text_alignment 0.10.1 → 0.11.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 710fa3dfe07e268c62de77c67abb52e1893a36a9c0fbdfbf7400328ab7f8358a
-  data.tar.gz: 31a086666978c4776a60b0ce19fde48beeb8a5bce599b7572f20089cfb5c7500
+  metadata.gz: 4d626b64acdca0630dc344e2f0f4c2152e481ca7c8209ac1aa8c025fb1ae7c0f
+  data.tar.gz: 972222f1f3a575cdb30cd83f886d6d3f36109b79a35cc2c970aa89cbe8fd007d
 SHA512:
-  metadata.gz: d0a752d2203f65e48b4f10ff29eb29a77504e7cef4da1dd3602e2ea5107a99b7edff4c3609c79c7dcba0f9cc458780dc4788ff8123a781a4d77ae89236170bf5
-  data.tar.gz: '094ff5f662ccdc2fc14f81549126362be49fd64610f994c6ba5c4457fbbbececc1931257589ebde49211c2bce744ca033bfe3673c8af5c8a01af7679bfb626cc'
+  metadata.gz: 288691e9ee650af3b9dfe5b47f1a3d05ff952292958ef733a5a5f47aeeaa73e6e1b7b0591afe3549b6cbabf04817bf406d3ca2ee55faf65b7e3816dfe52be644
+  data.tar.gz: 7f45baecc36f00310f868a596ca60bce709078bdc51a5f9438a916cd42062bb93dacadafb64e90e42276d0633914c978045fed4b597ea7411c8762761be6bd36

data/bin/align_annotations

@@ -26,9 +26,8 @@ def read_text(filename)
 	end
 end
 
-def align_denotations(denotations, source_text, target_text, debug = false, cm = nil)
-	alignment = TextAlignment::TextAlignment.new(source_text, target_text, denotations, cm)
-	cm = alignment.cultivation_map
+def align_denotations(denotations, source_text, alignment, debug = false)
+	alignment.align(source_text, denotations)
 	new_denotations = alignment.transform_hdenotations(denotations)
 
 	if debug
@@ -48,23 +47,22 @@ def align_denotations(denotations, source_text, target_text, debug = false, cm =
 	warn
 
 	# return target annotations
-	[new_denotations, cm]
+	new_denotations
 end
 
-def align_mannotations(source_annotations, target_text, debug = false)
-	target_annotations = {text:target_text}
+def align_mannotations(source_annotations, reference_text, alignment, debug = false)
+	target_annotations = {text:reference_text}
 
 	idnum_denotations = 0
 	idnum_relations = 0
 	idnum_attributes = 0
 	idnum_modifications = 0
 
-	cm = nil
 	source_annotations.each_with_index do |annotations, i|
 		if annotations.has_key?(:denotations) && !annotations[:denotations].empty?
 			ididx = {}
 			warn "[#{i}]-=-=-=-=-"
-			denotations, cm = align_denotations(annotations[:denotations], annotations[:text], target_text, debug, cm)
+			denotations = align_denotations(annotations[:denotations], annotations[:text], alignment, debug)
 
 			denotations.each do |d|
 				reid = 'T' + (idnum_denotations += 1).to_s
@@ -112,14 +110,16 @@ unless ARGV.length == 2
 end
 
 source_annotations = read_annotations(ARGV[0])
-target_text = read_text(ARGV[1])
+reference_text = read_text(ARGV[1])
+
+alignment = TextAlignment::TextAlignment.new(reference_text, true)
 
 target_annotations = if source_annotations.class == Array
-	align_mannotations(source_annotations, target_text, false)
+	align_mannotations(source_annotations, reference_text, alignment, false)
 else
-	denotations, cm = align_denotations(source_annotations[:denotations], source_annotations[:text], target_text, false)
-	# denotations = align_denotations(source_annotations[:denotations], source_annotations[:text], target_text, true)
-	source_annotations.merge({text:target_text, denotations:denotations})
+	denotations = align_denotations(source_annotations[:denotations], source_annotations[:text], alignment)
+	# denotations = align_denotations(source_annotations[:denotations], source_annotations[:text], reference_text, true)
+	source_annotations.merge({text:reference_text, denotations:denotations})
 end
 
 puts target_annotations.to_json

data/lib/text_alignment/char_mapping.rb

@@ -77,11 +77,11 @@ TextAlignment::CHAR_MAPPING = [
 
 
 class TextAlignment::CharMapping
-	attr_reader :str
+	attr_reader :mapped_text
 
-	def initialize(_str, char_mapping = nil)
+	def initialize(_text, char_mapping = nil)
 		char_mapping ||= TextAlignment::CHAR_MAPPING
-		@str, offset_mapping = enmap_str(_str, char_mapping)
+		@mapped_text, offset_mapping = enmap_text(_text, char_mapping)
 		@index_enmap = offset_mapping.to_h
 		@index_demap = offset_mapping.map{|m| m.reverse}.to_h
 	end
@@ -95,6 +95,8 @@ class TextAlignment::CharMapping
 	end
 
