text_alignment 0.11.0 → 0.11.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4d626b64acdca0630dc344e2f0f4c2152e481ca7c8209ac1aa8c025fb1ae7c0f
-  data.tar.gz: 972222f1f3a575cdb30cd83f886d6d3f36109b79a35cc2c970aa89cbe8fd007d
+  metadata.gz: ef59c0cd578ed453a67edeb3b29059f4b76c4c541f777fa35a06e76b299e2564
+  data.tar.gz: 5594c0f6eb1d52cc331c210fdf936e7cb09e30277f21933027e55b0c8cfa0e24
 SHA512:
-  metadata.gz: 288691e9ee650af3b9dfe5b47f1a3d05ff952292958ef733a5a5f47aeeaa73e6e1b7b0591afe3549b6cbabf04817bf406d3ca2ee55faf65b7e3816dfe52be644
-  data.tar.gz: 7f45baecc36f00310f868a596ca60bce709078bdc51a5f9438a916cd42062bb93dacadafb64e90e42276d0633914c978045fed4b597ea7411c8762761be6bd36
+  metadata.gz: ef5bae492d7e7b840c00943ac24e571392c4c992a085c6a63c8cb6db126ed9137ba94f8cd7af42a91e17aa327cb9a5ce24b909944c157f21beb8a88c8ce7528b
+  data.tar.gz: b6d84dcdc35399f91a0f6a5a24d84b11169de9d861a249acca52f0fec86e6f1fbf1bb4dbb47a5d43654fa43735b181aa64034447fb71ca675090818dcf67133a

data/bin/align_annotations

@@ -77,7 +77,9 @@ def align_mannotations(source_annotations, reference_text, alignment, debug = fa
 				annotations[:relations].each do |r|
 					reid = 'R' + (idnum_relations += 1).to_s
 					ididx[r[:id]] = reid
-					target_annotations[:relations] << r.dup.merge({id:reid, subj:ididx[r[:subj]], obj:ididx[r[:obj]]})
+					sid = ididx[r[:subj]]
+					oid = ididx[r[:obj]]
+					target_annotations[:relations] << r.dup.merge({id:reid, subj:sid, obj:oid}) unless sid.nil? || oid.nil?
 				end
 			end
 
@@ -86,7 +88,8 @@ def align_mannotations(source_annotations, reference_text, alignment, debug = fa
 				annotations[:attributes].each do |a|
 					reid = 'A' + (idnum_attributes += 1).to_s
 					ididx[a[:id]] = reid
-					target_annotations[:attributes] << a.dup.merge({id:reid, subj:ididx[a[:subj]]})
+					sid = ididx[a[:subj]]
+					target_annotations[:attributes] << a.dup.merge({id:reid, subj:sid}) unless sid.nil?
 				end
 			end
 
@@ -95,7 +98,8 @@ def align_mannotations(source_annotations, reference_text, alignment, debug = fa
 				annotations[:modifications].each do |m|
 					reid = 'M' + (idnum_modifications += 1).to_s
 					ididx[m[:id]] = reid
-					target_annotations[:modifications] << m.dup.merge({id:reid, obj:ididx[m[:obj]]})
+					oid = ididx[m[:obj]]
+					target_annotations[:modifications] << m.dup.merge({id:reid, obj:oid}) unless oid.nil?
 				end
 			end
 		end
@@ -115,11 +119,13 @@ reference_text = read_text(ARGV[1])
 alignment = TextAlignment::TextAlignment.new(reference_text, true)
 
 target_annotations = if source_annotations.class == Array
+	# align_mannotations(source_annotations, reference_text, alignment, true)
 	align_mannotations(source_annotations, reference_text, alignment, false)
 else
-	denotations = align_denotations(source_annotations[:denotations], source_annotations[:text], alignment)
-	# denotations = align_denotations(source_annotations[:denotations], source_annotations[:text], reference_text, true)
+	# denotations = align_denotations(source_annotations[:denotations], source_annotations[:text], alignment)
+	denotations = align_denotations(source_annotations[:denotations], source_annotations[:text], alignment, false)
 	source_annotations.merge({text:reference_text, denotations:denotations})
 end
 
