text_alignment 0.2.9 → 0.3.9

data/lib/text_alignment/mappings.rb

@@ -1,75 +1,74 @@
-module TextAlignment
+module TextAlignment; end unless defined? TextAlignment
 
-  TextAlignment::MAPPINGS = [
-    ["©", "(c)"],   #U+00A9 (Copyright Sign)
+TextAlignment::MAPPINGS = [
+	["©", "(c)"],   #U+00A9 (Copyright Sign)
 
-    ["α", "alpha"],   #U+03B1 (greek small letter alpha)
-    ["β", "beta"],    #U+03B2 (greek small letter beta)
-    ["γ", "gamma"],   #U+03B3 (greek small letter gamma)
-    ["δ", "delta"],   #U+03B4 (greek small letter delta)
-    ["ε", "epsilon"], #U+03B5 (greek small letter epsilon)
-    ["ζ", "zeta"],    #U+03B6 (greek small letter zeta)
-    ["η", "eta"],     #U+03B7 (greek small letter eta)
-    ["θ", "theta"],   #U+03B7 (greek small letter eta)
-    ["ι", "iota"],    #U+03B7 (greek small letter eta)
-    ["κ", "kappa"],   #U+03BA (greek small letter kappa)
-    ["λ", "lambda"],  #U+03BB (greek small letter lambda)
-    ["λ", "lamda"],  #U+03BB (greek small letter lambda)
-    ["μ", "mu"],      #U+03BC (greek small letter mu)
-    ["ν", "nu"],      #U+03BD (greek small letter nu)
-    ["ξ", "xi"],      #U+03BE (greek small letter xi)
-    ["ο", "omicron"], #U+03BF (greek small letter omicron)
-    ["π", "pi"],      #U+03C0 (greek small letter pi)
-    ["ρ", "rho"],     #U+03C1 (greek small letter rho)
-    ["σ", "sigma"],   #U+03C3 (greek small letter sigma)
-    ["τ", "tau"],     #U+03C4 (greek small letter tau)
-    ["υ", "upsilon"], #U+03C5 (greek small letter upsilon)
-    ["φ", "phi"],     #U+03C6 (greek small letter phi)
-    ["χ", "chi"],     #U+03C7 (greek small letter chi)
-    ["ψ", "psi"],     #U+03C8 (greek small letter psi)
-    ["ω", "omega"],   #U+03C9 (greek small letter omega)
+	["α", "alpha"],   #U+03B1 (greek small letter alpha)
+	["β", "beta"],    #U+03B2 (greek small letter beta)
+	["γ", "gamma"],   #U+03B3 (greek small letter gamma)
+	["δ", "delta"],   #U+03B4 (greek small letter delta)
+	["ε", "epsilon"], #U+03B5 (greek small letter epsilon)
+	["ζ", "zeta"],    #U+03B6 (greek small letter zeta)
+	["η", "eta"],     #U+03B7 (greek small letter eta)
+	["θ", "theta"],   #U+03B7 (greek small letter eta)
+	["ι", "iota"],    #U+03B7 (greek small letter eta)
+	["κ", "kappa"],   #U+03BA (greek small letter kappa)
+	["λ", "lambda"],  #U+03BB (greek small letter lambda)
+	["λ", "lamda"],  #U+03BB (greek small letter lambda)
+	["μ", "mu"],      #U+03BC (greek small letter mu)
+	["ν", "nu"],      #U+03BD (greek small letter nu)
+	["ξ", "xi"],      #U+03BE (greek small letter xi)
+	["ο", "omicron"], #U+03BF (greek small letter omicron)
+	["π", "pi"],      #U+03C0 (greek small letter pi)
+	["ρ", "rho"],     #U+03C1 (greek small letter rho)
+	["σ", "sigma"],   #U+03C3 (greek small letter sigma)
+	["τ", "tau"],     #U+03C4 (greek small letter tau)
+	["υ", "upsilon"], #U+03C5 (greek small letter upsilon)
+	["φ", "phi"],     #U+03C6 (greek small letter phi)
+	["χ", "chi"],     #U+03C7 (greek small letter chi)
