text_alignment 0.6 → 0.6.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: dc820991f5f694f154b94c369158909ccba3760829e0d881c7fd2e6ef7ddd149
-  data.tar.gz: 40ae6f2e388405426a77682bd1a3fb7a3c853076eced9b7301b632081dfd0a57
+  metadata.gz: fb5dd06236d0b1a8a9c8c5fcb92807a62bdd30e0648bcbd636b95b2a8a45b9b4
+  data.tar.gz: 9266b852993bfee999daa92e3f38ec93e2aec77171fee27c1fea6ac2a17e4d23
 SHA512:
-  metadata.gz: 5802241b4a8394d3c570c1d4b8f5e1d7706c72852e2d6e6fb23bda2f6e2972fa09f7001695db026667144e2af982eeb91ed0b700bd8151af6df794c98e3c069b
-  data.tar.gz: 8d7c93acbef6ab12bb2a0291444a7bcc73b0236bb5b0d06d274e95aa30c9ffc829965653b58270686147a9ac30ccf570518b3ad266120b320dfb20cd1620f5f9
+  metadata.gz: 7ee2a590fb31bcc27121a4a227d7fcefe2e8e80646bea3898bb86729ca3ca299e0aebcf23bea30e2391687e6ec0d6573c04a4605f728562482c7edbd0c0285e0
+  data.tar.gz: 73612c185fe533b0daa22d44e7776ed610025cb1bd874f05d95761079f95d1e8a06ead68c88b84bab4d33e8a676edff1e98880912254d9a7ecb5c4ead5eb01fb

data/lib/text_alignment/text_alignment.rb

@@ -91,8 +91,9 @@ class TextAlignment::TextAlignment
 						_str1 = str1[b1 ... e1]
 						_str2 = str2[b2 ... e2]
 						alignment = TextAlignment::MixedAlignment.new(_str1.downcase, _str2.downcase, mappings)
-						if alignment.similarity < 0.6
-							@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:0, end:e2}, alignment: :empty}
+						similarity = alignment_similarity(_str1, _str2, alignment)
+						if similarity < 0.6
+							@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:0, end:e2}, alignment: :empty, similarity: similarity}
 						else
 							@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:0, end:e2}, alignment:alignment}
 						end
@@ -114,8 +115,9 @@ class TextAlignment::TextAlignment
 					@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}
 				else
 					alignment = TextAlignment::MixedAlignment.new(_str1.downcase, _str2.downcase, mappings)
-					if alignment.similarity < 0.6
-						@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}
+					similarity = alignment_similarity(_str1, _str2, alignment)
+					if similarity < 0.6
+						@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty, similarity: similarity}
 					else
 						@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment:alignment}
 					end
@@ -143,8 +145,9 @@ class TextAlignment::TextAlignment
 					_str2 = str2[b2 ... e2]
 
 					alignment = TextAlignment::MixedAlignment.new(_str1.downcase, _str2.downcase, mappings)
-					if alignment.similarity < 0.6
-						@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}
+					similarity = alignment_similarity(_str1, _str2, alignment)
+					if similarity < 0.6
+						@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty, similarity: similarity}
 					else
 						@block_alignment[:blocks] << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment:alignment}
 					end
@@ -250,6 +253,7 @@ class TextAlignment::TextAlignment
 				"===== common =====\n" +
 				stext[a[:source][:begin] ... a[:source][:end]] + "\n\n"
 			when :empty
+				"xxxxx disparate texts (similarity: #{a[:similarity]})\n" +
 				"<<<<< string 1\n" +
 				stext[a[:source][:begin] ... a[:source][:end]] + "\n\n" +
 				">>>>> string 2\n" +
@@ -316,15 +320,15 @@ class TextAlignment::TextAlignment
 		pletters = TextAlignment::PADDING_LETTERS
 
 		# find the padding letter for str1
-		padding_letter1 = begin
+		@padding_letter1 = begin
 			i = pletters.index{|l| str2.index(l).nil?}
 			raise RuntimeError, "Could not find a padding letter for str1" if i.nil?
 			TextAlignment::PADDING_LETTERS[i]
 		end
 
 		# find the padding letter for str2
-		padding_letter2 = begin
-			i = pletters.index{|l| l != padding_letter1 && str1.index(l).nil?}
+		@padding_letter2 = begin
+			i = pletters.index{|l| l != @padding_letter1 && str1.index(l).nil?}
 			raise RuntimeError, "Could not find a padding letter for str2" if i.nil?
 			TextAlignment::PADDING_LETTERS[i]
 		end
@@ -335,12 +339,12 @@ class TextAlignment::TextAlignment
 			from = f[1]
 
 			if str2.index(f[0])
-				to   = f[0] + (padding_letter1 * (f[1].length - 1))
+				to   = f[0] + (@padding_letter1 * (f[1].length - 1))
 				str1.gsub!(from, to)
 			end
 
 			if str1.index(f[0])
-				to   = f[0] + (padding_letter2 * (f[1].length - 1))
+				to   = f[0] + (@padding_letter2 * (f[1].length - 1))
 				str2.gsub!(from, to)
 			end
 		end
@@ -349,4 +353,13 @@ class TextAlignment::TextAlignment
 		[str1, str2, mappings]
 	end
 
+	def alignment_similarity(_s1, _s2, alignment)
+		# compute the lcs only with non-whitespace letters
+		lcs = alignment.sdiff.count{|d| d.action == '=' && d.old_element =~ /\S/ && d.new_element =~ /\S/}
+
+		s1 = _s1.tr(@padding_letter1, ' ')
+		s2 = _s2.tr(@padding_letter2, ' ')
+		similarity  = 2 * lcs / (s1.scan(/\S/).count + s2.scan(/\S/).count).to_f
+	end
+
 end

data/lib/text_alignment/version.rb

@@ -1,3 +1,3 @@
 class TextAlignment
-	VERSION = '0.6'
+	VERSION = '0.6.1'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: text_alignment
 version: !ruby/object:Gem::Version
-  version: '0.6'
+  version: 0.6.1
 platform: ruby
 authors:
 - Jin-Dong Kim