+# pp alignment.block_alignment
 puts target_annotations.to_json

data/lib/text_alignment/anchor_finder.rb

@@ -7,11 +7,8 @@ module TextAlignment; end unless defined? TextAlignment
 class TextAlignment::AnchorFinder
 
 	def initialize(source_str, target_str, cultivation_map)
-		@s1, @s2 = if reverse?(source_str, target_str)
-			[target_str.downcase, source_str.downcase]
-		else
-			[source_str.downcase, target_str.downcase]
-		end
+		@s1 = source_str.downcase
+		@s2 = target_str.downcase
 
 		@cultivation_map = cultivation_map
 
@@ -19,19 +16,13 @@ class TextAlignment::AnchorFinder
 		@size_window = TextAlignment::SIZE_WINDOW
 		@sim_threshold = TextAlignment::TEXT_SIMILARITY_THRESHOLD
 		@pos_s1_final_possible_begin = @s1.length - @size_ngram - 1
+		@pos_s2_final_possible_end = @s2.length
 
 		# positions of last match
 		@pos_s1_last_match = 0
 		@pos_s2_last_match = 0
 	end
 
-	def reverse?(source_str = nil, target_str = nil)
-		unless source_str.nil?
-			@reverse_p = target_str.length < source_str.length
-		end
-		@reverse_p
-	end
-
 	def get_next_anchor
 		# To find the beginning positions of an anchor ngram in s1 and s2, beginning from the last positions matched
 		beg_s2 = for beg_s1 in @pos_s1_last_match .. @pos_s1_final_possible_begin
@@ -49,26 +40,23 @@ class TextAlignment::AnchorFinder
 		# To extend the block to the left
 		b1 = beg_s1
 		b2 = beg_s2
-		while b1 >= @pos_s1_last_match && b2 >= @pos_s2_last_match && @s1[b1] == @s2[b2]
+		left_boundary_b2 = [@pos_s2_last_match, (@cultivation_map.last_cultivated_position(b2) || 0)].max
+		while b1 > @pos_s1_last_match && b2 > left_boundary_b2 && @s1[b1 - 1] == @s2[b2 - 1]
 			b1 -= 1; b2 -= 1
 		end
-		b1 += 1; b2 += 1
 
 		# To extend the block to the right
 		e1 = beg_s1 + @size_ngram
 		e2 = beg_s2 + @size_ngram
-		while @s1[e1] && @s1[e1] == @s2[e2]
+		right_boundary_b2 = @cultivation_map.next_cultivated_position(e2) || @pos_s2_final_possible_end
+		while @s1[e1] && e2 < right_boundary_b2 && @s1[e1] == @s2[e2]
 			e1 += 1; e2 += 1
 		end
 
 		@pos_s1_last_match = e1
 		@pos_s2_last_match = e2
 
-		if reverse?
-			{source:{begin:b2 , end:e2}, target:{begin:b1, end:e1}}
-		else
-			{source:{begin:b1 , end:e1}, target:{begin:b2, end:e2}}
-		end
+		{source:{begin:b1 , end:e1}, target:{begin:b2, end:e2}}
 	end
 
 	private
@@ -91,13 +79,7 @@ class TextAlignment::AnchorFinder
 	# return nil if the anchor is too much frequent
 	def find_beg_s2_candidates(anchor, search_position)
 		candidates = []
-		while _beg_s2 = @s2.index(anchor, search_position)
-			search_again_position = @cultivation_map.search_again_position(_beg_s2)
-			unless search_again_position.nil?
-				search_position = search_again_position
-				next
-			end
-
+		while _beg_s2 = @cultivation_map.index(anchor, @s2, search_position)
 			candidates << _beg_s2
 
 			# for speed, skip anchor of high frequency

data/lib/text_alignment/cultivation_map.rb

@@ -4,16 +4,91 @@ class TextAlignment::CultivationMap
 	attr_reader :map
 
 	def initialize
-		@map = {}
+		@map = []
 	end
 
 	def cultivate(regions)
-		regions.each do |b, e|
-			(b ... e).each{|p| @map[p] = e}
+		@map += regions
+		@map.sort!{|a, b| a[0] <=> b[0]}
+		new_map = []
+		@map.each do |region|
+			if new_map.empty?
+				new_map << region
+			elsif new_map.last[1] > region[0]
+				raise "Overlapping regions: #{new_map.last} : #{region}"
+			elsif new_map.last[1] == region[0]
+				new_map.last[1] == region[1]
+			else
+				new_map << region
+			end
 		end
+		@map = new_map
 	end
 
