1 /*
2 +----------------------------------------------------------------------+
3 | PHP Version 7 |
4 +----------------------------------------------------------------------+
5 | Copyright (c) 1997-2018 The PHP Group |
6 +----------------------------------------------------------------------+
7 | This source file is subject to version 3.01 of the PHP license, |
8 | that is bundled with this package in the file LICENSE, and is |
9 | available through the world-wide-web at the following url: |
10 | http://www.php.net/license/3_01.txt |
11 | If you did not receive a copy of the PHP license and are unable to |
12 | obtain it through the world-wide-web, please send a note to |
13 | license@php.net so we can mail you a copy immediately. |
14 +----------------------------------------------------------------------+
15 | Author: Tsukada Takuya <tsukada@fminn.nagano.nagano.jp> |
16 | Rui Hirokawa <hirokawa@php.net> |
17 +----------------------------------------------------------------------+
18 */
19
20 /* $Id$ */
21
22 /*
23 * PHP 4 Multibyte String module "mbstring"
24 *
25 * History:
26 * 2000.5.19 Release php-4.0RC2_jstring-1.0
27 * 2001.4.1 Release php4_jstring-1.0.91
28 * 2001.4.30 Release php4_jstring-1.1 (contribute to The PHP Group)
29 * 2001.5.1 Renamed from jstring to mbstring (hirokawa@php.net)
30 */
31
32 /*
33 * PHP3 Internationalization support program.
34 *
35 * Copyright (c) 1999,2000 by the PHP3 internationalization team.
36 * All rights reserved.
37 *
38 * See README_PHP3-i18n-ja for more detail.
39 *
40 * Authors:
41 * Hironori Sato <satoh@jpnnet.com>
42 * Shigeru Kanemoto <sgk@happysize.co.jp>
43 * Tsukada Takuya <tsukada@fminn.nagano.nagano.jp>
44 * Rui Hirokawa <rui_hirokawa@ybb.ne.jp>
45 */
46
47 /* {{{ includes */
48 #ifdef HAVE_CONFIG_H
49 #include "config.h"
50 #endif
51
52 #include "php.h"
53 #include "php_ini.h"
54 #include "php_variables.h"
55 #include "mbstring.h"
56 #include "ext/standard/php_string.h"
57 #include "ext/standard/php_mail.h"
58 #include "ext/standard/exec.h"
59 #include "ext/standard/url.h"
60 #include "main/php_output.h"
61 #include "ext/standard/info.h"
62
63 #include "libmbfl/mbfl/mbfl_allocators.h"
64 #include "libmbfl/mbfl/mbfilter_pass.h"
65
66 #include "php_variables.h"
67 #include "php_globals.h"
68 #include "rfc1867.h"
69 #include "php_content_types.h"
70 #include "SAPI.h"
71 #include "php_unicode.h"
72 #include "TSRM.h"
73
74 #include "mb_gpc.h"
75
76 #if HAVE_MBREGEX
77 #include "php_mbregex.h"
78 #endif
79
80 #include "zend_multibyte.h"
81
82 #if HAVE_ONIG
83 #include "php_onig_compat.h"
84 #include <oniguruma.h>
85 #undef UChar
86 #elif HAVE_PCRE || HAVE_BUNDLED_PCRE
87 #include "ext/pcre/php_pcre.h"
88 #endif
89 /* }}} */
90
91 #if HAVE_MBSTRING
92
93 /* {{{ prototypes */
94 ZEND_DECLARE_MODULE_GLOBALS(mbstring)
95
96 static PHP_GINIT_FUNCTION(mbstring);
97 static PHP_GSHUTDOWN_FUNCTION(mbstring);
98
99 static void php_mb_populate_current_detect_order_list(void);
100
101 static int php_mb_encoding_translation(void);
102
103 static void php_mb_gpc_get_detect_order(const zend_encoding ***list, size_t *list_size);
104
105 static void php_mb_gpc_set_input_encoding(const zend_encoding *encoding);
106
107 /* }}} */
108
109 /* {{{ php_mb_default_identify_list */
110 typedef struct _php_mb_nls_ident_list {
111 enum mbfl_no_language lang;
112 const enum mbfl_no_encoding *list;
113 size_t list_size;
114 } php_mb_nls_ident_list;
115
116 static const enum mbfl_no_encoding php_mb_default_identify_list_ja[] = {
117 mbfl_no_encoding_ascii,
118 mbfl_no_encoding_jis,
119 mbfl_no_encoding_utf8,
120 mbfl_no_encoding_euc_jp,
121 mbfl_no_encoding_sjis
122 };
123
124 static const enum mbfl_no_encoding php_mb_default_identify_list_cn[] = {
125 mbfl_no_encoding_ascii,
126 mbfl_no_encoding_utf8,
127 mbfl_no_encoding_euc_cn,
128 mbfl_no_encoding_cp936
129 };
130
131 static const enum mbfl_no_encoding php_mb_default_identify_list_tw_hk[] = {
132 mbfl_no_encoding_ascii,
133 mbfl_no_encoding_utf8,
134 mbfl_no_encoding_euc_tw,
135 mbfl_no_encoding_big5
136 };
137
138 static const enum mbfl_no_encoding php_mb_default_identify_list_kr[] = {
139 mbfl_no_encoding_ascii,
140 mbfl_no_encoding_utf8,
141 mbfl_no_encoding_euc_kr,
142 mbfl_no_encoding_uhc
143 };
144
145 static const enum mbfl_no_encoding php_mb_default_identify_list_ru[] = {
146 mbfl_no_encoding_ascii,
147 mbfl_no_encoding_utf8,
148 mbfl_no_encoding_koi8r,
149 mbfl_no_encoding_cp1251,
150 mbfl_no_encoding_cp866
151 };
152
153 static const enum mbfl_no_encoding php_mb_default_identify_list_hy[] = {
154 mbfl_no_encoding_ascii,
155 mbfl_no_encoding_utf8,
156 mbfl_no_encoding_armscii8
157 };
158
159 static const enum mbfl_no_encoding php_mb_default_identify_list_tr[] = {
160 mbfl_no_encoding_ascii,
161 mbfl_no_encoding_utf8,
162 mbfl_no_encoding_cp1254,
163 mbfl_no_encoding_8859_9
164 };
165
166 static const enum mbfl_no_encoding php_mb_default_identify_list_ua[] = {
167 mbfl_no_encoding_ascii,
168 mbfl_no_encoding_utf8,
169 mbfl_no_encoding_koi8u
170 };
171
172 static const enum mbfl_no_encoding php_mb_default_identify_list_neut[] = {
173 mbfl_no_encoding_ascii,
174 mbfl_no_encoding_utf8
175 };
176
177
178 static const php_mb_nls_ident_list php_mb_default_identify_list[] = {
179 { mbfl_no_language_japanese, php_mb_default_identify_list_ja, sizeof(php_mb_default_identify_list_ja) / sizeof(php_mb_default_identify_list_ja[0]) },
180 { mbfl_no_language_korean, php_mb_default_identify_list_kr, sizeof(php_mb_default_identify_list_kr) / sizeof(php_mb_default_identify_list_kr[0]) },
181 { mbfl_no_language_traditional_chinese, php_mb_default_identify_list_tw_hk, sizeof(php_mb_default_identify_list_tw_hk) / sizeof(php_mb_default_identify_list_tw_hk[0]) },
182 { mbfl_no_language_simplified_chinese, php_mb_default_identify_list_cn, sizeof(php_mb_default_identify_list_cn) / sizeof(php_mb_default_identify_list_cn[0]) },
183 { mbfl_no_language_russian, php_mb_default_identify_list_ru, sizeof(php_mb_default_identify_list_ru) / sizeof(php_mb_default_identify_list_ru[0]) },
184 { mbfl_no_language_armenian, php_mb_default_identify_list_hy, sizeof(php_mb_default_identify_list_hy) / sizeof(php_mb_default_identify_list_hy[0]) },
185 { mbfl_no_language_turkish, php_mb_default_identify_list_tr, sizeof(php_mb_default_identify_list_tr) / sizeof(php_mb_default_identify_list_tr[0]) },
186 { mbfl_no_language_ukrainian, php_mb_default_identify_list_ua, sizeof(php_mb_default_identify_list_ua) / sizeof(php_mb_default_identify_list_ua[0]) },
187 { mbfl_no_language_neutral, php_mb_default_identify_list_neut, sizeof(php_mb_default_identify_list_neut) / sizeof(php_mb_default_identify_list_neut[0]) }
188 };
189
190 /* }}} */
191
192 /* {{{ mb_overload_def mb_ovld[] */
193 static const struct mb_overload_def mb_ovld[] = {
194 {MB_OVERLOAD_MAIL, "mail", "mb_send_mail", "mb_orig_mail"},
195 {MB_OVERLOAD_STRING, "strlen", "mb_strlen", "mb_orig_strlen"},
196 {MB_OVERLOAD_STRING, "strpos", "mb_strpos", "mb_orig_strpos"},
197 {MB_OVERLOAD_STRING, "strrpos", "mb_strrpos", "mb_orig_strrpos"},
198 {MB_OVERLOAD_STRING, "stripos", "mb_stripos", "mb_orig_stripos"},
199 {MB_OVERLOAD_STRING, "strripos", "mb_strripos", "mb_orig_strripos"},
200 {MB_OVERLOAD_STRING, "strstr", "mb_strstr", "mb_orig_strstr"},
201 {MB_OVERLOAD_STRING, "strrchr", "mb_strrchr", "mb_orig_strrchr"},
202 {MB_OVERLOAD_STRING, "stristr", "mb_stristr", "mb_orig_stristr"},
203 {MB_OVERLOAD_STRING, "substr", "mb_substr", "mb_orig_substr"},
204 {MB_OVERLOAD_STRING, "strtolower", "mb_strtolower", "mb_orig_strtolower"},
205 {MB_OVERLOAD_STRING, "strtoupper", "mb_strtoupper", "mb_orig_strtoupper"},
206 {MB_OVERLOAD_STRING, "substr_count", "mb_substr_count", "mb_orig_substr_count"},
207 {0, NULL, NULL, NULL}
208 };
209 /* }}} */
210
211 /* {{{ arginfo */
212 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_language, 0, 0, 0)
213 ZEND_ARG_INFO(0, language)
214 ZEND_END_ARG_INFO()
215
216 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_internal_encoding, 0, 0, 0)
217 ZEND_ARG_INFO(0, encoding)
218 ZEND_END_ARG_INFO()
219
220 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_http_input, 0, 0, 0)
221 ZEND_ARG_INFO(0, type)
222 ZEND_END_ARG_INFO()
223
224 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_http_output, 0, 0, 0)
225 ZEND_ARG_INFO(0, encoding)
226 ZEND_END_ARG_INFO()
227
228 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_detect_order, 0, 0, 0)
229 ZEND_ARG_INFO(0, encoding)
230 ZEND_END_ARG_INFO()
231
232 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_substitute_character, 0, 0, 0)
233 ZEND_ARG_INFO(0, substchar)
234 ZEND_END_ARG_INFO()
235
236 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_preferred_mime_name, 0, 0, 1)
237 ZEND_ARG_INFO(0, encoding)
238 ZEND_END_ARG_INFO()
239
240 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_parse_str, 0, 0, 1)
241 ZEND_ARG_INFO(0, encoded_string)
242 ZEND_ARG_INFO(1, result)
243 ZEND_END_ARG_INFO()
244
245 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_output_handler, 0, 0, 2)
246 ZEND_ARG_INFO(0, contents)
247 ZEND_ARG_INFO(0, status)
248 ZEND_END_ARG_INFO()
249
250 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strlen, 0, 0, 1)
251 ZEND_ARG_INFO(0, str)
252 ZEND_ARG_INFO(0, encoding)
253 ZEND_END_ARG_INFO()
254
255 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strpos, 0, 0, 2)
256 ZEND_ARG_INFO(0, haystack)
257 ZEND_ARG_INFO(0, needle)
258 ZEND_ARG_INFO(0, offset)
259 ZEND_ARG_INFO(0, encoding)
260 ZEND_END_ARG_INFO()
261
262 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strrpos, 0, 0, 2)
263 ZEND_ARG_INFO(0, haystack)
264 ZEND_ARG_INFO(0, needle)
265 ZEND_ARG_INFO(0, offset)
266 ZEND_ARG_INFO(0, encoding)
267 ZEND_END_ARG_INFO()
268
269 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_stripos, 0, 0, 2)
270 ZEND_ARG_INFO(0, haystack)
271 ZEND_ARG_INFO(0, needle)
272 ZEND_ARG_INFO(0, offset)
273 ZEND_ARG_INFO(0, encoding)
274 ZEND_END_ARG_INFO()
275
276 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strripos, 0, 0, 2)
277 ZEND_ARG_INFO(0, haystack)
278 ZEND_ARG_INFO(0, needle)
279 ZEND_ARG_INFO(0, offset)
280 ZEND_ARG_INFO(0, encoding)
281 ZEND_END_ARG_INFO()
282
283 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strstr, 0, 0, 2)
284 ZEND_ARG_INFO(0, haystack)
285 ZEND_ARG_INFO(0, needle)
286 ZEND_ARG_INFO(0, part)
287 ZEND_ARG_INFO(0, encoding)
288 ZEND_END_ARG_INFO()
289
290 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strrchr, 0, 0, 2)
291 ZEND_ARG_INFO(0, haystack)
292 ZEND_ARG_INFO(0, needle)
293 ZEND_ARG_INFO(0, part)
294 ZEND_ARG_INFO(0, encoding)
295 ZEND_END_ARG_INFO()
296
297 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_stristr, 0, 0, 2)
298 ZEND_ARG_INFO(0, haystack)
299 ZEND_ARG_INFO(0, needle)
300 ZEND_ARG_INFO(0, part)
301 ZEND_ARG_INFO(0, encoding)
302 ZEND_END_ARG_INFO()
303
304 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strrichr, 0, 0, 2)
305 ZEND_ARG_INFO(0, haystack)
306 ZEND_ARG_INFO(0, needle)
307 ZEND_ARG_INFO(0, part)
308 ZEND_ARG_INFO(0, encoding)
309 ZEND_END_ARG_INFO()
310
311 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_substr_count, 0, 0, 2)
312 ZEND_ARG_INFO(0, haystack)
313 ZEND_ARG_INFO(0, needle)
314 ZEND_ARG_INFO(0, encoding)
315 ZEND_END_ARG_INFO()
316
317 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_substr, 0, 0, 2)
318 ZEND_ARG_INFO(0, str)
319 ZEND_ARG_INFO(0, start)
320 ZEND_ARG_INFO(0, length)
321 ZEND_ARG_INFO(0, encoding)
322 ZEND_END_ARG_INFO()
323
324 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strcut, 0, 0, 2)
325 ZEND_ARG_INFO(0, str)
326 ZEND_ARG_INFO(0, start)
327 ZEND_ARG_INFO(0, length)
328 ZEND_ARG_INFO(0, encoding)
329 ZEND_END_ARG_INFO()
330
331 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strwidth, 0, 0, 1)
332 ZEND_ARG_INFO(0, str)
333 ZEND_ARG_INFO(0, encoding)
334 ZEND_END_ARG_INFO()
335
336 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strimwidth, 0, 0, 3)
337 ZEND_ARG_INFO(0, str)
338 ZEND_ARG_INFO(0, start)
339 ZEND_ARG_INFO(0, width)
340 ZEND_ARG_INFO(0, trimmarker)
341 ZEND_ARG_INFO(0, encoding)
342 ZEND_END_ARG_INFO()
343
344 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_convert_encoding, 0, 0, 2)
345 ZEND_ARG_INFO(0, str)
346 ZEND_ARG_INFO(0, to)
347 ZEND_ARG_INFO(0, from)
348 ZEND_END_ARG_INFO()
349
350 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_convert_case, 0, 0, 2)
351 ZEND_ARG_INFO(0, sourcestring)
352 ZEND_ARG_INFO(0, mode)
353 ZEND_ARG_INFO(0, encoding)
354 ZEND_END_ARG_INFO()
355
356 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strtoupper, 0, 0, 1)
357 ZEND_ARG_INFO(0, sourcestring)
358 ZEND_ARG_INFO(0, encoding)
359 ZEND_END_ARG_INFO()
360
361 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strtolower, 0, 0, 1)
362 ZEND_ARG_INFO(0, sourcestring)
363 ZEND_ARG_INFO(0, encoding)
364 ZEND_END_ARG_INFO()
365
366 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_detect_encoding, 0, 0, 1)
367 ZEND_ARG_INFO(0, str)
368 ZEND_ARG_INFO(0, encoding_list)
369 ZEND_ARG_INFO(0, strict)
370 ZEND_END_ARG_INFO()
371
372 ZEND_BEGIN_ARG_INFO(arginfo_mb_list_encodings, 0)
373 ZEND_END_ARG_INFO()
374
375 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_encoding_aliases, 0, 0, 1)
376 ZEND_ARG_INFO(0, encoding)
377 ZEND_END_ARG_INFO()
378
379 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_encode_mimeheader, 0, 0, 1)
380 ZEND_ARG_INFO(0, str)
381 ZEND_ARG_INFO(0, charset)
382 ZEND_ARG_INFO(0, transfer)
