[php-src] master: Merge branch 'PHP-8.5'

Author: Weilin Du (LamentXU123)
Date: 2026-08-18T17:51:12+08:00

Commit: https://github.com/php/php-src/commit/1f2319e42cbd99727c403c01ec22614f2356fe38
Raw diff: https://github.com/php/php-src/commit/1f2319e42cbd99727c403c01ec22614f2356fe38.diff

Merge branch 'PHP-8.5'

* PHP-8.5:
  Fix GH-23094: Use byte offsets in NumberFormatter parsing (#23318)

Changed paths:
  A  ext/intl/tests/gh23094.phpt
  M  NEWS
  M  ext/intl/formatter/formatter_parse.cpp


Diff:

diff --git a/NEWS b/NEWS
index 3db6bc520e84..9690787e59bf 100644
--- a/NEWS
+++ b/NEWS
@@ -34,6 +34,8 @@ PHP                                                                        NEWS
     the result without a terminating NUL. (iliaal)
   . Fixed a double-free when IntlGregorianCalendar construction fails after
     the ICU constructor adopts the TimeZone. (iliaal)
+  . Fixed bug GH-23094 (NumberFormatter parsing offsets use UTF-16 positions
+    for UTF-8 strings). (ColumbusLabs)
 
 - Phar:
   . Fixed Phar archives being automatically detected when ".phar" only occurs
diff --git a/ext/intl/formatter/formatter_parse.cpp b/ext/intl/formatter/formatter_parse.cpp
index 40107229e0ae..7a233f012d46 100644
--- a/ext/intl/formatter/formatter_parse.cpp
+++ b/ext/intl/formatter/formatter_parse.cpp
@@ -18,6 +18,7 @@
 
 #include <unicode/fmtable.h>
 #include <unicode/curramt.h>
+#include <unicode/ustring.h>
 #include "../intl_convertcpp.h"
 #include "formatter_class.h"
 #include "formatter_format.h"
@@ -31,6 +32,42 @@ extern "C" {
 
 #define ICU_LOCALE_BUG 1
 
+static bool numfmt_utf8_offset_to_utf16(const char *str, size_t str_len, int32_t *position, UErrorCode *status)
+{
+	int32_t utf16_position;
+
+	if (*position < 0 || (size_t) *position > str_len) {
+		return true;
+	}
+
+	*status = U_ZERO_ERROR;
+	u_strFromUTF8(nullptr, 0, &utf16_position, str, *position, status);
+	if (*status != U_BUFFER_OVERFLOW_ERROR && U_FAILURE(*status)) {
+		return false;
+	}
+	*status = U_ZERO_ERROR;
+
+	*position = utf16_position;
+	return true;
+}
+
+static int32_t numfmt_utf16_offset_to_utf8(const icu::UnicodeString &str, int32_t position)
+{
+	int32_t utf8_position;
+	UErrorCode status = U_ZERO_ERROR;
+
+	if (position < 0 || position > str.length()) {
+		return position;
+	}
+
+	u_strToUTF8(nullptr, 0, &utf8_position, str.getBuffer(), position, &status);
+	if (status != U_BUFFER_OVERFLOW_ERROR && U_FAILURE(status)) {
+		return position;
+	}
+
+	return utf8_position;
+}
+
 /* {{{ Parse a number. */
 U_CFUNC PHP_FUNCTION( numfmt_parse )
 {
@@ -65,6 +102,9 @@ U_CFUNC PHP_FUNCTION( numfmt_parse )
 	icu::UnicodeString ustr;
 	intl_stringFromChar(ustr, str, str_len, &INTL_DATA_ERROR_CODE(nfo));
 	INTL_METHOD_CHECK_STATUS( nfo, "String conversion to UTF-16 failed" );
+	if (zposition && !numfmt_utf8_offset_to_utf16(str, str_len, &position, &INTL_DATA_ERROR_CODE(nfo))) {
+		INTL_METHOD_CHECK_STATUS(nfo, "Invalid UTF-8 offset");
+	}
 
 #if ICU_LOCALE_BUG && defined(LC_NUMERIC)
 	/* need to copy here since setlocale may change it later */
@@ -122,6 +162,7 @@ U_CFUNC PHP_FUNCTION( numfmt_parse )
 	}
 
 	if (zposition) {
+		position = numfmt_utf16_offset_to_utf8(ustr, position);
 		ZEND_TRY_ASSIGN_REF_LONG(zposition, position);
 	}
 
@@ -167,6 +208,9 @@ U_CFUNC PHP_FUNCTION( numfmt_parse_currency )
 			RETURN_THROWS();
 		}
 		position = (int32_t) long_position;
+		if (!numfmt_utf8_offset_to_utf16(str, str_len, &position, &INTL_DATA_ERROR_CODE(nfo))) {
+			INTL_METHOD_CHECK_STATUS(nfo, "Invalid UTF-8 offset");
+		}
 	}
 
 	icu::ParsePosition pp(position);
@@ -178,7 +222,8 @@ U_CFUNC PHP_FUNCTION( numfmt_parse_currency )
 	}
 
 	if(zposition) {
-		ZEND_TRY_ASSIGN_REF_LONG(zposition, pp.getIndex());
+		position = numfmt_utf16_offset_to_utf8(ustr, pp.getIndex());
+		ZEND_TRY_ASSIGN_REF_LONG(zposition, position);
 	}
 
 	const double number = currAmt->getNumber().getDouble(INTL_DATA_ERROR_CODE(nfo));
diff --git a/ext/intl/tests/gh23094.phpt b/ext/intl/tests/gh23094.phpt
new file mode 100644
index 000000000000..9ace16fa481f
--- /dev/null
+++ b/ext/intl/tests/gh23094.phpt
@@ -0,0 +1,47 @@
+--TEST--
+GH-23094 NumberFormatter parse offsets use UTF-8 byte positions
+--EXTENSIONS--
+intl
+--FILE--
+<?php
+
+$prefix = "\u{1F600}";
+
+$formatter = new NumberFormatter('en_US', NumberFormatter::DECIMAL);
+$offset = strlen($prefix);
+var_dump($formatter->parse($prefix . '123', NumberFormatter::TYPE_INT32, $offset));
+var_dump($offset);
+
+$offset = 1;
+var_dump($formatter->parse("\u{00E9}123", NumberFormatter::TYPE_INT32, $offset));
+var_dump($offset);
+var_dump(intl_is_failure($formatter->getErrorCode()));
+
+$formatter = new NumberFormatter('en_US', NumberFormatter::CURRENCY);
+$offset = strlen($prefix);
+$currency = null;
+var_dump($formatter->parseCurrency($prefix . '$123.45', $currency, $offset));
+var_dump($currency);
+var_dump($offset);
+
+$offset = 1;
+$currency = null;
+var_dump($formatter->parseCurrency("\u{00E9}$123.45", $currency, $offset));
+var_dump($currency);
+var_dump($offset);
+var_dump(intl_is_failure($formatter->getErrorCode()));
+
+?>
+--EXPECT--
+int(123)
+int(7)
+bool(false)
+int(1)
+bool(true)
+float(123.45)
+string(3) "USD"
+int(11)
+bool(false)
+NULL
+int(1)
+bool(true)