text_alignment 0.4 → 0.4.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 35f83f76cd3e9ca59604327710d2eac4684b8abca9dbd167b2b391d98ea561f9
-  data.tar.gz: 77007a1bfcf72d0681ac9d21b35221e7d519a291383f3d0d8fa0574d631b65ea
+  metadata.gz: d56b44ee3cc3451b78ddb87134fefd6c043b4730f1170b07069a417bb162bad1
+  data.tar.gz: 67318d70bee5bb82127c98e79e0b45e5e337832a2b8d23469474215168579f20
 SHA512:
-  metadata.gz: 2f936569e072e6693f279b5d3a349965b286192d8f694fffbbf41110f66afbd5d80f865a9232c2bef60ec94920e0a814142c8756f0ba3ff29a109cbbe1f7abec
-  data.tar.gz: 955b85aa639f733361f354cc68ac3d8b14a17a14784f1e6ec24f240934ef0d160c6b8cf01232e1e4243c2eecffcaf1665999cc3e455bfb0e61c0dd905d2cb8b4
+  metadata.gz: 60ae2292556d15d860e0c7377be9c1fd5aec6741b26f562fee8c0c2a9a123d0b8c656b2f7f34b8f74f83ea9ac4878b53f17454e3b25d27d00cd98ecf8175a387
+  data.tar.gz: fe8fe545101cddb638c00ffb6f6785f8b2f5ddb109b4010739e399a2c50fe41164d791fe4a5599a76fca5553c494f78fd49630368d48e406a8211fc0516916f5

data/lib/text_alignment/anchor_finder.rb

@@ -1,18 +1,15 @@
 #!/usr/bin/env ruby
+require 'text_alignment/constants'
 require 'string-similarity'
 
 module TextAlignment; end unless defined? TextAlignment
 
-TextAlignment::SIZE_NGRAM = 8 unless defined? TextAlignment::SIZE_NGRAM
-TextAlignment::SIZE_WINDOW = 40 unless defined? TextAlignment::SIZE_WINDOW
-TextAlignment::TEXT_SIMILARITY_TRESHOLD = 0.8 unless defined? TextAlignment::TEXT_SIMILARITY_TRESHOLD
-
 class TextAlignment::AnchorFinder
 
 	def initialize(source_str, target_str, _size_ngram = nil, _size_window = nil, _text_similiarity_threshold = nil)
 		@size_ngram  = _size_ngram  || TextAlignment::SIZE_NGRAM
 		@size_window = _size_window || TextAlignment::SIZE_WINDOW
-		@sim_threshold = _size_window || TextAlignment::TEXT_SIMILARITY_TRESHOLD
+		@sim_threshold = _text_similiarity_threshold || TextAlignment::TEXT_SIMILARITY_THRESHOLD
 
 		@reverse = (target_str.length < source_str.length)
 

data/lib/text_alignment/approximate_fit.rb

@@ -1,13 +1,11 @@
 #!/usr/bin/env ruby
+require 'text_alignment/constants'
 require 'string-similarity'
 
 module TextAlignment; end unless defined? TextAlignment
 
 # approximate the location of str1 in str2
-TextAlignment::SIGNATURE_NGRAM = 7 unless defined? TextAlignment::SIGNATURE_NGRAM
 TextAlignment::MIN_LENGTH_FOR_APPROXIMATION = 50 unless defined? TextAlignment::MIN_LENGTH_FOR_APPROXIMATION
-TextAlignment::BUFFER_RATE = 0.1 unless defined? TextAlignment::BUFFER_RATE
-TextAlignment::TEXT_SIMILARITY_TRESHOLD = 0.7 unless defined? TextAlignment::TEXT_SIMILARITY_TRESHOLD
 
 class << TextAlignment
 
@@ -16,8 +14,8 @@ class << TextAlignment
 		raise ArgumentError, 'nil string' if str1.nil? || str2.nil?
 		return 0, str2.length if str2.length < TextAlignment::MIN_LENGTH_FOR_APPROXIMATION
 
-		ngram1 = (0 .. str1.length - TextAlignment::SIGNATURE_NGRAM).collect{|i| str1[i, TextAlignment::SIGNATURE_NGRAM]}
-		ngram2 = (0 .. str2.length - TextAlignment::SIGNATURE_NGRAM).collect{|i| str2[i, TextAlignment::SIGNATURE_NGRAM]}
+		ngram1 = (0 .. str1.length - TextAlignment::SIZE_NGRAM).collect{|i| str1[i, TextAlignment::SIZE_NGRAM]}
+		ngram2 = (0 .. str2.length - TextAlignment::SIZE_NGRAM).collect{|i| str2[i, TextAlignment::SIZE_NGRAM]}
 		ngram_shared = ngram1 & ngram2
 