383 ZEND_ARG_INFO(0, linefeed)
384 ZEND_ARG_INFO(0, indent)
385 ZEND_END_ARG_INFO()
386
387 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_decode_mimeheader, 0, 0, 1)
388 ZEND_ARG_INFO(0, string)
389 ZEND_END_ARG_INFO()
390
391 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_convert_kana, 0, 0, 1)
392 ZEND_ARG_INFO(0, str)
393 ZEND_ARG_INFO(0, option)
394 ZEND_ARG_INFO(0, encoding)
395 ZEND_END_ARG_INFO()
396
397 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_convert_variables, 0, 0, 3)
398 ZEND_ARG_INFO(0, to)
399 ZEND_ARG_INFO(0, from)
400 ZEND_ARG_VARIADIC_INFO(1, vars)
401 ZEND_END_ARG_INFO()
402
403 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_encode_numericentity, 0, 0, 2)
404 ZEND_ARG_INFO(0, string)
405 ZEND_ARG_INFO(0, convmap)
406 ZEND_ARG_INFO(0, encoding)
407 ZEND_ARG_INFO(0, is_hex)
408 ZEND_END_ARG_INFO()
409
410 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_decode_numericentity, 0, 0, 2)
411 ZEND_ARG_INFO(0, string)
412 ZEND_ARG_INFO(0, convmap)
413 ZEND_ARG_INFO(0, encoding)
414 ZEND_END_ARG_INFO()
415
416 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_send_mail, 0, 0, 3)
417 ZEND_ARG_INFO(0, to)
418 ZEND_ARG_INFO(0, subject)
419 ZEND_ARG_INFO(0, message)
420 ZEND_ARG_INFO(0, additional_headers)
421 ZEND_ARG_INFO(0, additional_parameters)
422 ZEND_END_ARG_INFO()
423
424 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_get_info, 0, 0, 0)
425 ZEND_ARG_INFO(0, type)
426 ZEND_END_ARG_INFO()
427
428 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_check_encoding, 0, 0, 0)
429 ZEND_ARG_INFO(0, var)
430 ZEND_ARG_INFO(0, encoding)
431 ZEND_END_ARG_INFO()
432
433 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_regex_encoding, 0, 0, 0)
434 ZEND_ARG_INFO(0, encoding)
435 ZEND_END_ARG_INFO()
436
437 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg, 0, 0, 2)
438 ZEND_ARG_INFO(0, pattern)
439 ZEND_ARG_INFO(0, string)
440 ZEND_ARG_INFO(1, registers)
441 ZEND_END_ARG_INFO()
442
443 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_eregi, 0, 0, 2)
444 ZEND_ARG_INFO(0, pattern)
445 ZEND_ARG_INFO(0, string)
446 ZEND_ARG_INFO(1, registers)
447 ZEND_END_ARG_INFO()
448
449 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_replace, 0, 0, 3)
450 ZEND_ARG_INFO(0, pattern)
451 ZEND_ARG_INFO(0, replacement)
452 ZEND_ARG_INFO(0, string)
453 ZEND_ARG_INFO(0, option)
454 ZEND_END_ARG_INFO()
455
456 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_eregi_replace, 0, 0, 3)
457 ZEND_ARG_INFO(0, pattern)
458 ZEND_ARG_INFO(0, replacement)
459 ZEND_ARG_INFO(0, string)
460 ZEND_ARG_INFO(0, option)
461 ZEND_END_ARG_INFO()
462
463 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_replace_callback, 0, 0, 3)
464 ZEND_ARG_INFO(0, pattern)
465 ZEND_ARG_INFO(0, callback)
466 ZEND_ARG_INFO(0, string)
467 ZEND_ARG_INFO(0, option)
468 ZEND_END_ARG_INFO()
469
470 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_split, 0, 0, 2)
471 ZEND_ARG_INFO(0, pattern)
472 ZEND_ARG_INFO(0, string)
473 ZEND_ARG_INFO(0, limit)
474 ZEND_END_ARG_INFO()
475
476 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_match, 0, 0, 2)
477 ZEND_ARG_INFO(0, pattern)
478 ZEND_ARG_INFO(0, string)
479 ZEND_ARG_INFO(0, option)
480 ZEND_END_ARG_INFO()
481
482 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search, 0, 0, 0)
483 ZEND_ARG_INFO(0, pattern)
484 ZEND_ARG_INFO(0, option)
485 ZEND_END_ARG_INFO()
486
487 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search_pos, 0, 0, 0)
488 ZEND_ARG_INFO(0, pattern)
489 ZEND_ARG_INFO(0, option)
490 ZEND_END_ARG_INFO()
491
492 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search_regs, 0, 0, 0)
493 ZEND_ARG_INFO(0, pattern)
494 ZEND_ARG_INFO(0, option)
495 ZEND_END_ARG_INFO()
496
497 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search_init, 0, 0, 1)
498 ZEND_ARG_INFO(0, string)
499 ZEND_ARG_INFO(0, pattern)
500 ZEND_ARG_INFO(0, option)
501 ZEND_END_ARG_INFO()
502
503 ZEND_BEGIN_ARG_INFO(arginfo_mb_ereg_search_getregs, 0)
504 ZEND_END_ARG_INFO()
505
506 ZEND_BEGIN_ARG_INFO(arginfo_mb_ereg_search_getpos, 0)
507 ZEND_END_ARG_INFO()
508
509 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search_setpos, 0, 0, 1)
510 ZEND_ARG_INFO(0, position)
511 ZEND_END_ARG_INFO()
512
513 ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_regex_set_options, 0, 0, 0)
514 ZEND_ARG_INFO(0, options)
515 ZEND_END_ARG_INFO()
516 /* }}} */
517
518 /* {{{ zend_function_entry mbstring_functions[] */
519 const zend_function_entry mbstring_functions[] = {
520 PHP_FE(mb_convert_case, arginfo_mb_convert_case)
521 PHP_FE(mb_strtoupper, arginfo_mb_strtoupper)
522 PHP_FE(mb_strtolower, arginfo_mb_strtolower)
523 PHP_FE(mb_language, arginfo_mb_language)
524 PHP_FE(mb_internal_encoding, arginfo_mb_internal_encoding)
525 PHP_FE(mb_http_input, arginfo_mb_http_input)
526 PHP_FE(mb_http_output, arginfo_mb_http_output)
527 PHP_FE(mb_detect_order, arginfo_mb_detect_order)
528 PHP_FE(mb_substitute_character, arginfo_mb_substitute_character)
529 PHP_FE(mb_parse_str, arginfo_mb_parse_str)
530 PHP_FE(mb_output_handler, arginfo_mb_output_handler)
531 PHP_FE(mb_preferred_mime_name, arginfo_mb_preferred_mime_name)
532 PHP_FE(mb_strlen, arginfo_mb_strlen)
533 PHP_FE(mb_strpos, arginfo_mb_strpos)
534 PHP_FE(mb_strrpos, arginfo_mb_strrpos)
535 PHP_FE(mb_stripos, arginfo_mb_stripos)
536 PHP_FE(mb_strripos, arginfo_mb_strripos)
537 PHP_FE(mb_strstr, arginfo_mb_strstr)
538 PHP_FE(mb_strrchr, arginfo_mb_strrchr)
539 PHP_FE(mb_stristr, arginfo_mb_stristr)
540 PHP_FE(mb_strrichr, arginfo_mb_strrichr)
541 PHP_FE(mb_substr_count, arginfo_mb_substr_count)
542 PHP_FE(mb_substr, arginfo_mb_substr)
543 PHP_FE(mb_strcut, arginfo_mb_strcut)
544 PHP_FE(mb_strwidth, arginfo_mb_strwidth)
545 PHP_FE(mb_strimwidth, arginfo_mb_strimwidth)
546 PHP_FE(mb_convert_encoding, arginfo_mb_convert_encoding)
547 PHP_FE(mb_detect_encoding, arginfo_mb_detect_encoding)
548 PHP_FE(mb_list_encodings, arginfo_mb_list_encodings)
549 PHP_FE(mb_encoding_aliases, arginfo_mb_encoding_aliases)
550 PHP_FE(mb_convert_kana, arginfo_mb_convert_kana)
551 PHP_FE(mb_encode_mimeheader, arginfo_mb_encode_mimeheader)
552 PHP_FE(mb_decode_mimeheader, arginfo_mb_decode_mimeheader)
553 PHP_FE(mb_convert_variables, arginfo_mb_convert_variables)
554 PHP_FE(mb_encode_numericentity, arginfo_mb_encode_numericentity)
555 PHP_FE(mb_decode_numericentity, arginfo_mb_decode_numericentity)
556 PHP_FE(mb_send_mail, arginfo_mb_send_mail)
557 PHP_FE(mb_get_info, arginfo_mb_get_info)
558 PHP_FE(mb_check_encoding, arginfo_mb_check_encoding)
559 #if HAVE_MBREGEX
560 PHP_MBREGEX_FUNCTION_ENTRIES
561 #endif
562 PHP_FE_END
563 };
564 /* }}} */
565
566 /* {{{ zend_module_entry mbstring_module_entry */
567 zend_module_entry mbstring_module_entry = {
568 STANDARD_MODULE_HEADER,
569 "mbstring",
570 mbstring_functions,
571 PHP_MINIT(mbstring),
572 PHP_MSHUTDOWN(mbstring),
573 PHP_RINIT(mbstring),
574 PHP_RSHUTDOWN(mbstring),
575 PHP_MINFO(mbstring),
576 PHP_MBSTRING_VERSION,
577 PHP_MODULE_GLOBALS(mbstring),
578 PHP_GINIT(mbstring),
579 PHP_GSHUTDOWN(mbstring),
580 NULL,
581 STANDARD_MODULE_PROPERTIES_EX
582 };
583 /* }}} */
584
585 /* {{{ static sapi_post_entry php_post_entries[] */
586 static sapi_post_entry php_post_entries[] = {
587 { DEFAULT_POST_CONTENT_TYPE, sizeof(DEFAULT_POST_CONTENT_TYPE)-1, sapi_read_standard_form_data, php_std_post_handler },
588 { MULTIPART_CONTENT_TYPE, sizeof(MULTIPART_CONTENT_TYPE)-1, NULL, rfc1867_post_handler },
589 { NULL, 0, NULL, NULL }
590 };
591 /* }}} */
592
593 #ifdef COMPILE_DL_MBSTRING
594 #ifdef ZTS
595 ZEND_TSRMLS_CACHE_DEFINE()
596 #endif
ZEND_GET_MODULE(mbstring)597 ZEND_GET_MODULE(mbstring)
598 #endif
599
600 static char *get_internal_encoding(void) {
601 if (PG(internal_encoding) && PG(internal_encoding)[0]) {
602 return PG(internal_encoding);
603 } else if (SG(default_charset)) {
604 return SG(default_charset);
605 }
606 return "";
607 }
608
get_input_encoding(void)609 static char *get_input_encoding(void) {
610 if (PG(input_encoding) && PG(input_encoding)[0]) {
611 return PG(input_encoding);
612 } else if (SG(default_charset)) {
613 return SG(default_charset);
614 }
615 return "";
616 }
617
get_output_encoding(void)618 static char *get_output_encoding(void) {
619 if (PG(output_encoding) && PG(output_encoding)[0]) {
620 return PG(output_encoding);
621 } else if (SG(default_charset)) {
622 return SG(default_charset);
623 }
624 return "";
625 }
626
627
628 /* {{{ allocators */
_php_mb_allocators_malloc(unsigned int sz)629 static void *_php_mb_allocators_malloc(unsigned int sz)
630 {
631 return emalloc(sz);
632 }
633
_php_mb_allocators_realloc(void * ptr,unsigned int sz)634 static void *_php_mb_allocators_realloc(void *ptr, unsigned int sz)
635 {
636 return erealloc(ptr, sz);
637 }
638
_php_mb_allocators_calloc(unsigned int nelems,unsigned int szelem)639 static void *_php_mb_allocators_calloc(unsigned int nelems, unsigned int szelem)
640 {
641 return ecalloc(nelems, szelem);
642 }
643
_php_mb_allocators_free(void * ptr)644 static void _php_mb_allocators_free(void *ptr)
645 {
646 efree(ptr);
647 }
648
_php_mb_allocators_pmalloc(unsigned int sz)649 static void *_php_mb_allocators_pmalloc(unsigned int sz)
650 {
651 return pemalloc(sz, 1);
652 }
653
_php_mb_allocators_prealloc(void * ptr,unsigned int sz)654 static void *_php_mb_allocators_prealloc(void *ptr, unsigned int sz)
655 {
656 return perealloc(ptr, sz, 1);
657 }
658
_php_mb_allocators_pfree(void * ptr)659 static void _php_mb_allocators_pfree(void *ptr)
660 {
661 pefree(ptr, 1);
662 }
663
664 static mbfl_allocators _php_mb_allocators = {
665 _php_mb_allocators_malloc,
666 _php_mb_allocators_realloc,
667 _php_mb_allocators_calloc,
668 _php_mb_allocators_free,
669 _php_mb_allocators_pmalloc,
670 _php_mb_allocators_prealloc,
671 _php_mb_allocators_pfree
672 };
673 /* }}} */
674
675 /* {{{ static sapi_post_entry mbstr_post_entries[] */
676 static sapi_post_entry mbstr_post_entries[] = {
677 { DEFAULT_POST_CONTENT_TYPE, sizeof(DEFAULT_POST_CONTENT_TYPE)-1, sapi_read_standard_form_data, php_mb_post_handler },
678 { MULTIPART_CONTENT_TYPE, sizeof(MULTIPART_CONTENT_TYPE)-1, NULL, rfc1867_post_handler },
679 { NULL, 0, NULL, NULL }
680 };
681 /* }}} */
682
683 /* {{{ static int php_mb_parse_encoding_list()
684 * Return FAILURE if input contains any illegal encoding, otherwise SUCCESS.
685 * Even if any illegal encoding is detected the result may contain a list
686 * of parsed encodings.
687 */
688 static int
php_mb_parse_encoding_list(const char * value,size_t value_length,const mbfl_encoding *** return_list,size_t * return_size,int persistent)689 php_mb_parse_encoding_list(const char *value, size_t value_length, const mbfl_encoding ***return_list, size_t *return_size, int persistent)
690 {
691 int bauto, ret = SUCCESS;
692 size_t n, size;
693 char *p, *p1, *p2, *endp, *tmpstr;
694 const mbfl_encoding **entry, **list;
695
696 list = NULL;
697 if (value == NULL || value_length <= 0) {
698 if (return_list) {
699 *return_list = NULL;
700 }
701 if (return_size) {
702 *return_size = 0;
703 }
704 return FAILURE;
705 } else {
706 /* copy the value string for work */
707 if (value[0]=='"' && value[value_length-1]=='"' && value_length>2) {
708 tmpstr = (char *)estrndup(value+1, value_length-2);
709 value_length -= 2;
710 }
711 else
712 tmpstr = (char *)estrndup(value, value_length);
713 if (tmpstr == NULL) {
714 return FAILURE;
715 }
716 /* count the number of listed encoding names */
717 endp = tmpstr + value_length;
718 n = 1;
719 p1 = tmpstr;
720 while ((p2 = (char*)php_memnstr(p1, ",", 1, endp)) != NULL) {
721 p1 = p2 + 1;
722 n++;
723 }
724 size = n + MBSTRG(default_detect_order_list_size);
725 /* make list */
726 list = (const mbfl_encoding **)pecalloc(size, sizeof(mbfl_encoding*), persistent);
727 if (list != NULL) {
728 entry = list;
729 n = 0;
730 bauto = 0;
731 p1 = tmpstr;
732 do {
733 p2 = p = (char*)php_memnstr(p1, ",", 1, endp);
734 if (p == NULL) {
735 p = endp;
736 }
737 *p = '\0';
738 /* trim spaces */
739 while (p1 < p && (*p1 == ' ' || *p1 == '\t')) {
740 p1++;
741 }
742 p--;
743 while (p > p1 && (*p == ' ' || *p == '\t')) {
744 *p = '\0';
745 p--;
746 }
747 /* convert to the encoding number and check encoding */
748 if (strcasecmp(p1, "auto") == 0) {
749 if (!bauto) {
750 const enum mbfl_no_encoding *src = MBSTRG(default_detect_order_list);
751 const size_t identify_list_size = MBSTRG(default_detect_order_list_size);
752 size_t i;
753 bauto = 1;
754 for (i = 0; i < identify_list_size; i++) {
755 *entry++ = mbfl_no2encoding(*src++);
756 n++;
757 }
758 }
759 } else {
760 const mbfl_encoding *encoding = mbfl_name2encoding(p1);
761 if (encoding) {
762 *entry++ = encoding;
763 n++;
764 } else {
765 ret = FAILURE;
766 }
767 }
768 p1 = p2 + 1;
769 } while (n < size && p2 != NULL);
770 if (n > 0) {
771 if (return_list) {
772 *return_list = list;
773 } else {
774 pefree(list, persistent);
775 }
776 } else {
777 pefree(list, persistent);
778 if (return_list) {
779 *return_list = NULL;
780 }
781 ret = FAILURE;
782 }
783 if (return_size) {
784 *return_size = n;
785 }
786 } else {
787 if (return_list) {
788 *return_list = NULL;
789 }
790 if (return_size) {
791 *return_size = 0;
792 }
793 ret = FAILURE;
794 }
795 efree(tmpstr);
796 }
797
798 return ret;
799 }
800 /* }}} */
801
802 /* {{{ static int php_mb_parse_encoding_array()
803 * Return FAILURE if input contains any illegal encoding, otherwise SUCCESS.
