ext/Intl: Implement a compatibility layer for ICU version-specific API difference

ext/intl/config.m4

@@ -9,6 +9,7 @@ if test "$PHP_INTL" != "no"; then
   INTL_COMMON_FLAGS="$ICU_CFLAGS -Wno-write-strings -D__STDC_LIMIT_MACROS -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -DZEND_ENABLE_STATIC_TSRMLS_CACHE=1"
   PHP_NEW_EXTENSION([intl], m4_normalize([
       intl_convert.c
+      intl_icu_compat.c
       intl_error.c
       php_intl.c
     ]),

ext/intl/config.w32

@@ -9,7 +9,7 @@ if (PHP_INTL != "no") {
 		CHECK_LIB("icuuc.lib", "intl", PHP_INTL) &&
 					CHECK_HEADER("unicode/utf.h", "CFLAGS_INTL")) {
 		// always build as shared - zend_strtod.c/ICU type conflict
-		EXTENSION("intl", "php_intl.c intl_convert.c intl_convertcpp.cpp intl_error.c ", true,
+		EXTENSION("intl", "php_intl.c intl_convert.c intl_icu_compat.c intl_convertcpp.cpp intl_error.c ", true,
 								"/I \"" + configure_module_dirname + "\" /DZEND_ENABLE_STATIC_TSRMLS_CACHE=1");
 		ADD_EXTENSION_DEP('intl', 'date');
 		ADD_SOURCES(configure_module_dirname + "/collator", "\

ext/intl/converter/converter.cpp

@@ -20,6 +20,8 @@
 #include <unicode/ucnv.h>
 #include <unicode/ustring.h>
 
+#include "../intl_icu_compat.h"
+
 extern "C" {
 #include "converter.h"
 #include "php_intl.h"
@@ -949,18 +951,10 @@ static zend_object *php_converter_clone_object(zend_object *object) {
 	zend_object *retval = php_converter_object_ctor(object->ce, &objval);
 	UErrorCode error = U_ZERO_ERROR;
 
-#if U_ICU_VERSION_MAJOR_NUM > 70
-	objval->src = ucnv_clone(oldobj->src, &error);
-#else
-	objval->src = ucnv_safeClone(oldobj->src, NULL, NULL, &error);
-#endif
+	objval->src = intl_icu_compat_ucnv_clone(oldobj->src, &error);
 	if (U_SUCCESS(error)) {
 		error = U_ZERO_ERROR;
-#if U_ICU_VERSION_MAJOR_NUM > 70
-		objval->dest = ucnv_clone(oldobj->dest, &error);
-#else
-		objval->dest = ucnv_safeClone(oldobj->dest, NULL, NULL, &error);
-#endif
+		objval->dest = intl_icu_compat_ucnv_clone(oldobj->dest, &error);
 	}
 
 	if (U_FAILURE(error)) {

ext/intl/grapheme/grapheme_string.cpp

@@ -345,7 +345,6 @@ U_CFUNC PHP_FUNCTION(grapheme_substr)
 	int32_t start = 0;
 	int iter_val;
 	UErrorCode status;
-	unsigned char u_break_iterator_buffer[U_BRK_SAFECLONE_BUFFERSIZE];
 	UBreakIterator* bi = nullptr;
 	int sub_str_start_pos, sub_str_end_pos;
 	int32_t (*iter_func)(UBreakIterator *);
@@ -407,7 +406,7 @@ U_CFUNC PHP_FUNCTION(grapheme_substr)
 		RETURN_FALSE;
 	}
 
-	bi = grapheme_get_break_iterator((void*)u_break_iterator_buffer, &status );
+	bi = grapheme_get_break_iterator(&status);
 
 	if( U_FAILURE(status) ) {
 		RETURN_FALSE;
@@ -729,7 +728,6 @@ U_CFUNC PHP_FUNCTION(grapheme_extract)
 	int32_t start = 0;
 	zend_long extract_type = GRAPHEME_EXTRACT_TYPE_COUNT;
 	UErrorCode status;
-	unsigned char u_break_iterator_buffer[U_BRK_SAFECLONE_BUFFERSIZE];
 	UBreakIterator* bi = nullptr;
 	int ret_pos;
 	zval *next = nullptr; /* return offset of next part of the string */
@@ -829,7 +827,7 @@ U_CFUNC PHP_FUNCTION(grapheme_extract)
 
