konjak 0.0.14 → 0.0.15

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 7b6348fb462dc7f40f7199c009882b589b3013f1
-  data.tar.gz: fcca5beda91ef23b2386e22367697e76d5e37428
+  metadata.gz: 219a842656b4743ac6f782fdca2167dcf85e51ab
+  data.tar.gz: 2a1c74e8771791b5ac9dab000a7d556f88a14335
 SHA512:
-  metadata.gz: d46050f4b09475b2fb426b60551ce2a4f1dd9697878d131033e9ce53936336d7024a4ed102cdd307d6ca4533f0f40d134bd2781a931d11abc47f19bbc32be9d4
-  data.tar.gz: 8597872f8d75b6ad1c28e48e86aca55a727eb5f4237182af87d1eece9ddf7657831ad2b5463cdd67d2739e7fbcfc65c935858fddbb0b30f94a31591b769f613a
+  metadata.gz: f26891f463c259af8eee10698208a6df50c92b49c5b0f0a5065feadcccd10fe45733b8598eec5e47d35286feb519156aa8b82600576eeccd2d6cf39ecaf29daf
+  data.tar.gz: c88fe8b6fdbab45f4f5ddcc8c851869e47b62fc7bfc800676dba0f63352e193ffb453b15f95244eb364398b5c841629f2ce2d9df5b9edeca2a7b23e7c5c8caad

data/lib/konjak/body.rb

@@ -1,8 +1,10 @@
 module Konjak
   class Body < StructuralElement
+    TAG_NAME = 'body'
+
     # childrens
     def translation_units
-      children.select {|c| c.name == 'tu' }.map {|tu| TranslationUnit.new(tu) }
+      children.select {|c| c.name == TranslationUnit::TAG_NAME }.map! {|tu| TranslationUnit.new(tu) }
     end
 
     # methods

data/lib/konjak/element.rb

@@ -1,7 +1,10 @@
+require 'mem'
 require 'delegate'
 
 module Konjak
   class Element < Delegator
+    include Mem
+
     def initialize(__element__)
       @__element__ = __element__
     end

data/lib/konjak/header.rb

@@ -1,5 +1,7 @@
 module Konjak
   class Header < StructuralElement
+    TAG_NAME = 'header'
+
     # required attrs
     tmx_attr_accessor(:creation_tool,         :creationtool,        required: true)
     tmx_attr_accessor(:creation_tool_version, :creationtoolversion, required: true)
@@ -18,15 +20,15 @@ module Konjak
 
     # childrens
     def notes
-      children.select {|c| c.name == 'note' }.map {|n| Note.new(n) }
+      children.select {|c| c.name == Note::TAG_NAME }.map! {|n| Note.new(n) }
     end
 
     def user_defined_encodings
-      children.select {|c| c.name == 'ude' }.map {|n| UserDefinedEncoding.new(n) }
+      children.select {|c| c.name == UserDefinedEncoding::TAG_NAME }.map! {|n| UserDefinedEncoding.new(n) }
     end
 
     def properties
-      children.select {|c| c.name == 'prop' }.map {|n| Property.new(n) }
+      children.select {|c| c.name == Property::TAG_NAME }.map! {|n| Property.new(n) }
     end
 
     # methods

data/lib/konjak/highlight.rb

@@ -11,7 +11,7 @@ module Konjak
     #     Zero, one or more of the following elements: <bpt>, <ept>, <it>, <ph>, and <hi>.
     #     They can be in any order, except that each <bpt> element must have a subsequent corresponding <ept> element.
     def can_contain?(element)
-      [Text, BeginPairedTag, EndPairedTag, IsolatedTag, Placeholder, Hilight].any? {|c| c === element }
+      [String, BeginPairedTag, EndPairedTag, IsolatedTag, Placeholder, Hilight].any? {|c| c === element }
     end
   end
 end

data/lib/konjak/html_segmentor.rb

@@ -1,18 +1,19 @@
 module Konjak
   class HtmlSegmentor < Segmentor
+    SEGMENTS_PATTERNS = [
+      %r{<(?<start>p|h1|h2|h3|h4|h5|h6|li|title|td)>(.*?)</\k<start>>}m,
+      %r{<(?<start>p|h1|h2|h3|h4|h5|h6|li|title|td) [^>]*?>(.*?)</\k<start>>}m,
+      %r{<div>(.*?)</div>}m,
+      %r{<div [^>]*?>(.*?)</div>}m
+    ]
+
     def segments
       segments = [content.dup]
 