804 * Even if any illegal encoding is detected the result may contain a list
805 * of parsed encodings.
806 */
807 static int
php_mb_parse_encoding_array(zval * array,const mbfl_encoding *** return_list,size_t * return_size,int persistent)808 php_mb_parse_encoding_array(zval *array, const mbfl_encoding ***return_list, size_t *return_size, int persistent)
809 {
810 zval *hash_entry;
811 HashTable *target_hash;
812 int i, n, size, bauto, ret = SUCCESS;
813 const mbfl_encoding **list, **entry;
814
815 list = NULL;
816 if (Z_TYPE_P(array) == IS_ARRAY) {
817 target_hash = Z_ARRVAL_P(array);
818 i = zend_hash_num_elements(target_hash);
819 size = i + MBSTRG(default_detect_order_list_size);
820 list = (const mbfl_encoding **)pecalloc(size, sizeof(mbfl_encoding*), persistent);
821 if (list != NULL) {
822 entry = list;
823 bauto = 0;
824 n = 0;
825 ZEND_HASH_FOREACH_VAL(target_hash, hash_entry) {
826 convert_to_string_ex(hash_entry);
827 if (strcasecmp(Z_STRVAL_P(hash_entry), "auto") == 0) {
828 if (!bauto) {
829 const enum mbfl_no_encoding *src = MBSTRG(default_detect_order_list);
830 const size_t identify_list_size = MBSTRG(default_detect_order_list_size);
831 size_t j;
832
833 bauto = 1;
834 for (j = 0; j < identify_list_size; j++) {
835 *entry++ = mbfl_no2encoding(*src++);
836 n++;
837 }
838 }
839 } else {
840 const mbfl_encoding *encoding = mbfl_name2encoding(Z_STRVAL_P(hash_entry));
841 if (encoding) {
842 *entry++ = encoding;
843 n++;
844 } else {
845 ret = FAILURE;
846 }
847 }
848 i--;
849 } ZEND_HASH_FOREACH_END();
850 if (n > 0) {
851 if (return_list) {
852 *return_list = list;
853 } else {
854 pefree(list, persistent);
855 }
856 } else {
857 pefree(list, persistent);
858 if (return_list) {
859 *return_list = NULL;
860 }
861 ret = FAILURE;
862 }
863 if (return_size) {
864 *return_size = n;
865 }
866 } else {
867 if (return_list) {
868 *return_list = NULL;
869 }
870 if (return_size) {
871 *return_size = 0;
872 }
873 ret = FAILURE;
874 }
875 }
876
877 return ret;
878 }
879 /* }}} */
880
881 /* {{{ zend_multibyte interface */
php_mb_zend_encoding_fetcher(const char * encoding_name)882 static const zend_encoding* php_mb_zend_encoding_fetcher(const char *encoding_name)
883 {
884 return (const zend_encoding*)mbfl_name2encoding(encoding_name);
885 }
886
php_mb_zend_encoding_name_getter(const zend_encoding * encoding)887 static const char *php_mb_zend_encoding_name_getter(const zend_encoding *encoding)
888 {
889 return ((const mbfl_encoding *)encoding)->name;
890 }
891
php_mb_zend_encoding_lexer_compatibility_checker(const zend_encoding * _encoding)892 static int php_mb_zend_encoding_lexer_compatibility_checker(const zend_encoding *_encoding)
893 {
894 const mbfl_encoding *encoding = (const mbfl_encoding*)_encoding;
895 if (encoding->flag & MBFL_ENCTYPE_SBCS) {
896 return 1;
897 }
898 if ((encoding->flag & (MBFL_ENCTYPE_MBCS | MBFL_ENCTYPE_GL_UNSAFE)) == MBFL_ENCTYPE_MBCS) {
899 return 1;
900 }
901 return 0;
902 }
903
php_mb_zend_encoding_detector(const unsigned char * arg_string,size_t arg_length,const zend_encoding ** list,size_t list_size)904 static const zend_encoding *php_mb_zend_encoding_detector(const unsigned char *arg_string, size_t arg_length, const zend_encoding **list, size_t list_size)
905 {
906 mbfl_string string;
907
908 if (!list) {
909 list = (const zend_encoding **)MBSTRG(current_detect_order_list);
910 list_size = MBSTRG(current_detect_order_list_size);
911 }
912
913 mbfl_string_init(&string);
914 string.no_language = MBSTRG(language);
915 string.val = (unsigned char *)arg_string;
916 string.len = arg_length;
917 return (const zend_encoding *) mbfl_identify_encoding2(&string, (const mbfl_encoding **)list, list_size, 0);
918 }
919
php_mb_zend_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)920 static size_t php_mb_zend_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)
921 {
922 mbfl_string string, result;
923 mbfl_buffer_converter *convd;
924 int status, loc;
925
926 /* new encoding */
927 /* initialize string */
928 mbfl_string_init(&string);
929 mbfl_string_init(&result);
930 string.no_encoding = ((const mbfl_encoding*)encoding_from)->no_encoding;
931 string.no_language = MBSTRG(language);
932 string.val = (unsigned char*)from;
933 string.len = from_length;
934
935 /* initialize converter */
936 convd = mbfl_buffer_converter_new2((const mbfl_encoding *)encoding_from, (const mbfl_encoding *)encoding_to, string.len);
937 if (convd == NULL) {
938 return -1;
939 }
940 mbfl_buffer_converter_illegal_mode(convd, MBSTRG(current_filter_illegal_mode));
941 mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
942
943 /* do it */
944 status = mbfl_buffer_converter_feed2(convd, &string, &loc);
945 if (status) {
946 mbfl_buffer_converter_delete(convd);
947 return (size_t)-1;
948 }
949
950 mbfl_buffer_converter_flush(convd);
951 if (!mbfl_buffer_converter_result(convd, &result)) {
952 mbfl_buffer_converter_delete(convd);
953 return (size_t)-1;
954 }
955
956 *to = result.val;
957 *to_length = result.len;
958
959 mbfl_buffer_converter_delete(convd);
960
961 return loc;
962 }
963
php_mb_zend_encoding_list_parser(const char * encoding_list,size_t encoding_list_len,const zend_encoding *** return_list,size_t * return_size,int persistent)964 static int php_mb_zend_encoding_list_parser(const char *encoding_list, size_t encoding_list_len, const zend_encoding ***return_list, size_t *return_size, int persistent)
965 {
966 return php_mb_parse_encoding_list(encoding_list, encoding_list_len, (const mbfl_encoding ***)return_list, return_size, persistent);
967 }
968
php_mb_zend_internal_encoding_getter(void)969 static const zend_encoding *php_mb_zend_internal_encoding_getter(void)
970 {
971 return (const zend_encoding *)MBSTRG(internal_encoding);
972 }
973
php_mb_zend_internal_encoding_setter(const zend_encoding * encoding)974 static int php_mb_zend_internal_encoding_setter(const zend_encoding *encoding)
975 {
976 MBSTRG(internal_encoding) = (const mbfl_encoding *)encoding;
977 return SUCCESS;
978 }
979
980 static zend_multibyte_functions php_mb_zend_multibyte_functions = {
981 "mbstring",
982 php_mb_zend_encoding_fetcher,
983 php_mb_zend_encoding_name_getter,
984 php_mb_zend_encoding_lexer_compatibility_checker,
985 php_mb_zend_encoding_detector,
986 php_mb_zend_encoding_converter,
987 php_mb_zend_encoding_list_parser,
988 php_mb_zend_internal_encoding_getter,
989 php_mb_zend_internal_encoding_setter
990 };
991 /* }}} */
992
993 static void *_php_mb_compile_regex(const char *pattern);
994 static int _php_mb_match_regex(void *opaque, const char *str, size_t str_len);
995 static void _php_mb_free_regex(void *opaque);
996
997 #if HAVE_ONIG
998 /* {{{ _php_mb_compile_regex */
_php_mb_compile_regex(const char * pattern)999 static void *_php_mb_compile_regex(const char *pattern)
1000 {
1001 php_mb_regex_t *retval;
1002 OnigErrorInfo err_info;
1003 int err_code;
1004
1005 if ((err_code = onig_new(&retval,
1006 (const OnigUChar *)pattern,
1007 (const OnigUChar *)pattern + strlen(pattern),
1008 ONIG_OPTION_IGNORECASE | ONIG_OPTION_DONT_CAPTURE_GROUP,
1009 ONIG_ENCODING_ASCII, &OnigSyntaxPerl, &err_info))) {
1010 OnigUChar err_str[ONIG_MAX_ERROR_MESSAGE_LEN];
1011 onig_error_code_to_str(err_str, err_code, err_info);
1012 php_error_docref(NULL, E_WARNING, "%s: %s", pattern, err_str);
1013 retval = NULL;
1014 }
1015 return retval;
1016 }
1017 /* }}} */
1018
1019 /* {{{ _php_mb_match_regex */
_php_mb_match_regex(void * opaque,const char * str,size_t str_len)1020 static int _php_mb_match_regex(void *opaque, const char *str, size_t str_len)
1021 {
1022 return onig_search((php_mb_regex_t *)opaque, (const OnigUChar *)str,
1023 (const OnigUChar*)str + str_len, (const OnigUChar *)str,
1024 (const OnigUChar*)str + str_len, NULL, ONIG_OPTION_NONE) >= 0;
1025 }
1026 /* }}} */
1027
1028 /* {{{ _php_mb_free_regex */
_php_mb_free_regex(void * opaque)1029 static void _php_mb_free_regex(void *opaque)
1030 {
1031 onig_free((php_mb_regex_t *)opaque);
1032 }
1033 /* }}} */
1034 #elif HAVE_PCRE || HAVE_BUNDLED_PCRE
1035 /* {{{ _php_mb_compile_regex */
_php_mb_compile_regex(const char * pattern)1036 static void *_php_mb_compile_regex(const char *pattern)
1037 {
1038 pcre *retval;
1039 const char *err_str;
1040 int err_offset;
1041
1042 if (!(retval = pcre_compile(pattern,
1043 PCRE_CASELESS, &err_str, &err_offset, NULL))) {
1044 php_error_docref(NULL, E_WARNING, "%s (offset=%d): %s", pattern, err_offset, err_str);
1045 }
1046 return retval;
1047 }
1048 /* }}} */
1049
1050 /* {{{ _php_mb_match_regex */
_php_mb_match_regex(void * opaque,const char * str,size_t str_len)1051 static int _php_mb_match_regex(void *opaque, const char *str, size_t str_len)
1052 {
1053 return pcre_exec((pcre *)opaque, NULL, str, (int)str_len, 0,
1054 0, NULL, 0) >= 0;
1055 }
1056 /* }}} */
1057
1058 /* {{{ _php_mb_free_regex */
_php_mb_free_regex(void * opaque)1059 static void _php_mb_free_regex(void *opaque)
1060 {
1061 pcre_free(opaque);
1062 }
1063 /* }}} */
1064 #endif
1065
1066 /* {{{ php_mb_nls_get_default_detect_order_list */
php_mb_nls_get_default_detect_order_list(enum mbfl_no_language lang,enum mbfl_no_encoding ** plist,size_t * plist_size)1067 static int php_mb_nls_get_default_detect_order_list(enum mbfl_no_language lang, enum mbfl_no_encoding **plist, size_t *plist_size)
1068 {
1069 size_t i;
1070
1071 *plist = (enum mbfl_no_encoding *) php_mb_default_identify_list_neut;
1072 *plist_size = sizeof(php_mb_default_identify_list_neut) / sizeof(php_mb_default_identify_list_neut[0]);
1073
1074 for (i = 0; i < sizeof(php_mb_default_identify_list) / sizeof(php_mb_default_identify_list[0]); i++) {
1075 if (php_mb_default_identify_list[i].lang == lang) {
1076 *plist = (enum mbfl_no_encoding *)php_mb_default_identify_list[i].list;
1077 *plist_size = php_mb_default_identify_list[i].list_size;
1078 return 1;
1079 }
1080 }
1081 return 0;
1082 }
1083 /* }}} */
1084
php_mb_rfc1867_substring_conf(const zend_encoding * encoding,char * start,int len,char quote)1085 static char *php_mb_rfc1867_substring_conf(const zend_encoding *encoding, char *start, int len, char quote)
1086 {
1087 char *result = emalloc(len + 2);
1088 char *resp = result;
1089 int i;
1090
1091 for (i = 0; i < len && start[i] != quote; ++i) {
1092 if (start[i] == '\\' && (start[i + 1] == '\\' || (quote && start[i + 1] == quote))) {
1093 *resp++ = start[++i];
1094 } else {
1095 size_t j = php_mb_mbchar_bytes_ex(start+i, (const mbfl_encoding *)encoding);
1096
1097 while (j-- > 0 && i < len) {
1098 *resp++ = start[i++];
1099 }
1100 --i;
1101 }
1102 }
1103
1104 *resp = '\0';
1105 return result;
1106 }
1107
php_mb_rfc1867_getword(const zend_encoding * encoding,char ** line,char stop)1108 static char *php_mb_rfc1867_getword(const zend_encoding *encoding, char **line, char stop) /* {{{ */
1109 {
1110 char *pos = *line, quote;
1111 char *res;
1112
1113 while (*pos && *pos != stop) {
1114 if ((quote = *pos) == '"' || quote == '\'') {
1115 ++pos;
1116 while (*pos && *pos != quote) {
1117 if (*pos == '\\' && pos[1] && pos[1] == quote) {
1118 pos += 2;
1119 } else {
1120 ++pos;
1121 }
1122 }
1123 if (*pos) {
1124 ++pos;
1125 }
1126 } else {
1127 pos += php_mb_mbchar_bytes_ex(pos, (const mbfl_encoding *)encoding);
1128
1129 }
1130 }
1131 if (*pos == '\0') {
1132 res = estrdup(*line);
1133 *line += strlen(*line);
1134 return res;
1135 }
1136
1137 res = estrndup(*line, pos - *line);
1138
1139 while (*pos == stop) {
1140 pos += php_mb_mbchar_bytes_ex(pos, (const mbfl_encoding *)encoding);
1141 }
1142
1143 *line = pos;
1144 return res;
1145 }
1146 /* }}} */
1147
php_mb_rfc1867_getword_conf(const zend_encoding * encoding,char * str)1148 static char *php_mb_rfc1867_getword_conf(const zend_encoding *encoding, char *str) /* {{{ */
1149 {
1150 while (*str && isspace(*(unsigned char *)str)) {
1151 ++str;
1152 }
1153
1154 if (!*str) {
1155 return estrdup("");
1156 }
1157
1158 if (*str == '"' || *str == '\'') {
1159 char quote = *str;
1160
1161 str++;
1162 return php_mb_rfc1867_substring_conf(encoding, str, strlen(str), quote);
1163 } else {
1164 char *strend = str;
1165
1166 while (*strend && !isspace(*(unsigned char *)strend)) {
1167 ++strend;
1168 }
1169 return php_mb_rfc1867_substring_conf(encoding, str, strend - str, 0);
1170 }
1171 }
1172 /* }}} */
1173
php_mb_rfc1867_basename(const zend_encoding * encoding,char * filename)1174 static char *php_mb_rfc1867_basename(const zend_encoding *encoding, char *filename) /* {{{ */
1175 {
1176 char *s, *s2;
1177 const size_t filename_len = strlen(filename);
1178
1179 /* The \ check should technically be needed for win32 systems only where
1180 * it is a valid path separator. However, IE in all it's wisdom always sends
1181 * the full path of the file on the user's filesystem, which means that unless
1182 * the user does basename() they get a bogus file name. Until IE's user base drops
1183 * to nill or problem is fixed this code must remain enabled for all systems. */
1184 s = php_mb_safe_strrchr_ex(filename, '\\', filename_len, (const mbfl_encoding *)encoding);
1185 s2 = php_mb_safe_strrchr_ex(filename, '/', filename_len, (const mbfl_encoding *)encoding);
1186
1187 if (s && s2) {
1188 if (s > s2) {
1189 return ++s;
1190 } else {
1191 return ++s2;
1192 }
1193 } else if (s) {
1194 return ++s;
1195 } else if (s2) {
1196 return ++s2;
1197 } else {
1198 return filename;
1199 }
1200 }
1201 /* }}} */
1202
1203 /* {{{ php.ini directive handler */
1204 /* {{{ static PHP_INI_MH(OnUpdate_mbstring_language) */
PHP_INI_MH(OnUpdate_mbstring_language)1205 static PHP_INI_MH(OnUpdate_mbstring_language)
1206 {
1207 enum mbfl_no_language no_language;
1208
1209 no_language = mbfl_name2no_language(ZSTR_VAL(new_value));
1210 if (no_language == mbfl_no_language_invalid) {
1211 MBSTRG(language) = mbfl_no_language_neutral;
1212 return FAILURE;
1213 }
1214 MBSTRG(language) = no_language;
1215 php_mb_nls_get_default_detect_order_list(no_language, &MBSTRG(default_detect_order_list), &MBSTRG(default_detect_order_list_size));
1216 return SUCCESS;
1217 }
1218 /* }}} */
1219
1220 /* {{{ static PHP_INI_MH(OnUpdate_mbstring_detect_order) */
PHP_INI_MH(OnUpdate_mbstring_detect_order)1221 static PHP_INI_MH(OnUpdate_mbstring_detect_order)
1222 {
1223 const mbfl_encoding **list;
1224 size_t size;
1225
1226 if (!new_value) {
1227 if (MBSTRG(detect_order_list)) {
1228 pefree(MBSTRG(detect_order_list), 1);
1229 }
1230 MBSTRG(detect_order_list) = NULL;
1231 MBSTRG(detect_order_list_size) = 0;
1232 return SUCCESS;
1233 }
1234
1235 if (FAILURE == php_mb_parse_encoding_list(ZSTR_VAL(new_value), ZSTR_LEN(new_value), &list, &size, 1)) {
1236 return FAILURE;
1237 }
1238
1239 if (MBSTRG(detect_order_list)) {
1240 pefree(MBSTRG(detect_order_list), 1);
1241 }
1242 MBSTRG(detect_order_list) = list;
1243 MBSTRG(detect_order_list_size) = size;
1244 return SUCCESS;
1245 }
1246 /* }}} */
1247
1248 /* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_input) */