 	def enmap_denotations(_denotations)
+		return nil if _denotations.nil?
+
 		denotations = _denotations.map do |d|
 			d.dup.merge(span:{begin:enmap_position(d[:span][:begin]), end:enmap_position(d[:span][:end])})
 		end
@@ -102,12 +104,12 @@ class TextAlignment::CharMapping
 
 	private
 
-	def enmap_str(_str, char_mapping)
-		str = _str.dup
+	def enmap_text(_text, char_mapping)
+		text = _text.dup
 
 		# To execute the single letter mapping
 		char_mapping.each do |one, long|
-			str.gsub!(one, long) if long.length == 1
+			text.gsub!(one, long) if long.length == 1
 		end
 
 		# To get the (location, length) index for replacements
@@ -116,18 +118,18 @@ class TextAlignment::CharMapping
 			next if long.length == 1
 
 			init_next = 0
-			while loc = str.index(long, init_next)
+			while loc = text.index(long, init_next)
 				loc_len << [loc, long.length]
 				init_next = loc + long.length
 			end
 
 			# a workaround to avoid messing-up due to embedding
-			str.gsub!(long, one * long.length)
+			text.gsub!(long, one * long.length)
 		end
 
 		# To get the (location, length) index for consecutive whitespace sequences
 		init_next = 0
-		while loc = str.index(/\s{2,}/, init_next)
+		while loc = text.index(/\s{2,}/, init_next)
 			len = $~[0].length
 			loc_len << [loc, len]
 			init_next = loc + len
@@ -148,20 +150,20 @@ class TextAlignment::CharMapping
 			init_next = loc + len
 		end
 
-		offset_mapping += (init_next .. str.length).map do |i|
+		offset_mapping += (init_next .. text.length).map do |i|
 			j += 1
 			[i, j - 1]
 		end
 
 		# To execute the long letter mapping
 		char_mapping.each do |one, long|
-			str.gsub!(one * long.length, one) if long.length > 1
+			text.gsub!(one * long.length, one) if long.length > 1
 		end
 
 		# To replace multi whitespace sequences to a space
-		str.gsub!(/\s{2,}/, ' ')
+		text.gsub!(/\s{2,}/, ' ')
 
-		[str, offset_mapping]
+		[text, offset_mapping]
 	end
 end
 
@@ -178,10 +180,10 @@ if __FILE__ == $0
 		denotations = annotations[:tracks].first[:denotations]
 	end
 
-	str_mapping = TextAlignment::CharMapping.new(annotations[:text])
-	str_mapped = str_mapping.str
-	denotations_mapped = str_mapping.enmap_denotations(denotations)
-	new_annotations = {text:str_mapped, denotations:denotations_mapped}
+	text_mapping = TextAlignment::CharMapping.new(annotations[:text])
+	text_mapped = text_mapping.mapped_text
+	denotations_mapped = text_mapping.enmap_denotations(denotations)
+	new_annotations = {text:text_mapped, denotations:denotations_mapped}
 
 	puts new_annotations.to_json
 end

data/lib/text_alignment/text_alignment.rb

@@ -10,40 +10,59 @@ class TextAlignment::TextAlignment
 	attr_reader :block_alignment
 	attr_reader :similarity
 	attr_reader :lost_annotations
-	attr_reader :cultivation_map
 
-	def initialize(_str1, _str2, _denotations = nil, _cultivation_map = nil)
-		raise ArgumentError, "nil string" if _str1.nil? || _str2.nil?
+	# Initialize with a reference text, again which texts will be aligned
+	def initialize(reference_text, to_prevent_overlap = false)
+		raise ArgumentError, "nil text" if reference_text.nil?
 
-		@block_alignment = {source_text: _str1, target_text: _str2, denotations: _denotations}
-		@original_str1 = _str1
-		@original_str2 = _str2
+		@original_rtext = reference_text
+		@rtext_mapping = TextAlignment::CharMapping.new(reference_text)
+		@to_prevent_overlap = to_prevent_overlap
 
-		@str1_mapping = TextAlignment::CharMapping.new(_str1)
-		@str2_mapping = TextAlignment::CharMapping.new(_str2)
+		@original_text = nil
+		@block_alignment = nil
+		@cultivation_map = TextAlignment::CultivationMap.new
+	end
+
+	def align(text, denotations = nil)
+		# To maintain the cultivation map
+		update_cultivation_map if @to_prevent_overlap
+
+		# In case the input text is the same as the previous one, reuse the previous text mapping
+		unless @original_text && @original_text == text
+			@original_text = text
+			@text_mapping = TextAlignment::CharMapping.new(text)
+		end
 
