GIT
 PATH
   remote: .
   specs:
-    analects (0.2.1)
+    analects (0.3.1)
       ice_nine (~> 0.11.0)
       inflecto (~> 0.0.2)
       plexus-rmmseg (~> 0.1.6)
@@ -178,6 +166,7 @@ GEM
     thread_safe (0.2.0)
       atomic (>= 1.1.7, < 2)
     timers (1.1.0)
+    ting (0.9.0)
     unparser (0.1.9)
       abstract_type (~> 0.0.7)
       adamantium (~> 0.1)
@@ -220,7 +209,6 @@ DEPENDENCIES
   rb-notifu (~> 0.0.4)
   rbench (~> 0.2.3)
   reek (~> 1.3.2)
-  rmmseg!
   rspec (~> 2.14.1)
   rspec-core (~> 2.14.8)
   rubocop (~> 0.18.1)
@@ -231,7 +219,6 @@ DEPENDENCIES
   rubysl-singleton (~> 2.0.0)
   simplecov (~> 0.8.2)
   terminal-notifier-guard (~> 1.5.3)
-  ting!
   yard (~> 0.8.7)
   yard-spellcheck (~> 0.1.5)
   yardstick (~> 0.9.9)

data/lib/analects.rb

@@ -28,6 +28,7 @@ require 'analects/version'
 require 'analects/encoding'
 require 'analects/cli/progress'
 require 'analects/cedict_loader'
+require 'analects/hsk_loader'
 require 'analects/chise_ids_loader'
 require 'analects/source'
 require 'analects/library'

data/lib/analects/cedict_loader.rb

@@ -6,7 +6,7 @@ module Analects
 
     attr_reader :headers
 
-    def initialize(io)
+    def initialize(io, library)
       @contents = io.read
       @headers = {}
       @contents.each_line do |line|
@@ -21,14 +21,25 @@ module Analects
       [:traditional, :simplified, :pinyin, :definitions]
     end
 
-    def each
-      if block_given?
-        @contents.each_line do |line|
-          yield process_contents(line) if line !~ /^#/
+    def each(&blk)
+      return to_enum(__method__) unless block_given?
+      @entries ||= @contents.each_line.map do |line|
+        process_contents(line) if line !~ /^#/
+      end.compact
+      @entries.each(&blk)
+    end
+
+    def find_by(qry)
+      qry.map {|field, value| lookup_index(field).fetch(value, [])}.inject {|r1, r2| r1 & r2}
+    end
+
+    def lookup_index(field)
+      @indexes ||= field_names.each_with_object({}) do |field, acc|
+        acc[field] = each_with_object({}) do |entry, acc|
+          (acc[entry[field_names.index(field)]] ||= []) << entry
         end
-      else
-        enum_for(:each)
       end
+      @indexes[field]
     end
 
     private

data/lib/analects/chise_ids_loader.rb

@@ -14,7 +14,7 @@ module Analects
       end
     end
 
-    def initialize(pathname, only_unicode = true)
+    def initialize(pathname, library, only_unicode = true)
       @contents = MultiFile.new(pathname.children.select{|ch| ch.to_s =~ /IDS-.*\.txt/})
       @only_unicode = only_unicode
     end

data/lib/analects/hsk_loader.rb

@@ -0,0 +1,27 @@
+# encoding: UTF-8
+
+module Analects
+  class HskLoader
+    extend Forwardable
+    include Enumerable
+
+    FIELDS = [:level, :simplified, :traditional, :pinyin]
+
+    class HskEntry < Struct.new(:library, *FIELDS)
+      def cedict
+        @cedict ||= library.cedict.loader.find_by(
+          traditional: traditional,
+          simplified: simplified,
+        )
+      end
+    end
+
+    def_delegators :@contents, :each
+
+    def initialize(io, library)
+      @contents = CSV(io.read).map do |level, simplified, traditional, pinyin|
+        HskEntry.new(library, level, simplified, traditional, pinyin)
+      end
+    end
+  end
+end

data/lib/analects/library.rb

@@ -47,17 +47,20 @@ module Analects
         }.merge(options_for :chise_ids)
       )
     end
-    # def hsk
-    #   @hsk ||= Source.new(
 