PHP_INI_MH(OnUpdate_mbstring_http_input)1249 static PHP_INI_MH(OnUpdate_mbstring_http_input)
1250 {
1251 const mbfl_encoding **list;
1252 size_t size;
1253
1254 if (!new_value || !ZSTR_VAL(new_value)) {
1255 if (MBSTRG(http_input_list)) {
1256 pefree(MBSTRG(http_input_list), 1);
1257 }
1258 if (SUCCESS == php_mb_parse_encoding_list(get_input_encoding(), strlen(get_input_encoding())+1, &list, &size, 1)) {
1259 MBSTRG(http_input_list) = list;
1260 MBSTRG(http_input_list_size) = size;
1261 return SUCCESS;
1262 }
1263 MBSTRG(http_input_list) = NULL;
1264 MBSTRG(http_input_list_size) = 0;
1265 return SUCCESS;
1266 }
1267
1268 if (FAILURE == php_mb_parse_encoding_list(ZSTR_VAL(new_value), ZSTR_LEN(new_value), &list, &size, 1)) {
1269 return FAILURE;
1270 }
1271
1272 if (MBSTRG(http_input_list)) {
1273 pefree(MBSTRG(http_input_list), 1);
1274 }
1275 MBSTRG(http_input_list) = list;
1276 MBSTRG(http_input_list_size) = size;
1277
1278 if (stage & (PHP_INI_STAGE_ACTIVATE | PHP_INI_STAGE_RUNTIME)) {
1279 php_error_docref("ref.mbstring", E_DEPRECATED, "Use of mbstring.http_input is deprecated");
1280 }
1281
1282 return SUCCESS;
1283 }
1284 /* }}} */
1285
1286 /* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_output) */
PHP_INI_MH(OnUpdate_mbstring_http_output)1287 static PHP_INI_MH(OnUpdate_mbstring_http_output)
1288 {
1289 const mbfl_encoding *encoding;
1290
1291 if (new_value == NULL || ZSTR_LEN(new_value) == 0) {
1292 encoding = mbfl_name2encoding(get_output_encoding());
1293 if (!encoding) {
1294 MBSTRG(http_output_encoding) = &mbfl_encoding_pass;
1295 MBSTRG(current_http_output_encoding) = &mbfl_encoding_pass;
1296 return SUCCESS;
1297 }
1298 } else {
1299 encoding = mbfl_name2encoding(ZSTR_VAL(new_value));
1300 if (!encoding) {
1301 MBSTRG(http_output_encoding) = &mbfl_encoding_pass;
1302 MBSTRG(current_http_output_encoding) = &mbfl_encoding_pass;
1303 return FAILURE;
1304 }
1305 }
1306 MBSTRG(http_output_encoding) = encoding;
1307 MBSTRG(current_http_output_encoding) = encoding;
1308
1309 if (stage & (PHP_INI_STAGE_ACTIVATE | PHP_INI_STAGE_RUNTIME)) {
1310 php_error_docref("ref.mbstring", E_DEPRECATED, "Use of mbstring.http_output is deprecated");
1311 }
1312
1313 return SUCCESS;
1314 }
1315 /* }}} */
1316
1317 /* {{{ static _php_mb_ini_mbstring_internal_encoding_set */
_php_mb_ini_mbstring_internal_encoding_set(const char * new_value,uint new_value_length)1318 int _php_mb_ini_mbstring_internal_encoding_set(const char *new_value, uint new_value_length)
1319 {
1320 const mbfl_encoding *encoding;
1321
1322 if (!new_value || !new_value_length || !(encoding = mbfl_name2encoding(new_value))) {
1323 /* falls back to UTF-8 if an unknown encoding name is given */
1324 encoding = mbfl_no2encoding(mbfl_no_encoding_utf8);
1325 }
1326 MBSTRG(internal_encoding) = encoding;
1327 MBSTRG(current_internal_encoding) = encoding;
1328 #if HAVE_MBREGEX
1329 {
1330 const char *enc_name = new_value;
1331 if (FAILURE == php_mb_regex_set_default_mbctype(enc_name)) {
1332 /* falls back to UTF-8 if an unknown encoding name is given */
1333 enc_name = "UTF-8";
1334 php_mb_regex_set_default_mbctype(enc_name);
1335 }
1336 php_mb_regex_set_mbctype(new_value);
1337 }
1338 #endif
1339 return SUCCESS;
1340 }
1341 /* }}} */
1342
1343 /* {{{ static PHP_INI_MH(OnUpdate_mbstring_internal_encoding) */
PHP_INI_MH(OnUpdate_mbstring_internal_encoding)1344 static PHP_INI_MH(OnUpdate_mbstring_internal_encoding)
1345 {
1346 if (stage & (PHP_INI_STAGE_ACTIVATE | PHP_INI_STAGE_RUNTIME)) {
1347 php_error_docref("ref.mbstring", E_DEPRECATED, "Use of mbstring.internal_encoding is deprecated");
1348 }
1349
1350 if (OnUpdateString(entry, new_value, mh_arg1, mh_arg2, mh_arg3, stage) == FAILURE) {
1351 return FAILURE;
1352 }
1353
1354 if (stage & (PHP_INI_STAGE_STARTUP | PHP_INI_STAGE_SHUTDOWN | PHP_INI_STAGE_RUNTIME)) {
1355 if (new_value && ZSTR_LEN(new_value)) {
1356 return _php_mb_ini_mbstring_internal_encoding_set(ZSTR_VAL(new_value), ZSTR_LEN(new_value));
1357 } else {
1358 return _php_mb_ini_mbstring_internal_encoding_set(get_internal_encoding(), strlen(get_internal_encoding())+1);
1359 }
1360 } else {
1361 /* the corresponding mbstring globals needs to be set according to the
1362 * ini value in the later stage because it never falls back to the
1363 * default value if 1. no value for mbstring.internal_encoding is given,
1364 * 2. mbstring.language directive is processed in per-dir or runtime
1365 * context and 3. call to the handler for mbstring.language is done
1366 * after mbstring.internal_encoding is handled. */
1367 return SUCCESS;
1368 }
1369 }
1370 /* }}} */
1371
1372 /* {{{ static PHP_INI_MH(OnUpdate_mbstring_substitute_character) */
PHP_INI_MH(OnUpdate_mbstring_substitute_character)1373 static PHP_INI_MH(OnUpdate_mbstring_substitute_character)
1374 {
1375 int c;
1376 char *endptr = NULL;
1377
1378 if (new_value != NULL) {
1379 if (strcasecmp("none", ZSTR_VAL(new_value)) == 0) {
1380 MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
1381 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
1382 } else if (strcasecmp("long", ZSTR_VAL(new_value)) == 0) {
1383 MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG;
1384 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG;
1385 } else if (strcasecmp("entity", ZSTR_VAL(new_value)) == 0) {
1386 MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY;
1387 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY;
1388 } else {
1389 MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1390 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1391 if (ZSTR_LEN(new_value) > 0) {
1392 c = strtol(ZSTR_VAL(new_value), &endptr, 0);
1393 if (*endptr == '\0') {
1394 MBSTRG(filter_illegal_substchar) = c;
1395 MBSTRG(current_filter_illegal_substchar) = c;
1396 }
1397 }
1398 }
1399 } else {
1400 MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1401 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1402 MBSTRG(filter_illegal_substchar) = 0x3f; /* '?' */
1403 MBSTRG(current_filter_illegal_substchar) = 0x3f; /* '?' */
1404 }
1405
1406 return SUCCESS;
1407 }
1408 /* }}} */
1409
1410 /* {{{ static PHP_INI_MH(OnUpdate_mbstring_encoding_translation) */
PHP_INI_MH(OnUpdate_mbstring_encoding_translation)1411 static PHP_INI_MH(OnUpdate_mbstring_encoding_translation)
1412 {
1413 if (new_value == NULL) {
1414 return FAILURE;
1415 }
1416
1417 OnUpdateBool(entry, new_value, mh_arg1, mh_arg2, mh_arg3, stage);
1418
1419 if (MBSTRG(encoding_translation)) {
1420 sapi_unregister_post_entry(php_post_entries);
1421 sapi_register_post_entries(mbstr_post_entries);
1422 } else {
1423 sapi_unregister_post_entry(mbstr_post_entries);
1424 sapi_register_post_entries(php_post_entries);
1425 }
1426
1427 return SUCCESS;
1428 }
1429 /* }}} */
1430
1431 /* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_output_conv_mimetypes */
PHP_INI_MH(OnUpdate_mbstring_http_output_conv_mimetypes)1432 static PHP_INI_MH(OnUpdate_mbstring_http_output_conv_mimetypes)
1433 {
1434 zend_string *tmp;
1435 void *re = NULL;
1436
1437 if (!new_value) {
1438 new_value = entry->orig_value;
1439 }
1440 tmp = php_trim(new_value, NULL, 0, 3);
1441
1442 if (ZSTR_LEN(tmp) > 0) {
1443 if (!(re = _php_mb_compile_regex(ZSTR_VAL(tmp)))) {
1444 zend_string_release(tmp);
1445 return FAILURE;
1446 }
1447 }
1448
1449 if (MBSTRG(http_output_conv_mimetypes)) {
1450 _php_mb_free_regex(MBSTRG(http_output_conv_mimetypes));
1451 }
1452
1453 MBSTRG(http_output_conv_mimetypes) = re;
1454
1455 zend_string_release(tmp);
1456 return SUCCESS;
1457 }
1458 /* }}} */
1459 /* }}} */
1460
1461 /* {{{ php.ini directive registration */
1462 PHP_INI_BEGIN()
1463 PHP_INI_ENTRY("mbstring.language", "neutral", PHP_INI_ALL, OnUpdate_mbstring_language)
1464 PHP_INI_ENTRY("mbstring.detect_order", NULL, PHP_INI_ALL, OnUpdate_mbstring_detect_order)
1465 PHP_INI_ENTRY("mbstring.http_input", NULL, PHP_INI_ALL, OnUpdate_mbstring_http_input)
1466 PHP_INI_ENTRY("mbstring.http_output", NULL, PHP_INI_ALL, OnUpdate_mbstring_http_output)
1467 STD_PHP_INI_ENTRY("mbstring.internal_encoding", NULL, PHP_INI_ALL, OnUpdate_mbstring_internal_encoding, internal_encoding_name, zend_mbstring_globals, mbstring_globals)
1468 PHP_INI_ENTRY("mbstring.substitute_character", NULL, PHP_INI_ALL, OnUpdate_mbstring_substitute_character)
1469 STD_PHP_INI_ENTRY("mbstring.func_overload", "0",
1470 PHP_INI_SYSTEM, OnUpdateLong, func_overload, zend_mbstring_globals, mbstring_globals)
1471
1472 STD_PHP_INI_BOOLEAN("mbstring.encoding_translation", "0",
1473 PHP_INI_SYSTEM | PHP_INI_PERDIR,
1474 OnUpdate_mbstring_encoding_translation,
1475 encoding_translation, zend_mbstring_globals, mbstring_globals)
1476 PHP_INI_ENTRY("mbstring.http_output_conv_mimetypes",
1477 "^(text/|application/xhtml\\+xml)",
1478 PHP_INI_ALL,
1479 OnUpdate_mbstring_http_output_conv_mimetypes)
1480
1481 STD_PHP_INI_BOOLEAN("mbstring.strict_detection", "0",
1482 PHP_INI_ALL,
1483 OnUpdateLong,
1484 strict_detection, zend_mbstring_globals, mbstring_globals)
PHP_INI_END()1485 PHP_INI_END()
1486 /* }}} */
1487
1488 /* {{{ module global initialize handler */
1489 static PHP_GINIT_FUNCTION(mbstring)
1490 {
1491 #if defined(COMPILE_DL_MBSTRING) && defined(ZTS)
1492 ZEND_TSRMLS_CACHE_UPDATE();
1493 #endif
1494
1495 mbstring_globals->language = mbfl_no_language_uni;
1496 mbstring_globals->internal_encoding = NULL;
1497 mbstring_globals->current_internal_encoding = mbstring_globals->internal_encoding;
1498 mbstring_globals->http_output_encoding = &mbfl_encoding_pass;
1499 mbstring_globals->current_http_output_encoding = &mbfl_encoding_pass;
1500 mbstring_globals->http_input_identify = NULL;
1501 mbstring_globals->http_input_identify_get = NULL;
1502 mbstring_globals->http_input_identify_post = NULL;
1503 mbstring_globals->http_input_identify_cookie = NULL;
1504 mbstring_globals->http_input_identify_string = NULL;
1505 mbstring_globals->http_input_list = NULL;
1506 mbstring_globals->http_input_list_size = 0;
1507 mbstring_globals->detect_order_list = NULL;
1508 mbstring_globals->detect_order_list_size = 0;
1509 mbstring_globals->current_detect_order_list = NULL;
1510 mbstring_globals->current_detect_order_list_size = 0;
1511 mbstring_globals->default_detect_order_list = (enum mbfl_no_encoding *) php_mb_default_identify_list_neut;
1512 mbstring_globals->default_detect_order_list_size = sizeof(php_mb_default_identify_list_neut) / sizeof(php_mb_default_identify_list_neut[0]);
1513 mbstring_globals->filter_illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1514 mbstring_globals->filter_illegal_substchar = 0x3f; /* '?' */
1515 mbstring_globals->current_filter_illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1516 mbstring_globals->current_filter_illegal_substchar = 0x3f; /* '?' */
1517 mbstring_globals->illegalchars = 0;
1518 mbstring_globals->func_overload = 0;
1519 mbstring_globals->encoding_translation = 0;
1520 mbstring_globals->strict_detection = 0;
1521 mbstring_globals->outconv = NULL;
1522 mbstring_globals->http_output_conv_mimetypes = NULL;
1523 #if HAVE_MBREGEX
1524 mbstring_globals->mb_regex_globals = php_mb_regex_globals_alloc();
1525 #endif
1526 }
1527 /* }}} */
1528
1529 /* {{{ PHP_GSHUTDOWN_FUNCTION */
PHP_GSHUTDOWN_FUNCTION(mbstring)1530 static PHP_GSHUTDOWN_FUNCTION(mbstring)
1531 {
1532 if (mbstring_globals->http_input_list) {
1533 free(mbstring_globals->http_input_list);
1534 }
1535 if (mbstring_globals->detect_order_list) {
1536 free(mbstring_globals->detect_order_list);
1537 }
1538 if (mbstring_globals->http_output_conv_mimetypes) {
1539 _php_mb_free_regex(mbstring_globals->http_output_conv_mimetypes);
1540 }
1541 #if HAVE_MBREGEX
1542 php_mb_regex_globals_free(mbstring_globals->mb_regex_globals);
1543 #endif
1544 }
1545 /* }}} */
1546
1547 /* {{{ PHP_MINIT_FUNCTION(mbstring) */
PHP_MINIT_FUNCTION(mbstring)1548 PHP_MINIT_FUNCTION(mbstring)
1549 {
1550 #if defined(COMPILE_DL_MBSTRING) && defined(ZTS)
1551 ZEND_TSRMLS_CACHE_UPDATE();
1552 #endif
1553 __mbfl_allocators = &_php_mb_allocators;
1554
1555 REGISTER_INI_ENTRIES();
1556
1557 /* This is a global handler. Should not be set in a per-request handler. */
1558 sapi_register_treat_data(mbstr_treat_data);
1559
1560 /* Post handlers are stored in the thread-local context. */
1561 if (MBSTRG(encoding_translation)) {
1562 sapi_register_post_entries(mbstr_post_entries);
1563 }
1564
1565 REGISTER_LONG_CONSTANT("MB_OVERLOAD_MAIL", MB_OVERLOAD_MAIL, CONST_CS | CONST_PERSISTENT);
1566 REGISTER_LONG_CONSTANT("MB_OVERLOAD_STRING", MB_OVERLOAD_STRING, CONST_CS | CONST_PERSISTENT);
1567 REGISTER_LONG_CONSTANT("MB_OVERLOAD_REGEX", MB_OVERLOAD_REGEX, CONST_CS | CONST_PERSISTENT);
1568
1569 REGISTER_LONG_CONSTANT("MB_CASE_UPPER", PHP_UNICODE_CASE_UPPER, CONST_CS | CONST_PERSISTENT);
1570 REGISTER_LONG_CONSTANT("MB_CASE_LOWER", PHP_UNICODE_CASE_LOWER, CONST_CS | CONST_PERSISTENT);
1571 REGISTER_LONG_CONSTANT("MB_CASE_TITLE", PHP_UNICODE_CASE_TITLE, CONST_CS | CONST_PERSISTENT);
1572
1573 #if HAVE_MBREGEX
1574 PHP_MINIT(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
1575 #endif
1576
1577 if (FAILURE == zend_multibyte_set_functions(&php_mb_zend_multibyte_functions)) {
1578 return FAILURE;
1579 }
1580
1581 php_rfc1867_set_multibyte_callbacks(
1582 php_mb_encoding_translation,
1583 php_mb_gpc_get_detect_order,
1584 php_mb_gpc_set_input_encoding,
1585 php_mb_rfc1867_getword,
1586 php_mb_rfc1867_getword_conf,
1587 php_mb_rfc1867_basename);
1588
1589 return SUCCESS;
1590 }
1591 /* }}} */
1592
1593 /* {{{ PHP_MSHUTDOWN_FUNCTION(mbstring) */
PHP_MSHUTDOWN_FUNCTION(mbstring)1594 PHP_MSHUTDOWN_FUNCTION(mbstring)
1595 {
1596 UNREGISTER_INI_ENTRIES();
1597
1598 zend_multibyte_restore_functions();
1599
1600 #if HAVE_MBREGEX
1601 PHP_MSHUTDOWN(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
1602 #endif
1603
1604 return SUCCESS;
1605 }
1606 /* }}} */
1607
1608 /* {{{ PHP_RINIT_FUNCTION(mbstring) */
PHP_RINIT_FUNCTION(mbstring)1609 PHP_RINIT_FUNCTION(mbstring)
1610 {
1611 zend_function *func, *orig;
1612 const struct mb_overload_def *p;
1613
1614 MBSTRG(current_internal_encoding) = MBSTRG(internal_encoding);
1615 MBSTRG(current_http_output_encoding) = MBSTRG(http_output_encoding);
1616 MBSTRG(current_filter_illegal_mode) = MBSTRG(filter_illegal_mode);
1617 MBSTRG(current_filter_illegal_substchar) = MBSTRG(filter_illegal_substchar);
1618
1619 MBSTRG(illegalchars) = 0;
1620
1621 php_mb_populate_current_detect_order_list();
1622
1623 /* override original function. */
1624 if (MBSTRG(func_overload)){
1625 p = &(mb_ovld[0]);
1626
1627 CG(compiler_options) |= ZEND_COMPILE_NO_BUILTIN_STRLEN;
1628 while (p->type > 0) {
1629 if ((MBSTRG(func_overload) & p->type) == p->type &&
1630 !zend_hash_str_exists(EG(function_table), p->save_func, strlen(p->save_func))
1631 ) {
1632 func = zend_hash_str_find_ptr(EG(function_table), p->ovld_func, strlen(p->ovld_func));
1633
1634 if ((orig = zend_hash_str_find_ptr(EG(function_table), p->orig_func, strlen(p->orig_func))) == NULL) {
