text_alignment 0.12.7 → 0.12.9

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 354d6971ab0c7d6e80af0633512d5f7257b5e6e1a4bef021bfc1e1eee6e546c1
-  data.tar.gz: bae6c626e170de96ae32beb24c9d074b1e56283e497dffc53c2da270c20de058
+  metadata.gz: 7da208b2cd252983fd0c7f8378130f7b13bba3df1c698f55e2133d54d9dab61d
+  data.tar.gz: 998405eb7c03b065faae083368a285cf526a193d7b59d22aafce643887b4150e
 SHA512:
-  metadata.gz: 3409a5c7419c43a311e76468a6783b6a6a808c12dd14343eace7c0242dc3cd79d01616069dce12d49f5db8cc77b293ab469efa07b6a5878e1bade341f24f1c1d
-  data.tar.gz: fe7df5b352f14989701c0dbf0bc94169b415a5c155b7e04f670803d154576e4a39ea7dddf3f320a9df8e8e9c3fb3bc67ddc051d7e3d939ca9bc7a58de1b3b952
+  metadata.gz: 246e0796040cedd989e12f7ef51f33a3209e0f82a396813ad9062323459e731c8d7b651a2461b732d90119c2e95586e1dbc296db91a99aeca574f4337aca6c8a
+  data.tar.gz: 06dedaa086dd878e41d26d18e427cf1d7c564583b34d047758a57e4508506cba6714abe9d8e04fa63c474ec2cc3fb9a585fef00312efe409c22b73528bdbde97

data/lib/text_alignment/anchor_finder.rb

@@ -32,12 +32,31 @@ class TextAlignment::AnchorFinder
 
 	def get_next_anchor
 		# To find the beginning positions of an anchor ngram in s1 and s2, beginning from the last positions matched
+		iterations = 0
 		beg_s2 = for beg_s1 in @pos_s1_last_match .. @pos_s1_final_possible_begin
+			iterations += 1
 
 			# To skip whitespace letters
 			next if [' ', "\n", "\t"].include? @s1[beg_s1]
 
+			# Skip positions that start with punctuation or numbers (likely poor anchors)
+			next if @s1[beg_s1] =~ /[^a-zA-Z]/
+
+			# Performance optimization: skip if we've had too many failed attempts recently
+			if iterations > 50 && @recent_failures && @recent_failures > 10
+				step_size = [@recent_failures / 5, 3].min
+				beg_s1 += step_size
+				next if beg_s1 > @pos_s1_final_possible_begin
+			end
+
 			_beg_s2 = get_beg_s2(beg_s1)
+
+			if _beg_s2.nil?
+				@recent_failures = (@recent_failures || 0) + 1
+			else
+				@recent_failures = 0  # Reset on success
+			end
+
 			break _beg_s2 unless _beg_s2.nil?
 		end
 
@@ -72,7 +91,16 @@ class TextAlignment::AnchorFinder
 		# to get the anchor to search for in s2
 		anchor = @s1[beg_s1, @size_ngram]
 
+		# Quick frequency check: skip very short or very common ngrams
+		return nil if anchor.length < @size_ngram
+		return nil if anchor =~ /^(.)\1+$/ # Skip repeating character patterns like "aaaaaaaa"
+
+		# Skip ngrams that are mostly whitespace or punctuation
+		non_alnum_count = anchor.count("^a-zA-Z0-9")
+		return nil if non_alnum_count > @size_ngram / 2
+
 		search_position = @to_ignore_text_order ? 0 : @pos_s2_last_match
+
 		beg_s2_candidates = find_beg_s2_candidates(anchor, search_position)
 		return nil if beg_s2_candidates.empty?
 
@@ -87,7 +115,7 @@ class TextAlignment::AnchorFinder
 			candidates << _beg_s2
 
 			# for speed, skip anchor of high frequency
-			if candidates.length > 5
+			if candidates.length > 3
 				candidates.clear
 				break
 			end

data/lib/text_alignment/char_mapping.rb

@@ -202,7 +202,7 @@ class TextAlignment::CharMapping
 
 		# To execute the long letter mapping
 		char_mapping.each do |one, long|
-			next unless text =~ /#{one}/
+			next unless text.include?(one)
 			text.gsub!(one * long.length, one) if long.length > 1
 		end
 
@@ -215,17 +215,29 @@ class TextAlignment::CharMapping
 	# To get squeeze positions of whitespaces to one
 	def get_positions_squeeze_ws_1(text)
 		rpositions = []
-		text.scan(/\s{2,}/) do |s|
-			loc = $~.begin(0)
-			len = $~.end(0) - loc
+		scanner    = StringScanner.new(text)
+
+		while scanner.scan_until(/\s{2,}/)
+			len = scanner.matched_size
+			loc = scanner.pos - len
 			rpositions << [loc, len, 1]
 		end
+
 		rpositions
 	end
 
 	# To get squeeze positions of whitespaces to zero
 	def get_positions_squeeze_ws_0(text)
-		text.enum_for(:scan, /\s+/).map{[b = $~.begin(0), $~.end(0) - b, 0]}
+		rpositions = []
+		scanner    = StringScanner.new(text)
+
+		while scanner.scan(/\s+/)
+			len   = scanner.matched_size
+			start = scanner.pos - len
+			rpositions << [start, len, 0]
+		end
+
+		rpositions
 	end
 
 	def squeeze_ws_1!(text)

data/lib/text_alignment/text_alignment.rb

@@ -38,7 +38,6 @@ class TextAlignment::TextAlignment
 			@original_text = text
 			@text_mapping = TextAlignment::CharMapping.new(text, nil, @to_ignore_whitespaces)
 		end
-
 		@mapped_text = @text_mapping.mapped_text
 
 		## To generate the block_alignment of the input text against the reference text
@@ -251,7 +250,7 @@ class TextAlignment::TextAlignment
 				if b2 < e2
 					_str2 = str2[b2 ... e2]
 
-					sum += if _str1.strip.empty? || _str2.strip.empty?
+					gap_result = 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
@@ -307,6 +306,8 @@ class TextAlignment::TextAlignment
 							end
 						end
 					end
+
+					sum += gap_result
 				elsif b2 > e2 # when out of order
 					# ToDo
 				end
@@ -317,6 +318,7 @@ class TextAlignment::TextAlignment
 			cblock.nil? ? sum : sum << cblock
 		end
 
+		blocks2
 	end
 
 	def whole_block_alignment(str1, str2, cultivation_map)

data/lib/text_alignment/version.rb

@@ -1,3 +1,3 @@
 class TextAlignment
-	VERSION = '0.12.7'
+	VERSION = '0.12.9'
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: text_alignment
 version: !ruby/object:Gem::Version
-  version: 0.12.7
+  version: 0.12.9
 platform: ruby
 authors:
 - Jin-Dong Kim
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2025-05-22 00:00:00.000000000 Z
+date: 2025-09-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: ruby-dictionary