-	def search_again_position(position)
-		@map[position]
+	def search_again_position(position, end_position = nil)
+		end_position ||= position
+		region = @map.bsearch{|r| end_position < r[1]}
+		if region.nil? || region[0] > position
+			nil
+		else
+			region[1]
+		end
+	end
+
+	def last_cultivated_position(position)
+		ridx = @map.rindex{|r| r[1] <= position}
+		ridx.nil? ? nil : @map[ridx][1]
+	end
+
+	def next_cultivated_position(position)
+		region = @map.bsearch{|r| position < r[0]}
+		region.nil? ? nil : region[0]
+	end
+
+	def in_regions(region)
+		@map.select{|r| (r[1] > region[0] && r[1] <= region[1]) || (r[0] < region[1] && r[0] >= region[0])}
+	end
+
+	def region_state(region)
+		closed_parts = in_regions(region)
+		if closed_parts.empty?
+			[:open, region]
+		else
+			if front_open?(region, closed_parts)
+				if rear_open?(region, closed_parts)
+					[:middle_closed, [closed_parts.first[1], closed_parts.last[0]]]
+				else
+					[:front_open, [region[0], closed_parts.first[0]]]
+				end
+			else
+				if rear_open?(region, closed_parts)
+					[:rear_open, [closed_parts.last[1], region[1]]]
+				else
+					[:closed, nil]
+				end
+			end
+		end
+	end
+
+	def index(target, string, position)
+		length = target.length
+		loop do
+			_begin = string.index(target, position)
+			break if _begin.nil?
+			position = search_again_position(_begin)
+			next unless position.nil?
+			break _begin if region_state([_begin, _begin + length])[0] == :open
+			position = _begin + 1
+		end
+	end
+
+	private
+
+	def front_open?(region, closed_parts)
+		closed_parts.first[0] > region[0]
+	end
+
+	def rear_open?(region, closed_parts)
+		closed_parts.last[1] < region[1]
 	end
 end

data/lib/text_alignment/mixed_alignment.rb

@@ -144,11 +144,17 @@ class TextAlignment::MixedAlignment
 	def compute_similarity(s1, s2, sdiff)
 		return 0 if sdiff.nil?
 
-		# compute the lcs only with non-whitespace letters
-		lcs = sdiff.count{|d| d.action == '=' && d.old_element =~ /\S/ && d.new_element =~ /\S/}
-		return 0 if lcs == 0
+		# recoverbility
+		count_nws =	sdiff.count{|d| d.old_element =~ /\S/}
+		count_nws_match =	sdiff.count{|d| d.action == '=' && d.old_element =~ /\S/}
 
-		similarity = lcs.to_f / [s1.scan(/\S/).count, s2.scan(/\S/).count].min
-	end
+		coverage = count_nws_match.to_f / count_nws
+
+		# fragmentation rate
+		count_ofrag = sdiff.count{|d| d.old_element =~ /\s/}
+		count_frag = sdiff.collect{|d| (d.action == '=') && (d.old_element =~/\s/) ? ' ' : d.action}.join.scan(/=+/).count
+		rate_frag = count_ofrag.to_f / count_frag
 
+		similarity = coverage * rate_frag
+	end
 end

data/lib/text_alignment/text_alignment.rb

@@ -15,8 +15,9 @@ class TextAlignment::TextAlignment
 	def initialize(reference_text, to_prevent_overlap = false)
 		raise ArgumentError, "nil text" if reference_text.nil?
 