1635 php_error_docref("ref.mbstring", E_WARNING, "mbstring couldn't find function %s.", p->orig_func);
1636 return FAILURE;
1637 } else {
1638 ZEND_ASSERT(orig->type == ZEND_INTERNAL_FUNCTION);
1639 zend_hash_str_add_mem(EG(function_table), p->save_func, strlen(p->save_func), orig, sizeof(zend_internal_function));
1640 function_add_ref(orig);
1641
1642 if (zend_hash_str_update_mem(EG(function_table), p->orig_func, strlen(p->orig_func), func, sizeof(zend_internal_function)) == NULL) {
1643 php_error_docref("ref.mbstring", E_WARNING, "mbstring couldn't replace function %s.", p->orig_func);
1644 return FAILURE;
1645 }
1646
1647 function_add_ref(func);
1648 }
1649 }
1650 p++;
1651 }
1652 }
1653 #if HAVE_MBREGEX
1654 PHP_RINIT(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
1655 #endif
1656 zend_multibyte_set_internal_encoding((const zend_encoding *)MBSTRG(internal_encoding));
1657
1658 return SUCCESS;
1659 }
1660 /* }}} */
1661
1662 /* {{{ PHP_RSHUTDOWN_FUNCTION(mbstring) */
PHP_RSHUTDOWN_FUNCTION(mbstring)1663 PHP_RSHUTDOWN_FUNCTION(mbstring)
1664 {
1665 const struct mb_overload_def *p;
1666 zend_function *orig;
1667
1668 if (MBSTRG(current_detect_order_list) != NULL) {
1669 efree(MBSTRG(current_detect_order_list));
1670 MBSTRG(current_detect_order_list) = NULL;
1671 MBSTRG(current_detect_order_list_size) = 0;
1672 }
1673 if (MBSTRG(outconv) != NULL) {
1674 MBSTRG(illegalchars) += mbfl_buffer_illegalchars(MBSTRG(outconv));
1675 mbfl_buffer_converter_delete(MBSTRG(outconv));
1676 MBSTRG(outconv) = NULL;
1677 }
1678
1679 /* clear http input identification. */
1680 MBSTRG(http_input_identify) = NULL;
1681 MBSTRG(http_input_identify_post) = NULL;
1682 MBSTRG(http_input_identify_get) = NULL;
1683 MBSTRG(http_input_identify_cookie) = NULL;
1684 MBSTRG(http_input_identify_string) = NULL;
1685
1686 /* clear overloaded function. */
1687 if (MBSTRG(func_overload)){
1688 p = &(mb_ovld[0]);
1689 while (p->type > 0) {
1690 if ((MBSTRG(func_overload) & p->type) == p->type &&
1691 (orig = zend_hash_str_find_ptr(EG(function_table), p->save_func, strlen(p->save_func)))) {
1692
1693 zend_hash_str_update_mem(EG(function_table), p->orig_func, strlen(p->orig_func), orig, sizeof(zend_internal_function));
1694 function_add_ref(orig);
1695 zend_hash_str_del(EG(function_table), p->save_func, strlen(p->save_func));
1696 }
1697 p++;
1698 }
1699 CG(compiler_options) &= ~ZEND_COMPILE_NO_BUILTIN_STRLEN;
1700 }
1701
1702 #if HAVE_MBREGEX
1703 PHP_RSHUTDOWN(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
1704 #endif
1705
1706 return SUCCESS;
1707 }
1708 /* }}} */
1709
1710 /* {{{ PHP_MINFO_FUNCTION(mbstring) */
PHP_MINFO_FUNCTION(mbstring)1711 PHP_MINFO_FUNCTION(mbstring)
1712 {
1713 php_info_print_table_start();
1714 php_info_print_table_row(2, "Multibyte Support", "enabled");
1715 php_info_print_table_row(2, "Multibyte string engine", "libmbfl");
1716 php_info_print_table_row(2, "HTTP input encoding translation", MBSTRG(encoding_translation) ? "enabled": "disabled");
1717 {
1718 char tmp[256];
1719 snprintf(tmp, sizeof(tmp), "%d.%d.%d", MBFL_VERSION_MAJOR, MBFL_VERSION_MINOR, MBFL_VERSION_TEENY);
1720 php_info_print_table_row(2, "libmbfl version", tmp);
1721 }
1722 #if HAVE_ONIG
1723 {
1724 char tmp[256];
1725 snprintf(tmp, sizeof(tmp), "%d.%d.%d", ONIGURUMA_VERSION_MAJOR, ONIGURUMA_VERSION_MINOR, ONIGURUMA_VERSION_TEENY);
1726 php_info_print_table_row(2, "oniguruma version", tmp);
1727 }
1728 #endif
1729 php_info_print_table_end();
1730
1731 php_info_print_table_start();
1732 php_info_print_table_header(1, "mbstring extension makes use of \"streamable kanji code filter and converter\", which is distributed under the GNU Lesser General Public License version 2.1.");
1733 php_info_print_table_end();
1734
1735 #if HAVE_MBREGEX
1736 PHP_MINFO(mb_regex)(ZEND_MODULE_INFO_FUNC_ARGS_PASSTHRU);
1737 #endif
1738
1739 DISPLAY_INI_ENTRIES();
1740 }
1741 /* }}} */
1742
1743 /* {{{ proto string mb_language([string language])
1744 Sets the current language or Returns the current language as a string */
PHP_FUNCTION(mb_language)1745 PHP_FUNCTION(mb_language)
1746 {
1747 zend_string *name = NULL;
1748
1749 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|S", &name) == FAILURE) {
1750 return;
1751 }
1752 if (name == NULL) {
1753 RETVAL_STRING((char *)mbfl_no_language2name(MBSTRG(language)));
1754 } else {
1755 zend_string *ini_name = zend_string_init("mbstring.language", sizeof("mbstring.language") - 1, 0);
1756 if (FAILURE == zend_alter_ini_entry(ini_name, name, PHP_INI_USER, PHP_INI_STAGE_RUNTIME)) {
1757 php_error_docref(NULL, E_WARNING, "Unknown language \"%s\"", ZSTR_VAL(name));
1758 RETVAL_FALSE;
1759 } else {
1760 RETVAL_TRUE;
1761 }
1762 zend_string_release(ini_name);
1763 }
1764 }
1765 /* }}} */
1766
1767 /* {{{ proto string mb_internal_encoding([string encoding])
1768 Sets the current internal encoding or Returns the current internal encoding as a string */
PHP_FUNCTION(mb_internal_encoding)1769 PHP_FUNCTION(mb_internal_encoding)
1770 {
1771 const char *name = NULL;
1772 size_t name_len;
1773 const mbfl_encoding *encoding;
1774
1775 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|s", &name, &name_len) == FAILURE) {
1776 return;
1777 }
1778 if (name == NULL) {
1779 name = MBSTRG(current_internal_encoding) ? MBSTRG(current_internal_encoding)->name: NULL;
1780 if (name != NULL) {
1781 RETURN_STRING(name);
1782 } else {
1783 RETURN_FALSE;
1784 }
1785 } else {
1786 encoding = mbfl_name2encoding(name);
1787 if (!encoding) {
1788 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", name);
1789 RETURN_FALSE;
1790 } else {
1791 MBSTRG(current_internal_encoding) = encoding;
1792 RETURN_TRUE;
1793 }
1794 }
1795 }
1796 /* }}} */
1797
1798 /* {{{ proto mixed mb_http_input([string type])
1799 Returns the input encoding */
PHP_FUNCTION(mb_http_input)1800 PHP_FUNCTION(mb_http_input)
1801 {
1802 char *typ = NULL;
1803 size_t typ_len;
1804 int retname;
1805 char *list, *temp;
1806 const mbfl_encoding *result = NULL;
1807
1808 retname = 1;
1809 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|s", &typ, &typ_len) == FAILURE) {
1810 return;
1811 }
1812 if (typ == NULL) {
1813 result = MBSTRG(http_input_identify);
1814 } else {
1815 switch (*typ) {
1816 case 'G':
1817 case 'g':
1818 result = MBSTRG(http_input_identify_get);
1819 break;
1820 case 'P':
1821 case 'p':
1822 result = MBSTRG(http_input_identify_post);
1823 break;
1824 case 'C':
1825 case 'c':
1826 result = MBSTRG(http_input_identify_cookie);
1827 break;
1828 case 'S':
1829 case 's':
1830 result = MBSTRG(http_input_identify_string);
1831 break;
1832 case 'I':
1833 case 'i':
1834 {
1835 const mbfl_encoding **entry = MBSTRG(http_input_list);
1836 const size_t n = MBSTRG(http_input_list_size);
1837 size_t i;
1838 array_init(return_value);
1839 for (i = 0; i < n; i++) {
1840 add_next_index_string(return_value, (*entry)->name);
1841 entry++;
1842 }
1843 retname = 0;
1844 }
1845 break;
1846 case 'L':
1847 case 'l':
1848 {
1849 const mbfl_encoding **entry = MBSTRG(http_input_list);
1850 const size_t n = MBSTRG(http_input_list_size);
1851 size_t i;
1852 list = NULL;
1853 for (i = 0; i < n; i++) {
1854 if (list) {
1855 temp = list;
1856 spprintf(&list, 0, "%s,%s", temp, (*entry)->name);
1857 efree(temp);
1858 if (!list) {
1859 break;
1860 }
1861 } else {
1862 list = estrdup((*entry)->name);
1863 }
1864 entry++;
1865 }
1866 }
1867 if (!list) {
1868 RETURN_FALSE;
1869 }
1870 RETVAL_STRING(list);
1871 efree(list);
1872 retname = 0;
1873 break;
1874 default:
1875 result = MBSTRG(http_input_identify);
1876 break;
1877 }
1878 }
1879
1880 if (retname) {
1881 if (result) {
1882 RETVAL_STRING(result->name);
1883 } else {
1884 RETVAL_FALSE;
1885 }
1886 }
1887 }
1888 /* }}} */
1889
1890 /* {{{ proto string mb_http_output([string encoding])
1891 Sets the current output_encoding or returns the current output_encoding as a string */
PHP_FUNCTION(mb_http_output)1892 PHP_FUNCTION(mb_http_output)
1893 {
1894 const char *name = NULL;
1895 size_t name_len;
1896 const mbfl_encoding *encoding;
1897
1898 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|s", &name, &name_len) == FAILURE) {
1899 return;
1900 }
1901
1902 if (name == NULL) {
1903 name = MBSTRG(current_http_output_encoding) ? MBSTRG(current_http_output_encoding)->name: NULL;
1904 if (name != NULL) {
1905 RETURN_STRING(name);
1906 } else {
1907 RETURN_FALSE;
1908 }
1909 } else {
1910 encoding = mbfl_name2encoding(name);
1911 if (!encoding) {
1912 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", name);
1913 RETURN_FALSE;
1914 } else {
1915 MBSTRG(current_http_output_encoding) = encoding;
1916 RETURN_TRUE;
1917 }
1918 }
1919 }
1920 /* }}} */
1921
1922 /* {{{ proto bool|array mb_detect_order([mixed encoding-list])
1923 Sets the current detect_order or Return the current detect_order as a array */
PHP_FUNCTION(mb_detect_order)1924 PHP_FUNCTION(mb_detect_order)
1925 {
1926 zval *arg1 = NULL;
1927
1928 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|z", &arg1) == FAILURE) {
1929 return;
1930 }
1931
1932 if (!arg1) {
1933 size_t i;
1934 size_t n = MBSTRG(current_detect_order_list_size);
1935 const mbfl_encoding **entry = MBSTRG(current_detect_order_list);
1936 array_init(return_value);
1937 for (i = 0; i < n; i++) {
1938 add_next_index_string(return_value, (*entry)->name);
1939 entry++;
1940 }
1941 } else {
1942 const mbfl_encoding **list = NULL;
1943 size_t size = 0;
1944 switch (Z_TYPE_P(arg1)) {
1945 case IS_ARRAY:
1946 if (FAILURE == php_mb_parse_encoding_array(arg1, &list, &size, 0)) {
1947 if (list) {
1948 efree(list);
1949 }
1950 RETURN_FALSE;
1951 }
1952 break;
1953 default:
1954 convert_to_string_ex(arg1);
1955 if (FAILURE == php_mb_parse_encoding_list(Z_STRVAL_P(arg1), Z_STRLEN_P(arg1), &list, &size, 0)) {
1956 if (list) {
1957 efree(list);
1958 }
1959 RETURN_FALSE;
1960 }
1961 break;
1962 }
1963
1964 if (list == NULL) {
1965 RETURN_FALSE;
1966 }
1967
1968 if (MBSTRG(current_detect_order_list)) {
1969 efree(MBSTRG(current_detect_order_list));
1970 }
1971 MBSTRG(current_detect_order_list) = list;
1972 MBSTRG(current_detect_order_list_size) = size;
1973 RETURN_TRUE;
1974 }
1975 }
1976 /* }}} */
1977
1978 /* {{{ proto mixed mb_substitute_character([mixed substchar])
1979 Sets the current substitute_character or returns the current substitute_character */
PHP_FUNCTION(mb_substitute_character)1980 PHP_FUNCTION(mb_substitute_character)
1981 {
1982 zval *arg1 = NULL;
1983
1984 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|z", &arg1) == FAILURE) {
1985 return;
1986 }
1987
1988 if (!arg1) {
1989 if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
1990 RETURN_STRING("none");
1991 } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG) {
1992 RETURN_STRING("long");
1993 } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY) {
1994 RETURN_STRING("entity");
1995 } else {
1996 RETURN_LONG(MBSTRG(current_filter_illegal_substchar));
1997 }
1998 } else {
1999 RETVAL_TRUE;
2000
2001 switch (Z_TYPE_P(arg1)) {
2002 case IS_STRING:
2003 if (strncasecmp("none", Z_STRVAL_P(arg1), Z_STRLEN_P(arg1)) == 0) {
2004 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
2005 } else if (strncasecmp("long", Z_STRVAL_P(arg1), Z_STRLEN_P(arg1)) == 0) {
2006 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG;
2007 } else if (strncasecmp("entity", Z_STRVAL_P(arg1), Z_STRLEN_P(arg1)) == 0) {
2008 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY;
2009 } else {
2010 convert_to_long_ex(arg1);
2011
2012 if (Z_LVAL_P(arg1) < 0xffff && Z_LVAL_P(arg1) > 0x0) {
2013 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
2014 MBSTRG(current_filter_illegal_substchar) = Z_LVAL_P(arg1);
2015 } else {
2016 php_error_docref(NULL, E_WARNING, "Unknown character.");
2017 RETURN_FALSE;
2018 }
2019 }
2020 break;
2021 default:
2022 convert_to_long_ex(arg1);
2023 if (Z_LVAL_P(arg1) < 0xffff && Z_LVAL_P(arg1) > 0x0) {
2024 MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
2025 MBSTRG(current_filter_illegal_substchar) = Z_LVAL_P(arg1);
2026 } else {
2027 php_error_docref(NULL, E_WARNING, "Unknown character.");
2028 RETURN_FALSE;
2029 }
2030 break;
2031 }
2032 }
2033 }
2034 /* }}} */
2035
2036 /* {{{ proto string mb_preferred_mime_name(string encoding)
2037 Return the preferred MIME name (charset) as a string */
PHP_FUNCTION(mb_preferred_mime_name)2038 PHP_FUNCTION(mb_preferred_mime_name)
2039 {
2040 enum mbfl_no_encoding no_encoding;
2041 char *name = NULL;
2042 size_t name_len;
2043
2044 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s", &name, &name_len) == FAILURE) {
2045 return;
2046 } else {
2047 no_encoding = mbfl_name2no_encoding(name);
2048 if (no_encoding == mbfl_no_encoding_invalid) {
2049 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", name);
2050 RETVAL_FALSE;
2051 } else {
2052 const char *preferred_name = mbfl_no2preferred_mime_name(no_encoding);
2053 if (preferred_name == NULL || *preferred_name == '\0') {
2054 php_error_docref(NULL, E_WARNING, "No MIME preferred name corresponding to \"%s\"", name);
2055 RETVAL_FALSE;
2056 } else {
2057 RETVAL_STRING((char *)preferred_name);
2058 }
2059 }
2060 }
2061 }
2062 /* }}} */
2063
2064 #define IS_SJIS1(c) ((((c)>=0x81 && (c)<=0x9f) || ((c)>=0xe0 && (c)<=0xf5)) ? 1 : 0)
2065 #define IS_SJIS2(c) ((((c)>=0x40 && (c)<=0x7e) || ((c)>=0x80 && (c)<=0xfc)) ? 1 : 0)
2066
2067 /* {{{ proto bool mb_parse_str(string encoded_string [, array result])
2068 Parses GET/POST/COOKIE data and sets global variables */
PHP_FUNCTION(mb_parse_str)2069 PHP_FUNCTION(mb_parse_str)
2070 {
2071 zval *track_vars_array = NULL;
2072 char *encstr = NULL;
2073 size_t encstr_len;
2074 php_mb_encoding_handler_info_t info;
2075 const mbfl_encoding *detected;
2076
2077 track_vars_array = NULL;
2078 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|z/", &encstr, &encstr_len, &track_vars_array) == FAILURE) {
2079 return;
2080 }
2081
2082 if (track_vars_array != NULL) {
2083 /* Clear out the array */
2084 zval_dtor(track_vars_array);
2085 array_init(track_vars_array);
2086 }
2087
2088 encstr = estrndup(encstr, encstr_len);
2089
2090 info.data_type = PARSE_STRING;
2091 info.separator = PG(arg_separator).input;
2092 info.report_errors = 1;
2093 info.to_encoding = MBSTRG(current_internal_encoding);
2094 info.to_language = MBSTRG(language);
2095 info.from_encodings = MBSTRG(http_input_list);
2096 info.num_from_encodings = MBSTRG(http_input_list_size);
2097 info.from_language = MBSTRG(language);
2098
2099 if (track_vars_array != NULL) {
2100 detected = _php_mb_encoding_handler_ex(&info, track_vars_array, encstr);
2101 } else {
2102 zval tmp;
2103 zend_array *symbol_table;
2104 if (zend_forbid_dynamic_call("mb_parse_str() with a single argument") == FAILURE) {
2105 efree(encstr);
2106 return;
2107 }
2108
2109 symbol_table = zend_rebuild_symbol_table();
2110 ZVAL_ARR(&tmp, symbol_table);
2111 detected = _php_mb_encoding_handler_ex(&info, &tmp, encstr);
