2 files changed, 66 insertions, 14 deletions
diff --git a/lib/unicode_normalize/normalize.rb b/lib/unicode_normalize/normalize.rb
index e67fad187a..0447df8de7 100644
--- a/lib/unicode_normalize/normalize.rb
+++ b/lib/unicode_normalize/normalize.rb
@@ -82,16 +82,22 @@ module UnicodeNormalize  # :nodoc:
 
   ## Canonical Ordering
   def self.canonical_ordering_one(string)
-    sorting = string.each_char.collect { |c| [c, CLASS_TABLE[c]] }
-    (sorting.length-2).downto(0) do |i| # almost, but not exactly bubble sort
-      (0..i).each do |j|
-        later_class = sorting[j+1].last
-        if 0<later_class and later_class<sorting[j].last
-          sorting[j], sorting[j+1] = sorting[j+1], sorting[j]
-        end
+    result = ''
+    unordered = []
+    chars = string.chars
+    n = chars.size
+    chars.each_with_index do |char, i|
+      ccc = CLASS_TABLE[char]
+      if ccc == 0
+        unordered.sort!.each { result << chars[it % n] }
+        unordered.clear
+        result << char
+      else
+        unordered << ccc * n + i
       end
     end
-    return sorting.collect(&:first).join('')
+    unordered.sort!.each { result << chars[it % n] }
+    result
   end
 
   ## Normalization Forms for Patterns (not whole Strings)
@@ -105,17 +111,22 @@ module UnicodeNormalize  # :nodoc:
     start = nfd_string[0]
     last_class = CLASS_TABLE[start]-1
     accents = ''
+    result = ''
     nfd_string[1..-1].each_char do |accent|
       accent_class = CLASS_TABLE[accent]
       if last_class<accent_class and composite = COMPOSITION_TABLE[start+accent]
         start = composite
+      elsif accent_class == 0
+        result << start << accents
+        start = accent
+        accents = ''
+        last_class = -1
       else
         accents << accent
         last_class = accent_class
       end
     end
-    accents = nfc_one(accents) if accents.length>1 # TODO: change from recursion to loop
-    hangul_comp_one(start+accents)
+    hangul_comp_one(result+start+accents)
   end
 
   def self.normalize(string, form = :nfc)
diff --git a/lib/unicode_normalize/tables.rb b/lib/unicode_normalize/tables.rb
index b5b708defd..dd5d3499b8 100644
--- a/lib/unicode_normalize/tables.rb
+++ b/lib/unicode_normalize/tables.rb
@@ -1,8 +1,8 @@
 # coding: us-ascii
 # frozen_string_literal: true
 
-Encoding::UNICODE_VERSION == "16.0.0" or
-  raise "Unicode version mismatch: 16.0.0 expected but #{Encoding::UNICODE_VERSION}"
+Encoding::UNICODE_VERSION == "17.0.0" or
+  raise "Unicode version mismatch: 17.0.0 expected but #{Encoding::UNICODE_VERSION}"
 
 # automatically generated by template/unicode_norm_gen.tmpl
 
@@ -99,7 +99,8 @@ module UnicodeNormalize  # :nodoc:
     "\u1A75-\u1A7C" \
     "\u1A7F" \
     "\u1AB0-\u1ABD" \
-    "\u1ABF-\u1ACE" \
+    "\u1ABF-\u1ADD" \
+    "\u1AE0-\u1AEB" \
     "\u1B34\u1B35" \
     "\u1B44" \
     "\u1B6B-\u1B73" \
@@ -154,6 +155,7 @@ module UnicodeNormalize  # :nodoc:
     "\u{10D24}-\u{10D27}" \
     "\u{10D69}-\u{10D6D}" \
     "\u{10EAB}\u{10EAC}" \
+    "\u{10EFA}\u{10EFB}" \
     "\u{10EFD}-\u{10EFF}" \
     "\u{10F46}-\u{10F50}" \
     "\u{10F82}-\u{10F85}" \
@@ -230,6 +232,10 @@ module UnicodeNormalize  # :nodoc:
     "\u{1E2EC}-\u{1E2EF}" \
     "\u{1E4EC}-\u{1E4EF}" \
     "\u{1E5EE}\u{1E5EF}" \
+    "\u{1E6E3}" \
+    "\u{1E6E6}" \
+    "\u{1E6EE}\u{1E6EF}" \
+    "\u{1E6F5}" \
     "\u{1E8D0}-\u{1E8D6}" \
     "\u{1E944}-\u{1E94A}" \
     "]"
@@ -1460,7 +1466,7 @@ module UnicodeNormalize  # :nodoc:
     "\u3280-\u33FF" \
     "\uA69C\uA69D" \
     "\uA770" \
-    "\uA7F2-\uA7F4" \
+    "\uA7F1-\uA7F4" \
     "\uA7F8\uA7F9" \
     "\uAB5C-\uAB5F" \
     "\uAB69" \
@@ -2019,6 +2025,33 @@ module UnicodeNormalize  # :nodoc:
     "\u1ACC"=>230,
     "\u1ACD"=>230,
     "\u1ACE"=>230,
+    "\u1ACF"=>230,
+    "\u1AD0"=>230,
+    "\u1AD1"=>230,
+    "\u1AD2"=>230,
+    "\u1AD3"=>230,
+    "\u1AD4"=>230,
+    "\u1AD5"=>230,
+    "\u1AD6"=>230,
+    "\u1AD7"=>230,
+    "\u1AD8"=>230,
+    "\u1AD9"=>230,
+    "\u1ADA"=>230,
+    "\u1ADB"=>230,
+    "\u1ADC"=>230,
+    "\u1ADD"=>220,
+    "\u1AE0"=>230,
+    "\u1AE1"=>230,
+    "\u1AE2"=>230,
+    "\u1AE3"=>230,
+    "\u1AE4"=>230,
+    "\u1AE5"=>230,
+    "\u1AE6"=>220,
+    "\u1AE7"=>230,
+    "\u1AE8"=>230,
+    "\u1AE9"=>230,
+    "\u1AEA"=>230,
+    "\u1AEB"=>234,
     "\u1B34"=>7,
     "\u1B44"=>9,
     "\u1B6B"=>230,
@@ -2293,6 +2326,8 @@ module UnicodeNormalize  # :nodoc:
     "\u{10D6D}"=>230,
     "\u{10EAB}"=>230,
     "\u{10EAC}"=>230,
+    "\u{10EFA}"=>220,
+    "\u{10EFB}"=>220,
     "\u{10EFD}"=>220,
     "\u{10EFE}"=>220,
     "\u{10EFF}"=>220,
@@ -2479,6 +2514,11 @@ module UnicodeNormalize  # :nodoc:
     "\u{1E4EF}"=>230,
     "\u{1E5EE}"=>230,
     "\u{1E5EF}"=>220,
+    "\u{1E6E3}"=>230,
+    "\u{1E6E6}"=>230,
+    "\u{1E6EE}"=>230,
+    "\u{1E6EF}"=>230,
+    "\u{1E6F5}"=>230,
     "\u{1E8D0}"=>220,
     "\u{1E8D1}"=>220,
     "\u{1E8D2}"=>220,
@@ -5922,6 +5962,7 @@ module UnicodeNormalize  # :nodoc:
     "\uA69C"=>"\u044A",
     "\uA69D"=>"\u044C",
     "\uA770"=>"\uA76F",
+    "\uA7F1"=>"S",
     "\uA7F2"=>"C",
     "\uA7F3"=>"F",
     "\uA7F4"=>"Q",