+	["ψ", "psi"],     #U+03C8 (greek small letter psi)
+	["ω", "omega"],   #U+03C9 (greek small letter omega)
 
-    ["Α", "Alpha"],   #U+0391 (greek capital letter alpha)
-    ["Β", "Beta"],    #U+0392 (greek capital letter beta)
-    ["Γ", "Gamma"],   #U+0393 (greek capital letter gamma)
-    ["Δ", "Delta"],   #U+0394 (greek capital letter delta)
-    ["Ε", "Epsilon"], #U+0395 (greek capital letter epsilon)
-    ["Ζ", "Zeta"],    #U+0396 (greek capital letter zeta)
-    ["Η", "Eta"],     #U+0397 (greek capital letter eta)
-    ["Θ", "Theta"],   #U+0398 (greek capital letter theta)
-    ["Ι", "Iota"],    #U+0399 (greek capital letter iota)
-    ["Κ", "Kappa"],   #U+039A (greek capital letter kappa)
-    ["Λ", "Lambda"],  #U+039B (greek capital letter lambda)
-    ["Λ", "Lamda"],  #U+039B (greek capital letter lambda)
-    ["Μ", "Mu"],      #U+039C (greek capital letter mu)
-    ["Ν", "Nu"],      #U+039D (greek capital letter nu)
-    ["Ξ", "Xi"],      #U+039E (greek capital letter xi)
-    ["Ο", "Omicron"], #U+039F (greek capital letter omicron)
-    ["Π", "Pi"],      #U+03A0 (greek capital letter pi)
-    ["Ρ", "Rho"],     #U+03A1 (greek capital letter rho)
-    ["Σ", "Sigma"],   #U+03A3 (greek capital letter sigma)
-    ["Τ", "Tau"],     #U+03A4 (greek capital letter tau)
-    ["Υ", "Upsilon"], #U+03A5 (greek capital letter upsilon)
-    ["Φ", "Phi"],     #U+03A6 (greek capital letter phi)
-    ["Χ", "Chi"],     #U+03A7 (greek capital letter chi)
-    ["Ψ", "Psi"],     #U+03A8 (greek capital letter Psi)
-    ["Ω", "Omega"],   #U+03A9 (greek capital letter omega)
+	["Α", "Alpha"],   #U+0391 (greek capital letter alpha)
+	["Β", "Beta"],    #U+0392 (greek capital letter beta)
+	["Γ", "Gamma"],   #U+0393 (greek capital letter gamma)
+	["Δ", "Delta"],   #U+0394 (greek capital letter delta)
+	["Ε", "Epsilon"], #U+0395 (greek capital letter epsilon)
+	["Ζ", "Zeta"],    #U+0396 (greek capital letter zeta)
+	["Η", "Eta"],     #U+0397 (greek capital letter eta)
+	["Θ", "Theta"],   #U+0398 (greek capital letter theta)
+	["Ι", "Iota"],    #U+0399 (greek capital letter iota)
+	["Κ", "Kappa"],   #U+039A (greek capital letter kappa)
+	["Λ", "Lambda"],  #U+039B (greek capital letter lambda)
+	["Λ", "Lamda"],  #U+039B (greek capital letter lambda)
+	["Μ", "Mu"],      #U+039C (greek capital letter mu)
+	["Ν", "Nu"],      #U+039D (greek capital letter nu)
+	["Ξ", "Xi"],      #U+039E (greek capital letter xi)
+	["Ο", "Omicron"], #U+039F (greek capital letter omicron)
+	["Π", "Pi"],      #U+03A0 (greek capital letter pi)
+	["Ρ", "Rho"],     #U+03A1 (greek capital letter rho)
+	["Σ", "Sigma"],   #U+03A3 (greek capital letter sigma)
+	["Τ", "Tau"],     #U+03A4 (greek capital letter tau)
+	["Υ", "Upsilon"], #U+03A5 (greek capital letter upsilon)
+	["Φ", "Phi"],     #U+03A6 (greek capital letter phi)
+	["Χ", "Chi"],     #U+03A7 (greek capital letter chi)
+	["Ψ", "Psi"],     #U+03A8 (greek capital letter Psi)
+	["Ω", "Omega"],   #U+03A9 (greek capital letter omega)
 