2112 }
2113
2114 MBSTRG(http_input_identify) = detected;
2115
2116 RETVAL_BOOL(detected);
2117
2118 if (encstr != NULL) efree(encstr);
2119 }
2120 /* }}} */
2121
2122 /* {{{ proto string mb_output_handler(string contents, int status)
2123 Returns string in output buffer converted to the http_output encoding */
PHP_FUNCTION(mb_output_handler)2124 PHP_FUNCTION(mb_output_handler)
2125 {
2126 char *arg_string;
2127 size_t arg_string_len;
2128 zend_long arg_status;
2129 mbfl_string string, result;
2130 const char *charset;
2131 char *p;
2132 const mbfl_encoding *encoding;
2133 int last_feed, len;
2134 unsigned char send_text_mimetype = 0;
2135 char *s, *mimetype = NULL;
2136
2137 if (zend_parse_parameters(ZEND_NUM_ARGS(), "sl", &arg_string, &arg_string_len, &arg_status) == FAILURE) {
2138 return;
2139 }
2140
2141 encoding = MBSTRG(current_http_output_encoding);
2142
2143 /* start phase only */
2144 if ((arg_status & PHP_OUTPUT_HANDLER_START) != 0) {
2145 /* delete the converter just in case. */
2146 if (MBSTRG(outconv)) {
2147 MBSTRG(illegalchars) += mbfl_buffer_illegalchars(MBSTRG(outconv));
2148 mbfl_buffer_converter_delete(MBSTRG(outconv));
2149 MBSTRG(outconv) = NULL;
2150 }
2151 if (encoding == &mbfl_encoding_pass) {
2152 RETURN_STRINGL(arg_string, arg_string_len);
2153 }
2154
2155 /* analyze mime type */
2156 if (SG(sapi_headers).mimetype &&
2157 _php_mb_match_regex(
2158 MBSTRG(http_output_conv_mimetypes),
2159 SG(sapi_headers).mimetype,
2160 strlen(SG(sapi_headers).mimetype))) {
2161 if ((s = strchr(SG(sapi_headers).mimetype,';')) == NULL){
2162 mimetype = estrdup(SG(sapi_headers).mimetype);
2163 } else {
2164 mimetype = estrndup(SG(sapi_headers).mimetype,s-SG(sapi_headers).mimetype);
2165 }
2166 send_text_mimetype = 1;
2167 } else if (SG(sapi_headers).send_default_content_type) {
2168 mimetype = SG(default_mimetype) ? SG(default_mimetype) : SAPI_DEFAULT_MIMETYPE;
2169 }
2170
2171 /* if content-type is not yet set, set it and activate the converter */
2172 if (SG(sapi_headers).send_default_content_type || send_text_mimetype) {
2173 charset = encoding->mime_name;
2174 if (charset) {
2175 len = spprintf( &p, 0, "Content-Type: %s; charset=%s", mimetype, charset );
2176 if (sapi_add_header(p, len, 0) != FAILURE) {
2177 SG(sapi_headers).send_default_content_type = 0;
2178 }
2179 }
2180 /* activate the converter */
2181 MBSTRG(outconv) = mbfl_buffer_converter_new2(MBSTRG(current_internal_encoding), encoding, 0);
2182 if (send_text_mimetype){
2183 efree(mimetype);
2184 }
2185 }
2186 }
2187
2188 /* just return if the converter is not activated. */
2189 if (MBSTRG(outconv) == NULL) {
2190 RETURN_STRINGL(arg_string, arg_string_len);
2191 }
2192
2193 /* flag */
2194 last_feed = ((arg_status & PHP_OUTPUT_HANDLER_END) != 0);
2195 /* mode */
2196 mbfl_buffer_converter_illegal_mode(MBSTRG(outconv), MBSTRG(current_filter_illegal_mode));
2197 mbfl_buffer_converter_illegal_substchar(MBSTRG(outconv), MBSTRG(current_filter_illegal_substchar));
2198
2199 /* feed the string */
2200 mbfl_string_init(&string);
2201 /* these are not needed. convd has encoding info.
2202 string.no_language = MBSTRG(language);
2203 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2204 */
2205 string.val = (unsigned char *)arg_string;
2206 string.len = arg_string_len;
2207 mbfl_buffer_converter_feed(MBSTRG(outconv), &string);
2208 if (last_feed) {
2209 mbfl_buffer_converter_flush(MBSTRG(outconv));
2210 }
2211 /* get the converter output, and return it */
2212 mbfl_buffer_converter_result(MBSTRG(outconv), &result);
2213 // TODO: avoid reallocation ???
2214 RETVAL_STRINGL((char *)result.val, result.len); /* the string is already strdup()'ed */
2215 efree(result.val);
2216
2217 /* delete the converter if it is the last feed. */
2218 if (last_feed) {
2219 MBSTRG(illegalchars) += mbfl_buffer_illegalchars(MBSTRG(outconv));
2220 mbfl_buffer_converter_delete(MBSTRG(outconv));
2221 MBSTRG(outconv) = NULL;
2222 }
2223 }
2224 /* }}} */
2225
2226 /* {{{ proto int mb_strlen(string str [, string encoding])
2227 Get character numbers of a string */
PHP_FUNCTION(mb_strlen)2228 PHP_FUNCTION(mb_strlen)
2229 {
2230 int n;
2231 mbfl_string string;
2232 char *enc_name = NULL;
2233 size_t enc_name_len, string_len;
2234
2235 mbfl_string_init(&string);
2236
2237 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|s", (char **)&string.val, &string_len, &enc_name, &enc_name_len) == FAILURE) {
2238 return;
2239 }
2240
2241 if (ZEND_SIZE_T_UINT_OVFL(string_len)) {
2242 php_error_docref(NULL, E_WARNING, "String overflows the max allowed length of %u", UINT_MAX);
2243 return;
2244 }
2245
2246 string.len = (uint32_t)string_len;
2247
2248 string.no_language = MBSTRG(language);
2249 if (enc_name == NULL) {
2250 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2251 } else {
2252 string.no_encoding = mbfl_name2no_encoding(enc_name);
2253 if (string.no_encoding == mbfl_no_encoding_invalid) {
2254 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2255 RETURN_FALSE;
2256 }
2257 }
2258
2259 n = mbfl_strlen(&string);
2260 if (n >= 0) {
2261 RETVAL_LONG(n);
2262 } else {
2263 RETVAL_FALSE;
2264 }
2265 }
2266 /* }}} */
2267
2268 /* {{{ proto int mb_strpos(string haystack, string needle [, int offset [, string encoding]])
2269 Find position of first occurrence of a string within another */
PHP_FUNCTION(mb_strpos)2270 PHP_FUNCTION(mb_strpos)
2271 {
2272 int n, reverse = 0;
2273 zend_long offset = 0, slen;
2274 mbfl_string haystack, needle;
2275 char *enc_name = NULL;
2276 size_t enc_name_len, haystack_len, needle_len;
2277
2278 mbfl_string_init(&haystack);
2279 mbfl_string_init(&needle);
2280 haystack.no_language = MBSTRG(language);
2281 haystack.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2282 needle.no_language = MBSTRG(language);
2283 needle.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2284
2285 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|ls", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &offset, &enc_name, &enc_name_len) == FAILURE) {
2286 return;
2287 }
2288
2289 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2290 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2291 return;
2292 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2293 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2294 return;
2295 }
2296
2297 haystack.len = (uint32_t)haystack_len;
2298 needle.len = (uint32_t)needle_len;
2299
2300 if (enc_name != NULL) {
2301 haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2302 if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2303 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2304 RETURN_FALSE;
2305 }
2306 }
2307
2308 slen = mbfl_strlen(&haystack);
2309 if (offset < 0) {
2310 offset += slen;
2311 }
2312 if (offset < 0 || offset > slen) {
2313 php_error_docref(NULL, E_WARNING, "Offset not contained in string");
2314 RETURN_FALSE;
2315 }
2316 if (needle.len == 0) {
2317 php_error_docref(NULL, E_WARNING, "Empty delimiter");
2318 RETURN_FALSE;
2319 }
2320
2321 n = mbfl_strpos(&haystack, &needle, offset, reverse);
2322 if (n >= 0) {
2323 RETVAL_LONG(n);
2324 } else {
2325 switch (-n) {
2326 case 1:
2327 break;
2328 case 2:
2329 php_error_docref(NULL, E_WARNING, "Needle has not positive length");
2330 break;
2331 case 4:
2332 php_error_docref(NULL, E_WARNING, "Unknown encoding or conversion error");
2333 break;
2334 case 8:
2335 php_error_docref(NULL, E_NOTICE, "Argument is empty");
2336 break;
2337 default:
2338 php_error_docref(NULL, E_WARNING, "Unknown error in mb_strpos");
2339 break;
2340 }
2341 RETVAL_FALSE;
2342 }
2343 }
2344 /* }}} */
2345
2346 /* {{{ proto int mb_strrpos(string haystack, string needle [, int offset [, string encoding]])
2347 Find position of last occurrence of a string within another */
PHP_FUNCTION(mb_strrpos)2348 PHP_FUNCTION(mb_strrpos)
2349 {
2350 int n;
2351 mbfl_string haystack, needle;
2352 char *enc_name = NULL;
2353 size_t enc_name_len, haystack_len, needle_len;
2354 zval *zoffset = NULL;
2355 long offset = 0, str_flg;
2356 char *enc_name2 = NULL;
2357 int enc_name_len2;
2358
2359 mbfl_string_init(&haystack);
2360 mbfl_string_init(&needle);
2361 haystack.no_language = MBSTRG(language);
2362 haystack.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2363 needle.no_language = MBSTRG(language);
2364 needle.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2365
2366 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|zs", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &zoffset, &enc_name, &enc_name_len) == FAILURE) {
2367 return;
2368 }
2369
2370 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2371 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2372 return;
2373 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2374 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2375 return;
2376 }
2377
2378 haystack.len = (uint32_t)haystack_len;
2379 needle.len = (uint32_t)needle_len;
2380
2381 if (zoffset) {
2382 if (Z_TYPE_P(zoffset) == IS_STRING) {
2383 enc_name2 = Z_STRVAL_P(zoffset);
2384 enc_name_len2 = Z_STRLEN_P(zoffset);
2385 str_flg = 1;
2386
2387 if (enc_name2 != NULL) {
2388 switch (*enc_name2) {
2389 case '0':
2390 case '1':
2391 case '2':
2392 case '3':
2393 case '4':
2394 case '5':
2395 case '6':
2396 case '7':
2397 case '8':
2398 case '9':
2399 case ' ':
2400 case '-':
2401 case '.':
2402 break;
2403 default :
2404 str_flg = 0;
2405 break;
2406 }
2407 }
2408
2409 if (str_flg) {
2410 convert_to_long_ex(zoffset);
2411 offset = Z_LVAL_P(zoffset);
2412 } else {
2413 enc_name = enc_name2;
2414 enc_name_len = enc_name_len2;
2415 }
2416 } else {
2417 convert_to_long_ex(zoffset);
2418 offset = Z_LVAL_P(zoffset);
2419 }
2420 }
2421
2422 if (enc_name != NULL) {
2423 haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2424 if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2425 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2426 RETURN_FALSE;
2427 }
2428 }
2429
2430 if (haystack.len <= 0) {
2431 RETURN_FALSE;
2432 }
2433 if (needle.len <= 0) {
2434 RETURN_FALSE;
2435 }
2436
2437 {
2438 int haystack_char_len = mbfl_strlen(&haystack);
2439 if ((offset > 0 && offset > haystack_char_len) ||
2440 (offset < 0 && -offset > haystack_char_len)) {
2441 php_error_docref(NULL, E_WARNING, "Offset is greater than the length of haystack string");
2442 RETURN_FALSE;
2443 }
2444 }
2445
2446 n = mbfl_strpos(&haystack, &needle, offset, 1);
2447 if (n >= 0) {
2448 RETVAL_LONG(n);
2449 } else {
2450 RETVAL_FALSE;
2451 }
2452 }
2453 /* }}} */
2454
2455 /* {{{ proto int mb_stripos(string haystack, string needle [, int offset [, string encoding]])
2456 Finds position of first occurrence of a string within another, case insensitive */
PHP_FUNCTION(mb_stripos)2457 PHP_FUNCTION(mb_stripos)
2458 {
2459 int n = -1;
2460 zend_long offset = 0;
2461 mbfl_string haystack, needle;
2462 const char *from_encoding = MBSTRG(current_internal_encoding)->mime_name;
2463 size_t from_encoding_len, haystack_len, needle_len;
2464
2465 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|ls", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &offset, &from_encoding, &from_encoding_len) == FAILURE) {
2466 return;
2467 }
2468
2469 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2470 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2471 return;
2472 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2473 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2474 return;
2475 }
2476
2477 haystack.len = (uint32_t)haystack_len;
2478 needle.len = (uint32_t)needle_len;
2479
2480 if (needle.len == 0) {
2481 php_error_docref(NULL, E_WARNING, "Empty delimiter");
2482 RETURN_FALSE;
2483 }
2484 n = php_mb_stripos(0, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, offset, from_encoding);
2485
2486 if (n >= 0) {
2487 RETVAL_LONG(n);
2488 } else {
2489 RETVAL_FALSE;
2490 }
2491 }
2492 /* }}} */
2493
2494 /* {{{ proto int mb_strripos(string haystack, string needle [, int offset [, string encoding]])
2495 Finds position of last occurrence of a string within another, case insensitive */
PHP_FUNCTION(mb_strripos)2496 PHP_FUNCTION(mb_strripos)
2497 {
2498 int n = -1;
2499 zend_long offset = 0;
2500 mbfl_string haystack, needle;
2501 const char *from_encoding = MBSTRG(current_internal_encoding)->mime_name;
2502 size_t from_encoding_len, haystack_len, needle_len;
2503
2504 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|ls", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &offset, &from_encoding, &from_encoding_len) == FAILURE) {
2505 return;
2506 }
2507
2508 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2509 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2510 return;
2511 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2512 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2513 return;
2514 }
2515
2516 haystack.len = (uint32_t)haystack_len;
2517 needle.len = (uint32_t)needle_len;
2518
2519 n = php_mb_stripos(1, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, offset, from_encoding);
2520
2521 if (n >= 0) {
2522 RETVAL_LONG(n);
2523 } else {
2524 RETVAL_FALSE;
2525 }
2526 }
2527 /* }}} */
2528
2529 /* {{{ proto string mb_strstr(string haystack, string needle[, bool part[, string encoding]])
2530 Finds first occurrence of a string within another */
PHP_FUNCTION(mb_strstr)2531 PHP_FUNCTION(mb_strstr)
2532 {
2533 int n, len, mblen;
2534 mbfl_string haystack, needle, result, *ret = NULL;
2535 char *enc_name = NULL;
2536 size_t enc_name_len, haystack_len, needle_len;
2537 zend_bool part = 0;
2538
2539 mbfl_string_init(&haystack);
2540 mbfl_string_init(&needle);
2541 haystack.no_language = MBSTRG(language);
2542 haystack.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2543 needle.no_language = MBSTRG(language);
2544 needle.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2545
2546 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|bs", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &part, &enc_name, &enc_name_len) == FAILURE) {
2547 return;
2548 }
2549
2550 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2551 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2552 return;
2553 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2554 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2555 return;
2556 }
2557
2558 haystack.len = (uint32_t)haystack_len;
2559 needle.len = (uint32_t)needle_len;
2560
2561 if (enc_name != NULL) {
2562 haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2563 if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2564 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2565 RETURN_FALSE;
2566 }
2567 }
2568
2569 if (needle.len <= 0) {
2570 php_error_docref(NULL, E_WARNING, "Empty delimiter");
2571 RETURN_FALSE;
2572 }
2573 n = mbfl_strpos(&haystack, &needle, 0, 0);
2574 if (n >= 0) {
2575 mblen = mbfl_strlen(&haystack);
2576 if (part) {
2577 ret = mbfl_substr(&haystack, &result, 0, n);
2578 if (ret != NULL) {
2579 // TODO: avoid reallocation ???
