ext/mbstring: Update to Unicode 15.1

NEWS

@@ -123,6 +123,7 @@ PHP                                                                        NEWS
 - MBString:
   . Added mb_trim, mb_ltrim and mb_rtrim. (Yuya Hamada)
   . Added mb_ucfirst and mb_lcfirst. (Yuya Hamada)
+  . Updated Unicode data tables to Unicode 15.1. (Ayesh Karunaratne)
 
 - MySQLnd:
   . Fixed bug GH-13440 (PDO quote bottleneck). (nielsdos)

UPGRADING

@@ -684,6 +684,9 @@ PHP 8.4 UPGRADE NOTES
     $domain name is empty or too long, and if $variant is not
     INTL_IDNA_VARIANT_UTS46.
 
+- MBString:
+  . Unicode data tables have been updated to Unicode 15.1.
+
 - OpenSSL:
   . The OpenSSL extension now requires at least OpenSSL 1.1.1.
 

ext/mbstring/libmbfl/mbfl/eaw_table.h

@@ -58,14 +58,13 @@ static const struct {
 	{ 0x2e80, 0x2e99 },
 	{ 0x2e9b, 0x2ef3 },
 	{ 0x2f00, 0x2fd5 },
-	{ 0x2ff0, 0x2ffb },
-	{ 0x3000, 0x303e },
+	{ 0x2ff0, 0x303e },
 	{ 0x3041, 0x3096 },
 	{ 0x3099, 0x30ff },
 	{ 0x3105, 0x312f },
 	{ 0x3131, 0x318e },
 	{ 0x3190, 0x31e3 },
-	{ 0x31f0, 0x321e },
+	{ 0x31ef, 0x321e },
 	{ 0x3220, 0x3247 },
 	{ 0x3250, 0x4dbf },
 	{ 0x4e00, 0xa48c },
@@ -88,7 +87,9 @@ static const struct {
 	{ 0x1aff5, 0x1affb },
 	{ 0x1affd, 0x1affe },
 	{ 0x1b000, 0x1b122 },
+	{ 0x1b132, 0x1b132 },
 	{ 0x1b150, 0x1b152 },
+	{ 0x1b155, 0x1b155 },
 	{ 0x1b164, 0x1b167 },
 	{ 0x1b170, 0x1b2fb },
 	{ 0x1f004, 0x1f004 },
@@ -122,23 +123,21 @@ static const struct {
 	{ 0x1f6cc, 0x1f6cc },
 	{ 0x1f6d0, 0x1f6d2 },
 	{ 0x1f6d5, 0x1f6d7 },
-	{ 0x1f6dd, 0x1f6df },
+	{ 0x1f6dc, 0x1f6df },
 	{ 0x1f6eb, 0x1f6ec },
 	{ 0x1f6f4, 0x1f6fc },
 	{ 0x1f7e0, 0x1f7eb },
 	{ 0x1f7f0, 0x1f7f0 },
 	{ 0x1f90c, 0x1f93a },
 	{ 0x1f93c, 0x1f945 },
 	{ 0x1f947, 0x1f9ff },
-	{ 0x1fa70, 0x1fa74 },
-	{ 0x1fa78, 0x1fa7c },
-	{ 0x1fa80, 0x1fa86 },
-	{ 0x1fa90, 0x1faac },
-	{ 0x1fab0, 0x1faba },
-	{ 0x1fac0, 0x1fac5 },
-	{ 0x1fad0, 0x1fad9 },
-	{ 0x1fae0, 0x1fae7 },
-	{ 0x1faf0, 0x1faf6 },
+	{ 0x1fa70, 0x1fa7c },
+	{ 0x1fa80, 0x1fa88 },
+	{ 0x1fa90, 0x1fabd },
+	{ 0x1fabf, 0x1fac5 },
+	{ 0x1face, 0x1fadb },
+	{ 0x1fae0, 0x1fae8 },
+	{ 0x1faf0, 0x1faf8 },
 	{ 0x20000, 0x2fffd },
 	{ 0x30000, 0x3fffd },
 };

ext/mbstring/tests/unicode_versions.phpt

@@ -0,0 +1,28 @@
+--TEST--
+mbstring Unicode Data tests
+--EXTENSIONS--
+mbstring
+--FILE--
+<?php
+
+print "ASCII (PHP): " .      mb_strwidth('PHP',    'UTF-8')    . "\n";
+
+print "Vietnamese (Xin chào): " . mb_strwidth('Xin chào',    'UTF-8')    . "\n";
+
+print "Traditional Chinese (你好): " . mb_strwidth('你好',    'UTF-8')    . "\n";
+
+print "Sinhalese (අයේෂ්): " . mb_strwidth('අයේෂ්',    'UTF-8')    . "\n";
+
+print "Emoji (\u{1F418}): " . mb_strwidth("\u{1F418}", 'UTF-8') . "\n";
+
+// New in Unicode 15.0, width=2
+print "Emoji (\u{1F6DC}): " . mb_strwidth("\u{1F6DC}", 'UTF-8') . "\n";
+
+?>
+--EXPECT--
+ASCII (PHP): 3
+Vietnamese (Xin chào): 8
+Traditional Chinese (你好): 4
+Sinhalese (අයේෂ්): 5
+Emoji (🐘): 2
+Emoji (🛜): 2

ext/mbstring/ucgendat/ucgendat.php

@@ -367,14 +367,22 @@ function parseSpecialCasing(UnicodeData $data, string $input) : void {
 
 function parseDerivedCoreProperties(UnicodeData $data, string $input) : void {
     foreach (parseDataFile($input) as $fields) {
-        if (count($fields) != 2) {
-            throw new Exception("Line does not contain 2 fields");
+        $fieldCount = count($fields);
+        if ($fieldCount != 2 && $fieldCount !== 3) {
+            throw new Exception("Line does not contain 2 or 3 fields");
         }
 
-        $property = $fields[1];
-        if ($property != 'Cased' && $property != 'Case_Ignorable') {
+        $usedProperties = ['Cased', 'Case_Ignorable'];
+        if (isset($fields[2]) && in_array($fields[2], $usedProperties, true)) {
+            $property = $fields[2];
+        }
+        elseif (!in_array($fields[1], $usedProperties, true)) {
             continue;
         }
+        else{
+            $property = $fields[1];
+        }
+
 
         $range = explode('..', $fields[0]);
         if (count($range) == 2) {