konjak 0.0.21 → 0.0.22

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: f16761c5f15cb70d150d182162126a265ec0fa1b
-  data.tar.gz: 3441bf4468073d0612fbb175e8f757702d8d9edb
+  metadata.gz: c6d01369d26197b67672a4260930011ef66880bd
+  data.tar.gz: 9e2596af227e206039d32f3107fb3ee318adbb25
 SHA512:
-  metadata.gz: 1e04e5a40af148be17c0d535e823eec36effa2edd02faae8165e6d3475f8f410b287e7a4fb5ba1b59fd287167ff5d3ec8f1ecfa89f8d1b643c33e51b472132f5
-  data.tar.gz: 521d6ff59812122d838e191fcfa45815879500726d44ea217a96addf50491aa3c7e0f97b087113ad4716e7c55fe9d595e293c5c166c0d26c8fcbb36e63ee7b07
+  metadata.gz: 5248cd97806b0f7ad9b487838dc03ae65d631ec85d69dac8b8c1d5f84cc2265e8b9303ce9be543537939349a94dd13f465fe953cc6fc2ff2de640f26320c7ec6
+  data.tar.gz: 0e37939c78bfd23a9be0f86f89c06a0edfbf83140a0dd0d5d1c9debd4ef0beeb7f59c610c3b278115249bc852492122ea1887cf8ea91e3b5da00fad4300cab43

data/lib/konjak/tmx_segmentor/strategy.rb

@@ -8,7 +8,15 @@ module Konjak
   class TmxSegmentor < Segmentor
     class Strategy
       Edge = Struct.new(:prev,  :current)
-      Node = Struct.new(:range, :segment)
+      Node = Struct.new(:range, :segments) do
+        def <=>(other)
+          [range.begin, -max_segment_size] <=> [other.range.begin, -other.max_segment_size]
+        end
+
+        def max_segment_size
+          segments.max_by {|s| s.text.size }.text.size
+        end
+      end
       Node::None = -1
 
       include Mem
@@ -25,19 +33,21 @@ module Konjak
         return [@text] if nodes.empty?
 
         segments = []
-        prev_text_index = 0
+        prev_text_end = 0
+        prev_segment = nil
         max_cost_path.each do |node|
           range     = node.range
-          segment   = node.segment
-          prev_text = @text[prev_text_index...range.begin]
+          segment   = select_next_segment(prev_segment, node)
+          prev_text = @text[prev_text_end...range.begin]
 
           segments << prev_text unless prev_text.empty?
 
           segments << SegmentString.new(@text[range.begin, range.size], segment)
 
-          prev_text_index = range.end
+          prev_segment    = segment
+          prev_text_end = range.end
         end
-        after_text = @text[prev_text_index..-1]
+        after_text = @text[prev_text_end..-1]
         segments << after_text unless after_text.empty?
         segments
       end
@@ -48,10 +58,11 @@ module Konjak
         costs      = nodes.map {|node| [node, node.range.size] }.to_h
 
         edges.each do |edge|
-          node, node2  = edge.prev, edge.current
+          node, node2    = edge.prev, edge.current
+          node2_cost     = costs[node2]
           new_node2_cost = costs[node] + calc_edge_cost(edge)
 
-          if costs[node2] < new_node2_cost
+          if node2_cost < new_node2_cost
             costs[node2] = new_node2_cost
             prev_nodes[node2] = node
           end
@@ -59,7 +70,7 @@ module Konjak
 
         node, _ = costs.max_by {|_, cost| cost }
 
-        max_cost_node_indexes = Enumerator.new {|y|
+        Enumerator.new {|y|
           loop do
             break if node == Node::None
             y << node
@@ -90,16 +101,18 @@ module Konjak
       def nodes
         return @nodes if @nodes
 
-        @nodes = []
+        @nodes = {}
 
         translation_units.each {|tu|
           segment = tu.variant(@lang).segment
           @text.scan(compile_pattern(segment)) {
-            @nodes << Node.new(($~.begin(0)...$~.end(0)), segment)
+            r = $~.begin(0)...$~.end(0)
+            @nodes[r] ||= []
+            @nodes[r] << segment
           }
         }
-
-        @nodes.sort_by! {|node| [node.range.begin, -node.segment.text.size] }
+        @nodes = @nodes.map {|(r, segments)| Node.new(r, segments) }
+        @nodes.sort!
 
         @nodes
       end
@@ -107,7 +120,8 @@ module Konjak
       def default_options
         {
           translation_unit_filter: -> (tu) { true },
-          calc_edge_cost:  -> (edge) { edge.current.size }
+          calc_edge_cost:  -> (edge) { edge.current.max_segment_size },
+          select_next_segment: -> (_, node) { node.segments.first }
         }
       end
 
@@ -119,6 +133,10 @@ module Konjak
         @options[:translation_unit_filter]
       end
 
+      def select_next_segment(prev_segment, node)
+        @options[:select_next_segment].call(prev_segment, node)
+      end
+
       def translation_units
         @translation_units ||= @tmx.body.translation_units.select(&translation_unit_filter)
       end

data/lib/konjak/version.rb

@@ -1,3 +1,3 @@
 module Konjak
-  VERSION = "0.0.21"
+  VERSION = "0.0.22"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: konjak
 version: !ruby/object:Gem::Version
-  version: 0.0.21
+  version: 0.0.22
 platform: ruby
 authors:
 - Seiei Higa
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-10-16 00:00:00.000000000 Z
+date: 2015-10-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport