text_alignment 0.4.3 → 0.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 52bc948955e2df858e397b14eabb4411f73b3ff1e4d879ff4b7015d3b5e03308
-  data.tar.gz: fd20caec51c95bdc475e0698a52bb7fdebc9e22c43bb47267a883bcc75862268
+  metadata.gz: 9f5f7f27c8628123530d51d0a68060aa6fb850bcef8c7089c8bf990f7257a80b
+  data.tar.gz: 45f768df4e7d89c931295985adb31df9e725156e5a85d7f78a5b7cd26d00be4d
 SHA512:
-  metadata.gz: dbcb7ab70a64d4a398a5c5761cc5b2f5de6835ccc0e2d0854556f03ef91d0c0294986cc2ff1273788e6b7b0c73dfdf86fd16ee1ef8ce35ecc11d61f8eaab9521
-  data.tar.gz: 01d21cdcc0ab81d61e08ff1f52360ba35973756fd5060ce866391ff622d4cf87da945dba43a622a0581d63ee96c8723e1cb28991bfa02f4e1e803896bdc64d7f
+  metadata.gz: 4004293fa10eb247110764d16a24900a10e714072227dd6e5626d2123deca4bcec00b7255e1affaba4d7dda75d7e42049aabf6b1e6c51f4c005124443b5f9ffc
+  data.tar.gz: eb5c8a2c89c8973242bb77457e0f8d9922486d9743e7c39ee12ea4025ece6d888be5308eaddec076e5a21c84149dcb31bc0ed6fcd68881fb6c8c4063e49fb64d

data/bin/align_annotations

@@ -103,65 +103,56 @@ target_annotations = if source_annotations.class == Array
 else
 	alignment = TextAlignment::TextAlignment.new(source_annotations[:text], target_text)
 
-	pp alignment
+	# pp alignment
 
 	# verification
-	source_text = source_annotations[:text]
-	puts "=====BEGIN"
-	(0 ... source_text.rstrip.length).each do |p|
-		t = alignment.transform_begin_position(p)
-		if t.nil?
-			print source_text[p]
-		else
-			print '.'
-		end
-	end
-	puts
-	puts "=====END"
-
-	puts "=====BEGIN"
-	(0 .. source_text.rstrip.length).each do |p|
-		t = alignment.transform_end_position(p)
-		if t.nil?
-			print source_text[p]
-		else
-			print '.'
-		end
-	end
-	puts
-	puts "=====END"
-
-	# alignment.block_alignments.each do |a|
-	# 	if a[:alignment].nil? || a[:alignment] == :empty
-	# 		# p [a[:source], a[:target]]
-	# 		# p a[:alignment]
+	# source_text = source_annotations[:text]
+	# puts "=====BEGIN"
+	# (0 ... source_text.rstrip.length).each do |p|
+	# 	t = alignment.transform_begin_position(p)
+	# 	if t.nil?
+	# 		print source_text[p]
+	# 	else
+	# 		print '.'
+	# 	end
+	# end
+	# puts
+	# puts "=====END"
+
+	# puts "=====BEGIN"
+	# (0 .. source_text.rstrip.length).each do |p|
+	# 	t = alignment.transform_end_position(p)
+	# 	if t.nil?
+	# 		print source_text[p]
 	# 	else
-	# 		p [a[:source], a[:target]]
-	# 		p a[:alignment].similarity
-	# 		puts "--"
-	# 		puts source_annotations[:text][a[:source][:begin] ... a[:source][:end]]
-	# 		puts "--"
-	# 		puts target_text[a[:target][:begin] ... a[:target][:end]]
-	# 		puts "======"
+	# 		print '.'
 	# 	end
 	# end
+	# puts
+	# puts "=====END"
+
+	source_text = source_annotations[:text]
+
+	# puts "[block alignment]"
+	puts alignment.alignment_table
+	# puts "====="
 	# exit
 