-		@original_rtext = reference_text
+		@original_reference_text = reference_text
 		@rtext_mapping = TextAlignment::CharMapping.new(reference_text)
+		@mapped_reference_text = @rtext_mapping.mapped_text
 		@to_prevent_overlap = to_prevent_overlap
 
 		@original_text = nil
@@ -34,21 +35,19 @@ class TextAlignment::TextAlignment
 			@text_mapping = TextAlignment::CharMapping.new(text)
 		end
 
-		text_mapped = @text_mapping.mapped_text
+		@mapped_text = @text_mapping.mapped_text
 		denotations_mapped = @text_mapping.enmap_denotations(denotations)
 
-		rtext_mapped = @rtext_mapping.mapped_text
-
 		## To generate the block_alignment of the input text against the reference text
-
 		# Initialization
-		@block_alignment = {text: @original_text, reference_text: @original_rtext, denotations: denotations}
+		# @block_alignment = {text: @original_text, reference_text: @original_reference_text, denotations: denotations}
+		@block_alignment = {text: @mapped_text, reference_text: @mapped_reference_text, denotations: denotations}
 
 		# Generation
-		@block_alignment[:blocks] = if r = whole_block_alignment(text_mapped, rtext_mapped, @cultivation_map)
+		@block_alignment[:blocks] = if r = whole_block_alignment(@mapped_text, @mapped_reference_text, @cultivation_map)
 			r
 		else
-			find_block_alignment(text_mapped, rtext_mapped, denotations_mapped, @cultivation_map)
+			find_block_alignment(@mapped_text, @mapped_reference_text, denotations_mapped, @cultivation_map)
 		end
 	end
 
@@ -134,7 +133,7 @@ class TextAlignment::TextAlignment
 			source = {begin:d.begin, end:d.end}
 			d.begin = transform_begin_position(d.begin);
 			d.end = transform_end_position(d.end);
-			raise "invalid transform" unless !d.begin.nil? && !d.end.nil? && d.begin >= 0 && d.end > d.begin && d.end <= @original_rtext.length
+			raise "invalid transform" unless !d.begin.nil? && !d.end.nil? && d.begin >= 0 && d.end > d.begin && d.end <= @original_reference_text.length
 		rescue
 			@lost_annotations << {source: source, target:{begin:d.begin, end:d.end}}
 			d.begin = nil
@@ -150,7 +149,7 @@ class TextAlignment::TextAlignment
 
 		r = hdenotations.collect do |d|
 			t = transform_a_span(d[:span])
-			raise "invalid transform" unless !t[:begin].nil? && !t[:end].nil? && t[:begin] >= 0 && t[:end] > t[:begin] && t[:end] <= @original_rtext.length
+			raise "invalid transform" unless !t[:begin].nil? && !t[:end].nil? && t[:begin] >= 0 && t[:end] > t[:begin] && t[:end] <= @original_reference_text.length
 			new_d = d.dup.merge({span:t})
 		rescue
 			@lost_annotations << {source: d[:span], target:t}
@@ -161,8 +160,8 @@ class TextAlignment::TextAlignment
 	end
 
 	def alignment_show
-		stext = @block_alignment[:text]
-		ttext = @block_alignment[:reference_text]
+		stext = @mapped_text
+		ttext = @mapped_reference_text
 
 		show = ''
 		@block_alignment[:blocks].each do |a|
@@ -218,7 +217,7 @@ class TextAlignment::TextAlignment
 
 				"***** local mismatch [#{a[:source][:begin]} - #{a[:source][:end]}] [#{a[:target][:begin]} - #{a[:target][:end]}] (similarity: #{a[:similarity]})\n" +
 				"[#{astr1}]\n" +
-				"[#{astr2}]\n\n"
+				"[#{astr2.gsub("\n", " ")}]\n\n"
 			end
 		end
 		show
@@ -257,142 +256,109 @@ class TextAlignment::TextAlignment
 		# puts "-=-=-=-=-"
 		# puts
 
-		## to fill the gaps
-		last_block = nil
-		blocks2 = blocks.inject([]) do |sum, block|
-			b1 = last_block ? last_block[:source][:end] : 0
-			e1 = block[:source][:begin]
+		## To fill the gaps
+		## lblock: last block, cblock: current block
+		lblock = nil
+		blocks2 = (blocks + [nil]).inject([]) do |sum, cblock|
+			b1 = lblock.nil? ? 0 : lblock[:source][:end]
+			e1 = cblock.nil? ? str1.length : cblock[:source][:begin]
 