2580 RETVAL_STRINGL((char *)ret->val, ret->len);
2581 efree(ret->val);
2582 } else {
2583 RETVAL_FALSE;
2584 }
2585 } else {
2586 len = (mblen - n);
2587 ret = mbfl_substr(&haystack, &result, n, len);
2588 if (ret != NULL) {
2589 // TODO: avoid reallocation ???
2590 RETVAL_STRINGL((char *)ret->val, ret->len);
2591 efree(ret->val);
2592 } else {
2593 RETVAL_FALSE;
2594 }
2595 }
2596 } else {
2597 RETVAL_FALSE;
2598 }
2599 }
2600 /* }}} */
2601
2602 /* {{{ proto string mb_strrchr(string haystack, string needle[, bool part[, string encoding]])
2603 Finds the last occurrence of a character in a string within another */
PHP_FUNCTION(mb_strrchr)2604 PHP_FUNCTION(mb_strrchr)
2605 {
2606 int n, len, mblen;
2607 mbfl_string haystack, needle, result, *ret = NULL;
2608 char *enc_name = NULL;
2609 size_t enc_name_len, haystack_len, needle_len;
2610 zend_bool part = 0;
2611
2612 mbfl_string_init(&haystack);
2613 mbfl_string_init(&needle);
2614 haystack.no_language = MBSTRG(language);
2615 haystack.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2616 needle.no_language = MBSTRG(language);
2617 needle.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2618
2619 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|bs", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &part, &enc_name, &enc_name_len) == FAILURE) {
2620 return;
2621 }
2622
2623 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2624 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2625 return;
2626 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2627 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2628 return;
2629 }
2630
2631 haystack.len = (uint32_t)haystack_len;
2632 needle.len = (uint32_t)needle_len;
2633
2634 if (enc_name != NULL) {
2635 haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2636 if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2637 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2638 RETURN_FALSE;
2639 }
2640 }
2641
2642 if (haystack.len <= 0) {
2643 RETURN_FALSE;
2644 }
2645 if (needle.len <= 0) {
2646 RETURN_FALSE;
2647 }
2648 n = mbfl_strpos(&haystack, &needle, 0, 1);
2649 if (n >= 0) {
2650 mblen = mbfl_strlen(&haystack);
2651 if (part) {
2652 ret = mbfl_substr(&haystack, &result, 0, n);
2653 if (ret != NULL) {
2654 // TODO: avoid reallocation ???
2655 RETVAL_STRINGL((char *)ret->val, ret->len);
2656 efree(ret->val);
2657 } else {
2658 RETVAL_FALSE;
2659 }
2660 } else {
2661 len = (mblen - n);
2662 ret = mbfl_substr(&haystack, &result, n, len);
2663 if (ret != NULL) {
2664 // TODO: avoid reallocation ???
2665 RETVAL_STRINGL((char *)ret->val, ret->len);
2666 efree(ret->val);
2667 } else {
2668 RETVAL_FALSE;
2669 }
2670 }
2671 } else {
2672 RETVAL_FALSE;
2673 }
2674 }
2675 /* }}} */
2676
2677 /* {{{ proto string mb_stristr(string haystack, string needle[, bool part[, string encoding]])
2678 Finds first occurrence of a string within another, case insensitive */
PHP_FUNCTION(mb_stristr)2679 PHP_FUNCTION(mb_stristr)
2680 {
2681 zend_bool part = 0;
2682 size_t from_encoding_len, len, mblen, haystack_len, needle_len;
2683 int n;
2684 mbfl_string haystack, needle, result, *ret = NULL;
2685 const char *from_encoding = MBSTRG(current_internal_encoding)->mime_name;
2686 mbfl_string_init(&haystack);
2687 mbfl_string_init(&needle);
2688 haystack.no_language = MBSTRG(language);
2689 haystack.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2690 needle.no_language = MBSTRG(language);
2691 needle.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2692
2693
2694 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|bs", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &part, &from_encoding, &from_encoding_len) == FAILURE) {
2695 return;
2696 }
2697
2698 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2699 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2700 return;
2701 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2702 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2703 return;
2704 }
2705
2706 haystack.len = (uint32_t)haystack_len;
2707 needle.len = (uint32_t)needle_len;
2708
2709 if (!needle.len) {
2710 php_error_docref(NULL, E_WARNING, "Empty delimiter");
2711 RETURN_FALSE;
2712 }
2713
2714 haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(from_encoding);
2715 if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2716 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", from_encoding);
2717 RETURN_FALSE;
2718 }
2719
2720 n = php_mb_stripos(0, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, 0, from_encoding);
2721
2722 if (n <0) {
2723 RETURN_FALSE;
2724 }
2725
2726 mblen = mbfl_strlen(&haystack);
2727
2728 if (part) {
2729 ret = mbfl_substr(&haystack, &result, 0, n);
2730 if (ret != NULL) {
2731 // TODO: avoid reallocation ???
2732 RETVAL_STRINGL((char *)ret->val, ret->len);
2733 efree(ret->val);
2734 } else {
2735 RETVAL_FALSE;
2736 }
2737 } else {
2738 len = (mblen - n);
2739 ret = mbfl_substr(&haystack, &result, n, len);
2740 if (ret != NULL) {
2741 // TODO: avoid reallocaton ???
2742 RETVAL_STRINGL((char *)ret->val, ret->len);
2743 efree(ret->val);
2744 } else {
2745 RETVAL_FALSE;
2746 }
2747 }
2748 }
2749 /* }}} */
2750
2751 /* {{{ proto string mb_strrichr(string haystack, string needle[, bool part[, string encoding]])
2752 Finds the last occurrence of a character in a string within another, case insensitive */
PHP_FUNCTION(mb_strrichr)2753 PHP_FUNCTION(mb_strrichr)
2754 {
2755 zend_bool part = 0;
2756 int n, len, mblen;
2757 size_t from_encoding_len, haystack_len, needle_len;
2758 mbfl_string haystack, needle, result, *ret = NULL;
2759 const char *from_encoding = MBSTRG(current_internal_encoding)->name;
2760 mbfl_string_init(&haystack);
2761 mbfl_string_init(&needle);
2762 haystack.no_language = MBSTRG(language);
2763 haystack.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2764 needle.no_language = MBSTRG(language);
2765 needle.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2766
2767
2768 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|bs", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &part, &from_encoding, &from_encoding_len) == FAILURE) {
2769 return;
2770 }
2771
2772 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2773 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2774 return;
2775 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2776 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2777 return;
2778 }
2779
2780 haystack.len = (uint32_t)haystack_len;
2781 needle.len = (uint32_t)needle_len;
2782
2783 haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(from_encoding);
2784 if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2785 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", from_encoding);
2786 RETURN_FALSE;
2787 }
2788
2789 n = php_mb_stripos(1, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, 0, from_encoding);
2790
2791 if (n <0) {
2792 RETURN_FALSE;
2793 }
2794
2795 mblen = mbfl_strlen(&haystack);
2796
2797 if (part) {
2798 ret = mbfl_substr(&haystack, &result, 0, n);
2799 if (ret != NULL) {
2800 // TODO: avoid reallocation ???
2801 RETVAL_STRINGL((char *)ret->val, ret->len);
2802 efree(ret->val);
2803 } else {
2804 RETVAL_FALSE;
2805 }
2806 } else {
2807 len = (mblen - n);
2808 ret = mbfl_substr(&haystack, &result, n, len);
2809 if (ret != NULL) {
2810 // TODO: avoid reallocation ???
2811 RETVAL_STRINGL((char *)ret->val, ret->len);
2812 efree(ret->val);
2813 } else {
2814 RETVAL_FALSE;
2815 }
2816 }
2817 }
2818 /* }}} */
2819
2820 /* {{{ proto int mb_substr_count(string haystack, string needle [, string encoding])
2821 Count the number of substring occurrences */
PHP_FUNCTION(mb_substr_count)2822 PHP_FUNCTION(mb_substr_count)
2823 {
2824 int n;
2825 mbfl_string haystack, needle;
2826 char *enc_name = NULL;
2827 size_t enc_name_len, haystack_len, needle_len;
2828
2829 mbfl_string_init(&haystack);
2830 mbfl_string_init(&needle);
2831 haystack.no_language = MBSTRG(language);
2832 haystack.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2833 needle.no_language = MBSTRG(language);
2834 needle.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2835
2836 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|s", (char **)&haystack.val, &haystack_len, (char **)&needle.val, &needle_len, &enc_name, &enc_name_len) == FAILURE) {
2837 return;
2838 }
2839
2840 if (ZEND_SIZE_T_UINT_OVFL(haystack_len)) {
2841 php_error_docref(NULL, E_WARNING, "Haystack length overflows the max allowed length of %u", UINT_MAX);
2842 return;
2843 } else if (ZEND_SIZE_T_UINT_OVFL(needle_len)) {
2844 php_error_docref(NULL, E_WARNING, "Needle length overflows the max allowed length of %u", UINT_MAX);
2845 return;
2846 }
2847
2848 haystack.len = (uint32_t)haystack_len;
2849 needle.len = (uint32_t)needle_len;
2850
2851 if (enc_name != NULL) {
2852 haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2853 if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2854 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2855 RETURN_FALSE;
2856 }
2857 }
2858
2859 if (needle.len <= 0) {
2860 php_error_docref(NULL, E_WARNING, "Empty substring");
2861 RETURN_FALSE;
2862 }
2863
2864 n = mbfl_substr_count(&haystack, &needle);
2865 if (n >= 0) {
2866 RETVAL_LONG(n);
2867 } else {
2868 RETVAL_FALSE;
2869 }
2870 }
2871 /* }}} */
2872
2873 /* {{{ proto string mb_substr(string str, int start [, int length [, string encoding]])
2874 Returns part of a string */
PHP_FUNCTION(mb_substr)2875 PHP_FUNCTION(mb_substr)
2876 {
2877 char *str, *encoding = NULL;
2878 zend_long from, len;
2879 int mblen;
2880 size_t str_len, encoding_len;
2881 zend_bool len_is_null = 1;
2882 mbfl_string string, result, *ret;
2883
2884 if (zend_parse_parameters(ZEND_NUM_ARGS(), "sl|l!s", &str, &str_len, &from, &len, &len_is_null, &encoding, &encoding_len) == FAILURE) {
2885 return;
2886 }
2887
2888 mbfl_string_init(&string);
2889 string.no_language = MBSTRG(language);
2890 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2891
2892 if (encoding) {
2893 string.no_encoding = mbfl_name2no_encoding(encoding);
2894 if (string.no_encoding == mbfl_no_encoding_invalid) {
2895 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", encoding);
2896 RETURN_FALSE;
2897 }
2898 }
2899
2900 string.val = (unsigned char *)str;
2901 string.len = str_len;
2902
2903 if (len_is_null) {
2904 len = str_len;
2905 }
2906
2907 /* measures length */
2908 mblen = 0;
2909 if (from < 0 || len < 0) {
2910 mblen = mbfl_strlen(&string);
2911 }
2912
2913 /* if "from" position is negative, count start position from the end
2914 * of the string
2915 */
2916 if (from < 0) {
2917 from = mblen + from;
2918 if (from < 0) {
2919 from = 0;
2920 }
2921 }
2922
2923 /* if "length" position is negative, set it to the length
2924 * needed to stop that many chars from the end of the string
2925 */
2926 if (len < 0) {
2927 len = (mblen - from) + len;
2928 if (len < 0) {
2929 len = 0;
2930 }
2931 }
2932
2933 if (((MBSTRG(func_overload) & MB_OVERLOAD_STRING) == MB_OVERLOAD_STRING)
2934 && (from >= mbfl_strlen(&string))) {
2935 RETURN_FALSE;
2936 }
2937
2938 if (from > INT_MAX) {
2939 from = INT_MAX;
2940 }
2941 if (len > INT_MAX) {
2942 len = INT_MAX;
2943 }
2944
2945 ret = mbfl_substr(&string, &result, from, len);
2946 if (NULL == ret) {
2947 RETURN_FALSE;
2948 }
2949
2950 // TODO: avoid reallocation ???
2951 RETVAL_STRINGL((char *)ret->val, ret->len); /* the string is already strdup()'ed */
2952 efree(ret->val);
2953 }
2954 /* }}} */
2955
2956 /* {{{ proto string mb_strcut(string str, int start [, int length [, string encoding]])
2957 Returns part of a string */
PHP_FUNCTION(mb_strcut)2958 PHP_FUNCTION(mb_strcut)
2959 {
2960 char *encoding = NULL;
2961 zend_long from, len;
2962 size_t encoding_len, string_len;
2963 zend_bool len_is_null = 1;
2964 mbfl_string string, result, *ret;
2965
2966 mbfl_string_init(&string);
2967 string.no_language = MBSTRG(language);
2968 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
2969
2970 if (zend_parse_parameters(ZEND_NUM_ARGS(), "sl|l!s", (char **)&string.val, &string_len, &from, &len, &len_is_null, &encoding, &encoding_len) == FAILURE) {
2971 return;
2972 }
2973
2974 if (ZEND_SIZE_T_UINT_OVFL(string_len)) {
2975 php_error_docref(NULL, E_WARNING, "String length overflows the max allowed length of %u", UINT_MAX);
2976 return;
2977 }
2978
2979 string.len = (uint32_t)string_len;
2980
2981 if (encoding) {
2982 string.no_encoding = mbfl_name2no_encoding(encoding);
2983 if (string.no_encoding == mbfl_no_encoding_invalid) {
2984 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", encoding);
2985 RETURN_FALSE;
2986 }
2987 }
2988
2989 if (len_is_null) {
2990 len = string.len;
2991 }
2992
2993 /* if "from" position is negative, count start position from the end
2994 * of the string
2995 */
2996 if (from < 0) {
2997 from = string.len + from;
2998 if (from < 0) {
2999 from = 0;
3000 }
3001 }
3002
3003 /* if "length" position is negative, set it to the length
3004 * needed to stop that many chars from the end of the string
3005 */
3006 if (len < 0) {
3007 len = (string.len - from) + len;
3008 if (len < 0) {
3009 len = 0;
3010 }
3011 }
3012
3013 if ((unsigned int)from > string.len) {
3014 RETURN_FALSE;
3015 }
3016
3017 ret = mbfl_strcut(&string, &result, from, len);
3018 if (ret == NULL) {
3019 RETURN_FALSE;
3020 }
3021
3022 // TODO: avoid reallocation ???
