konjak 0.0.12 → 0.0.13

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 2b3737eb910538d760657c5e527b35be1fc9599f
-  data.tar.gz: 85f858f6a152cf53b32e6b831b737d8729712437
+  metadata.gz: cd181dadc5d9128c27be3312f50274acffbcfce8
+  data.tar.gz: 29fa12fb01c4409b7587496413b0fda0c46ab0a4
 SHA512:
-  metadata.gz: 38622d0a0d77bb334554e0cb5fde0cf8226c1eab06bfec5e86e5a202174c61acd9b83de34ec297299acd4974b1cc85198500591685daaf4e2d7adf77c435ed1f
-  data.tar.gz: bb75de764a2a6722e5435be2f3ca3acd6d39147013630892a8072fadeefadc3bee28f3614327668d98f5843d3ac74756600b1168dfa56def824b70a8253e4430
+  metadata.gz: 188010cd203698454c7f2204c20e6a6e19a18a7eb25640d0c9555e011568ff87fe96ba521ab04c1c42975c59636e634992144a825980f9398abb12e72f913996
+  data.tar.gz: ba9cf03f27fe8599815060771ea437fb852b95e0a5f3f1e06524e9ef0c5930b0a373c9fca08cbffd83c44b5238acd51403058775a83fe328a3550c7424cd88ee

data/lib/konjak/segment/gtt.rb

@@ -1,18 +1,26 @@
+require 'mem'
+
 module Konjak
   class Segment < StructuralElement
     module GTT
+      include Mem
+
       Tag = Struct.new(:gtt, :html)
 
       def compile_gtt_html_pattern
         regexp = Regexp.escape(text)
         gtt_tag_ns.each do |n|
-          regexp = regexp.sub(/\\\{#{n}\\\}/)    { "(?<n#{n}><(?<_#{n}>\\w+)[^>]*>)" }
-          regexp = regexp.gsub(/\\\{#{n}\\\}/)   { "\\k<n#{n}>" }
-          regexp = regexp.gsub(/\\\{\/#{n}\\\}/) { "</\\k<_#{n}>>" }
+          regexp.sub!(/\\\{#{n}\\\}/)    { "(?<n#{n}><(?<_#{n}>\\w+)[^>]*>)" }
+          regexp.gsub!(/\\\{#{n}\\\}/)   { "\\k<n#{n}>" }
+          regexp.gsub!(/\\\{\/#{n}\\\}/) { "</\\k<_#{n}>>" }
         end
-        regexp = regexp.gsub(/(?:\\\s|\n)/m) { '\s' }
+        regexp.gsub!(/(?:\\\s|\n)/m)        { '\s' }
+        regexp.gsub!(/(?:\\s)+/m)           {|s| s + '++' }
+        regexp.gsub!(/^(?<s>(?:\\s)+)\+\+/) { $~[:s] }
+        regexp.gsub!(/(?<s>(?:\\s)+)\+\+$/) { $~[:s] }
         Regexp.compile(regexp)
       end
+      memoize :compile_gtt_html_pattern
 
       def extract_gtt_tags_from(text)
         m = text.match(compile_gtt_html_pattern)

data/lib/konjak/segment.rb

@@ -1,9 +1,11 @@
 require 'konjak/segment/gtt'
+require 'mem'
 
 module Konjak
   # container
   class Segment < StructuralElement
     include GTT
+    include Mem
 
     # children
     def text
@@ -17,9 +19,13 @@ module Konjak
 
     def compile_pattern
       regexp = Regexp.escape(text)
-      regexp = regexp.gsub(/(?:\\\s|\n)/m) { '\s' }
+      regexp = regexp.gsub(/(?:\\\s|\n)/m)        { '\s' }
+      regexp = regexp.gsub(/(?:\\s)+/m)           {|s| s + '++' }
+      regexp = regexp.gsub(/^(?<s>(?:\\s)+)\+\+/) { $~[:s] }
+      regexp = regexp.gsub(/(?<s>(?:\\s)+)\+\+$/) { $~[:s] }
       Regexp.compile(regexp)
     end
+    memoize :compile_pattern
 
     def translation_unit
       TranslationUnit.new(translation_unit_variant.parent)

data/lib/konjak/tmx_segmentor/gtt_html_strategy.rb

@@ -7,21 +7,8 @@ module Konjak
 
       private
 
-      def split(translation_unit, text)
-        segment = translation_unit.variant(@lang).segment
-        pattern = segment.compile_gtt_html_pattern
-
-        texts = []
-        while true
-          head, match, tail = text.partition(pattern)
-          break if match.empty? || text.length < min_segment_length
-          texts << head unless head.empty?
-
-          texts << SegmentString.new(match, segment)
-
-          text = tail
-        end
-        texts << text
+      def compile_pattern(segment)
+        segment.compile_gtt_html_pattern
       end
     end
   end

data/lib/konjak/tmx_segmentor/strategy.rb

@@ -16,10 +16,13 @@ module Konjak
       def segmentize(text)
         segments = [text]
         translation_units.each do |translation_unit|
+          segment = translation_unit.variant(@lang).segment
+
           segments.map! {|text|
+            next text if text.length < min_segment_length
             next text if text.is_a?(SegmentString)
 
-            split(translation_unit, text)
+            split(segment, text)
           }.flatten!
         end
         segments
@@ -35,6 +38,23 @@ module Konjak
         @options[:min_segment_length]
       end
 
+      def split(segment, text)
+        texts = []
+        while true
+          break if text.length < min_segment_length
+
+          head, match, tail = text.partition(compile_pattern(segment))
+          break if match.empty?
+
+          texts << head unless head.empty?
+
+          texts << SegmentString.new(match, segment)
+
+          text = tail
+        end
+        texts << text
+      end
+
       def translation_units
         @tmx.body.translation_units.sort_by {|tu|
           -tu.variant(@lang).segment.text.length

data/lib/konjak/tmx_segmentor/text_strategy.rb

@@ -7,21 +7,8 @@ module Konjak
 
       private
 
-      def split(translation_unit, text)
-        segment = translation_unit.variant(@lang).segment
-        pattern = segment.compile_pattern
-
-        texts = []
-        while true
-          head, match, tail = text.partition(pattern)
-          break if match.empty? || text.length < min_segment_length
-          texts << head unless head.empty?
-
-          texts << SegmentString.new(match, segment)
-
-          text = tail
-        end
-        texts << text
+      def compile_pattern(segment)
+        segment.compile_pattern
       end
     end
   end

data/lib/konjak/version.rb

@@ -1,3 +1,3 @@
 module Konjak
-  VERSION = "0.0.12"
+  VERSION = "0.0.13"
 end

data/spec/konjak_translate_spec.rb

@@ -19,8 +19,10 @@ EXPECT
 
   context 'when blanks between words is not exactly match' do
     let(:doc) { <<DOC }
-this is data
-(with a non-standard character: ).
+this is data         (with a non-standard
+
+
+character: ).
 this is data (with a non-standard character: ).
 DOC
 
@@ -54,7 +56,7 @@ EXPECT
     context 'when blanks between words is not exactly match' do
       let(:doc) { <<GTT_HTML }
 This is
-<a href="http://example.com">example</a>.
+         <a href="http://example.com">example</a>.
 And This
 is
 <b>example</b>. Yey.

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: konjak
 version: !ruby/object:Gem::Version
-  version: 0.0.12
+  version: 0.0.13
 platform: ruby
 authors:
 - Seiei Higa