       begin
         size = segments.size
 
-        segments_patterns = [
-          %r{<(?<start>p|h1|h2|h3|h4|h5|h6|li|title|td)>(.*?)</\k<start>>}m,
-          %r{<(?<start>p|h1|h2|h3|h4|h5|h6|li|title|td) [^>]*?>(.*?)</\k<start>>}m,
-          %r{<div>(.*?)</div>}m,
-          %r{<div [^>]*?>(.*?)</div>}m
-        ]
-        segments_patterns.each do |pattern|
+        SEGMENTS_PATTERNS.each do |pattern|
           segments.map! do |s|
             s.partition(pattern)
           end

data/lib/konjak/map.rb

@@ -1,5 +1,7 @@
 module Konjak
   class Map < StructuralElement
+    TAG_NAME = 'map'
+
     # required attrs
     tmx_attr_accessor(:unicode, required: true)
 

data/lib/konjak/note.rb

@@ -1,17 +1,19 @@
 module Konjak
   class Note < StructuralElement
+    TAG_NAME = 'note'
+
     # optional attrs
     tmx_attr_accessor(:xml_lang,   :'xml:lang')
     tmx_attr_accessor(:o_encoding, :"o-encoding")
 
     # childrens
     def text
-      Text.new(super)
+      super
     end
 
     # methods
     def can_contain?(element)
-      Text === element
+      String === element
     end
   end
 end

data/lib/konjak/polytex_segmentor.rb

@@ -1,31 +1,33 @@
 module Konjak
   class PolytexSegmentor < Segmentor
+
+    SEGMENTS_PATTERNS = [
+      /\\begin\{(?<start>[^\}]+)\}([\n.]*?)\\end\{\k<start>\}/m,
+      /(?<=\\chapter\{)[^\}]+(?=\})/,
+      /(?<=\\section\{)[^\}]+(?=\})/,
+      /(?<=\\subsection\{)[^\}]+(?=\})/,
+      /\\footnote\{(?<gr>\\(?!footnote)[^\{]+\{[^\}]+\}(?:\{[^\}]+\})?\g<gr>|[^{])+\}/m,
+      /(?<=\\footnote\{)(?<gr>\\(?!footnote)[^\{]+\{[^\}]+\}(?:\{[^\}]+\})?\g<gr>|[^{])+(?=\})/m,
+      /(?<=\\codecaption\{).+(?= \\|\}$)/,
+      /(?<=\\caption\{).+(?=\\label\{.*\}\}$)/,
+      /(?<=\n)^.*$(?=\n)/m,
+      /# .*$/,
+      /(?<=^).+?[\.\?\!](?= |\n|\t)/,
+      /(?<=\()[^\.\n]+[\.\?\!](?=\))/,
+      /^ (?=[\w\\]+)/,
+      /^\s+% .*$/,
+      /^$/,
+      /\\noindent /,
+      /\\item /,
+    ]
+
     def segments
       segments = [content.dup]
 
       begin
         size = segments.size
 
-        segments_patterns = [
-          /\\begin\{(?<start>[^\}]+)\}([\n.]*?)\\end\{\k<start>\}/m,
-          /(?<=\\chapter\{)[^\}]+(?=\})/,
-          /(?<=\\section\{)[^\}]+(?=\})/,
-          /(?<=\\subsection\{)[^\}]+(?=\})/,
-          /\\footnote\{(?<gr>\\(?!footnote)[^\{]+\{[^\}]+\}(?:\{[^\}]+\})?\g<gr>|[^{])+\}/m,
-          /(?<=\\footnote\{)(?<gr>\\(?!footnote)[^\{]+\{[^\}]+\}(?:\{[^\}]+\})?\g<gr>|[^{])+(?=\})/m,
-          /(?<=\\codecaption\{).+(?= \\|\}$)/,
-          /(?<=\\caption\{).+(?=\\label\{.*\}\}$)/,
-          /(?<=\n)^.*$(?=\n)/m,
-          /# .*$/,
-          /(?<=^).+?[\.\?\!](?= |\n|\t)/,
-          /(?<=\()[^\.\n]+[\.\?\!](?=\))/,
-          /^ (?=[\w\\]+)/,
-          /^\s+% .*$/,
-          /^$/,
-          /\\noindent /,
-          /\\item /,
-        ]
-        segments_patterns.each do |pattern|
+        SEGMENTS_PATTERNS.each do |pattern|
           segments.map! do |s|
             s.partition(pattern)
           end
@@ -37,4 +39,5 @@ module Konjak
       segments
     end
   end
+
 end

data/lib/konjak/property.rb

@@ -1,5 +1,7 @@
 module Konjak
   class Property < StructuralElement
+    TAG_NAME = 'prop'
+
     # required attrs
     tmx_attr_accessor(:type, required: true)
 