-    ["ϕ", "phi"],     #U+03D5 (greek phi symbol)
+	["ϕ", "phi"],     #U+03D5 (greek phi symbol)
 
-    ["×", "x"],       #U+00D7 (multiplication sign)
-    ["•", "*"],       #U+2022 (bullet)
-    [" ", " "],       #U+2009 (thin space)
-    [" ", " "],       #U+200A (hair space)
-    [" ", " "],       #U+00A0 (no-break space)
-    ["　", " "],       #U+3000 (ideographic space)
-    ["−", "-"],       #U+2212 (minus sign)
-    ["–", "-"],       #U+2013 (en dash)
-    ["′", "'"],       #U+2032 (prime)
-    ["‘", "'"],       #U+2018 (left single quotation mark)
-    ["’", "'"],       #U+2019 (right single quotation mark)
-    ["“", '"'],       #U+201C (left double quotation mark)
-    ["”", '"'],        #U+201D (right double quotation mark)
-    ['"', "''"]
+	["×", "x"],       #U+00D7 (multiplication sign)
+	["•", "*"],       #U+2022 (bullet)
+	[" ", " "],       #U+2009 (thin space)
+	[" ", " "],       #U+200A (hair space)
+	[" ", " "],       #U+00A0 (no-break space)
+	["　", " "],       #U+3000 (ideographic space)
+	["−", "-"],       #U+2212 (minus sign)
+	["–", "-"],       #U+2013 (en dash)
+	["′", "'"],       #U+2032 (prime)
+	["‘", "'"],       #U+2018 (left single quotation mark)
+	["’", "'"],       #U+2019 (right single quotation mark)
+	["“", '"'],       #U+201C (left double quotation mark)
+	["”", '"'],        #U+201D (right double quotation mark)
+	['"', "''"]
   ]
-end

data/lib/text_alignment/mixed_alignment.rb

@@ -0,0 +1,193 @@
+#!/usr/bin/env ruby
+require 'diff-lcs'
+require 'text_alignment/lcs_min'
+require 'text_alignment/find_divisions'
+require 'text_alignment/lcs_comparison'
+require 'text_alignment/lcs_alignment'
+require 'text_alignment/lcs_cdiff'
+require 'text_alignment/glcs_alignment'
+require 'text_alignment/mappings'
+
+module TextAlignment; end unless defined? TextAlignment
+
+TextAlignment::NOMATCH_CHARS = "@^|#$%&_" unless defined? TextAlignment::NOMATCH_CHARS
+
+class TextAlignment::MixedAlignment
+	attr_reader :sdiff
+	attr_reader :position_map_begin, :position_map_end
+	attr_reader :common_elements, :mapped_elements
+	attr_reader :similarity
+	attr_reader :str1_match_initial, :str1_match_final, :str2_match_initial, :str2_match_final
+
+	def initialize(str1, str2, mappings = [])
+		raise ArgumentError, "nil string" if str1.nil? || str2.nil?
+		raise ArgumentError, "nil mappings" if mappings.nil?
+
+		## preprocessing
+		str1 = str1.dup
+		str2 = str2.dup
+		mappings = mappings.dup
+
+		## find the first nomatch character
+		TextAlignment::NOMATCH_CHARS.each_char do |c|
+			if str2.index(c).nil?
+				@nomatch_char1 = c
+				break
+			end
+		end
+		raise RuntimeError, "Cannot find nomatch character" if @nomatch_char1.nil?
+
+		## find the first nomatch character
+		TextAlignment::NOMATCH_CHARS.each_char do |c|
+			if c != @nomatch_char1 && str1.index(c).nil?
+				@nomatch_char2 = c
+				break
+			end
+		end
+		raise RuntimeError, "Cannot find nomatch character" if @nomatch_char2.nil?
+
+		# single character mappings
+		character_mappings = mappings.select{|m| m[0].length == 1 && m[1].length == 1}
+		characters_from = character_mappings.collect{|m| m[0]}.join
+		characters_to   = character_mappings.collect{|m| m[1]}.join
+		characters_to.gsub!(/-/, '\-')
+
+		str1.tr!(characters_from, characters_to)
+		str2.tr!(characters_from, characters_to)
+
+		mappings.delete_if{|m| m[0].length == 1 && m[1].length == 1}
+
+		# ASCII foldings
+		ascii_foldings = mappings.select{|m| m[0].length == 1 && m[1].length > 1}
+		ascii_foldings.each do |f|
+			from = f[1]
+
+			if str2.index(f[0])
+				to   = f[0] + (@nomatch_char1 * (f[1].length - 1))
+				str1.gsub!(from, to)
+			end
+
+			if str1.index(f[0])
+				to   = f[0] + (@nomatch_char2 * (f[1].length - 1))
+				str2.gsub!(from, to)
+			end
+		end
+		mappings.delete_if{|m| m[0].length == 1 && m[1].length > 1}
+
+		_compute_mixed_alignment(str1, str2, mappings)
+	end
+
+	def transform_begin_position(begin_position)
+		@position_map_begin[begin_position]
+	end
+
+	def transform_end_position(end_position)
+		@position_map_end[end_position]
+	end
+
+	def transform_a_span(span)
+		{begin: @position_map_begin[span[:begin]], end: @position_map_end[span[:end]]}
+	end
+
+	def transform_spans(spans)
+		spans.map{|span| transform_a_span(span)}
+	end
+
+	def transform_denotations!(denotations)
+		denotations.map!{|d| d.begin = @position_map_begin[d.begin]; d.end = @position_map_end[d.end]; d} unless denotations.nil?
+	end
+
+	def transform_hdenotations(hdenotations)
+		return nil if hdenotations.nil?
+		hdenotations.collect{|d| d.dup.merge({span:transform_a_span(d[:span])})}
+	end
+
+	private
+
+	def _compute_mixed_alignment(str1, str2, mappings = [])
+		lcsmin = TextAlignment::LCSMin.new(str1, str2)
+		lcs = lcsmin.lcs
+		@sdiff = lcsmin.sdiff
+
+		if @sdiff.nil?
+			@similarity = 0
+			return
+		end
+
+		cmp = TextAlignment::LCSComparison.new(str1, str2, lcs, @sdiff)
+		@similarity         = cmp.similarity
+		@str1_match_initial = cmp.str1_match_initial
+		@str1_match_final   = cmp.str1_match_final
+		@str2_match_initial = cmp.str2_match_initial
+		@str2_match_final   = cmp.str2_match_final
+
+		posmap_begin, posmap_end = {}, {}
+		@common_elements, @mapped_elements = [], []
+
+		addition, deletion = [], []
+
+		@sdiff.each do |h|
+			case h.action
+			when '='
+				p1, p2 = h.old_position, h.new_position
+
+				@common_elements << [str1[p1], str2[p2]]
+				posmap_begin[p1], posmap_end[p1] = p2, p2
+
+				if !addition.empty? && deletion.empty?
+					posmap_end[p1] = p2 - addition.length unless p1 == 0
+				elsif addition.empty? && !deletion.empty?
+					deletion.each{|p| posmap_begin[p], posmap_end[p] = p2, p2}
+				elsif !addition.empty? && !deletion.empty?
+					if addition.length > 1 || deletion.length > 1
+						galign = TextAlignment::GLCSAlignment.new(str1[deletion[0] .. deletion[-1]], str2[addition[0] .. addition[-1]], mappings)
+						galign.position_map_begin.each {|k, v| posmap_begin[k + deletion[0]] = v.nil? ? nil : v + addition[0]}
+						galign.position_map_end.each   {|k, v|   posmap_end[k + deletion[0]] = v.nil? ? nil : v + addition[0]}
+						posmap_begin[p1], posmap_end[p1] = p2, p2
+						@common_elements += galign.common_elements
+						@mapped_elements += galign.mapped_elements
+					else
+						posmap_begin[deletion[0]], posmap_end[deletion[0]] = addition[0], addition[0]
+						deletion[1..-1].each{|p| posmap_begin[p], posmap_end[p] = nil, nil}
+						@mapped_elements << [str1[deletion[0], deletion.length], str2[addition[0], addition.length]]
+					end
+				end
+
+				addition.clear; deletion.clear
+
+			when '!'
+				deletion << h.old_position
+				addition << h.new_position
+			when '-'
+				deletion << h.old_position
+			when '+'
+				addition << h.new_position
+			end
+		end
+
+		p1, p2 = str1.length, str2.length
+		posmap_begin[p1], posmap_end[p1] = p2, p2
+
+		if !addition.empty? && deletion.empty?
+			posmap_end[p1] = p2 - addition.length unless p1 == 0
+		elsif addition.empty? && !deletion.empty?
+			deletion.each{|p| posmap_begin[p], posmap_end[p] = p2, p2}
+		elsif !addition.empty? && !deletion.empty?
+			if addition.length > 1 && deletion.length > 1
+				galign = TextAlignment::GLCSAlignment.new(str1[deletion[0] .. deletion[-1]], str2[addition[0] .. addition[-1]], mappings)
+				galign.position_map_begin.each {|k, v| posmap_begin[k + deletion[0]] = v.nil? ? nil : v + addition[0]}
+				galign.position_map_end.each   {|k, v|   posmap_end[k + deletion[0]] = v.nil? ? nil : v + addition[0]}
+				posmap_begin[p1], posmap_end[p1] = p2, p2
+				@common_elements += galign.common_elements
+				@mapped_elements += galign.mapped_elements
+			else
+				posmap_begin[deletion[0]], posmap_end[deletion[0]] = addition[0], addition[0]
+				deletion[1..-1].each{|p| posmap_begin[p], posmap_end[p] = nil, nil}
+				@mapped_elements << [str1[deletion[0], deletion.length], str2[addition[0], addition.length]]
+			end
+		end
+
+		@position_map_begin = posmap_begin.sort.to_h
+		@position_map_end = posmap_end.sort.to_h
+	end
+end

data/lib/text_alignment/text_alignment.rb

@@ -1,181 +1,239 @@
 #!/usr/bin/env ruby
-require 'diff-lcs'
-require 'text_alignment/lcs_min'
-require 'text_alignment/find_divisions'
-require 'text_alignment/lcs_comparison'
-require 'text_alignment/lcs_alignment'
-require 'text_alignment/lcs_cdiff'
-require 'text_alignment/glcs_alignment'
-require 'text_alignment/mappings'
+require 'text_alignment/anchor_finder'
+require 'text_alignment/mixed_alignment'
 
 module TextAlignment; end unless defined? TextAlignment
 
-TextAlignment::SIGNATURE_NGRAM = 5 unless defined? TextAlignment::SIGNATURE_NGRAM
-TextAlignment::NOMATCH_CHARS = "@^|#$%&_" unless defined? TextAlignment::NOMATCH_CHARS
+TextAlignment::SIGNATURE_NGRAM = 7 unless defined? TextAlignment::SIGNATURE_NGRAM
+TextAlignment::BUFFER_RATE = 0.1 unless defined? TextAlignment::BUFFER_RATE
 
 class TextAlignment::TextAlignment
-  attr_reader :sdiff
-  attr_reader :position_map_begin, :position_map_end
-  attr_reader :common_elements, :mapped_elements
-  attr_reader :similarity
-  attr_reader :str1_match_initial, :str1_match_final, :str2_match_initial, :str2_match_final
-
-  def initialize(str1, str2, mappings = [])
-    raise ArgumentError, "nil string" if str1.nil? || str2.nil?
-    raise ArgumentError, "nil mappings" if mappings.nil?
-
-    ## preprocessing
-    str1 = str1.dup
-    str2 = str2.dup
-    mappings = mappings.dup
-
-    ## find the first nomatch character
-    TextAlignment::NOMATCH_CHARS.each_char do |c|
-      if str2.index(c).nil?
-        @nomatch_char1 = c
-        break
-      end
-    end
-    raise RuntimeError, "Cannot find nomatch character" if @nomatch_char1.nil? 
-
-    ## find the first nomatch character
-    TextAlignment::NOMATCH_CHARS.each_char do |c|
-      if c != @nomatch_char1 && str1.index(c).nil?
-        @nomatch_char2 = c
-        break
-      end
-    end
-    raise RuntimeError, "Cannot find nomatch character" if @nomatch_char2.nil? 
