Merge branch 'PHP-7.4'

This commit is contained in:
Nikita Popov 2019-04-17 14:06:05 +02:00
commit 1d53d6df7e
15 changed files with 209 additions and 123 deletions

View File

@ -11,7 +11,7 @@ if (!extension_loaded("mbstring")) {
?>
--INI--
zend.multibyte=1
mbstring.internal_encoding=big5
internal_encoding=big5
--FILE--
<?php
echo '\'hello';

View File

@ -11,7 +11,7 @@ if (!extension_loaded("mbstring")) {
?>
--INI--
zend.multibyte=1
mbstring.internal_encoding=SJIS
internal_encoding=SJIS
--FILE--
<?php
declare(encoding='Shift_JIS');

View File

@ -11,7 +11,7 @@ if (!extension_loaded("mbstring")) {
?>
--INI--
zend.multibyte=1
mbstring.internal_encoding=iso-8859-1
internal_encoding=iso-8859-1
--FILE--
<?php
print "Hello World\n";

View File

@ -12,7 +12,7 @@ if (!extension_loaded("mbstring")) {
--INI--
zend.multibyte=1
zend.script_encoding=Shift_JIS
mbstring.internal_encoding=Shift_JIS
internal_encoding=Shift_JIS
--FILE--
<?php
function —\Ž\”\($ˆø<CB86>”)

View File

@ -13,7 +13,7 @@ if (!extension_loaded("mbstring")) {
zend.multibyte=1
mbstring.encoding_translation = On
zend.script_encoding=Shift_JIS
mbstring.internal_encoding=UTF-8
internal_encoding=UTF-8
--FILE--
<?php
function —\Ž\”\($ˆø<CB86>”)

View File

@ -338,37 +338,25 @@ PHP_MINFO_FUNCTION(miconv)
}
/* }}} */
static char *get_internal_encoding(void) {
static const char *get_internal_encoding(void) {
if (ICONVG(internal_encoding) && ICONVG(internal_encoding)[0]) {
return ICONVG(internal_encoding);
} else if (PG(internal_encoding) && PG(internal_encoding)[0]) {
return PG(internal_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
return php_get_internal_encoding();
}
static char *get_input_encoding(void) {
static const char *get_input_encoding(void) {
if (ICONVG(input_encoding) && ICONVG(input_encoding)[0]) {
return ICONVG(input_encoding);
} else if (PG(input_encoding) && PG(input_encoding)[0]) {
return PG(input_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
return php_get_input_encoding();
}
static char *get_output_encoding(void) {
static const char *get_output_encoding(void) {
if (ICONVG(output_encoding) && ICONVG(output_encoding)[0]) {
return ICONVG(output_encoding);
} else if (PG(output_encoding) && PG(output_encoding)[0]) {
return PG(output_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
return php_get_output_encoding();
}
@ -2064,7 +2052,7 @@ static void _php_iconv_show_error(php_iconv_err_t err, const char *out_charset,
Returns the character count of str */
PHP_FUNCTION(iconv_strlen)
{
char *charset = get_internal_encoding();
const char *charset = get_internal_encoding();
size_t charset_len = 0;
zend_string *str;
@ -2096,7 +2084,7 @@ PHP_FUNCTION(iconv_strlen)
Returns specified part of a string */
PHP_FUNCTION(iconv_substr)
{
char *charset = get_internal_encoding();
const char *charset = get_internal_encoding();
size_t charset_len = 0;
zend_string *str;
zend_long offset, length = 0;
@ -2135,7 +2123,7 @@ PHP_FUNCTION(iconv_substr)
Finds position of first occurrence of needle within part of haystack beginning with offset */
PHP_FUNCTION(iconv_strpos)
{
char *charset = get_internal_encoding();
const char *charset = get_internal_encoding();
size_t charset_len = 0, haystk_len;
zend_string *haystk;
zend_string *ndl;
@ -2190,7 +2178,7 @@ PHP_FUNCTION(iconv_strpos)
Finds position of last occurrence of needle within part of haystack beginning with offset */
PHP_FUNCTION(iconv_strrpos)
{
char *charset = get_internal_encoding();
const char *charset = get_internal_encoding();
size_t charset_len = 0;
zend_string *haystk;
zend_string *ndl;
@ -2331,7 +2319,7 @@ PHP_FUNCTION(iconv_mime_encode)
PHP_FUNCTION(iconv_mime_decode)
{
zend_string *encoded_str;
char *charset = get_internal_encoding();
const char *charset = get_internal_encoding();
size_t charset_len = 0;
zend_long mode = 0;
@ -2371,7 +2359,7 @@ PHP_FUNCTION(iconv_mime_decode)
PHP_FUNCTION(iconv_mime_decode_headers)
{
zend_string *encoded_str;
char *charset = get_internal_encoding();
const char *charset = get_internal_encoding();
size_t charset_len = 0;
zend_long mode = 0;
char *enc_str_tmp;

View File

@ -58,11 +58,6 @@ MBSTRING_API SAPI_TREAT_DATA_FUNC(mbstr_treat_data)
const mbfl_encoding *detected;
php_mb_encoding_handler_info_t info;
if (arg != PARSE_STRING) {
char *value = MBSTRG(internal_encoding_name);
_php_mb_ini_mbstring_internal_encoding_set(value, value ? strlen(value): 0);
}
if (!MBSTRG(encoding_translation)) {
php_default_treat_data(arg, str, destArray);
return;

View File

@ -596,34 +596,6 @@ ZEND_TSRMLS_CACHE_DEFINE()
ZEND_GET_MODULE(mbstring)
#endif
static char *get_internal_encoding(void) {
if (PG(internal_encoding) && PG(internal_encoding)[0]) {
return PG(internal_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
}
static char *get_input_encoding(void) {
if (PG(input_encoding) && PG(input_encoding)[0]) {
return PG(input_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
}
static char *get_output_encoding(void) {
if (PG(output_encoding) && PG(output_encoding)[0]) {
return PG(output_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
}
/* {{{ allocators */
static void *_php_mb_allocators_malloc(size_t sz)
{
@ -1270,77 +1242,70 @@ static PHP_INI_MH(OnUpdate_mbstring_detect_order)
}
/* }}} */
/* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_input) */
static PHP_INI_MH(OnUpdate_mbstring_http_input)
{
static int _php_mb_ini_mbstring_http_input_set(const char *new_value, size_t new_value_length) {
const mbfl_encoding **list;
size_t size;
if (!new_value || !ZSTR_VAL(new_value)) {
if (MBSTRG(http_input_list)) {
pefree(MBSTRG(http_input_list), 1);
}
if (SUCCESS == php_mb_parse_encoding_list(get_input_encoding(), strlen(get_input_encoding())+1, &list, &size, 1)) {
MBSTRG(http_input_list) = list;
MBSTRG(http_input_list_size) = size;
return SUCCESS;
}
MBSTRG(http_input_list) = NULL;
MBSTRG(http_input_list_size) = 0;
return SUCCESS;
}
if (FAILURE == php_mb_parse_encoding_list(ZSTR_VAL(new_value), ZSTR_LEN(new_value), &list, &size, 1)) {
if (FAILURE == php_mb_parse_encoding_list(new_value, new_value_length, &list, &size, 1)) {
return FAILURE;
}
if (MBSTRG(http_input_list)) {
pefree(MBSTRG(http_input_list), 1);
}
MBSTRG(http_input_list) = list;
MBSTRG(http_input_list_size) = size;
return SUCCESS;
}
/* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_input) */
static PHP_INI_MH(OnUpdate_mbstring_http_input)
{
if (stage & (PHP_INI_STAGE_ACTIVATE | PHP_INI_STAGE_RUNTIME)) {
php_error_docref("ref.mbstring", E_DEPRECATED, "Use of mbstring.http_input is deprecated");
}
return SUCCESS;
if (!new_value || !ZSTR_VAL(new_value)) {
const char *encoding = php_get_input_encoding();
MBSTRG(http_input_set) = 0;
_php_mb_ini_mbstring_http_input_set(encoding, strlen(encoding));
return SUCCESS;
}
MBSTRG(http_input_set) = 1;
return _php_mb_ini_mbstring_http_input_set(ZSTR_VAL(new_value), ZSTR_LEN(new_value));
}
/* }}} */
static int _php_mb_ini_mbstring_http_output_set(const char *new_value) {
const mbfl_encoding *encoding = mbfl_name2encoding(new_value);
if (!encoding) {
return FAILURE;
}
MBSTRG(http_output_encoding) = encoding;
MBSTRG(current_http_output_encoding) = encoding;
return SUCCESS;
}
/* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_output) */
static PHP_INI_MH(OnUpdate_mbstring_http_output)
{
const mbfl_encoding *encoding;
if (new_value == NULL || ZSTR_LEN(new_value) == 0) {
encoding = mbfl_name2encoding(get_output_encoding());
if (!encoding) {
MBSTRG(http_output_encoding) = &mbfl_encoding_pass;
MBSTRG(current_http_output_encoding) = &mbfl_encoding_pass;
return SUCCESS;
}
} else {
encoding = mbfl_name2encoding(ZSTR_VAL(new_value));
if (!encoding) {
MBSTRG(http_output_encoding) = &mbfl_encoding_pass;
MBSTRG(current_http_output_encoding) = &mbfl_encoding_pass;
return FAILURE;
}
}
MBSTRG(http_output_encoding) = encoding;
MBSTRG(current_http_output_encoding) = encoding;
if (stage & (PHP_INI_STAGE_ACTIVATE | PHP_INI_STAGE_RUNTIME)) {
php_error_docref("ref.mbstring", E_DEPRECATED, "Use of mbstring.http_output is deprecated");
}
return SUCCESS;
if (new_value == NULL || ZSTR_LEN(new_value) == 0) {
MBSTRG(http_output_set) = 0;
_php_mb_ini_mbstring_http_output_set(php_get_output_encoding());
return SUCCESS;
}
MBSTRG(http_output_set) = 1;
return _php_mb_ini_mbstring_http_output_set(ZSTR_VAL(new_value));
}
/* }}} */
/* {{{ static _php_mb_ini_mbstring_internal_encoding_set */
int _php_mb_ini_mbstring_internal_encoding_set(const char *new_value, size_t new_value_length)
static int _php_mb_ini_mbstring_internal_encoding_set(const char *new_value, size_t new_value_length)
{
const mbfl_encoding *encoding;
@ -1376,20 +1341,13 @@ static PHP_INI_MH(OnUpdate_mbstring_internal_encoding)
return FAILURE;
}
if (stage & (PHP_INI_STAGE_STARTUP | PHP_INI_STAGE_SHUTDOWN | PHP_INI_STAGE_RUNTIME)) {
if (new_value && ZSTR_LEN(new_value)) {
return _php_mb_ini_mbstring_internal_encoding_set(ZSTR_VAL(new_value), ZSTR_LEN(new_value));
} else {
return _php_mb_ini_mbstring_internal_encoding_set(get_internal_encoding(), strlen(get_internal_encoding())+1);
}
if (new_value && ZSTR_LEN(new_value)) {
MBSTRG(internal_encoding_set) = 1;
return _php_mb_ini_mbstring_internal_encoding_set(ZSTR_VAL(new_value), ZSTR_LEN(new_value));
} else {
/* the corresponding mbstring globals needs to be set according to the
* ini value in the later stage because it never falls back to the
* default value if 1. no value for mbstring.internal_encoding is given,
* 2. mbstring.language directive is processed in per-dir or runtime
* context and 3. call to the handler for mbstring.language is done
* after mbstring.internal_encoding is handled. */
return SUCCESS;
const char *encoding = php_get_internal_encoding();
MBSTRG(internal_encoding_set) = 0;
return _php_mb_ini_mbstring_internal_encoding_set(encoding, strlen(encoding));
}
}
/* }}} */
@ -1511,6 +1469,24 @@ PHP_INI_BEGIN()
PHP_INI_END()
/* }}} */
static void mbstring_internal_encoding_changed_hook() {
/* One of the internal_encoding / input_encoding / output_encoding ini settings changed. */
if (!MBSTRG(internal_encoding_set)) {
const char *encoding = php_get_internal_encoding();
_php_mb_ini_mbstring_internal_encoding_set(encoding, strlen(encoding));
}
if (!MBSTRG(http_output_set)) {
const char *encoding = php_get_output_encoding();
_php_mb_ini_mbstring_http_output_set(encoding);
}
if (!MBSTRG(http_input_set)) {
const char *encoding = php_get_input_encoding();
_php_mb_ini_mbstring_http_input_set(encoding, strlen(encoding));
}
}
/* {{{ module global initialize handler */
static PHP_GINIT_FUNCTION(mbstring)
{
@ -1550,6 +1526,9 @@ ZEND_TSRMLS_CACHE_UPDATE();
#endif
mbstring_globals->last_used_encoding_name = NULL;
mbstring_globals->last_used_encoding = NULL;
mbstring_globals->internal_encoding_set = 0;
mbstring_globals->http_output_set = 0;
mbstring_globals->http_input_set = 0;
}
/* }}} */
@ -1581,6 +1560,11 @@ ZEND_TSRMLS_CACHE_UPDATE();
REGISTER_INI_ENTRIES();
/* We assume that we're the only user of the hook. */
ZEND_ASSERT(php_internal_encoding_changed == NULL);
php_internal_encoding_changed = mbstring_internal_encoding_changed_hook;
mbstring_internal_encoding_changed_hook();
/* This is a global handler. Should not be set in a per-request handler. */
sapi_register_treat_data(mbstr_treat_data);
@ -1680,6 +1664,10 @@ PHP_RSHUTDOWN_FUNCTION(mbstring)
MBSTRG(last_used_encoding_name) = NULL;
}
MBSTRG(internal_encoding_set) = 0;
MBSTRG(http_output_set) = 0;
MBSTRG(http_input_set) = 0;
#if HAVE_MBREGEX
PHP_RSHUTDOWN(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
#endif
@ -1763,6 +1751,7 @@ PHP_FUNCTION(mb_internal_encoding)
RETURN_FALSE;
} else {
MBSTRG(current_internal_encoding) = encoding;
MBSTRG(internal_encoding_set) = 1;
RETURN_TRUE;
}
}
@ -1886,6 +1875,7 @@ PHP_FUNCTION(mb_http_output)
php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", name);
RETURN_FALSE;
} else {
MBSTRG(http_output_set) = 1;
MBSTRG(current_http_output_encoding) = encoding;
RETURN_TRUE;
}

View File

@ -130,9 +130,6 @@ MBSTRING_API size_t php_mb_mbchar_bytes(const char *s);
MBSTRING_API size_t php_mb_stripos(int mode, const char *old_haystack, size_t old_haystack_len, const char *old_needle, size_t old_needle_len, zend_long offset, zend_string *from_encoding);
MBSTRING_API int php_mb_check_encoding(const char *input, size_t length, const char *enc);
/* internal use only */
int _php_mb_ini_mbstring_internal_encoding_set(const char *new_value, size_t new_value_length);
ZEND_BEGIN_MODULE_GLOBALS(mbstring)
char *internal_encoding_name;
const mbfl_encoding *internal_encoding;
@ -168,6 +165,10 @@ ZEND_BEGIN_MODULE_GLOBALS(mbstring)
#endif
zend_string *last_used_encoding_name;
const mbfl_encoding *last_used_encoding;
/* Whether an explicit internal_encoding / http_output / http_input encoding was set. */
zend_bool internal_encoding_set;
zend_bool http_output_set;
zend_bool http_input_set;
ZEND_END_MODULE_GLOBALS(mbstring)
#define MBSTRG(v) ZEND_MODULE_GLOBALS_ACCESSOR(mbstring, v)

View File

@ -50,7 +50,7 @@ string(6) "EUC-JP"
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(6) "EUC-JP"
string(0) ""
string(0) ""
Setting INI

