xref: /PHP-5.6/Zend/zend_multibyte.c (revision 3537e95d)
1 /*
2    +----------------------------------------------------------------------+
3    | Zend Engine                                                          |
4    +----------------------------------------------------------------------+
5    | Copyright (c) 1998-2016 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 TSRMLS_DC)28 static const zend_encoding *dummy_encoding_fetcher(const char *encoding_name TSRMLS_DC)
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 TSRMLS_DC)43 static const zend_encoding *dummy_encoding_detector(const unsigned char *string, size_t length, const zend_encoding **list, size_t list_size TSRMLS_DC)
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 TSRMLS_DC)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 TSRMLS_DC)
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 TSRMLS_DC)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 TSRMLS_DC)
54 {
55 	*return_list = pemalloc(0, persistent);
56 	*return_size = 0;
57 	return SUCCESS;
58 }
59 
dummy_internal_encoding_getter(TSRMLS_D)60 static const zend_encoding *dummy_internal_encoding_getter(TSRMLS_D)
61 {
62 	return NULL;
63 }
64 
dummy_internal_encoding_setter(const zend_encoding * encoding TSRMLS_DC)65 static int dummy_internal_encoding_setter(const zend_encoding *encoding TSRMLS_DC)
66 {
67 	return FAILURE;
68 }
69 
70 static zend_multibyte_functions multibyte_functions = {
71 	NULL,
72 	dummy_encoding_fetcher,
73 	dummy_encoding_name_getter,
74 	dummy_encoding_lexer_compatibility_checker,
75 	dummy_encoding_detector,
76 	dummy_encoding_converter,
77 	dummy_encoding_list_parser,
78 	dummy_internal_encoding_getter,
79 	dummy_internal_encoding_setter
80 };
81 
82 ZEND_API const zend_encoding *zend_multibyte_encoding_utf32be = (const zend_encoding*)"UTF-32BE";
83 ZEND_API const zend_encoding *zend_multibyte_encoding_utf32le = (const zend_encoding*)"UTF-32LE";
84 ZEND_API const zend_encoding *zend_multibyte_encoding_utf16be = (const zend_encoding*)"UTF-16BE";
85 ZEND_API const zend_encoding *zend_multibyte_encoding_utf16le = (const zend_encoding*)"UTF-32LE";
86 ZEND_API const zend_encoding *zend_multibyte_encoding_utf8 = (const zend_encoding*)"UTF-8";
87 
zend_multibyte_set_functions(const zend_multibyte_functions * functions TSRMLS_DC)88 ZEND_API int zend_multibyte_set_functions(const zend_multibyte_functions *functions TSRMLS_DC)
89 {
90 	zend_multibyte_encoding_utf32be = functions->encoding_fetcher("UTF-32BE" TSRMLS_CC);
91 	if (!zend_multibyte_encoding_utf32be) {
92 		return FAILURE;
93 	}
94 	zend_multibyte_encoding_utf32le = functions->encoding_fetcher("UTF-32LE" TSRMLS_CC);
95 	if (!zend_multibyte_encoding_utf32le) {
96 		return FAILURE;
97 	}
98 	zend_multibyte_encoding_utf16be = functions->encoding_fetcher("UTF-16BE" TSRMLS_CC);
99 	if (!zend_multibyte_encoding_utf16be) {
100 		return FAILURE;
101 	}
102 	zend_multibyte_encoding_utf16le = functions->encoding_fetcher("UTF-16LE" TSRMLS_CC);
103 	if (!zend_multibyte_encoding_utf16le) {
104 		return FAILURE;
105 	}
106 	zend_multibyte_encoding_utf8 = functions->encoding_fetcher("UTF-8" TSRMLS_CC);
107 	if (!zend_multibyte_encoding_utf8) {
108 		return FAILURE;
109 	}
110 
111 	multibyte_functions = *functions;
112 
113 	/* As zend_multibyte_set_functions() gets called after ini settings were
114 	 * populated, we need to reinitialize script_encoding here.
