xref: /PHP-7.2/Zend/zend_multibyte.c (revision 7a7ec01a)
1 /*
2    +----------------------------------------------------------------------+
3    | Zend Engine                                                          |
4    +----------------------------------------------------------------------+
5    | Copyright (c) 1998-2018 Zend Technologies Ltd. (http://www.zend.com) |
6    +----------------------------------------------------------------------+
7    | This source file is subject to version 2.00 of the Zend license,     |
8    | that is bundled with this package in the file LICENSE, and is        |
9    | available through the world-wide-web at                              |
10    | http://www.zend.com/license/2_00.txt.                                |
11    | If you did not receive a copy of the Zend license and are unable to  |
12    | obtain it through the world-wide-web, please send a note to          |
13    | license@zend.com so we can mail you a copy immediately.              |
14    +----------------------------------------------------------------------+
15    | Authors: Masaki Fujimoto <fujimoto@php.net>                          |
16    |          Rui Hirokawa <hirokawa@php.net>                             |
17    +----------------------------------------------------------------------+
18 */
19 
20 /* $Id$ */
21 
22 #include "zend.h"
23 #include "zend_compile.h"
24 #include "zend_operators.h"
25 #include "zend_multibyte.h"
26 #include "zend_ini.h"
27 
dummy_encoding_fetcher(const char * encoding_name)28 static const zend_encoding *dummy_encoding_fetcher(const char *encoding_name)
29 {
30 	return NULL;
31 }
32 
dummy_encoding_name_getter(const zend_encoding * encoding)33 static const char *dummy_encoding_name_getter(const zend_encoding *encoding)
34 {
35 	return (const char*)encoding;
36 }
37 
dummy_encoding_lexer_compatibility_checker(const zend_encoding * encoding)38 static int dummy_encoding_lexer_compatibility_checker(const zend_encoding *encoding)
39 {
40 	return 0;
41 }
42 
dummy_encoding_detector(const unsigned char * string,size_t length,const zend_encoding ** list,size_t list_size)43 static const zend_encoding *dummy_encoding_detector(const unsigned char *string, size_t length, const zend_encoding **list, size_t list_size)
44 {
45 	return NULL;
46 }
47 
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)48 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)
49 {
50 	return (size_t)-1;
51 }
52 
dummy_encoding_list_parser(const char * encoding_list,size_t encoding_list_len,const zend_encoding *** return_list,size_t * return_size,int persistent)53 static int dummy_encoding_list_parser(const char *encoding_list, size_t encoding_list_len, const zend_encoding ***return_list, size_t *return_size, int persistent)
54 {
55 	*return_list = pemalloc(0, persistent);
56 	*return_size = 0;
57 	return SUCCESS;
58 }
59 
dummy_internal_encoding_getter(void)60 static const zend_encoding *dummy_internal_encoding_getter(void)
61 {
62 	return NULL;
63 }
64 
dummy_internal_encoding_setter(const zend_encoding * encoding)65 static int dummy_internal_encoding_setter(const zend_encoding *encoding)
66 {
67 	return FAILURE;
68 }
69 
70 static zend_multibyte_functions multibyte_functions_dummy;
71 static zend_multibyte_functions multibyte_functions = {
72 	NULL,
73 	dummy_encoding_fetcher,
74 	dummy_encoding_name_getter,
75 	dummy_encoding_lexer_compatibility_checker,
76 	dummy_encoding_detector,
77 	dummy_encoding_converter,
78 	dummy_encoding_list_parser,
79 	dummy_internal_encoding_getter,
80 	dummy_internal_encoding_setter
81 };
82 
83 ZEND_API const zend_encoding *zend_multibyte_encoding_utf32be = (const zend_encoding*)"UTF-32BE";
84 ZEND_API const zend_encoding *zend_multibyte_encoding_utf32le = (const zend_encoding*)"UTF-32LE";
85 ZEND_API const zend_encoding *zend_multibyte_encoding_utf16be = (const zend_encoding*)"UTF-16BE";
86 ZEND_API const zend_encoding *zend_multibyte_encoding_utf16le = (const zend_encoding*)"UTF-32LE";
87 ZEND_API const zend_encoding *zend_multibyte_encoding_utf8 = (const zend_encoding*)"UTF-8";
88 
zend_multibyte_set_functions(const zend_multibyte_functions * functions)89 ZEND_API int zend_multibyte_set_functions(const zend_multibyte_functions *functions)
90 {
91 	zend_multibyte_encoding_utf32be = functions->encoding_fetcher("UTF-32BE");
92 	if (!zend_multibyte_encoding_utf32be) {
93 		return FAILURE;
94 	}
95 	zend_multibyte_encoding_utf32le = functions->encoding_fetcher("UTF-32LE");
96 	if (!zend_multibyte_encoding_utf32le) {
97 		return FAILURE;
98 	}
99 	zend_multibyte_encoding_utf16be = functions->encoding_fetcher("UTF-16BE");
100 	if (!zend_multibyte_encoding_utf16be) {
101 		return FAILURE;
102 	}
103 	zend_multibyte_encoding_utf16le = functions->encoding_fetcher("UTF-16LE");
104 	if (!zend_multibyte_encoding_utf16le) {
105 		return FAILURE;
106 	}
107 	zend_multibyte_encoding_utf8 = functions->encoding_fetcher("UTF-8");
108 	if (!zend_multibyte_encoding_utf8) {
109 		return FAILURE;
110 	}
111 
112 	multibyte_functions_dummy = multibyte_functions;
113 	multibyte_functions = *functions;
114 
115 	/* As zend_multibyte_set_functions() gets called after ini settings were
116 	 * populated, we need to reinitialize script_encoding here.