-
-    # single character mappings
-    character_mappings = mappings.select{|m| m[0].length == 1 && m[1].length == 1}
-    characters_from = character_mappings.collect{|m| m[0]}.join
-    characters_to   = character_mappings.collect{|m| m[1]}.join
-    characters_to.gsub!(/-/, '\-')
-
-    str1.tr!(characters_from, characters_to)
-    str2.tr!(characters_from, characters_to)
-
-    mappings.delete_if{|m| m[0].length == 1 && m[1].length == 1}
-
-    # ASCII foldings
-    ascii_foldings = mappings.select{|m| m[0].length == 1 && m[1].length > 1}
-    ascii_foldings.each do |f|
-      from = f[1]
-
-      if str2.index(f[0])
-        to   = f[0] + (@nomatch_char1 * (f[1].length - 1))
-        str1.gsub!(from, to)
-      end
-
-      if str1.index(f[0])
-        to   = f[0] + (@nomatch_char2 * (f[1].length - 1))
-        str2.gsub!(from, to)
-      end
-    end
-    mappings.delete_if{|m| m[0].length == 1 && m[1].length > 1}
-
-    _compute_mixed_alignment(str1, str2, mappings)
-  end
-
-  def transform_a_span(span)
-    {begin: @position_map_begin[span[:begin]], end: @position_map_end[span[:end]]}
-  end
-
-  def transform_spans(spans)
-    spans.map{|span| transform_a_span(span)}
-  end
-
-  def transform_denotations!(denotations)
-    denotations.map!{|d| d.begin = @position_map_begin[d.begin]; d.end = @position_map_end[d.end]; d} unless denotations.nil?
-  end
-
-  def transform_hdenotations(hdenotations)
-    return nil if hdenotations.nil?
-    hdenotations.collect{|d| d.dup.merge({span:transform_a_span(d[:span])})}
-  end
-
-  private
-
-  def _compute_mixed_alignment(str1, str2, mappings = [])
-    lcsmin = TextAlignment::LCSMin.new(str1, str2)
-    lcs = lcsmin.lcs
-    @sdiff = lcsmin.sdiff
-
-    cmp = TextAlignment::LCSComparison.new(str1, str2, lcs, @sdiff)
-    @similarity         = cmp.similarity
-    @str1_match_initial = cmp.str1_match_initial
-    @str1_match_final   = cmp.str1_match_final
-    @str2_match_initial = cmp.str2_match_initial
-    @str2_match_final   = cmp.str2_match_final
-
-    posmap_begin, posmap_end = {}, {}
-    @common_elements, @mapped_elements = [], []
-
-    addition, deletion = [], []
-
-    @sdiff.each do |h|
-      case h.action
-      when '='
-        p1, p2 = h.old_position, h.new_position
-
-        @common_elements << [str1[p1], str2[p2]]
-        posmap_begin[p1], posmap_end[p1] = p2, p2
-
-        if !addition.empty? && deletion.empty?
-          posmap_end[p1] = p2 - addition.length unless p1 == 0
-        elsif addition.empty? && !deletion.empty?
-          deletion.each{|p| posmap_begin[p], posmap_end[p] = p2, p2}
-        elsif !addition.empty? && !deletion.empty?
-          if addition.length > 1 || deletion.length > 1
-            galign = TextAlignment::GLCSAlignment.new(str1[deletion[0] .. deletion[-1]], str2[addition[0] .. addition[-1]], mappings)
-            galign.position_map_begin.each {|k, v| posmap_begin[k + deletion[0]] = v.nil? ? nil : v + addition[0]}
-            galign.position_map_end.each   {|k, v|   posmap_end[k + deletion[0]] = v.nil? ? nil : v + addition[0]}
-            posmap_begin[p1], posmap_end[p1] = p2, p2
-            @common_elements += galign.common_elements
-            @mapped_elements += galign.mapped_elements
-          else
-            posmap_begin[deletion[0]], posmap_end[deletion[0]] = addition[0], addition[0]
-            deletion[1..-1].each{|p| posmap_begin[p], posmap_end[p] = nil, nil}
-            @mapped_elements << [str1[deletion[0], deletion.length], str2[addition[0], addition.length]]
-          end
-        end
-
-        addition.clear; deletion.clear
-
-      when '!'