115 	 */
116 	{
117 		const char *value = zend_ini_string("zend.script_encoding", sizeof("zend.script_encoding"), 0);
118 		zend_multibyte_set_script_encoding_by_string(value, strlen(value) TSRMLS_CC);
119 	}
120 	return SUCCESS;
121 }
122 
zend_multibyte_get_functions(TSRMLS_D)123 ZEND_API const zend_multibyte_functions *zend_multibyte_get_functions(TSRMLS_D)
124 {
125 	return multibyte_functions.provider_name ? &multibyte_functions: NULL;
126 }
127 
zend_multibyte_fetch_encoding(const char * name TSRMLS_DC)128 ZEND_API const zend_encoding *zend_multibyte_fetch_encoding(const char *name TSRMLS_DC)
129 {
130 	return multibyte_functions.encoding_fetcher(name TSRMLS_CC);
131 }
132 
zend_multibyte_get_encoding_name(const zend_encoding * encoding)133 ZEND_API const char *zend_multibyte_get_encoding_name(const zend_encoding *encoding)
134 {
135 	return multibyte_functions.encoding_name_getter(encoding);
136 }
137 
zend_multibyte_check_lexer_compatibility(const zend_encoding * encoding)138 ZEND_API int zend_multibyte_check_lexer_compatibility(const zend_encoding *encoding)
139 {
140 	return multibyte_functions.lexer_compatibility_checker(encoding);
141 }
142 
zend_multibyte_encoding_detector(const unsigned char * string,size_t length,const zend_encoding ** list,size_t list_size TSRMLS_DC)143 ZEND_API const zend_encoding *zend_multibyte_encoding_detector(const unsigned char *string, size_t length, const zend_encoding **list, size_t list_size TSRMLS_DC)
144 {
145 	return multibyte_functions.encoding_detector(string, length, list, list_size TSRMLS_CC);
146 }
147 
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 TSRMLS_DC)148 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 TSRMLS_DC)
149 {
150 	return multibyte_functions.encoding_converter(to, to_length, from, from_length, encoding_to, encoding_from TSRMLS_CC);
151 }
152 
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 TSRMLS_DC)153 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 TSRMLS_DC)
154 {
155 	return multibyte_functions.encoding_list_parser(encoding_list, encoding_list_len, return_list, return_size, persistent TSRMLS_CC);
156 }
157 
zend_multibyte_get_internal_encoding(TSRMLS_D)158 ZEND_API const zend_encoding *zend_multibyte_get_internal_encoding(TSRMLS_D)
159 {
160 	return multibyte_functions.internal_encoding_getter(TSRMLS_C);
161 }
162 
zend_multibyte_get_script_encoding(TSRMLS_D)163 ZEND_API const zend_encoding *zend_multibyte_get_script_encoding(TSRMLS_D)
164 {
165 	return LANG_SCNG(script_encoding);
166 }
167 
zend_multibyte_set_script_encoding(const zend_encoding ** encoding_list,size_t encoding_list_size TSRMLS_DC)168 ZEND_API int zend_multibyte_set_script_encoding(const zend_encoding **encoding_list, size_t encoding_list_size TSRMLS_DC)
169 {
170 	if (CG(script_encoding_list)) {
171 		free(CG(script_encoding_list));
172 	}
173 	CG(script_encoding_list) = encoding_list;
174 	CG(script_encoding_list_size) = encoding_list_size;
175 	return SUCCESS;
176 }
177 
zend_multibyte_set_internal_encoding(const zend_encoding * encoding TSRMLS_DC)178 ZEND_API int zend_multibyte_set_internal_encoding(const zend_encoding *encoding TSRMLS_DC)
179 {
180 	return multibyte_functions.internal_encoding_setter(encoding TSRMLS_CC);
181 }
182 
zend_multibyte_set_script_encoding_by_string(const char * new_value,size_t new_value_length TSRMLS_DC)183 ZEND_API int zend_multibyte_set_script_encoding_by_string(const char *new_value, size_t new_value_length TSRMLS_DC)
184 {
185 	const zend_encoding **list = 0;
186 	size_t size = 0;
187 
188 	if (!new_value) {
189 		zend_multibyte_set_script_encoding(NULL, 0 TSRMLS_CC);
190 		return SUCCESS;
191 	}
192 
193 	if (FAILURE == zend_multibyte_parse_encoding_list(new_value, new_value_length, &list, &size, 1 TSRMLS_CC)) {
194 		return FAILURE;
195 	}
196 
197 	if (size == 0) {
198 		pefree(list, 1);
199 		return FAILURE;
200 	}
201 
202 	if (FAILURE == zend_multibyte_set_script_encoding(list, size TSRMLS_CC)) {
203 		return FAILURE;
204 	}
205 
206 	return SUCCESS;
207 }
208 
209 /*
210  * Local variables:
211  * tab-width: 4
212  * c-basic-offset: 4
213  * End:
214  * vim600: sw=4 ts=4 tw=78
215  * vim<600: sw=4 ts=4 tw=78
216  */
217