117 	 */
118 	{
119 		const char *value = zend_ini_string("zend.script_encoding", sizeof("zend.script_encoding") - 1, 0);
120 		zend_multibyte_set_script_encoding_by_string(value, strlen(value));
121 	}
122 	return SUCCESS;
123 }
124 
zend_multibyte_restore_functions(void)125 ZEND_API void zend_multibyte_restore_functions(void)
126 {
127 	multibyte_functions = multibyte_functions_dummy;
128 }
129 
zend_multibyte_get_functions(void)130 ZEND_API const zend_multibyte_functions *zend_multibyte_get_functions(void)
131 {
132 	return multibyte_functions.provider_name ? &multibyte_functions: NULL;
133 }
134 
zend_multibyte_fetch_encoding(const char * name)135 ZEND_API const zend_encoding *zend_multibyte_fetch_encoding(const char *name)
136 {
137 	return multibyte_functions.encoding_fetcher(name);
138 }
139 
zend_multibyte_get_encoding_name(const zend_encoding * encoding)140 ZEND_API const char *zend_multibyte_get_encoding_name(const zend_encoding *encoding)
141 {
142 	return multibyte_functions.encoding_name_getter(encoding);
143 }
144 
zend_multibyte_check_lexer_compatibility(const zend_encoding * encoding)145 ZEND_API int zend_multibyte_check_lexer_compatibility(const zend_encoding *encoding)
146 {
147 	return multibyte_functions.lexer_compatibility_checker(encoding);
148 }
149 
zend_multibyte_encoding_detector(const unsigned char * string,size_t length,const zend_encoding ** list,size_t list_size)150 ZEND_API const zend_encoding *zend_multibyte_encoding_detector(const unsigned char *string, size_t length, const zend_encoding **list, size_t list_size)
151 {
152 	return multibyte_functions.encoding_detector(string, length, list, list_size);
153 }
154 
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)155 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)
156 {
157 	return multibyte_functions.encoding_converter(to, to_length, from, from_length, encoding_to, encoding_from);
158 }
159 
zend_multibyte_parse_encoding_list(const char * encoding_list,size_t encoding_list_len,const zend_encoding *** return_list,size_t * return_size,int persistent)160 ZEND_API int zend_multibyte_parse_encoding_list(const char *encoding_list, size_t encoding_list_len, const zend_encoding ***return_list, size_t *return_size, int persistent)
161 {
162 	return multibyte_functions.encoding_list_parser(encoding_list, encoding_list_len, return_list, return_size, persistent);
163 }
164 
zend_multibyte_get_internal_encoding(void)165 ZEND_API const zend_encoding *zend_multibyte_get_internal_encoding(void)
166 {
167 	return multibyte_functions.internal_encoding_getter();
168 }
169 
zend_multibyte_get_script_encoding(void)170 ZEND_API const zend_encoding *zend_multibyte_get_script_encoding(void)
171 {
172 	return LANG_SCNG(script_encoding);
173 }
174 
zend_multibyte_set_script_encoding(const zend_encoding ** encoding_list,size_t encoding_list_size)175 ZEND_API int zend_multibyte_set_script_encoding(const zend_encoding **encoding_list, size_t encoding_list_size)
176 {
177 	if (CG(script_encoding_list)) {
178 		free((char*)CG(script_encoding_list));
179 	}
180 	CG(script_encoding_list) = encoding_list;
181 	CG(script_encoding_list_size) = encoding_list_size;
182 	return SUCCESS;
183 }
184 
zend_multibyte_set_internal_encoding(const zend_encoding * encoding)185 ZEND_API int zend_multibyte_set_internal_encoding(const zend_encoding *encoding)
186 {
187 	return multibyte_functions.internal_encoding_setter(encoding);
188 }
189 
zend_multibyte_set_script_encoding_by_string(const char * new_value,size_t new_value_length)190 ZEND_API int zend_multibyte_set_script_encoding_by_string(const char *new_value, size_t new_value_length)
191 {
192 	const zend_encoding **list = 0;
193 	size_t size = 0;
194 
195 	if (!new_value) {
196 		zend_multibyte_set_script_encoding(NULL, 0);
197 		return SUCCESS;
198 	}
199 
200 	if (FAILURE == zend_multibyte_parse_encoding_list(new_value, new_value_length, &list, &size, 1)) {
201 		return FAILURE;
202 	}
203 
204 	if (size == 0) {
205 		pefree((void*)list, 1);
206 		return FAILURE;
207 	}
208 
209 	if (FAILURE == zend_multibyte_set_script_encoding(list, size)) {
210 		return FAILURE;
211 	}
212 
213 	return SUCCESS;
214 }
215 
216 /*
217  * Local variables:
218  * tab-width: 4
219  * c-basic-offset: 4
220  * End:
221  * vim600: sw=4 ts=4 fdm=marker
222  * vim<600: sw=4 ts=4
223  */
224