@@ -9,14 +11,14 @@ module Konjak
 
     # childrens
     def text
-      Text.new(super)
+      super
     end
 
     # methods
     def can_contain?(element)
       # FIXME
       #    Tool-specific data or text.
-      Text === element
+      String === element
     end
 
     def unpublished?

data/lib/konjak/segment/gtt.rb

@@ -18,7 +18,6 @@ module Konjak
         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

@@ -4,26 +4,29 @@ require 'mem'
 module Konjak
   # container
   class Segment < StructuralElement
+    TAG_NAME = 'seg'
+    WHITE_SPACE_PATTERN_TEXT = '\s'
+    POSSESSIVE_QUALIFIER = '++'
+
     include GTT
     include Mem
 
     # children
     def text
-      Text.new(super)
+      super
     end
 
     # methods
     def can_contain?(element)
-      [Text, BeginPairedTag, EndPairedTag, IsolatedTag, Placeholder, Highlight].any? {|c| c === element }
+      [String, BeginPairedTag, EndPairedTag, IsolatedTag, Placeholder, Highlight].any? {|c| c === element }
     end
 
     def compile_pattern
       regexp = Regexp.escape(text)
-      regexp.gsub!(/(?<!^)\\\s/)          { '\s' }
-      regexp.gsub!(/(?<!^)(?:\\s)+(?!$)/) {|s| s + '++' }
+      regexp.gsub!(/(?<!^)\\\s/)          { WHITE_SPACE_PATTERN_TEXT }
+      regexp.gsub!(/(?<!^)(?:\\s)+(?!$)/) {|s| s + POSSESSIVE_QUALIFIER }
       Regexp.compile(regexp)
     end
-    memoize :compile_pattern
 
     def translation_unit
       TranslationUnit.new(translation_unit_variant.parent)

data/lib/konjak/sub_flow.rb

@@ -12,7 +12,7 @@ module Konjak
     #    They can be in any order, except that each <bpt> element must have a subsequent corresponding <ept> element.
 
     def can_contain?(element)
-      [Text, BeginPairedTag, EndPairedTag, IsolatedTag, Placeholder, Hilight].any? {|c| c === element }
+      [String, BeginPairedTag, EndPairedTag, IsolatedTag, Placeholder, Hilight].any? {|c| c === element }
     end
   end
 end

data/lib/konjak/tmx.rb

@@ -7,12 +7,12 @@ module Konjak
 
     # required element
     def header
-      Header.new(root.at_xpath('header'))
+      Header.new(root.at_xpath(Header::TAG_NAME))
     end
 
     # required element
     def body
-      Body.new(root.at_xpath('body'))
+      Body.new(root.at_xpath(Body::TAG_NAME))
     end
 
     # FIXME

data/lib/konjak/tmx_segmentor/strategy.rb

@@ -18,11 +18,13 @@ module Konjak
         translation_units.each do |translation_unit|
           segment = translation_unit.variant(@lang).segment
 
+          pat = compile_pattern(segment)
+
           segments.map! {|text|
             next text if text.length < min_segment_length
             next text if text.is_a?(SegmentString)
 
-            split(segment, text)
+            split(pat, segment, text)
           }.flatten!
         end
         segments
@@ -38,13 +40,16 @@ module Konjak
         @options[:min_segment_length]
       end
 
-      def split(segment, text)
+      def split(pat, segment, text)
         texts = []
         while true
           break if text.length < min_segment_length
 
-          head, match, tail = text.partition(compile_pattern(segment))
-          break if match.empty?
+          break unless text =~ pat
+
+          head  = $`
+          match = $&
+          tail  = $'
 
           texts << head unless head.empty?
 
