From 61a590e9bea64ddfd465a5e6f224bc2979d841e9 Mon Sep 17 00:00:00 2001
From: Xueming Shen <sherman@openjdk.org>
Date: Tue, 1 Jul 2025 00:58:43 +0000
Subject: [PATCH] 8354490: Pattern.CANON_EQ causes a pattern to not match a
 string with a UNICODE variation

Reviewed-by: rriggs, naoto
---
 .../share/classes/java/util/regex/Pattern.java    |  5 +----
 test/jdk/java/util/regex/RegExTest.java           | 15 +++++++++++++++
 2 files changed, 16 insertions(+), 4 deletions(-)

diff --git a/src/java.base/share/classes/java/util/regex/Pattern.java b/src/java.base/share/classes/java/util/regex/Pattern.java
index b7f03c1b0af..fd5627e2a00 100644
--- a/src/java.base/share/classes/java/util/regex/Pattern.java
+++ b/src/java.base/share/classes/java/util/regex/Pattern.java
@@ -4154,7 +4154,7 @@ loop:   for(int x=0, offset=0; x<nCodePoints; x++, offset+=len) {
                     if (predicate.is(ch0))
                         return next.match(matcher, j, seq);
                 } else {
-                    while (i + n < j) {
+                    while (i + n <= j) {
                         String nfc = Normalizer.normalize(
                             seq.toString().substring(i, j), Normalizer.Form.NFC);
                         if (nfc.codePointCount(0, nfc.length()) == 1) {
@@ -4163,13 +4163,10 @@ loop:   for(int x=0, offset=0; x<nCodePoints; x++, offset+=len) {
                                 return true;
                             }
                         }
-
                         ch0 = Character.codePointBefore(seq, j);
                         j -= Character.charCount(ch0);
                     }
                 }
-                if (j < matcher.to)
-                    return false;
             } else {
                 matcher.hitEnd = true;
             }
diff --git a/test/jdk/java/util/regex/RegExTest.java b/test/jdk/java/util/regex/RegExTest.java
index 3338b4f5c2d..2bd2dc3e38d 100644
--- a/test/jdk/java/util/regex/RegExTest.java
+++ b/test/jdk/java/util/regex/RegExTest.java
@@ -2324,6 +2324,21 @@ public class RegExTest {
         check(p, "test\u00e4\u0300\u0323", true);
 
         Object[][] data = new Object[][] {
+        // JDK-8354490
+        // emoji + emoji_component pair forms a single grapheme but remains
+        // as 2 separate characters in nfc. match & find should still work with the
+        // CANON_EQ flag, as long as the character class is appropriately specified.
+        {"^[^/]*\\.[^/]*$", "\u2764\ufe0ffile.txt",    "m", true},
+        { "\\p{IsEmoji}",     "ab\u2764\ufe0fcd",      "f", true },
+        { "[\\p{IsEmoji}]",   "ab\u2764\ufe0fcd",      "f", true },
+        { "\\p{IsEmoji}\\p{IsEmoji_Component}",       "\u2764\ufe0f", "m", true },
+        { "[\\p{IsEmoji}\\p{IsEmoji_Component}]{2}",  "\u2764\ufe0f", "m", true },
+        // greek with extra combining character
+        {"\\p{IsGreek}", "\u1f80\u0345", "f", true},
+        {"[\\p{IsGreek}]", "\u1f80\u0345", "f", true},
+        {"\\p{IsGreek}\\p{IsAlphabetic}",             "\u1f80\u0345", "m", true},
+        {"\\p{IsAlphabetic}*",                        "\u1f80\u0345", "m", true},
+        {"[\\p{IsAlphabetic}]*",                      "\u1f80\u0345", "m", true},
 
         // JDK-4867170
         { "[\u1f80-\u1f82]", "ab\u1f80cd",             "f", true },