-			sum += if b1 == e1
-				[block]
-			else
-				b2 = last_block ? last_block[:target][:end] : 0
-				e2 = block[:target][:begin]
-
-				if b2 == e2
-					[
-						{source:{begin:b1, end:e1}, alignment: :empty},
-						block
-					]
+			if b1 < e1
+				b2 = lblock.nil? ? 0 : lblock[:target][:end]
+				e2 = cblock.nil? ? str2.length : cblock[:target][:begin]
+				_str1 = str1[b1 ... e1]
+				_str2 = str2[b2 ... e2]
+
+				sum += if _str1.strip.empty? || _str2.strip.empty?
+					[{source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}]
 				else
 					len_buffer = ((e1 - b1) * (1 + TextAlignment::BUFFER_RATE)).to_i + TextAlignment::BUFFER_MIN
-
-					if b1 == 0 && b2 == 0
-						b2 = e2 - len_buffer if e2 > len_buffer
-					end
-
-					_str1 = str1[b1 ... e1]
-					_str2 = str2[b2 ... e2]
-
-					if _str1.strip.empty? || _str2.strip.empty?
-						[
-							{source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty},
-							block
-						]
-					elsif ((e2 - b2) - (e1 - b1)) > len_buffer
-						la_block1 = local_alignment_blocks(str1, b1, e1, str2, b2, b2 + len_buffer, denotations)
-						la_block2 = local_alignment_blocks(str1, b1, e1, str2, e2 - len_buffer, e2, denotations)
-						[la_block2, la_block2].max{|a, b| a.first[:similarity] <=> b.first[:similarity]} << block
-					else
-						local_alignment_blocks(str1, b1, e1, str2, b2, e2, denotations) << block
+					region_state, state_region = cultivation_map.region_state([b2, e2])
+					case region_state
+					when :closed
+						[]
+					when :front_open
+						oe2 = state_region[1]
+						me2 = (oe2 - b2) > len_buffer ? b2 + len_buffer : oe2
+						local_alignment(str1, b1, e1, str2, b2, me2, denotations, cultivation_map)
+					when :rear_open
+						ob2 = state_region[0]
+						mb2 = (e2 - ob2) > len_buffer ? e2 - len_buffer : ob2
+						local_alignment(str1, b1, e1, str2, mb2, e2, denotations, cultivation_map)
+					when :middle_closed
+						oe2 = state_region[0]
+						me2 = (oe2 - b2) > len_buffer ? b2 + len_buffer : oe2
+						attempt1 = local_alignment(str1, b1, e1, str2, b2, me2, denotations, cultivation_map)
+						if attempt1.empty?
+							ob2 = state_region[1]
+							mb2 = (e2 - ob2) > len_buffer ? e2 - len_buffer : ob2
+							local_alignment(str1, b1, e1, str2, mb2, e2, denotations, cultivation_map)
+						else
+							attempt1
+						end
+					else # :open
+						if (e2 - b2) > len_buffer
+							attempt1 = local_alignment(str1, b1, e1, str2, b2, b2 + len_buffer, denotations, cultivation_map)
+							if attempt1.empty?
+								local_alignment(str1, b1, e1, str2, e2 - len_buffer, e2, denotations, cultivation_map)
+							else
+								attempt1
+							end
+						else
+							local_alignment(str1, b1, e1, str2, b2, e2, denotations, cultivation_map)
+						end
 					end
 				end
 			end
 
-			last_block = block
-			sum
+			lblock = cblock
+			cblock.nil? ? sum : sum << cblock
 		end
 
-		# the last step
-		blocks2 += if last_block.nil?
-			local_alignment_blocks(str1, 0, str1.length, str2, 0, str2.length, denotations)
-		else
-			b1 = last_block[:source][:end]
-			if b1 < str1.length
-				e1 = str1.length
-				b2 = last_block[:target][:end]
-
-				_str1 = str1[b1 ... e1]
-				if _str1.strip.empty?
-					[{source:{begin:b1, end:e1}, alignment: :empty}]
-				else
-					if b2 < str2.length
-						len_buffer = ((e1 - b1) * (1 + TextAlignment::BUFFER_RATE)).to_i + TextAlignment::BUFFER_MIN
-						e2 = (str2.length - b2) > len_buffer ? b2 + len_buffer : str2.length
-
-						local_alignment_blocks(str1, b1, e1, str2, b2, e2, denotations)
-					else
-						[{source:{begin:b1, end:e1}, alignment: :empty}]
-					end
-				end
-			else
-				[]
-			end
-		end
 	end
 