@@ -62,7 +67,6 @@ module Konjak
           tu.variant(@lang).segment.text.length < min_segment_length
         }
       end
-      memoize :translation_units
     end
   end
 end

data/lib/konjak/translation_unit.rb

@@ -1,5 +1,7 @@
 module Konjak
   class TranslationUnit < StructuralElement
+    TAG_NAME = 'tu'
+
     # optional attrs
     tmx_attr_accessor(:tuid)
     tmx_attr_accessor(:o_encoding,            :"o-encoding")
@@ -19,8 +21,9 @@ module Konjak
 
     # childrens
     def variants
-      children.select {|c| c.name == 'tuv' }.map {|tuv| TranslationUnitVariant.new(tuv) }
+      children.select {|c| c.name == TranslationUnitVariant::TAG_NAME }.map! {|tuv| TranslationUnitVariant.new(tuv) }
     end
+    memoize :variants
 
     # methods
     def can_contain?(element)

data/lib/konjak/translation_unit_variant.rb

@@ -1,5 +1,7 @@
 module Konjak
   class TranslationUnitVariant < StructuralElement
+    TAG_NAME = 'tuv'
+
     # required attrs
     tmx_attr_accessor(:xml_lang, :'xml:lang', required: true)
 
@@ -18,16 +20,17 @@ module Konjak
 
     # childrens
     def notes
-      children.select {|c| c.name == 'note' }.map {|n| Note.new(n) }
+      children.select {|c| c.name == 'note' }.map! {|n| Note.new(n) }
     end
 
     def properties
-      children.select {|c| c.name == 'prop' }.map {|n| Property.new(n) }
+      children.select {|c| c.name == 'prop' }.map! {|n| Property.new(n) }
     end
 
     def segment
-      Segment.new(children.detect {|c| c.name == 'seg' })
+      Segment.new(children.detect {|c| c.name == Segment::TAG_NAME })
     end
+    memoize :segment
 
     # methods
 

data/lib/konjak/user_defined_encoding.rb

@@ -1,5 +1,7 @@
 module Konjak
   class UserDefinedEncoding < StructuralElement
+    TAG_NAME = 'ude'
+
     # required attrs
     tmx_attr_accessor(:name, required: true)
 
@@ -10,7 +12,7 @@ module Konjak
 
     # childrens
     def maps
-      children.select {|c| c.name == 'map' }.map {|n| Map.new(n) }
+      children.select {|c| c.name == Map::TAG_NAME }.map! {|n| Map.new(n) }
     end
 
     # methods

data/lib/konjak/version.rb

@@ -1,3 +1,3 @@
 module Konjak
-  VERSION = "0.0.14"
+  VERSION = "0.0.15"
 end

data/lib/konjak.rb

@@ -3,7 +3,6 @@ require 'konjak/parser'
 
 # not elements
 require 'konjak/code_data'
-require 'konjak/text'
 
 # elements
 require 'konjak/element'

data/spec/konjak_parse_spec.rb

@@ -38,7 +38,7 @@ describe Konjak do
 
         its(:xml_lang) { is_expected.to eq 'en' }
         its(:o_encoding) { is_expected.to eq 'iso-8859-1' }
-        its(:text) { is_expected.to be_instance_of Konjak::Text }
+        its(:text) { is_expected.to be_instance_of String }
 
         describe 'text' do
           subject { super().text }
@@ -90,7 +90,7 @@ describe Konjak do
         its(:xml_lang)   { is_expected.to eq 'en' }
         its(:o_encoding) { is_expected.to eq 'iso-8859-1' }
         its(:type)       { is_expected.to eq 'RTFPreamble' }
-        its(:text)       { is_expected.to be_instance_of Konjak::Text }
+        its(:text)       { is_expected.to be_instance_of String }
 
         describe '.text' do
           subject { super().text }
@@ -141,7 +141,7 @@ describe Konjak do
           describe '.segment' do
             subject { super().segment }
 
-            its(:text) { is_expected.to be_instance_of Konjak::Text }
+            its(:text) { is_expected.to be_instance_of String }
 
             describe '.text' do
               subject { super().text }

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: konjak
 version: !ruby/object:Gem::Version
-  version: 0.0.14
+  version: 0.0.15
 platform: ruby
 authors:
 - Seiei Higa
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2015-06-25 00:00:00.000000000 Z
+date: 2015-08-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mem
@@ -161,7 +161,6 @@ files:
 - lib/konjak/segmentor.rb
 - lib/konjak/structural_element.rb
 - lib/konjak/sub_flow.rb
-- lib/konjak/text.rb
 - lib/konjak/tmx.rb
 - lib/konjak/tmx_segmentor.rb
 - lib/konjak/tmx_segmentor/gtt_html_strategy.rb
@@ -216,4 +215,3 @@ test_files:
 - spec/konjak_translate_spec.rb
 - spec/spec_helper.rb
 - spec/support/equal_xml_matcher.rb
-has_rdoc: 

data/lib/konjak/text.rb

@@ -1,4 +0,0 @@
-module Konjak
-  class Text < String
-  end
-end