[Feature #19908] Update Unicode headers to 15.1.0

Merged: https://github.com/ruby/ruby/pull/12798

NEWS.md

@@ -17,6 +17,11 @@ Note: We're only listing outstanding class updates.
       Also, `Binding#local_variable_get` and `Binding#local_variable_set` reject to handle numbered parameters.
       [[Bug #21049]]
 
+* String
+
+    * Update Unicode to Version 15.1.0 and Emoji Version 15.1. [[Feature #19908]]
+        (also applies to Regexp)
+
 ## Stdlib updates
 
 The following bundled gems are promoted from default gems.
@@ -78,4 +83,5 @@ The following bundled gems are updated.
 
 ## JIT
 
+[Feature #19908]: https://bugs.ruby-lang.org/issues/19908
 [Bug #21049]:     https://bugs.ruby-lang.org/issues/21049

common.mk

@@ -20,7 +20,7 @@ mflags = $(MFLAGS)
 gnumake_recursive =
 enable_shared = $(ENABLE_SHARED:no=)
 
-UNICODE_VERSION = 15.0.0
+UNICODE_VERSION = 15.1.0
 UNICODE_EMOJI_VERSION_0 = $(UNICODE_VERSION)///
 UNICODE_EMOJI_VERSION_1 = $(UNICODE_EMOJI_VERSION_0:.0///=)
 UNICODE_EMOJI_VERSION = $(UNICODE_EMOJI_VERSION_1:///=)

doc/regexp/unicode_properties.rdoc

@@ -79,6 +79,9 @@ Older versions may not support all of these.
 - <tt>\p{Hyphen}</tt>
 - <tt>\p{IDS_Binary_Operator}</tt>, <tt>\p{IDSB}</tt>
 - <tt>\p{IDS_Trinary_Operator}</tt>, <tt>\p{IDST}</tt>
+- <tt>\p{IDS_Unary_Operator}</tt>, <tt>\p{IDSU}</tt>
+- <tt>\p{ID_Compat_Math_Continue}</tt>
+- <tt>\p{ID_Compat_Math_Start}</tt>
 - <tt>\p{Ideographic}</tt>, <tt>\p{Ideo}</tt>
 - <tt>\p{Join_Control}</tt>, <tt>\p{Join_C}</tt>
 - <tt>\p{Logical_Order_Exception}</tt>, <tt>\p{LOE}</tt>
@@ -120,6 +123,9 @@ Older versions may not support all of these.
 - <tt>\p{Grapheme_Link}</tt>, <tt>\p{Gr_Link}</tt>
 - <tt>\p{ID_Continue}</tt>, <tt>\p{IDC}</tt>
 - <tt>\p{ID_Start}</tt>, <tt>\p{IDS}</tt>
+- <tt>\p{InCB_Consonant}</tt>
+- <tt>\p{InCB_Extend}</tt>
+- <tt>\p{InCB_Linker}</tt>
 - <tt>\p{Lowercase}</tt>, <tt>\p{Lower}</tt>
 - <tt>\p{Math}</tt>
 - <tt>\p{Uppercase}</tt>, <tt>\p{Upper}</tt>
@@ -348,6 +354,7 @@ Older versions may not support all of these.
 - <tt>\p{In_CJK_Unified_Ideographs_Extension_F}</tt>
 - <tt>\p{In_CJK_Unified_Ideographs_Extension_G}</tt>
 - <tt>\p{In_CJK_Unified_Ideographs_Extension_H}</tt>
+- <tt>\p{In_CJK_Unified_Ideographs_Extension_I}</tt>
 - <tt>\p{In_Carian}</tt>
 - <tt>\p{In_Caucasian_Albanian}</tt>
 - <tt>\p{In_Chakma}</tt>
@@ -658,6 +665,7 @@ Older versions may not support all of these.
 - <tt>\p{Age_13_0}</tt>
 - <tt>\p{Age_14_0}</tt>
 - <tt>\p{Age_15_0}</tt>
+- <tt>\p{Age_15_1}</tt>
 - <tt>\p{Age_1_1}</tt>
 - <tt>\p{Age_2_0}</tt>
 - <tt>\p{Age_2_1}</tt>

enc/unicode/15.1.0/casefold.h

@@ -3,14 +3,14 @@
 
 #if defined ONIG_UNICODE_VERSION_STRING && !( \
       ONIG_UNICODE_VERSION_MAJOR == 15 && \
-      ONIG_UNICODE_VERSION_MINOR == 0 && \
+      ONIG_UNICODE_VERSION_MINOR == 1 && \
       ONIG_UNICODE_VERSION_TEENY == 0 && \
       1)
 # error ONIG_UNICODE_VERSION_STRING mismatch
 #endif
-#define ONIG_UNICODE_VERSION_STRING "15.0.0"
+#define ONIG_UNICODE_VERSION_STRING "15.1.0"
 #define ONIG_UNICODE_VERSION_MAJOR 15
-#define ONIG_UNICODE_VERSION_MINOR 0
+#define ONIG_UNICODE_VERSION_MINOR 1
 #define ONIG_UNICODE_VERSION_TEENY 0
 
 static const CaseFold_11_Type CaseFold_11_Table[] = {

spec/ruby/library/rbconfig/unicode_emoji_version_spec.rb

@@ -14,10 +14,16 @@ describe "RbConfig::CONFIG['UNICODE_EMOJI_VERSION']" do
     end
   end
 
-  # Caution: ruby_version_is means is_or_later
-  ruby_version_is "3.2" do
+  ruby_version_is "3.2"..."3.4" do
     it "is 15.0" do
       RbConfig::CONFIG['UNICODE_EMOJI_VERSION'].should == "15.0"
     end
   end
+
+  # Caution: ruby_version_is means is_or_later
+  ruby_version_is "3.5" do
+    it "is 15.1" do
+      RbConfig::CONFIG['UNICODE_EMOJI_VERSION'].should == "15.1"
+    end
+  end
 end

spec/ruby/library/rbconfig/unicode_version_spec.rb

@@ -14,10 +14,16 @@ describe "RbConfig::CONFIG['UNICODE_VERSION']" do
     end
   end
 
-  # Caution: ruby_version_is means is_or_later
-  ruby_version_is "3.2" do
+  ruby_version_is "3.2"..."3.4" do
     it "is 15.0.0" do
       RbConfig::CONFIG['UNICODE_VERSION'].should == "15.0.0"
     end
   end
+
+  # Caution: ruby_version_is means is_or_later
+  ruby_version_is "3.5" do
+    it "is 15.1.0" do
+      RbConfig::CONFIG['UNICODE_VERSION'].should == "15.1.0"
+    end
+  end
 end

test/ruby/test_regexp.rb

@@ -1511,6 +1511,19 @@ class TestRegexp < Test::Unit::TestCase
                        "CJK UNIFIED IDEOGRAPH-31350..CJK UNIFIED IDEOGRAPH-323AF")
   end
 
+  def test_unicode_age_15_1
+    @matches   = %w"15.1"
+    @unmatches = %w"15.0"
+
+    # https://www.unicode.org/Public/15.1.0/ucd/DerivedAge.txt
+    assert_unicode_age("\u{2FFC}".."\u{2FFF}",
+                       "IDEOGRAPHIC DESCRIPTION CHARACTER SURROUND FROM RIGHT..IDEOGRAPHIC DESCRIPTION CHARACTER ROTATION")
+    assert_unicode_age("\u{31EF}",
+                       "IDEOGRAPHIC DESCRIPTION CHARACTER SUBTRACTION")
+    assert_unicode_age("\u{2EBF0}".."\u{2EE5D}",
+                       "CJK UNIFIED IDEOGRAPH-2EBF0..CJK UNIFIED IDEOGRAPH-2EE5D")
+  end
+
   UnicodeAgeRegexps = Hash.new do |h, age|
     h[age] = [/\A\p{age=#{age}}+\z/u, /\A\P{age=#{age}}+\z/u].freeze
   end