 	bi = nullptr;
 	status = U_ZERO_ERROR;
-	bi = grapheme_get_break_iterator(u_break_iterator_buffer, &status );
+	bi = grapheme_get_break_iterator(&status);
 
 	ubrk_setUText(bi, &ut, &status);
 	/* if the caller put us in the middle of a grapheme, we can't detect it in all cases since we
@@ -855,7 +853,6 @@ U_CFUNC PHP_FUNCTION(grapheme_str_split)
 	zend_string *str;
 	zend_long split_len = 1;
 
-	unsigned char u_break_iterator_buffer[U_BRK_SAFECLONE_BUFFERSIZE];
 	UErrorCode ustatus = U_ZERO_ERROR;
 	int32_t pos, current, i, end_len = 0;
 	UBreakIterator* bi;
@@ -891,7 +888,7 @@ U_CFUNC PHP_FUNCTION(grapheme_str_split)
 
 	bi = nullptr;
 	ustatus = U_ZERO_ERROR;
-	bi = grapheme_get_break_iterator((void*)u_break_iterator_buffer, &ustatus );
+	bi = grapheme_get_break_iterator(&ustatus);
 
 	if( U_FAILURE(ustatus) ) {
 		RETURN_FALSE;
@@ -1031,17 +1028,15 @@ U_CFUNC PHP_FUNCTION(grapheme_levenshtein)
 		goto out_ustring2;
 	}
 
-	unsigned char u_break_iterator_buffer1[U_BRK_SAFECLONE_BUFFERSIZE];
-	unsigned char u_break_iterator_buffer2[U_BRK_SAFECLONE_BUFFERSIZE];
-	bi1 = grapheme_get_break_iterator(u_break_iterator_buffer1, &ustatus);
+	bi1 = grapheme_get_break_iterator(&ustatus);
 	if (U_FAILURE(ustatus)) {
 		intl_error_set_code(NULL, ustatus);
 		intl_error_set_custom_msg(NULL, "Error on grapheme_get_break_iterator for argument #1 ($string1)");
 		RETVAL_FALSE;
 		goto out_bi1;
 	}
 
-	bi2 = grapheme_get_break_iterator(u_break_iterator_buffer2, &ustatus);
+	bi2 = grapheme_get_break_iterator(&ustatus);
 	if (U_FAILURE(ustatus)) {
 		intl_error_set_code(NULL, ustatus);
 		intl_error_set_custom_msg(NULL, "Error on grapheme_get_break_iterator for argument #2 ($string2)");
@@ -1144,7 +1139,6 @@ U_CFUNC PHP_FUNCTION(grapheme_strrev)
 	char *pstr, *end, *p;
 	zend_string *ret;
 	int32_t pos = 0, current = 0, end_len = 0;
-	unsigned char u_break_iterator_buffer[U_BRK_SAFECLONE_BUFFERSIZE];
 
 	ZEND_PARSE_PARAMETERS_START(1, 1)
 		Z_PARAM_STR(string)
@@ -1168,7 +1162,7 @@ U_CFUNC PHP_FUNCTION(grapheme_strrev)
 	bi = nullptr;
 	ustatus = U_ZERO_ERROR;
 
-	bi = grapheme_get_break_iterator((void*)u_break_iterator_buffer, &ustatus );
+	bi = grapheme_get_break_iterator(&ustatus);
 	ret = zend_string_alloc(ZSTR_LEN(string), 0);
 	p = ZSTR_VAL(ret);
 

ext/intl/grapheme/grapheme_util.cpp

@@ -35,6 +35,8 @@ extern "C" {
 #include <unicode/ubrk.h>
 #include <unicode/usearch.h>
 
+#include "../intl_icu_compat.h"
+
 ZEND_EXTERN_MODULE_GLOBALS( intl )
 