 		# If there is no shared n-gram found, it may mean there is no serious overlap between the two strings
@@ -45,7 +43,7 @@ class << TextAlignment
 			text_similarity = text_similarity(str1, str2[fit_begin ... fit_end])
 			cache["#{fit_begin}-#{fit_end}"] = text_similarity
 
-			break if text_similarity > TextAlignment::TEXT_SIMILARITY_TRESHOLD
+			break if text_similarity > TextAlignment::TEXT_SIMILARITY_THRESHOLD
 			fit_begin, fit_end = nil, nil
 		end
 		return fit_begin, fit_end if fit_begin && fit_end && fit_begin < fit_end

data/lib/text_alignment/constants.rb

@@ -0,0 +1,7 @@
+module TextAlignment; end unless defined? TextAlignment
+
+TextAlignment::SIZE_NGRAM = 8 unless defined? TextAlignment::SIZE_NGRAM
+TextAlignment::SIZE_WINDOW = 60 unless defined? TextAlignment::SIZE_WINDOW
+TextAlignment::BUFFER_RATE = 0.1 unless defined? TextAlignment::BUFFER_RATE
+TextAlignment::BUFFER_MIN = 20 unless defined? TextAlignment::BUFFER_MIN
+TextAlignment::TEXT_SIMILARITY_THRESHOLD = 0.87 unless defined? TextAlignment::TEXT_SIMILARITY_THRESHOLD

data/lib/text_alignment/glcs_alignment_fast.rb

@@ -9,8 +9,6 @@ require 'text_alignment/mappings'
 
 module TextAlignment; end unless defined? TextAlignment
 
-TextAlignment::SIGNATURE_NGRAM = 5 unless defined? TextAlignment::SIGNATURE_NGRAM
-
 class TextAlignment::GLCSTextAlignment
 	attr_reader :position_map_begin, :position_map_end
 	attr_reader :common_elements, :mapped_elements

data/lib/text_alignment/text_alignment.rb

@@ -1,22 +1,22 @@
 #!/usr/bin/env ruby
+require 'text_alignment/constants'
 require 'text_alignment/anchor_finder'
 require 'text_alignment/mixed_alignment'
 
 module TextAlignment; end unless defined? TextAlignment
 
-TextAlignment::SIGNATURE_NGRAM = 7 unless defined? TextAlignment::SIGNATURE_NGRAM
-TextAlignment::BUFFER_RATE = 0.1 unless defined? TextAlignment::BUFFER_RATE
-TextAlignment::BUFFER_MIN = 20 unless defined? TextAlignment::BUFFER_MIN
-
-
 class TextAlignment::TextAlignment
 	attr_reader :block_alignments
 	attr_reader :similarity
 	attr_reader :lost_annotations
 
-	def initialize(str1, str2, mappings = nil)
+	def initialize(str1, str2, _size_ngram = nil, _size_window = nil, _text_similiarity_threshold = nil)
 		raise ArgumentError, "nil string" if str1.nil? || str2.nil?
 
+		size_ngram  = _size_ngram  || TextAlignment::SIZE_NGRAM
+		size_window = _size_window || TextAlignment::SIZE_WINDOW
+		sim_threshold = _text_similiarity_threshold || TextAlignment::TEXT_SIMILARITY_THRESHOLD
+
 		mappings ||= TextAlignment::MAPPINGS
 
 		# try exact match
@@ -26,7 +26,7 @@ class TextAlignment::TextAlignment
 			return @block_alignments
 		end
 
-		anchor_finder = TextAlignment::AnchorFinder.new(str1, str2)
+		anchor_finder = TextAlignment::AnchorFinder.new(str1, str2, size_ngram, size_window, sim_threshold)
 
 		# To collect matched blocks
 		mblocks = []

data/lib/text_alignment/version.rb

@@ -1,3 +1,3 @@
 class TextAlignment
-	VERSION = '0.4'
+	VERSION = '0.4.2'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: text_alignment
 version: !ruby/object:Gem::Version
-  version: '0.4'
+  version: 0.4.2
 platform: ruby
 authors:
 - Jin-Dong Kim
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-09-29 00:00:00.000000000 Z
+date: 2020-10-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ruby-dictionary
@@ -77,6 +77,7 @@ files:
 - lib/text_alignment.rb
 - lib/text_alignment/anchor_finder.rb
 - lib/text_alignment/approximate_fit.rb
+- lib/text_alignment/constants.rb
 - lib/text_alignment/find_divisions.rb
 - lib/text_alignment/glcs_alignment.rb
 - lib/text_alignment/glcs_alignment_fast.rb