text_alignment 0.3.16 → 0.3.21

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 61f98e83ee9c1d220dd228be6bb708b79f36d1c691f04dcb14d4af55f398b6da
-  data.tar.gz: f692e98a27a555baab2797ebe37407ad7133916db172a977e95415b9004e471c
+  metadata.gz: 65e1d9b45ff59ac0a233b7656d2aca99d7e4e1051b1a03a0c7726521d4f2b280
+  data.tar.gz: 710a3b68c5263f26572727e6e9591ebd5fdb095af4633bd5037c61eae0bb5cb6
 SHA512:
-  metadata.gz: 2dd8f865c245601c362e335df4e26413501fa682a97010b8aebd3ebc01864ae4772f6e716725331f9c6bc8f688818d665ef7a21384906211efc0e630b46f2313
-  data.tar.gz: 4c43199f474b94c825d8ec8ca2085b06107ec34df3d5ee988294f7423caef317c893bbba9879637d197fb2e7ae426c9e43c67cf042b7339959b638d5e5f60d01
+  metadata.gz: 598df22e41bbbe0a84b6e1a6a4e631ab0d8166810afd652086595feecbf0808a886685f42e5466626cbb1d6950dd9f1181be776b9938d6174dc7735c3ace24cd
+  data.tar.gz: f7dedfb7e64919129f816fbba24dbd1c2e2a056c242a0865915b8a611f594399b17d051d004a846796bba1c2e89c6fb2f17116cd118ca6217cf1a5dff4f6d4d8

data/bin/align_annotations

@@ -103,6 +103,8 @@ target_annotations = if source_annotations.class == Array
 else
 	alignment = TextAlignment::TextAlignment.new(source_annotations[:text], target_text)
 
+	pp alignment
+
 	# verification
 	source_text = source_annotations[:text]
 	puts "=====BEGIN"
@@ -129,9 +131,6 @@ else
 	puts
 	puts "=====END"
 
-	pp alignment
-
-	exit
 	# alignment.block_alignments.each do |a|
 	# 	if a[:alignment].nil? || a[:alignment] == :empty
 	# 		# p [a[:source], a[:target]]

data/lib/text_alignment/anchor_finder.rb

@@ -3,7 +3,7 @@ require 'string-similarity'
 
 module TextAlignment; end unless defined? TextAlignment
 
-TextAlignment::SIZE_NGRAM = 10 unless defined? TextAlignment::SIZE_NGRAM
+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
 
@@ -32,11 +32,11 @@ class TextAlignment::AnchorFinder
 		while @beg_s1 < (@s1.length - @size_ngram)
 			anchor = @s1[@beg_s1, @size_ngram]
 
-			search_position = 0
-			# search_position = @end_s2_prev
+			# search_position = 0
+			search_position = @end_s2_prev
 			while @beg_s2 = @s2.index(anchor, search_position)
 				# if both the begining points are sufficiantly close to the end points of the last match
-				break if @end_s1_prev && (@beg_s1 - @end_s1_prev < 5) && (@beg_s2 >= @end_s2_prev) && (@beg_s2 - @end_s2_prev < 5)
+				break if @beg_s1 > 0 && @beg_s2 > 0 && (@beg_s1 - @end_s1_prev < 5) && (@beg_s2 >= @end_s2_prev) && (@beg_s2 - @end_s2_prev < 5)
 
 				left_window_s1, left_window_s2 = get_left_windows
 				break if left_window_s1 && (text_similarity(left_window_s1, left_window_s2) > TextAlignment::TEXT_SIMILARITY_TRESHOLD)

data/lib/text_alignment/text_alignment.rb

@@ -40,17 +40,20 @@ class TextAlignment::TextAlignment
 			end
 		end
 
-		mblocks.each do |b|
-			p [b[:source], b[:target]]
-			puts "---"
-			puts str1[b[:source][:begin] ... b[:source][:end]]
-			puts "---"
-			puts str2[b[:target][:begin] ... b[:target][:end]]
-			puts "====="
-			puts
-		end
-		puts "-=-=-=-=-"
-		puts
+		# pp mblocks
+		# puts "-----"
+		# puts
+		# mblocks.each do |b|
+		# 	p [b[:source], b[:target]]
+		# 	puts "---"
+		# 	puts str1[b[:source][:begin] ... b[:source][:end]]
+		# 	puts "---"
+		# 	puts str2[b[:target][:begin] ... b[:target][:end]]
+		# 	puts "====="
+		# 	puts
+		# end
+		# puts "-=-=-=-=-"
+		# puts
 
 		## To find block alignments
 		@block_alignments = []
@@ -78,6 +81,8 @@ class TextAlignment::TextAlignment
 
 						@block_alignments << {source:{begin:0, end:b1}, target:{begin:0, end:b2}, alignment: :empty} if b1 > 0
 
+						_str1 = str1[b1 ... e1]
+						_str2 = str2[b2 ... e2]
 						alignment = TextAlignment::MixedAlignment.new(_str1.downcase, _str2.downcase, mappings)
 						if alignment.similarity < 0.6
 							@block_alignments << {source:{begin:b1, end:e1}, target:{begin:0, end:e2}, alignment: :empty}
@@ -113,24 +118,11 @@ class TextAlignment::TextAlignment
 		end
 
 		# Final step
-		if mblocks[-1][:source][:end] < str1.length
-			b1 = mblocks[-1][:source][:end]
-			b2 = mblocks[-1][:target][:end]
-
-			if mblocks[-1][:target][:end] < str2.length
-
-			else
-				e1 = str1.length
-				e2 = str2.length
-				@block_alignments << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}
-			end
-		end
-
 		if  mblocks[-1][:source][:end] < str1.length && mblocks[-1][:target][:end] < str2.length
 			b1 = mblocks[-1][:source][:end]
 			b2 = mblocks[-1][:target][:end]
-			_str1 = str1[b1 ... -1]
-			_str2 = str2[b2 ... -1]
+			_str1 = str1[b1 ... str1.length]
+			_str2 = str2[b2 ... str2.length]
 
 			unless _str1.strip.empty?
 				if _str2.strip.empty?
@@ -174,9 +166,8 @@ class TextAlignment::TextAlignment
 				nil
 			end
 		else
-			p begin_position
-			puts "-----"
-			block_alignment[:alignment].transform_begin_position(begin_position - block_alignment[:source][:begin]) + block_alignment[:target][:begin]
+			r = block_alignment[:alignment].transform_begin_position(begin_position - block_alignment[:source][:begin])
+			r.nil? ? nil : r + block_alignment[:target][:begin]
 		end
 	end
 
@@ -194,7 +185,8 @@ class TextAlignment::TextAlignment
 				nil
 			end
 		else
-			block_alignment[:alignment].transform_end_position(end_position - block_alignment[:source][:begin]) + block_alignment[:target][:begin]
+			r = block_alignment[:alignment].transform_end_position(end_position - block_alignment[:source][:begin])
+			r.nil? ? nil : r + block_alignment[:target][:begin]
 		end
 	end
 

data/lib/text_alignment/version.rb

@@ -1,3 +1,3 @@
 class TextAlignment
-	VERSION = '0.3.16'
+	VERSION = '0.3.21'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: text_alignment
 version: !ruby/object:Gem::Version
-  version: 0.3.16
+  version: 0.3.21
 platform: ruby
 authors:
 - Jin-Dong Kim
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-09-17 00:00:00.000000000 Z
+date: 2020-09-22 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ruby-dictionary