-    #   ).merge(options_for :hsk)
-    # end
+    def hsk
+      @hsk ||= Source.new( {
+          data_file: 'hsk.csv'
+        }.merge(options_for :hsk)
+      )
+    end
 
     private
 
     def options_for(name)
       {
         name: name,
+        library: self,
         url: Analects.const_get("#{name.to_s.upcase}_URL"),
         loader: Analects.const_get("#{Inflecto.camelize name}Loader"),
         data_dir: data_dir

data/lib/analects/source.rb

@@ -7,12 +7,13 @@ module Analects
       @options = options
     end
 
+    def library   ; options[:library]           ; end
     def name      ; options[:name]              ; end
     def url       ; options[:url]               ; end
     def retrieval ; Array(options[:retrieval])  ; end
 
     def loader
-      @loader ||= options[:loader].new(Pathname(location))
+      @loader ||= options[:loader].new(Pathname(location), library)
     end
 
     def data_dir

data/lib/analects/version.rb

@@ -1,3 +1,3 @@
 module Analects
-  VERSION = '0.2.1'
+  VERSION = '0.3.1'
 end

data/spec/analects/cedict_loader_spec.rb

@@ -14,7 +14,7 @@ describe Analects::CedictLoader do
 '
   end
 
-  let ( :cedict_loader ) { Analects::CedictLoader.new( StringIO.new(contents) ) }
+  let ( :cedict_loader ) { Analects::CedictLoader.new( StringIO.new(contents), nil ) }
 
   it "should parse headers" do
     cedict_loader.headers.should == { 'charset' => 'UTF-8', 'entries' => '104941' }

data/spec/analects/chise_ids_loader_spec.rb

@@ -3,7 +3,7 @@ require 'spec_helper'
 
 describe Analects::ChiseIdsLoader do
   subject(:loader) do
-    Analects::ChiseIdsLoader.new(Analects::ROOT.join('spec/test_data/chise_ids'), only_unicode)
+    Analects::ChiseIdsLoader.new(Analects::ROOT.join('spec/test_data/chise_ids'), nil, only_unicode)
   end
 
   let(:only_unicode) { false }

data/spec/analects/library_spec.rb

@@ -1,19 +1,19 @@
 require 'tempfile'
-require 'spec_helper'
 require 'securerandom'
+require 'pathname'
+
+require 'spec_helper'
 
 describe Analects::Library do
   subject(:library) {
     described_class.new(options)
   }
-  let(:options) { {} }
+  let(:data_dir) { File.join(Dir.tmpdir, 'analects-' + SecureRandom.hex(16)) }
+  let(:options) {
+    { data_dir: data_dir }
+  }
 
   context 'with a data_dir specified' do
-    let(:data_dir) { File.join(Dir.tmpdir, 'analects-' + SecureRandom.hex(16)) }
-    let(:options) {
-      { data_dir: data_dir }
-    }
-
     it 'should set that data dir on the sources' do
       subject.cedict.data_dir.should == data_dir
     end
@@ -23,7 +23,6 @@ describe Analects::Library do
     subject(:cedict) { library.cedict }
 
     its(:name)     { should == :cedict }
-    its(:location) { should == File.join(Dir.home, '.analects/cedict_1_0_ts_utf-8_mdbg.txt') }
 
     it "should download and unpack the CEDICT archive" do
       cedict.should_receive(:retrieve_http).once.with(Analects::CEDICT_URL).and_return(:a_stream)
@@ -42,7 +41,6 @@ describe Analects::Library do
 
     its( :name )      { should == :chise_ids }
     its( :retrieval ) { should == [ :git ] }
-    its(:location   ) { should == File.join(Dir.home, '.analects/chise_ids') }
     its( :url       ) { should == Analects::CHISE_IDS_URL}
   end
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: analects
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.3.1
 platform: ruby
 authors:
 - Arne Brasseur
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-03-16 00:00:00.000000000 Z
+date: 2014-04-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rspec
@@ -169,6 +169,7 @@ files:
 - lib/analects/chise_ids_loader.rb
 - lib/analects/cli/progress.rb
 - lib/analects/encoding.rb
+- lib/analects/hsk_loader.rb
 - lib/analects/library.rb
 - lib/analects/models/kangxi_radical.rb
 - lib/analects/models/zi.rb
@@ -208,7 +209,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.2.1
+rubygems_version: 2.0.14
 signing_key: 
 specification_version: 4
 summary: Toolkit for Mandarin language learning apps