 /* }}} */
@@ -105,7 +107,6 @@ U_CFUNC int32_t grapheme_strpos_utf16(char *haystack, size_t haystack_len, char
 {
 	UChar *uhaystack = NULL, *uneedle = NULL;
 	int32_t uhaystack_len = 0, uneedle_len = 0, char_pos, ret_pos, offset_pos = 0;
-	unsigned char u_break_iterator_buffer[U_BRK_SAFECLONE_BUFFERSIZE];
 	UBreakIterator* bi = NULL;
 	UErrorCode status;
 	UStringSearch* src = NULL;
@@ -125,7 +126,7 @@ U_CFUNC int32_t grapheme_strpos_utf16(char *haystack, size_t haystack_len, char
 
 	/* get a pointer to the haystack taking into account the offset */
 	status = U_ZERO_ERROR;
-	bi = grapheme_get_break_iterator(u_break_iterator_buffer, &status );
+	bi = grapheme_get_break_iterator(&status);
 	STRPOS_CHECK_STATUS(status, "Failed to get iterator");
 	status = U_ZERO_ERROR;
 	ubrk_setText(bi, uhaystack, uhaystack_len, &status);
@@ -235,12 +236,11 @@ U_CFUNC zend_long grapheme_ascii_check(const unsigned char *day, size_t len)
 /* {{{ grapheme_split_string: find and optionally return grapheme boundaries */
 U_CFUNC int32_t grapheme_split_string(const UChar *text, int32_t text_length, int boundary_array[], int boundary_array_len )
 {
-	unsigned char u_break_iterator_buffer[U_BRK_SAFECLONE_BUFFERSIZE];
 	UErrorCode		status = U_ZERO_ERROR;
 	int ret_len, pos;
 	UBreakIterator* bi;
 
-	bi = grapheme_get_break_iterator((void*)u_break_iterator_buffer, &status );
+	bi = grapheme_get_break_iterator(&status);
 
 	if( U_FAILURE(status) ) {
 		return -1;
@@ -375,7 +375,7 @@ U_CFUNC zend_long grapheme_strrpos_ascii(char *haystack, size_t haystack_len, ch
 /* }}} */
 
 /* {{{ grapheme_get_break_iterator: get a clone of the global character break iterator */
-U_CFUNC UBreakIterator* grapheme_get_break_iterator(void *stack_buffer, UErrorCode *status )
+U_CFUNC UBreakIterator* grapheme_get_break_iterator(UErrorCode *status )
 {
 	UBreakIterator *global_break_iterator = INTL_G( grapheme_iterator );
 
@@ -390,12 +390,6 @@ U_CFUNC UBreakIterator* grapheme_get_break_iterator(void *stack_buffer, UErrorCo
 		INTL_G(grapheme_iterator) = global_break_iterator;
 	}
 
-#if U_ICU_VERSION_MAJOR_NUM >= 69
-	return ubrk_clone(global_break_iterator, status);
-#else
-	int32_t buffer_size = U_BRK_SAFECLONE_BUFFERSIZE;
-
-	return ubrk_safeClone(global_break_iterator, stack_buffer, &buffer_size, status);
-#endif
+	return intl_icu_compat_ubrk_clone(global_break_iterator, status);
 }
 /* }}} */

ext/intl/grapheme/grapheme_util.h

@@ -22,7 +22,7 @@ extern "C" {
 #include "intl_convert.h"
 
 /* get_break_interator: get a break iterator from the global structure */
-UBreakIterator* grapheme_get_break_iterator(void *stack_buffer, UErrorCode *status );
+UBreakIterator* grapheme_get_break_iterator(UErrorCode *status );
 
 zend_long grapheme_ascii_check(const unsigned char *day, size_t len);
 void grapheme_substr_ascii(char *str, size_t str_len, int32_t f, int32_t l, char **sub_str, int32_t *sub_str_len);
@@ -37,7 +37,7 @@ int32_t grapheme_count_graphemes(UBreakIterator *bi, UChar *string, int32_t stri
 
 int32_t grapheme_get_haystack_offset(UBreakIterator* bi, int32_t offset);
 