 	# verification of source denotations
-	puts "[Invalid source denotations]"
-	source_annotations[:denotations] do |d|
-		p d unless d[:span][:begin] && d[:span][:end] && d[:span][:begin] < d[:span][:end] && d[:span][:begin] >= 0 && d[:span][:end] < source_text.length
-	end
-	puts "====="
-	puts
+	# puts "[Invalid source denotations]"
+	# source_annotations[:denotations] do |d|
+	# 	p d unless d[:span][:begin] && d[:span][:end] && d[:span][:begin] < d[:span][:end] && d[:span][:begin] >= 0 && d[:span][:end] < source_text.length
+	# end
+	# puts "====="
+	# puts
 
 	denotations = alignment.transform_hdenotations(source_annotations[:denotations])
-	puts "[Invalid transformation]"
-	denotations.each do |d|
-		p d unless d[:span][:begin] && d[:span][:end] && d[:span][:begin] < d[:span][:end] && d[:span][:begin] >= 0 && d[:span][:end] < target_text.length
-	end
-	puts "====="
-	puts
+	# puts "[Invalid transformation]"
+	# denotations.each do |d|
+	# 	p d unless d[:span][:begin] && d[:span][:end] && d[:span][:begin] < d[:span][:end] && d[:span][:begin] >= 0 && d[:span][:end] < target_text.length
+	# end
+	# puts "====="
+	# puts
 
 	lost_annotations += alignment.lost_annotations if alignment.lost_annotations
 

data/lib/text_alignment/text_alignment.rb

@@ -15,6 +15,9 @@ class TextAlignment::TextAlignment
 	def initialize(_str1, _str2, _size_ngram = nil, _size_window = nil, _text_similiarity_threshold = nil)
 		raise ArgumentError, "nil string" if _str1.nil? || _str2.nil?
 
+		@ostr1 = _str1
+		@ostr2 = _str2
+
 		str1, str2, mappings = string_preprocessing(_str1, _str2)
 
 		# try exact match
@@ -24,6 +27,13 @@ class TextAlignment::TextAlignment
 			return @block_alignments
 		end
 
+		# try exact match
+		block_begin = str2.downcase.index(str1.downcase)
+		unless block_begin.nil?
+			@block_alignments = [{source:{begin:0, end:str1.length}, target:{begin:block_begin, end:block_begin + str1.length}, delta:block_begin}]
+			return @block_alignments
+		end
+
 		anchor_finder = TextAlignment::AnchorFinder.new(str1, str2, _size_ngram, _size_window, _text_similiarity_threshold)
 
 		# To collect matched blocks
@@ -91,7 +101,7 @@ class TextAlignment::TextAlignment
 				end
 			end
 		end
-		@block_alignments << mblocks[0]
+		@block_alignments << mblocks[0].merge(alignment: :block)
 
 		(1 ... mblocks.length).each do |i|
 			b1 = mblocks[i - 1][:source][:end]
@@ -112,7 +122,7 @@ class TextAlignment::TextAlignment
 					end
 				end
 			end
-			@block_alignments << mblocks[i]
+			@block_alignments << mblocks[i].merge(alignment: :block)
 		end
 
 		# Final step
@@ -154,7 +164,7 @@ class TextAlignment::TextAlignment
 		i = @block_alignments.index{|b| b[:source][:end] > begin_position}
 		block_alignment = @block_alignments[i]
 
-		b = if block_alignment[:alignment].nil?
+		b = if block_alignment[:alignment] == :block
 			begin_position + block_alignment[:delta]
 		elsif block_alignment[:alignment] == :empty
 			if begin_position == block_alignment[:source][:begin]
@@ -173,7 +183,7 @@ class TextAlignment::TextAlignment
 		i = @block_alignments.index{|b| b[:source][:end] >= end_position}
 		block_alignment = @block_alignments[i]
 
-		e = if block_alignment[:alignment].nil?
+		e = if block_alignment[:alignment] == :block
 			end_position + block_alignment[:delta]
 		elsif block_alignment[:alignment] == :empty
 			if end_position == block_alignment[:source][:end]
@@ -230,8 +240,126 @@ class TextAlignment::TextAlignment
 		r
 	end
 