 	def whole_block_alignment(str1, str2, cultivation_map)
-		## Block exact match
-		search_position = 0
-
-		block_begin = begin
-			_block_begin = str2.index(str1, search_position)
-			break if _block_begin.nil?
-			search_position = cultivation_map.search_again_position(_block_begin)
-			_block_begin
-		end until search_position.nil?
-
-		unless block_begin.nil?
-			return [{source:{begin:0, end:str1.length}, target:{begin:block_begin, end:block_begin + str1.length}, delta:block_begin, alignment: :block}]
-		end
+		block_begin = cultivation_map.index(str1, str2, 0)
+		return [{source:{begin:0, end:str1.length}, target:{begin:block_begin, end:block_begin + str1.length}, delta:block_begin, alignment: :block}] unless block_begin.nil?
 
-		search_position = 0
+		block_begin = cultivation_map.index(str1.downcase, str2.downcase, 0)
+		return [{source:{begin:0, end:str1.length}, target:{begin:block_begin, end:block_begin + str1.length}, delta:block_begin, alignment: :block}] unless block_begin.nil?
 
-		dstr1 = str1.downcase
-		dstr2 = str2.downcase
-		block_begin = begin
-			_block_begin = dstr2.index(dstr1, search_position)
-			break if _block_begin.nil?
-			search_position = cultivation_map.search_again_position(_block_begin)
-			_block_begin
-		end until search_position.nil?
+		nil
+	end
 
-		unless block_begin.nil?
-			return [{source:{begin:0, end:str1.length}, target:{begin:block_begin, end:block_begin + str1.length}, delta:block_begin, alignment: :block}]
+	def local_alignment(str1, b1, e1, str2, b2, e2, denotations = nil, cultivation_map)
+		tblocks = term_based_alignment(str1, b1, e1, str2, b2, e2, denotations, cultivation_map)
+		if tblocks.empty?
+			lcs_alignment(str1, b1, e1, str2, b2, e2, cultivation_map)
+		else
+			tblocks
 		end
-
-		nil
 	end
 
-	def local_alignment_blocks(str1, b1, e1, str2, b2, e2, denotations = nil)
-		block2 = str2[b2 ... e2]
+	def term_based_alignment(str1, b1, e1, str2, b2, e2, denotations = nil, cultivation_map)
+		str2_block = str2[0 ... e2]
 
 		## term-based alignment
 		tblocks = if denotations
-			ds_in_scope = denotations.select{|d| d[:span][:begin] >= b1 && d[:span][:end] <= e1}.
+			denotations_in_scope = denotations.select{|d| d[:span][:begin] >= b1 && d[:span][:end] <= e1}.
 							sort{|d1, d2| d1[:span][:begin] <=> d2[:span][:begin] || d2[:span][:end] <=> d1[:span][:end] }.
 							map{|d| d.merge(lex:str1[d[:span][:begin] ... d[:span][:end]])}
 
-			position = 0
-			_tblocks = ds_in_scope.map do |term|
-				lex = term[:lex]
-				r = block2.index(lex, position)
-				if r.nil?
-					position = nil
-					break
-				end
-				position = r + lex.length
-				{source:term[:span], target:{begin:r + b2, end:r + b2 + lex.length}, alignment: :term, similarity: 0.9, delta: r + b2 - term[:span][:begin]}
+			search_position = b2
+			_tblocks = denotations_in_scope.map do |denotation|
+				lex = denotation[:lex]
+				term_begin = cultivation_map.index(lex, str2_block, search_position)
+				break [] if term_begin.nil? # break the loop if a missing term is found
+				search_position = term_begin + lex.length
+				{source:denotation[:span], target:{begin:term_begin, end:term_begin + lex.length}, alignment: :term, similarity: 0.9, delta: term_begin - denotation[:span][:begin]}
 			end
 
-			# missing term found
-			_tblocks = [] if position.nil?
-
 			# redundant matching found
-			unless position.nil?
-				ds_in_scope.each do |term|
-					lex = term[:lex]
-					look_forward = block2.index(lex, position)
+			unless _tblocks.empty?
+				search_position = _tblocks.last[:target][:end]
+				denotations_in_scope.each do |term|
+					look_forward = cultivation_map.index(term[:lex], str2_block, search_position)
 					unless look_forward.nil?
 						_tblocks = []
 						break
@@ -405,72 +371,37 @@ class TextAlignment::TextAlignment
 			[]
 		end
 