-UBreakIterator* grapheme_get_break_iterator(void *stack_buffer, UErrorCode *status );
+UBreakIterator* grapheme_get_break_iterator(UErrorCode *status );
 #ifdef __cplusplus
 }
 #endif

ext/intl/intl_icu_compat.c

@@ -0,0 +1,114 @@
+/*
+   +----------------------------------------------------------------------+
+   | Copyright (c) The PHP Group                                          |
+   +----------------------------------------------------------------------+
+   | This source file is subject to the Modified BSD License that is      |
+   | bundled with this package in the file LICENSE, and is available      |
+   | through the World Wide Web at <https://www.php.net/license/>.        |
+   |                                                                      |
+   | SPDX-License-Identifier: BSD-3-Clause                                |
+   +----------------------------------------------------------------------+
+   | Authors: Weilin Du <weilindu@php.net>                                |
+   +----------------------------------------------------------------------+
+ */
+
+#ifdef HAVE_CONFIG_H
+#include <config.h>
+#endif
+
+#include <stddef.h>
+
+#include "intl_icu_compat.h"
+#include <unicode/ubrk.h>
+#include <unicode/ucnv.h>
+#include <unicode/uspoof.h>
+#include <unicode/uset.h>
+
+UConverter *intl_icu_compat_ucnv_clone(const UConverter *converter, UErrorCode *status)
+{
+#if INTL_ICU_HAS_UCNV_CLONE
+	return ucnv_clone(converter, status);
+#else
+	return ucnv_safeClone(converter, NULL, NULL, status);
+#endif
+}
+
+UBreakIterator *intl_icu_compat_ubrk_clone(const UBreakIterator *break_iterator, UErrorCode *status)
+{
+#if INTL_ICU_HAS_UBRK_CLONE
+	return ubrk_clone(break_iterator, status);
+#else
+	return ubrk_safeClone(break_iterator, NULL, NULL, status);
+#endif
+}
+
+void intl_icu_compat_uspoof_init_checker(USpoofChecker *checker, IntlIcuSpoofCheckResult **check_result, UErrorCode *status)
+{
+#if INTL_ICU_HAS_SPOOFCHECKER_CHECK_RESULT
+	/* ICU 58 removed WSC/MSC handling in favor of TR39 restriction levels.
+	 * Keep the default highly restrictive behavior and allocate the extended
+	 * check result used by uspoof_check2*(). */
+	uspoof_setRestrictionLevel(checker, USPOOF_HIGHLY_RESTRICTIVE);
+	*check_result = uspoof_openCheckResult(status);
+#else
+	int checks = uspoof_getChecks(checker, status);
+
+	/* Single-script enforcement rejects legitimate mixed-script languages like Japanese. */
+	uspoof_setChecks(checker, checks & ~USPOOF_SINGLE_SCRIPT, status);
+	*check_result = NULL;
+#endif
+}
+
+void intl_icu_compat_uspoof_close_check_result(IntlIcuSpoofCheckResult *check_result)
+{
+#if INTL_ICU_HAS_SPOOFCHECKER_CHECK_RESULT
+	if (check_result) {
+		uspoof_closeCheckResult(check_result);
+	}
+#else
+	(void) check_result;
+#endif
+}
+
+int32_t intl_icu_compat_uspoof_check_utf8(const USpoofChecker *checker, const char *text, int32_t length, IntlIcuSpoofCheckResult *check_result, UErrorCode *status)
+{
+#if INTL_ICU_HAS_SPOOFCHECKER_CHECK_RESULT
+	return uspoof_check2UTF8(checker, text, length, check_result, status);
+#else
+	(void) check_result;
+	return uspoof_checkUTF8(checker, text, length, NULL, status);
+#endif
+}
+
+UBool intl_icu_compat_uspoof_check_result_mismatch(IntlIcuSpoofCheckResult *check_result, int32_t checks, int32_t *result_checks, UErrorCode *status)
+{
+#if INTL_ICU_HAS_SPOOFCHECKER_CHECK_RESULT
+	*result_checks = uspoof_getCheckResultChecks(check_result, status);
+	return *result_checks != checks;
+#else
+	(void) check_result;
+	(void) status;
+	*result_checks = checks;
+	return 0;
+#endif
+}
+
+UBool intl_icu_compat_uspoof_is_allowed_chars_pattern_option(int64_t pattern_option)
+{
+	return pattern_option == 0
+		|| pattern_option == USET_IGNORE_SPACE
+#if INTL_ICU_HAS_USET_SIMPLE_CASE_INSENSITIVE
+		|| pattern_option == (USET_IGNORE_SPACE|USET_SIMPLE_CASE_INSENSITIVE)
+#endif
+		|| pattern_option == (USET_IGNORE_SPACE|USET_CASE_INSENSITIVE)
+		|| pattern_option == (USET_IGNORE_SPACE|USET_ADD_CASE_MAPPINGS);
+}
+
+const char *intl_icu_compat_uspoof_allowed_chars_pattern_option_error_message(void)
+{
+#if INTL_ICU_HAS_USET_SIMPLE_CASE_INSENSITIVE
+	return "must be a valid pattern option, 0 or (SpoofChecker::IGNORE_SPACE|(<none> or SpoofChecker::CASE_INSENSITIVE or SpoofChecker::ADD_CASE_MAPPINGS or SpoofChecker::SIMPLE_CASE_INSENSITIVE))";
+#else
+	return "must be a valid pattern option, 0 or (SpoofChecker::IGNORE_SPACE|(<none> or SpoofChecker::CASE_INSENSITIVE or SpoofChecker::ADD_CASE_MAPPINGS))";
+#endif
+}