-	private
+	def alignment_table
+		table = <<-TABLE
+			<table class='text_alignment_table'>
+				<thead>
+					<tr>
+						<th class='text_alignment_left' style='width:50%'>Text 1</th>
+						<th class='text_alignment_rigt'>Text 2</th>
+					</tr>
+				</thead>
+				<tbody>
+		TABLE
+
+		@block_alignments.each do |a|
+			table += alignment_table_th(a)
+			table += "<tr>\n" + case a[:alignment]
+			when :block
+				"<td colspan='2' class='text_alignment_common'>" +
+				@ostr1[a[:source][:begin] ... a[:source][:end]] +
+				"</td>\n"
+			when :empty
+				"<td class='text_alignment_left'>"  + @ostr1[a[:source][:begin] ... a[:source][:end]] + "</td>\n" +
+				"<td class='text_alignment_right'>" + @ostr2[a[:target][:begin] ... a[:target][:end]] + "</td>\n"
+			else
+				base = a[:source][:begin]
+				astr1 = a[:alignment].sdiff.map do |c|
+					case c.action
+					when '='
+						@ostr1[c.old_position + base]
+					when '+'
+						'_'
+					when '-'
+						@ostr1[c.old_position + base]
+					when '!'
+						@ostr1[c.old_position + base] + '_'
+					end
+				end.join('')
+
+				base = a[:target][:begin]
+				astr2 = a[:alignment].sdiff.map do |c|
+					case c.action
+					when '='
+						@ostr2[c.new_position + base]
+					when '+'
+						@ostr2[c.new_position + base]
+					when '-'
+						'_'
+					when '!'
+						'_' + @ostr2[c.new_position + base]
+					end
+				end.join('')
+
+				"<td class='text_alignment_left'>"  + astr1 + "</td>\n" +
+				"<td class='text_alignment_right'>" + astr2 + "</td>\n"
+			end + "</tr>\n"
+		end
+		table += '</tbody></table>'
+	end
+
+	def alignment_table_th(a)
+		"<tr>" +
+		"<th class='text_alignment_left'>#{a[:source][:begin]} - #{a[:source][:end]}</th>" +
+		"<th class='text_alignment_right'>#{a[:target][:begin]} - #{a[:target][:end]}</th>" +
+		"</tr>"
+	end
+
+	def alignment_show
+		show = ''
+		@block_alignments.each do |a|
+			show += case a[:alignment]
+			when :block
+				"===== common =====\n" +
+				@ostr1[a[:source][:begin] ... a[:source][:end]] + "\n\n"
+			when :empty
+				puts "<<<<< string 1"
+				p @ostr1[a[:source][:begin] ... a[:source][:end]]
+				puts
+				puts ">>>>> string 2"
+				p @ostr2[a[:target][:begin] ... a[:target][:end]]
+				puts
+			else
+				puts "***** local mismatch"
+				astr1 = ''
+				astr2 = ''
+
+				base = a[:source][:begin]
+				astr1 = a[:alignment].sdiff.map do |c|
+					case c.action
+					when '='
+						@ostr1[c.old_position + base]
+					when '+'
+						'_'
+					when '-'
+						@ostr1[c.old_position + base]
+					when '!'
+						@ostr1[c.old_position + base] + '_'
+					end
+				end.join('')
+
+				base = a[:target][:begin]
+				astr2 = a[:alignment].sdiff.map do |c|
+					case c.action
+					when '='
+						@ostr2[c.new_position + base]
+					when '+'
+						@ostr2[c.new_position + base]
+					when '-'
+						'_'
+					when '!'
+						'_' + @ostr2[c.new_position + base]
+					end
+				end.join('')
 
+				puts '[' + astr1 + ']'
+				puts '[' + astr2 + ']'
+				puts
+			end
+		end
+	end
+
+	private
 
 	def string_preprocessing(_str1, _str2)
 		str1 = _str1.dup

data/lib/text_alignment/version.rb

@@ -1,3 +1,3 @@
 class TextAlignment
-	VERSION = '0.4.3'
+	VERSION = '0.5'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: text_alignment
 version: !ruby/object:Gem::Version
-  version: 0.4.3
+  version: '0.5'
 platform: ruby
 authors:
 - Jin-Dong Kim
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-10-02 00:00:00.000000000 Z
+date: 2020-10-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ruby-dictionary