-		if tblocks.empty?
-			if b1 == 0 && e1 == str1.length
-				if (e1 > 2000) || (e2 > 2000)
-					[{source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}]
-				else
-					block1 = str1[b1 ... e1]
-					block2 = str2[b2 ... e2]
-
-					## character-based alignment
-					alignment = TextAlignment::MixedAlignment.new(block1.downcase, block2.downcase)
-					if alignment.sdiff.nil?
-						[{source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}]
-					else
-						[{source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: alignment, similarity: alignment.similarity}]
-					end
-				end
-			else
-				block1 = str1[b1 ... e1]
-				block2 = str2[b2 ... e2]
+		ltblock = nil
+		tblocks2 = (tblocks + [nil]).inject([]) do |sum, ctblock|
+			tb1 = ltblock.nil? ? b1 : ltblock[:source][:end]
+			te1 = ctblock.nil? ? e1 : ctblock[:source][:begin]
 
-				## character-based alignment
-				alignment = TextAlignment::MixedAlignment.new(block1.downcase, block2.downcase)
-				if alignment.sdiff.nil?
-					[{source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}]
-				else
-					[{source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: alignment, similarity: alignment.similarity}]
-				end
+			if te1 > tb1
+				tb2 = ltblock.nil? ? b2 : ltblock[:target][:end]
+				te2 = ctblock.nil? ? e2 : ctblock[:target][:begin]
+				sum << {source:{begin:tb1, end:te1}, target:{begin:tb2, end:te2}, alignment: :empty}
 			end
-		else
-			last_tblock = nil
-			lblocks = tblocks.inject([]) do |sum, tblock|
-				tb1 = last_tblock ? last_tblock[:source][:end] : b1
-				te1 = tblock[:source][:begin]
 
-				sum += if te1 == tb1
-					[tblock]
-				else
-					tb2 = last_tblock ? last_tblock[:target][:end] : b2
-					te2 = tblock[:target][:begin]
-
-					if b2 == e2
-						[
-							{source:{begin:tb1, end:te1}, alignment: :empty},
-							tblock
-						]
-					else
-						[
-							{source:{begin:tb1, end:te1}, target:{begin:tb2, end:te2}, alignment: :empty},
-							tblock
-						]
-					end
-				end
+			ltblock = ctblock
+			ctblock.nil? ? sum : sum << ctblock
+		end
 
-				last_tblock = tblock
-				sum
-			end
+		tblocks2
+	end
 
-			if last_tblock[:source][:end] < e1
-				if last_tblock[:target][:end] < e2
-					lblocks << {source:{begin:last_tblock[:source][:end], end:e1}, target:{begin:last_tblock[:target][:end], end:e2}, alignment: :empty}
-				else
-					lblocks << {source:{begin:last_tblock[:source][:end], end:e1}, alignment: :empty}
-				end
-			end
+	def lcs_alignment(str1, b1, e1, str2, b2, e2, cultivation_map)
+		source = {begin:b1, end:e1}
+		target = {begin:b2, end:e2}
 
-			lblocks
+		if (e1 - b1) > 2000
+			[{source:source, target:target, alignment: :empty}]
+		else
+			alignment = TextAlignment::MixedAlignment.new(str1[b1 ... e1].downcase, str2[b2 ... e2].downcase)
+			if alignment.similarity < 0.5
+				[{source:source, target:target, alignment: :empty}]
+			else
+				[{source:source, target:target, alignment: alignment, similarity: alignment.similarity}]
+			end
 		end
 	end
 

data/lib/text_alignment/version.rb

@@ -1,3 +1,3 @@
 class TextAlignment
-	VERSION = '0.11.0'
+	VERSION = '0.11.1'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: text_alignment
 version: !ruby/object:Gem::Version
-  version: 0.11.0
+  version: 0.11.1
 platform: ruby
 authors:
 - Jin-Dong Kim
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-03-04 00:00:00.000000000 Z
+date: 2021-03-08 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ruby-dictionary