ext/intl/intl_icu_compat.h

@@ -0,0 +1,56 @@
+/*
+   +----------------------------------------------------------------------+
+   | Copyright (c) The PHP Group                                          |
+   +----------------------------------------------------------------------+
+   | This source file is subject to the Modified BSD License that is      |
+   | bundled with this package in the file LICENSE, and is available      |
+   | through the World Wide Web at <https://www.php.net/license/>.        |
+   |                                                                      |
+   | SPDX-License-Identifier: BSD-3-Clause                                |
+   +----------------------------------------------------------------------+
+   | Authors: Weilin Du <weilindu@php.net>                                |
+   +----------------------------------------------------------------------+
+ */
+
+#ifndef INTL_ICU_COMPAT_H
+#define INTL_ICU_COMPAT_H
+
+#include <unicode/ubrk.h>
+#include <unicode/ucnv.h>
+#include <unicode/uspoof.h>
+#include <unicode/utypes.h>
+#include <unicode/uversion.h>
+
+#define INTL_ICU_VERSION_AT_LEAST(major, minor) \
+	(U_ICU_VERSION_MAJOR_NUM > (major) || \
+		(U_ICU_VERSION_MAJOR_NUM == (major) && U_ICU_VERSION_MINOR_NUM >= (minor)))
+
+#define INTL_ICU_HAS_UBRK_CLONE INTL_ICU_VERSION_AT_LEAST(69, 0)
+#define INTL_ICU_HAS_UCNV_CLONE INTL_ICU_VERSION_AT_LEAST(71, 0)
+#define INTL_ICU_HAS_SPOOFCHECKER_CHECK_RESULT INTL_ICU_VERSION_AT_LEAST(58, 0)
+#define INTL_ICU_HAS_USET_SIMPLE_CASE_INSENSITIVE INTL_ICU_VERSION_AT_LEAST(73, 0)
+
+#if INTL_ICU_HAS_SPOOFCHECKER_CHECK_RESULT
+typedef USpoofCheckResult IntlIcuSpoofCheckResult;
+#else
+typedef void IntlIcuSpoofCheckResult;
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+UConverter *intl_icu_compat_ucnv_clone(const UConverter *converter, UErrorCode *status);
+UBreakIterator *intl_icu_compat_ubrk_clone(const UBreakIterator *break_iterator, UErrorCode *status);
+void intl_icu_compat_uspoof_init_checker(USpoofChecker *checker, IntlIcuSpoofCheckResult **check_result, UErrorCode *status);
+void intl_icu_compat_uspoof_close_check_result(IntlIcuSpoofCheckResult *check_result);
+int32_t intl_icu_compat_uspoof_check_utf8(const USpoofChecker *checker, const char *text, int32_t length, IntlIcuSpoofCheckResult *check_result, UErrorCode *status);
+UBool intl_icu_compat_uspoof_check_result_mismatch(IntlIcuSpoofCheckResult *check_result, int32_t checks, int32_t *result_checks, UErrorCode *status);
+UBool intl_icu_compat_uspoof_is_allowed_chars_pattern_option(int64_t pattern_option);
+const char *intl_icu_compat_uspoof_allowed_chars_pattern_option_error_message(void);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif