/src/php-src/Zend/zend_multibyte.c

Source
/*
   +----------------------------------------------------------------------+
   | Zend Engine                                                          |
   +----------------------------------------------------------------------+
   | Copyright © Zend Technologies Ltd., a subsidiary company of          |
   |     Perforce Software, Inc., and Contributors.                       |
   +----------------------------------------------------------------------+
   | 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: Masaki Fujimoto <fujimoto@php.net>                          |
   |          Rui Hirokawa <hirokawa@php.net>                             |
   +----------------------------------------------------------------------+
*/

#include "zend.h"
#include "zend_compile.h"
#include "zend_operators.h"
#include "zend_multibyte.h"
#include "zend_ini.h"

static const zend_encoding *dummy_encoding_fetcher(const char *encoding_name)
{
  return NULL;
}

static const char *dummy_encoding_name_getter(const zend_encoding *encoding)
{
  return (const char*)encoding;
}

static bool dummy_encoding_lexer_compatibility_checker(const zend_encoding *encoding)
{
  return false;
}

static const zend_encoding *dummy_encoding_detector(const unsigned char *string, size_t length, const zend_encoding **list, size_t list_size)
{
  return NULL;
}

static size_t dummy_encoding_converter(unsigned char **to, size_t *to_length, const unsigned char *from, size_t from_length, const zend_encoding *encoding_to, const zend_encoding *encoding_from)
{
  return (size_t)-1;
}

static zend_result dummy_encoding_list_parser(const char *encoding_list, size_t encoding_list_len, const zend_encoding ***return_list, size_t *return_size, bool persistent)
{
  *return_list = pemalloc(0, persistent);
  *return_size = 0;
  return SUCCESS;
}

static const zend_encoding *dummy_internal_encoding_getter(void)
{
  return NULL;
}

static zend_result dummy_internal_encoding_setter(const zend_encoding *encoding)
{
  return FAILURE;
}

static zend_multibyte_functions multibyte_functions_dummy;
static zend_multibyte_functions multibyte_functions = {
  NULL,
  dummy_encoding_fetcher,
  dummy_encoding_name_getter,
  dummy_encoding_lexer_compatibility_checker,
  dummy_encoding_detector,
  dummy_encoding_converter,
  dummy_encoding_list_parser,
  dummy_internal_encoding_getter,
  dummy_internal_encoding_setter
};

ZEND_API const zend_encoding *zend_multibyte_encoding_utf32be = (const zend_encoding*)"UTF-32BE";
ZEND_API const zend_encoding *zend_multibyte_encoding_utf32le = (const zend_encoding*)"UTF-32LE";
ZEND_API const zend_encoding *zend_multibyte_encoding_utf16be = (const zend_encoding*)"UTF-16BE";
ZEND_API const zend_encoding *zend_multibyte_encoding_utf16le = (const zend_encoding*)"UTF-32LE";
ZEND_API const zend_encoding *zend_multibyte_encoding_utf8 = (const zend_encoding*)"UTF-8";

ZEND_API zend_result zend_multibyte_set_functions(const zend_multibyte_functions *functions)
{
  zend_multibyte_encoding_utf32be = functions->encoding_fetcher("UTF-32BE");
  if (!zend_multibyte_encoding_utf32be) {
    return FAILURE;
  }
  zend_multibyte_encoding_utf32le = functions->encoding_fetcher("UTF-32LE");
  if (!zend_multibyte_encoding_utf32le) {
    return FAILURE;
  }
  zend_multibyte_encoding_utf16be = functions->encoding_fetcher("UTF-16BE");
  if (!zend_multibyte_encoding_utf16be) {
    return FAILURE;
  }
  zend_multibyte_encoding_utf16le = functions->encoding_fetcher("UTF-16LE");
  if (!zend_multibyte_encoding_utf16le) {
    return FAILURE;
  }
  zend_multibyte_encoding_utf8 = functions->encoding_fetcher("UTF-8");
  if (!zend_multibyte_encoding_utf8) {
    return FAILURE;
  }

  multibyte_functions_dummy = multibyte_functions;
  multibyte_functions = *functions;

  /* As zend_multibyte_set_functions() gets called after ini settings were
   * populated, we need to reinitialize script_encoding here.
   */
  {
    const zend_string *value = zend_ini_str_literal("zend.script_encoding");
    zend_multibyte_set_script_encoding_by_string(ZSTR_VAL(value), ZSTR_LEN(value));
  }
  return SUCCESS;
}

ZEND_API void zend_multibyte_restore_functions(void)
{
  multibyte_functions = multibyte_functions_dummy;
}

ZEND_API const zend_multibyte_functions *zend_multibyte_get_functions(void)
{
  return multibyte_functions.provider_name ? &multibyte_functions: NULL;
}

ZEND_API const zend_encoding *zend_multibyte_fetch_encoding(const char *name)
{
  return multibyte_functions.encoding_fetcher(name);
}

ZEND_API const char *zend_multibyte_get_encoding_name(const zend_encoding *encoding)
{
  return multibyte_functions.encoding_name_getter(encoding);
}

ZEND_API int zend_multibyte_check_lexer_compatibility(const zend_encoding *encoding)
{
  return multibyte_functions.lexer_compatibility_checker(encoding);
}

ZEND_API const zend_encoding *zend_multibyte_encoding_detector(const unsigned char *string, size_t length, const zend_encoding **list, size_t list_size)
{
  return multibyte_functions.encoding_detector(string, length, list, list_size);
}

ZEND_API size_t zend_multibyte_encoding_converter(unsigned char **to, size_t *to_length, const unsigned char *from, size_t from_length, const zend_encoding *encoding_to, const zend_encoding *encoding_from)
{
  return multibyte_functions.encoding_converter(to, to_length, from, from_length, encoding_to, encoding_from);
}

ZEND_API zend_result zend_multibyte_parse_encoding_list(const char *encoding_list, size_t encoding_list_len, const zend_encoding ***return_list, size_t *return_size, bool persistent)
{
  return multibyte_functions.encoding_list_parser(encoding_list, encoding_list_len, return_list, return_size, persistent);
}

ZEND_API const zend_encoding *zend_multibyte_get_internal_encoding(void)
{
  return multibyte_functions.internal_encoding_getter();
}

ZEND_API const zend_encoding *zend_multibyte_get_script_encoding(void)
{
  return LANG_SCNG(script_encoding);
}

ZEND_API zend_result zend_multibyte_set_script_encoding(const zend_encoding **encoding_list, size_t encoding_list_size)
{
  if (CG(script_encoding_list)) {
    free((char*)CG(script_encoding_list));
  }
  CG(script_encoding_list) = encoding_list;
  CG(script_encoding_list_size) = encoding_list_size;
  return SUCCESS;
}

ZEND_API zend_result zend_multibyte_set_internal_encoding(const zend_encoding *encoding)
{
  return multibyte_functions.internal_encoding_setter(encoding);
}

ZEND_API zend_result zend_multibyte_set_script_encoding_by_string(const char *new_value, size_t new_value_length)
{
  const zend_encoding **list = 0;
  size_t size = 0;

  if (!new_value) {
    zend_multibyte_set_script_encoding(NULL, 0);
    return SUCCESS;
  }

  if (FAILURE == zend_multibyte_parse_encoding_list(new_value, new_value_length, &list, &size, true)) {
    return FAILURE;
  }

  if (size == 0) {
    pefree((void*)list, 1);
    return FAILURE;
  }

  if (FAILURE == zend_multibyte_set_script_encoding(list, size)) {
    return FAILURE;
  }

  return SUCCESS;
}

Coverage Report

Created: 2026-06-02 06:36

Line	Count	Source
1		/*
2		+----------------------------------------------------------------------+
3		\| Zend Engine \|
4		+----------------------------------------------------------------------+
5		\| Copyright © Zend Technologies Ltd., a subsidiary company of \|
6		\| Perforce Software, Inc., and Contributors. \|
7		+----------------------------------------------------------------------+
8		\| This source file is subject to the Modified BSD License that is \|
9		\| bundled with this package in the file LICENSE, and is available \|
10		\| through the World Wide Web at <https://www.php.net/license/>. \|
11		\| \|
12		\| SPDX-License-Identifier: BSD-3-Clause \|
13		+----------------------------------------------------------------------+
14		\| Authors: Masaki Fujimoto <fujimoto@php.net> \|
15		\| Rui Hirokawa <hirokawa@php.net> \|
16		+----------------------------------------------------------------------+
17		*/
18
19		#include "zend.h"
20		#include "zend_compile.h"
21		#include "zend_operators.h"
22		#include "zend_multibyte.h"
23		#include "zend_ini.h"
24
25		static const zend_encoding dummy_encoding_fetcher(const char encoding_name)
26	0	{
27	0	return NULL;
28	0	}
29
30		static const char dummy_encoding_name_getter(const zend_encoding encoding)
31	0	{
32	0	return (const char*)encoding;
33	0	}
34
35		static bool dummy_encoding_lexer_compatibility_checker(const zend_encoding *encoding)
36	0	{
37	0	return false;
38	0	}
39
40		static const zend_encoding dummy_encoding_detector(const unsigned char string, size_t length, const zend_encoding **list, size_t list_size)
41	0	{
42	0	return NULL;
43	0	}
44
45		static size_t dummy_encoding_converter(unsigned char *to, size_t to_length, const unsigned char from, size_t from_length, const zend_encoding encoding_to, const zend_encoding *encoding_from)
46	0	{
47	0	return (size_t)-1;
48	0	}
49
50		static zend_result dummy_encoding_list_parser(const char encoding_list, size_t encoding_list_len, const zend_encoding *return_list, size_t return_size, bool persistent)
51	10	{
52	10	*return_list = pemalloc(0, persistent);
53	10	*return_size = 0;
54	10	return SUCCESS;
55	10	}
56
57		static const zend_encoding *dummy_internal_encoding_getter(void)
58	0	{
59	0	return NULL;
60	0	}
61
62		static zend_result dummy_internal_encoding_setter(const zend_encoding *encoding)
63	0	{
64	0	return FAILURE;
65	0	}
66
67		static zend_multibyte_functions multibyte_functions_dummy;
68		static zend_multibyte_functions multibyte_functions = {
69		NULL,
70		dummy_encoding_fetcher,
71		dummy_encoding_name_getter,
72		dummy_encoding_lexer_compatibility_checker,
73		dummy_encoding_detector,
74		dummy_encoding_converter,
75		dummy_encoding_list_parser,
76		dummy_internal_encoding_getter,
77		dummy_internal_encoding_setter
78		};
79
80		ZEND_API const zend_encoding zend_multibyte_encoding_utf32be = (const zend_encoding)"UTF-32BE";
81		ZEND_API const zend_encoding zend_multibyte_encoding_utf32le = (const zend_encoding)"UTF-32LE";
82		ZEND_API const zend_encoding zend_multibyte_encoding_utf16be = (const zend_encoding)"UTF-16BE";
83		ZEND_API const zend_encoding zend_multibyte_encoding_utf16le = (const zend_encoding)"UTF-32LE";
84		ZEND_API const zend_encoding zend_multibyte_encoding_utf8 = (const zend_encoding)"UTF-8";
85
86		ZEND_API zend_result zend_multibyte_set_functions(const zend_multibyte_functions *functions)
87	0	{
88	0	zend_multibyte_encoding_utf32be = functions->encoding_fetcher("UTF-32BE");
89	0	if (!zend_multibyte_encoding_utf32be) {
90	0	return FAILURE;
91	0	}
92	0	zend_multibyte_encoding_utf32le = functions->encoding_fetcher("UTF-32LE");
93	0	if (!zend_multibyte_encoding_utf32le) {
94	0	return FAILURE;
95	0	}
96	0	zend_multibyte_encoding_utf16be = functions->encoding_fetcher("UTF-16BE");
97	0	if (!zend_multibyte_encoding_utf16be) {
98	0	return FAILURE;
99	0	}
100	0	zend_multibyte_encoding_utf16le = functions->encoding_fetcher("UTF-16LE");
101	0	if (!zend_multibyte_encoding_utf16le) {
102	0	return FAILURE;
103	0	}
104	0	zend_multibyte_encoding_utf8 = functions->encoding_fetcher("UTF-8");
105	0	if (!zend_multibyte_encoding_utf8) {
106	0	return FAILURE;
107	0	}
108
109	0	multibyte_functions_dummy = multibyte_functions;
110	0	multibyte_functions = *functions;
111
112		/* As zend_multibyte_set_functions() gets called after ini settings were
113		* populated, we need to reinitialize script_encoding here.
114		*/
115	0	{
116	0	const zend_string *value = zend_ini_str_literal("zend.script_encoding");
117	0	zend_multibyte_set_script_encoding_by_string(ZSTR_VAL(value), ZSTR_LEN(value));
118	0	}
119	0	return SUCCESS;
120	0	}
121
122		ZEND_API void zend_multibyte_restore_functions(void)
123	0	{
124	0	multibyte_functions = multibyte_functions_dummy;
125	0	}
126
127		ZEND_API const zend_multibyte_functions *zend_multibyte_get_functions(void)
128	1	{
129	1	return multibyte_functions.provider_name ? &multibyte_functions: NULL;
130	1	}
131
132		ZEND_API const zend_encoding zend_multibyte_fetch_encoding(const char name)
133	0	{
134	0	return multibyte_functions.encoding_fetcher(name);
135	0	}
136
137		ZEND_API const char zend_multibyte_get_encoding_name(const zend_encoding encoding)
138	0	{
139	0	return multibyte_functions.encoding_name_getter(encoding);
140	0	}
141
142		ZEND_API int zend_multibyte_check_lexer_compatibility(const zend_encoding *encoding)
143	0	{
144	0	return multibyte_functions.lexer_compatibility_checker(encoding);
145	0	}
146
147		ZEND_API const zend_encoding zend_multibyte_encoding_detector(const unsigned char string, size_t length, const zend_encoding **list, size_t list_size)
148	0	{
149	0	return multibyte_functions.encoding_detector(string, length, list, list_size);
150	0	}
151
152		ZEND_API size_t zend_multibyte_encoding_converter(unsigned char *to, size_t to_length, const unsigned char from, size_t from_length, const zend_encoding encoding_to, const zend_encoding *encoding_from)
153	0	{
154	0	return multibyte_functions.encoding_converter(to, to_length, from, from_length, encoding_to, encoding_from);
155	0	}
156
157		ZEND_API zend_result zend_multibyte_parse_encoding_list(const char encoding_list, size_t encoding_list_len, const zend_encoding *return_list, size_t return_size, bool persistent)
158	10	{
159	10	return multibyte_functions.encoding_list_parser(encoding_list, encoding_list_len, return_list, return_size, persistent);
160	10	}
161
162		ZEND_API const zend_encoding *zend_multibyte_get_internal_encoding(void)
163	0	{
164	0	return multibyte_functions.internal_encoding_getter();
165	0	}
166
167		ZEND_API const zend_encoding *zend_multibyte_get_script_encoding(void)
168	0	{
169	0	return LANG_SCNG(script_encoding);
170	0	}
171
172		ZEND_API zend_result zend_multibyte_set_script_encoding(const zend_encoding **encoding_list, size_t encoding_list_size)
173	0	{
174	0	if (CG(script_encoding_list)) {
175	0	free((char*)CG(script_encoding_list));
176	0	}
177	0	CG(script_encoding_list) = encoding_list;
178	0	CG(script_encoding_list_size) = encoding_list_size;
179	0	return SUCCESS;
180	0	}
181
182		ZEND_API zend_result zend_multibyte_set_internal_encoding(const zend_encoding *encoding)
183	0	{
184	0	return multibyte_functions.internal_encoding_setter(encoding);
185	0	}
186
187		ZEND_API zend_result zend_multibyte_set_script_encoding_by_string(const char *new_value, size_t new_value_length)
188	0	{
189	0	const zend_encoding **list = 0;
190	0	size_t size = 0;
191
192	0	if (!new_value) {
193	0	zend_multibyte_set_script_encoding(NULL, 0);
194	0	return SUCCESS;
195	0	}
196
197	0	if (FAILURE == zend_multibyte_parse_encoding_list(new_value, new_value_length, &list, &size, true)) {
198	0	return FAILURE;
199	0	}
200
201	0	if (size == 0) {
202	0	pefree((void*)list, 1);
203	0	return FAILURE;
204	0	}
205
206	0	if (FAILURE == zend_multibyte_set_script_encoding(list, size)) {
207	0	return FAILURE;
208	0	}
209
210	0	return SUCCESS;
211	0	}