-		str1 = @str1_mapping.str
-		denotations = @str1_mapping.enmap_denotations(_denotations)
+		text_mapped = @text_mapping.mapped_text
+		denotations_mapped = @text_mapping.enmap_denotations(denotations)
 
-		str2 = @str2_mapping.str
+		rtext_mapped = @rtext_mapping.mapped_text
 
-		@cultivation_map = _cultivation_map || TextAlignment::CultivationMap.new
+		## To generate the block_alignment of the input text against the reference text
 
-		@block_alignment[:blocks] = if r = whole_block_alignment(str1, str2, @cultivation_map)
-			# whole block alignment
+		# Initialization
+		@block_alignment = {text: @original_text, reference_text: @original_rtext, denotations: denotations}
+
+		# Generation
+		@block_alignment[:blocks] = if r = whole_block_alignment(text_mapped, rtext_mapped, @cultivation_map)
 			r
 		else
-			find_block_alignment(str1, str2, denotations, @cultivation_map)
+			find_block_alignment(text_mapped, rtext_mapped, denotations_mapped, @cultivation_map)
 		end
+	end
 
+	def update_cultivation_map
+		return if @block_alignment.nil? || @block_alignment[:blocks].nil?
+
+		## To update the cultivation map
 		newly_cultivated_regions = @block_alignment[:blocks].collect do |b|
 			if b[:alignment] == :block || b[:alignment] == :term
 				[b[:target][:begin], b[:target][:end]]
 			else
 				nil
 			end
-		end.compact
-		newly_cultivated_regions_condensed = newly_cultivated_regions.inject([]) do |condensed, region|
+		end.compact.inject([]) do |condensed, region|
 			if condensed.empty? || (condensed.last.last + 1 < region.first)
 				condensed.push region
 			else
@@ -52,11 +71,11 @@ class TextAlignment::TextAlignment
 			condensed
 		end
 
-		@cultivation_map.cultivate(newly_cultivated_regions_condensed)
+		@cultivation_map.cultivate(newly_cultivated_regions)
 	end
 
 	def transform_begin_position(_begin_position)
-		begin_position = @str1_mapping.enmap_position(_begin_position)
+		begin_position = @text_mapping.enmap_position(_begin_position)
 
 		i = @block_alignment[:blocks].index{|b| b[:source][:end] > begin_position}
 		block = @block_alignment[:blocks][i]
@@ -74,11 +93,11 @@ class TextAlignment::TextAlignment
 			r.nil? ? nil : r + block[:target][:begin]
 		end
 
-		@str2_mapping.demap_position(b)
+		@rtext_mapping.demap_position(b)
 	end
 
 	def transform_end_position(_end_position)
-		end_position = @str1_mapping.enmap_position(_end_position)
+		end_position = @text_mapping.enmap_position(_end_position)
 
 		i = @block_alignment[:blocks].index{|b| b[:source][:end] >= end_position}
 		block = @block_alignment[:blocks][i]
@@ -96,7 +115,7 @@ class TextAlignment::TextAlignment
 			r.nil? ? nil : r + block[:target][:begin]
 		end
 
-		@str2_mapping.demap_position(e)
+		@rtext_mapping.demap_position(e)
 	end
 
 	def transform_a_span(span)
@@ -115,7 +134,7 @@ 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_str2.length
+			raise "invalid transform" unless !d.begin.nil? && !d.end.nil? && d.begin >= 0 && d.end > d.begin && d.end <= @original_rtext.length
 		rescue
 			@lost_annotations << {source: source, target:{begin:d.begin, end:d.end}}
 			d.begin = nil
@@ -131,7 +150,7 @@ 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_str2.length
+			raise "invalid transform" unless !t[:begin].nil? && !t[:end].nil? && t[:begin] >= 0 && t[:end] > t[:begin] && t[:end] <= @original_rtext.length
 			new_d = d.dup.merge({span:t})
 		rescue
 			@lost_annotations << {source: d[:span], target:t}
@@ -142,8 +161,8 @@ class TextAlignment::TextAlignment
 	end
 
 	def alignment_show
-		stext = @block_alignment[:source_text]
-		ttext = @block_alignment[:target_text]
+		stext = @block_alignment[:text]
+		ttext = @block_alignment[:reference_text]
 
 		show = ''
 		@block_alignment[:blocks].each do |a|

data/lib/text_alignment/version.rb

@@ -1,3 +1,3 @@
 class TextAlignment
-	VERSION = '0.10.1'
+	VERSION = '0.11.0'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: text_alignment
 version: !ruby/object:Gem::Version
-  version: 0.10.1
+  version: 0.11.0
 platform: ruby
 authors:
 - Jin-Dong Kim
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2021-03-03 00:00:00.000000000 Z
+date: 2021-03-04 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ruby-dictionary