View File

@ -0,0 +1,66 @@
--TEST--
Check that "internal_encoding" ini is picked up by mbstring
--INI--
internal_encoding=iso-8859-1
--FILE--
<?php
var_dump(mb_internal_encoding());
var_dump(mb_strlen("\xc3\xb6"));
ini_set('mbstring.internal_encoding', 'utf-8');
var_dump(mb_internal_encoding());
var_dump(mb_strlen("\xc3\xb6"));
// mbstring.internal_encoding is set, this has no effect
ini_set('internal_encoding', 'iso-8859-2');
var_dump(mb_internal_encoding());
var_dump(mb_strlen("\xc3\xb6"));
// mbstring.internal_encoding is unset, pick up internal_encoding again
ini_set('mbstring.internal_encoding', '');
var_dump(mb_internal_encoding());
var_dump(mb_strlen("\xc3\xb6"));
mb_internal_encoding('utf-8');
var_dump(mb_internal_encoding());
var_dump(mb_strlen("\xc3\xb6"));
// mb_internal_encoding() is set, this has no effect
ini_set('internal_encoding', 'iso-8859-3');
var_dump(mb_internal_encoding());
var_dump(mb_strlen("\xc3\xb6"));
// mbstring.internal_encoding is unset, pick up internal_encoding again
ini_set('mbstring.internal_encoding', '');
var_dump(mb_internal_encoding());
var_dump(mb_strlen("\xc3\xb6"));
?>
--EXPECTF--
string(10) "ISO-8859-1"
int(2)
Deprecated: ini_set(): Use of mbstring.internal_encoding is deprecated in %s on line %d
string(5) "UTF-8"
int(1)
string(5) "UTF-8"
int(1)
Deprecated: ini_set(): Use of mbstring.internal_encoding is deprecated in %s on line %d
string(10) "ISO-8859-2"
int(2)
string(5) "UTF-8"
int(1)
string(5) "UTF-8"
int(1)
Deprecated: ini_set(): Use of mbstring.internal_encoding is deprecated in %s on line %d
string(10) "ISO-8859-3"
int(2)

View File

@ -47,7 +47,7 @@ string(10) "ISO-8859-1"
string(0) ""
string(0) ""
string(0) ""
string(5) "UTF-8"
string(10) "ISO-8859-1"
bool(true)
string(5) "UTF-8"
Done

View File

@ -578,12 +578,44 @@ static PHP_INI_DISP(display_errors_mode)
}
/* }}} */
PHPAPI const char *php_get_internal_encoding() {
if (PG(internal_encoding) && PG(internal_encoding)[0]) {
return PG(internal_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
}
PHPAPI const char *php_get_input_encoding() {
if (PG(input_encoding) && PG(input_encoding)[0]) {
return PG(input_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
}
PHPAPI const char *php_get_output_encoding() {
if (PG(output_encoding) && PG(output_encoding)[0]) {
return PG(output_encoding);
} else if (SG(default_charset)) {
return SG(default_charset);
}
return "";
}
PHPAPI void (*php_internal_encoding_changed)(void) = NULL;
/* {{{ PHP_INI_MH
*/
static PHP_INI_MH(OnUpdateDefaultCharset)
{
if (new_value) {
OnUpdateString(entry, new_value, mh_arg1, mh_arg2, mh_arg3, stage);
if (php_internal_encoding_changed) {
php_internal_encoding_changed();
}
#ifdef PHP_WIN32
php_win32_cp_do_update(ZSTR_VAL(new_value));
#endif
@ -598,6 +630,9 @@ static PHP_INI_MH(OnUpdateInternalEncoding)
{
if (new_value) {
OnUpdateString(entry, new_value, mh_arg1, mh_arg2, mh_arg3, stage);
if (php_internal_encoding_changed) {
php_internal_encoding_changed();
}
#ifdef PHP_WIN32
php_win32_cp_do_update(ZSTR_VAL(new_value));
#endif
@ -612,6 +647,9 @@ static PHP_INI_MH(OnUpdateInputEncoding)
{
if (new_value) {
OnUpdateString(entry, new_value, mh_arg1, mh_arg2, mh_arg3, stage);
if (php_internal_encoding_changed) {
php_internal_encoding_changed();
}
#ifdef PHP_WIN32
php_win32_cp_do_update(NULL);
#endif
@ -626,6 +664,9 @@ static PHP_INI_MH(OnUpdateOutputEncoding)
{
if (new_value) {
OnUpdateString(entry, new_value, mh_arg1, mh_arg2, mh_arg3, stage);
if (php_internal_encoding_changed) {
php_internal_encoding_changed();
}
#ifdef PHP_WIN32
php_win32_cp_do_update(NULL);
#endif

View File

@ -371,6 +371,11 @@ PHPAPI int php_mergesort(void *base, size_t nmemb, size_t size, int (*cmp)(const
PHPAPI void php_register_pre_request_shutdown(void (*func)(void *), void *userdata);
PHPAPI void php_com_initialize(void);
PHPAPI char *php_get_current_user(void);
PHPAPI const char *php_get_internal_encoding(void);
PHPAPI const char *php_get_input_encoding(void);
PHPAPI const char *php_get_output_encoding(void);
PHPAPI extern void (*php_internal_encoding_changed)(void);
END_EXTERN_C()
/* PHP-named Zend macro wrappers */