-        deletion << h.old_position
-        addition << h.new_position
-      when '-'
-        deletion << h.old_position
-      when '+'
-        addition << h.new_position
-      end
-    end
-
-    p1, p2 = str1.length, str2.length
-    posmap_begin[p1], posmap_end[p1] = p2, p2
-
-    if !addition.empty? && deletion.empty?
-      posmap_end[p1] = p2 - addition.length unless p1 == 0
-    elsif addition.empty? && !deletion.empty?
-      deletion.each{|p| posmap_begin[p], posmap_end[p] = p2, p2}
-    elsif !addition.empty? && !deletion.empty?
-      if addition.length > 1 && deletion.length > 1
-        galign = TextAlignment::GLCSAlignment.new(str1[deletion[0] .. deletion[-1]], str2[addition[0] .. addition[-1]], mappings)
-        galign.position_map_begin.each {|k, v| posmap_begin[k + deletion[0]] = v.nil? ? nil : v + addition[0]}
-        galign.position_map_end.each   {|k, v|   posmap_end[k + deletion[0]] = v.nil? ? nil : v + addition[0]}
-        posmap_begin[p1], posmap_end[p1] = p2, p2
-        @common_elements += galign.common_elements
-        @mapped_elements += galign.mapped_elements
-      else
-        posmap_begin[deletion[0]], posmap_end[deletion[0]] = addition[0], addition[0]
-        deletion[1..-1].each{|p| posmap_begin[p], posmap_end[p] = nil, nil}
-        @mapped_elements << [str1[deletion[0], deletion.length], str2[addition[0], addition.length]]
-      end
-    end
-
-    @position_map_begin = posmap_begin.sort.to_h
-    @position_map_end = posmap_end.sort.to_h
-  end
+	attr_reader :block_alignments
+	attr_reader :similarity
+	attr_reader :lost_annotations
+
+	def initialize(str1, str2, mappings = nil)
+		raise ArgumentError, "nil string" if str1.nil? || str2.nil?
+
+		mappings ||= TextAlignment::MAPPINGS
+
+		# try exact match
+		block_begin = str2.index(str1)
+		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)
+
+		# To collect matched blocks
+		mblocks = []
+		while anchor = anchor_finder.get_next_anchor
+			last = mblocks.last
+			if last && (anchor[:source][:begin] == last[:source][:end] + 1) && (anchor[:target][:begin] == last[:target][:end] + 1)
+				last[:source][:end] = anchor[:source][:end]
+				last[:target][:end] = anchor[:target][:end]
+			else
+				mblocks << anchor
+			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
+
+		## To find block alignments
+		@block_alignments = []
+		return if mblocks.empty?
+
+		# Initial step
+		if mblocks[0][:source][:begin] > 0
+			e1 = mblocks[0][:source][:begin]
+			e2 = mblocks[0][:target][:begin]
+
+			if mblocks[0][:target][:begin] == 0
+				@block_alignments << {source:{begin:0, end:e1}, target:{begin:0, end:0}, alignment: :empty}
+			else
+				_str1 = str1[0 ... e1]
+				_str2 = str2[0 ... e2]
+
+				unless _str1.strip.empty?
+					if _str2.strip.empty?
+						@block_alignments << {source:{begin:0, end:e1}, target:{begin:0, end:e2}, alignment: :empty}
+					else
+						len_min = [_str1.length, _str2.length].min
+						len_buffer = (len_min * (1 + TextAlignment::BUFFER_RATE)).to_i
+						b1 = _str1.length < len_buffer ? 0 : e1 - len_buffer
+						b2 = _str2.length < len_buffer ? 0 : e2 - len_buffer
+
+						@block_alignments << {source:{begin:0, end:b1}, target:{begin:0, end:b2}, alignment: :empty} if b1 > 0
+
+						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}
+						else
+							@block_alignments << {source:{begin:b1, end:e1}, target:{begin:0, end:e2}, alignment:alignment}
+						end
+					end
+				end
+			end
+		end
+		@block_alignments << mblocks[0]
+
+		(1 ... mblocks.length).each do |i|
+			b1 = mblocks[i - 1][:source][:end]
+			b2 = mblocks[i - 1][:target][:end]
+			e1 = mblocks[i][:source][:begin]
+			e2 = mblocks[i][:target][:begin]
+			_str1 = str1[b1 ... e1]
+			_str2 = str2[b2 ... e2]
+			unless _str1.strip.empty?
+				if _str2.strip.empty?
+					@block_alignments << {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_alignments << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}
+					else
+						@block_alignments << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment:alignment}
+					end
+				end
+			end
+			@block_alignments << mblocks[i]
+		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]
+
+			unless _str1.strip.empty?
+				if _str2.strip.empty?
+					@block_alignments << {source:{begin:b1, end:str1.length}, target:{begin:b2, end:str2.length}, alignment: :empty}
+				else
+					len_min = [_str1.length, _str2.length].min
+					len_buffer = (len_min * (1 + TextAlignment::BUFFER_RATE)).to_i
+					e1 = _str1.length < len_buffer ? str1.length : b1 + len_buffer
+					e2 = _str2.length < len_buffer ? str1.length : b2 + len_buffer
+
+					alignment = TextAlignment::MixedAlignment.new(_str1.downcase, _str2.downcase, mappings)
+					if alignment.similarity < 0.6
+						@block_alignments << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment: :empty}
+					else
+						@block_alignments << {source:{begin:b1, end:e1}, target:{begin:b2, end:e2}, alignment:alignment}
+					end
+
+					@block_alignments << {source:{begin:e1, end:-1}, target:{begin:e2, end:-1}, alignment: :empty} if e1 < str1.length
+				end
+			end
+		end
+
+		@block_alignments.each do |a|
+			a[:delta] = a[:target][:begin] - a[:source][:begin]
+		end
+	end
+
+	def transform_begin_position(begin_position)
+		i = @block_alignments.index{|b| b[:source][:end] > begin_position}
+		block_alignment = @block_alignments[i]
+
+		b = if block_alignment[:alignment].nil?
+			begin_position + block_alignment[:delta]
+		elsif block_alignment[:alignment] == :empty
+			if begin_position == block_alignment[:source][:begin]
+				block_alignment[:target][:begin]
+			else
+				raise "lost annotation"
+			end
+		else
+			block_alignment[:alignment].transform_begin_position(begin_position - block_alignment[:source][:begin]) + block_alignment[:target][:begin]
+		end
+	end
+
+	def transform_end_position(end_position)
+		i = @block_alignments.index{|b| b[:source][:end] >= end_position}
+		block_alignment = @block_alignments[i]
+
+		e = if block_alignment[:alignment].nil?
+			end_position + block_alignment[:delta]
+		elsif block_alignment[:alignment] == :empty
+			if end_position == block_alignment[:source][:end]
+				block_alignment[:target][:end]
+			else
+				raise "lost annotation"
+			end
+		else
+			block_alignment[:alignment].transform_end_position(end_position - block_alignment[:source][:begin]) + block_alignment[:target][:begin]
+		end
+	end
+
+	def transform_a_span(span)
+		{begin: transform_begin_position(span[:begin]), end: transform_end_position(span[:end])}
+	end
+
+	def transform_spans(spans)
+		spans.map{|span| transform_a_span(span)}
+	end
+
+	def transform_denotations!(denotations)
+		puts "hereherehere========"
+		return nil if denotations.nil?
+		@lost_annotations = []
+
+		pp denotations
+		puts "-----"
+
+		denotations.each do |d|
+			begin
+				d.begin = transform_begin_position(d.begin);
+				d.end = transform_end_position(d.end);
+			rescue
+				@lost_annotations << d
+				d.begin = nil
+				d.end = nil
+			end
+		end
+
+		pp denotations
+	end
+
+	def transform_hdenotations(hdenotations)
+		return nil if hdenotations.nil?
+		@lost_annotations = []
+
+		r = hdenotations.collect do |d|
+			new_d = begin
+				d.dup.merge({span:transform_a_span(d[:span])})
+			rescue
+				@lost_annotations << d
+				nil
+			end
+		end.compact
+
+		r
+	end
+
 end