3023 RETVAL_STRINGL((char *)ret->val, ret->len); /* the string is already strdup()'ed */
3024 efree(ret->val);
3025 }
3026 /* }}} */
3027
3028 /* {{{ proto int mb_strwidth(string str [, string encoding])
3029 Gets terminal width of a string */
PHP_FUNCTION(mb_strwidth)3030 PHP_FUNCTION(mb_strwidth)
3031 {
3032 int n;
3033 mbfl_string string;
3034 char *enc_name = NULL;
3035 size_t enc_name_len, string_len;
3036
3037 mbfl_string_init(&string);
3038
3039 string.no_language = MBSTRG(language);
3040 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
3041
3042 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|s", (char **)&string.val, &string_len, &enc_name, &enc_name_len) == FAILURE) {
3043 return;
3044 }
3045
3046 if (ZEND_SIZE_T_UINT_OVFL(string_len)) {
3047 php_error_docref(NULL, E_WARNING, "String length overflows the max allowed length of %u", UINT_MAX);
3048 return;
3049 }
3050
3051 string.len = (uint32_t)string_len;
3052
3053 if (enc_name != NULL) {
3054 string.no_encoding = mbfl_name2no_encoding(enc_name);
3055 if (string.no_encoding == mbfl_no_encoding_invalid) {
3056 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", enc_name);
3057 RETURN_FALSE;
3058 }
3059 }
3060
3061 n = mbfl_strwidth(&string);
3062 if (n >= 0) {
3063 RETVAL_LONG(n);
3064 } else {
3065 RETVAL_FALSE;
3066 }
3067 }
3068 /* }}} */
3069
3070 /* {{{ proto string mb_strimwidth(string str, int start, int width [, string trimmarker [, string encoding]])
3071 Trim the string in terminal width */
PHP_FUNCTION(mb_strimwidth)3072 PHP_FUNCTION(mb_strimwidth)
3073 {
3074 char *str, *trimmarker = NULL, *encoding = NULL;
3075 zend_long from, width, swidth;
3076 size_t str_len, trimmarker_len, encoding_len;
3077 mbfl_string string, result, marker, *ret;
3078
3079 if (zend_parse_parameters(ZEND_NUM_ARGS(), "sll|ss", &str, &str_len, &from, &width, &trimmarker, &trimmarker_len, &encoding, &encoding_len) == FAILURE) {
3080 return;
3081 }
3082
3083 mbfl_string_init(&string);
3084 mbfl_string_init(&marker);
3085 string.no_language = MBSTRG(language);
3086 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
3087 marker.no_language = MBSTRG(language);
3088 marker.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
3089 marker.val = NULL;
3090 marker.len = 0;
3091
3092 if (encoding) {
3093 string.no_encoding = marker.no_encoding = mbfl_name2no_encoding(encoding);
3094 if (string.no_encoding == mbfl_no_encoding_invalid) {
3095 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", encoding);
3096 RETURN_FALSE;
3097 }
3098 }
3099
3100 string.val = (unsigned char *)str;
3101 string.len = str_len;
3102
3103 if ((from < 0) || (width < 0)) {
3104 swidth = mbfl_strwidth(&string);
3105 }
3106
3107 if (from < 0) {
3108 from += swidth;
3109 }
3110
3111 if (from < 0 || (size_t)from > str_len) {
3112 php_error_docref(NULL, E_WARNING, "Start position is out of range");
3113 RETURN_FALSE;
3114 }
3115
3116 if (width < 0) {
3117 width = swidth + width - from;
3118 }
3119
3120 if (width < 0) {
3121 php_error_docref(NULL, E_WARNING, "Width is out of range");
3122 RETURN_FALSE;
3123 }
3124
3125 if (trimmarker) {
3126 marker.val = (unsigned char *)trimmarker;
3127 marker.len = trimmarker_len;
3128 }
3129
3130 ret = mbfl_strimwidth(&string, &marker, &result, from, width);
3131
3132 if (ret == NULL) {
3133 RETURN_FALSE;
3134 }
3135 // TODO: avoid reallocation ???
3136 RETVAL_STRINGL((char *)ret->val, ret->len); /* the string is already strdup()'ed */
3137 efree(ret->val);
3138 }
3139 /* }}} */
3140
3141 /* {{{ MBSTRING_API char *php_mb_convert_encoding() */
php_mb_convert_encoding(const char * input,size_t length,const char * _to_encoding,const char * _from_encodings,size_t * output_len)3142 MBSTRING_API char * php_mb_convert_encoding(const char *input, size_t length, const char *_to_encoding, const char *_from_encodings, size_t *output_len)
3143 {
3144 mbfl_string string, result, *ret;
3145 const mbfl_encoding *from_encoding, *to_encoding;
3146 mbfl_buffer_converter *convd;
3147 size_t size;
3148 const mbfl_encoding **list;
3149 char *output=NULL;
3150
3151 if (output_len) {
3152 *output_len = 0;
3153 }
3154 if (!input) {
3155 return NULL;
3156 }
3157 /* new encoding */
3158 if (_to_encoding && strlen(_to_encoding)) {
3159 to_encoding = mbfl_name2encoding(_to_encoding);
3160 if (!to_encoding) {
3161 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", _to_encoding);
3162 return NULL;
3163 }
3164 } else {
3165 to_encoding = MBSTRG(current_internal_encoding);
3166 }
3167
3168 /* initialize string */
3169 mbfl_string_init(&string);
3170 mbfl_string_init(&result);
3171 from_encoding = MBSTRG(current_internal_encoding);
3172 string.no_encoding = from_encoding->no_encoding;
3173 string.no_language = MBSTRG(language);
3174 string.val = (unsigned char *)input;
3175 string.len = length;
3176
3177 /* pre-conversion encoding */
3178 if (_from_encodings) {
3179 list = NULL;
3180 size = 0;
3181 php_mb_parse_encoding_list(_from_encodings, strlen(_from_encodings), &list, &size, 0);
3182 if (size == 1) {
3183 from_encoding = *list;
3184 string.no_encoding = from_encoding->no_encoding;
3185 } else if (size > 1) {
3186 /* auto detect */
3187 from_encoding = mbfl_identify_encoding2(&string, list, size, MBSTRG(strict_detection));
3188 if (from_encoding) {
3189 string.no_encoding = from_encoding->no_encoding;
3190 } else {
3191 php_error_docref(NULL, E_WARNING, "Unable to detect character encoding");
3192 from_encoding = &mbfl_encoding_pass;
3193 to_encoding = from_encoding;
3194 string.no_encoding = from_encoding->no_encoding;
3195 }
3196 } else {
3197 php_error_docref(NULL, E_WARNING, "Illegal character encoding specified");
3198 }
3199 if (list != NULL) {
3200 efree((void *)list);
3201 }
3202 }
3203
3204 /* initialize converter */
3205 convd = mbfl_buffer_converter_new2(from_encoding, to_encoding, string.len);
3206 if (convd == NULL) {
3207 php_error_docref(NULL, E_WARNING, "Unable to create character encoding converter");
3208 return NULL;
3209 }
3210 mbfl_buffer_converter_illegal_mode(convd, MBSTRG(current_filter_illegal_mode));
3211 mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
3212
3213 /* do it */
3214 ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
3215 if (ret) {
3216 if (output_len) {
3217 *output_len = ret->len;
3218 }
3219 output = (char *)ret->val;
3220 }
3221
3222 MBSTRG(illegalchars) += mbfl_buffer_illegalchars(convd);
3223 mbfl_buffer_converter_delete(convd);
3224 return output;
3225 }
3226 /* }}} */
3227
3228 /* {{{ proto string mb_convert_encoding(string str, string to-encoding [, mixed from-encoding])
3229 Returns converted string in desired encoding */
PHP_FUNCTION(mb_convert_encoding)3230 PHP_FUNCTION(mb_convert_encoding)
3231 {
3232 char *arg_str, *arg_new;
3233 size_t str_len, new_len;
3234 zval *arg_old = NULL;
3235 size_t size, l, n;
3236 char *_from_encodings = NULL, *ret, *s_free = NULL;
3237
3238 zval *hash_entry;
3239 HashTable *target_hash;
3240
3241 if (zend_parse_parameters(ZEND_NUM_ARGS(), "ss|z", &arg_str, &str_len, &arg_new, &new_len, &arg_old) == FAILURE) {
3242 return;
3243 }
3244
3245 if (arg_old) {
3246 switch (Z_TYPE_P(arg_old)) {
3247 case IS_ARRAY:
3248 target_hash = Z_ARRVAL_P(arg_old);
3249 _from_encodings = NULL;
3250
3251 ZEND_HASH_FOREACH_VAL(target_hash, hash_entry) {
3252
3253 convert_to_string_ex(hash_entry);
3254
3255 if ( _from_encodings) {
3256 l = strlen(_from_encodings);
3257 n = strlen(Z_STRVAL_P(hash_entry));
3258 _from_encodings = erealloc(_from_encodings, l+n+2);
3259 memcpy(_from_encodings + l, ",", 1);
3260 memcpy(_from_encodings + l + 1, Z_STRVAL_P(hash_entry), Z_STRLEN_P(hash_entry) + 1);
3261 } else {
3262 _from_encodings = estrdup(Z_STRVAL_P(hash_entry));
3263 }
3264 } ZEND_HASH_FOREACH_END();
3265
3266 if (_from_encodings != NULL && !strlen(_from_encodings)) {
3267 efree(_from_encodings);
3268 _from_encodings = NULL;
3269 }
3270 s_free = _from_encodings;
3271 break;
3272 default:
3273 convert_to_string(arg_old);
3274 _from_encodings = Z_STRVAL_P(arg_old);
3275 break;
3276 }
3277 }
3278
3279 /* new encoding */
3280 ret = php_mb_convert_encoding(arg_str, str_len, arg_new, _from_encodings, &size);
3281 if (ret != NULL) {
3282 // TODO: avoid reallocation ???
3283 RETVAL_STRINGL(ret, size); /* the string is already strdup()'ed */
3284 efree(ret);
3285 } else {
3286 RETVAL_FALSE;
3287 }
3288
3289 if ( s_free) {
3290 efree(s_free);
3291 }
3292 }
3293 /* }}} */
3294
3295 /* {{{ proto string mb_convert_case(string sourcestring, int mode [, string encoding])
3296 Returns a case-folded version of sourcestring */
PHP_FUNCTION(mb_convert_case)3297 PHP_FUNCTION(mb_convert_case)
3298 {
3299 const char *from_encoding = MBSTRG(current_internal_encoding)->mime_name;
3300 char *str;
3301 size_t str_len, from_encoding_len;
3302 zend_long case_mode = 0;
3303 char *newstr;
3304 size_t ret_len;
3305
3306 RETVAL_FALSE;
3307 if (zend_parse_parameters(ZEND_NUM_ARGS(), "sl|s!", &str, &str_len,
3308 &case_mode, &from_encoding, &from_encoding_len) == FAILURE) {
3309 return;
3310 }
3311
3312 newstr = php_unicode_convert_case(case_mode, str, (size_t) str_len, &ret_len, from_encoding);
3313
3314 if (newstr) {
3315 // TODO: avoid reallocation ???
3316 RETVAL_STRINGL(newstr, ret_len);
3317 efree(newstr);
3318 }
3319 }
3320 /* }}} */
3321
3322 /* {{{ proto string mb_strtoupper(string sourcestring [, string encoding])
3323 * Returns a uppercased version of sourcestring
3324 */
PHP_FUNCTION(mb_strtoupper)3325 PHP_FUNCTION(mb_strtoupper)
3326 {
3327 const char *from_encoding = MBSTRG(current_internal_encoding)->mime_name;
3328 char *str;
3329 size_t str_len, from_encoding_len;
3330 char *newstr;
3331 size_t ret_len;
3332
3333 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|s!", &str, &str_len,
3334 &from_encoding, &from_encoding_len) == FAILURE) {
3335 return;
3336 }
3337 newstr = php_unicode_convert_case(PHP_UNICODE_CASE_UPPER, str, (size_t) str_len, &ret_len, from_encoding);
3338
3339 if (newstr) {
3340 // TODO: avoid reallocation ???
3341 RETVAL_STRINGL(newstr, ret_len);
3342 efree(newstr);
3343 return;
3344 }
3345 RETURN_FALSE;
3346 }
3347 /* }}} */
3348
3349 /* {{{ proto string mb_strtolower(string sourcestring [, string encoding])
3350 * Returns a lowercased version of sourcestring
3351 */
PHP_FUNCTION(mb_strtolower)3352 PHP_FUNCTION(mb_strtolower)
3353 {
3354 const char *from_encoding = MBSTRG(current_internal_encoding)->mime_name;
3355 char *str;
3356 size_t str_len, from_encoding_len;
3357 char *newstr;
3358 size_t ret_len;
3359
3360 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|s!", &str, &str_len,
3361 &from_encoding, &from_encoding_len) == FAILURE) {
3362 return;
3363 }
3364 newstr = php_unicode_convert_case(PHP_UNICODE_CASE_LOWER, str, (size_t) str_len, &ret_len, from_encoding);
3365
3366 if (newstr) {
3367 // TODO: avoid reallocation ???
3368 RETVAL_STRINGL(newstr, ret_len);
3369 efree(newstr);
3370 return;
3371 }
3372 RETURN_FALSE;
3373 }
3374 /* }}} */
3375
3376 /* {{{ proto string mb_detect_encoding(string str [, mixed encoding_list [, bool strict]])
3377 Encodings of the given string is returned (as a string) */
PHP_FUNCTION(mb_detect_encoding)3378 PHP_FUNCTION(mb_detect_encoding)
3379 {
3380 char *str;
3381 size_t str_len;
3382 zend_bool strict=0;
3383 zval *encoding_list = NULL;
3384
3385 mbfl_string string;
3386 const mbfl_encoding *ret;
3387 const mbfl_encoding **elist, **list;
3388 size_t size;
3389
3390 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|z!b", &str, &str_len, &encoding_list, &strict) == FAILURE) {
3391 return;
3392 }
3393
3394 /* make encoding list */
3395 list = NULL;
3396 size = 0;
3397 if (encoding_list) {
3398 switch (Z_TYPE_P(encoding_list)) {
3399 case IS_ARRAY:
3400 if (FAILURE == php_mb_parse_encoding_array(encoding_list, &list, &size, 0)) {
3401 if (list) {
3402 efree(list);
3403 list = NULL;
3404 size = 0;
3405 }
3406 }
3407 break;
3408 default:
3409 convert_to_string(encoding_list);
3410 if (FAILURE == php_mb_parse_encoding_list(Z_STRVAL_P(encoding_list), Z_STRLEN_P(encoding_list), &list, &size, 0)) {
3411 if (list) {
3412 efree(list);
3413 list = NULL;
3414 size = 0;
3415 }
3416 }
3417 break;
3418 }
3419 if (size <= 0) {
3420 php_error_docref(NULL, E_WARNING, "Illegal argument");
3421 }
3422 }
3423
3424 if (ZEND_NUM_ARGS() < 3) {
3425 strict = (zend_bool)MBSTRG(strict_detection);
3426 }
3427
3428 if (size > 0 && list != NULL) {
3429 elist = list;
3430 } else {
3431 elist = MBSTRG(current_detect_order_list);
3432 size = MBSTRG(current_detect_order_list_size);
3433 }
3434
3435 mbfl_string_init(&string);
3436 string.no_language = MBSTRG(language);
3437 string.val = (unsigned char *)str;
3438 string.len = str_len;
3439 ret = mbfl_identify_encoding2(&string, elist, size, strict);
3440
3441 if (list != NULL) {
3442 efree((void *)list);
3443 }
3444
3445 if (ret == NULL) {
3446 RETURN_FALSE;
3447 }
3448
3449 RETVAL_STRING((char *)ret->name);
3450 }
3451 /* }}} */
3452
3453 /* {{{ proto mixed mb_list_encodings()
3454 Returns an array of all supported entity encodings */
PHP_FUNCTION(mb_list_encodings)3455 PHP_FUNCTION(mb_list_encodings)
3456 {
3457 const mbfl_encoding **encodings;
3458 const mbfl_encoding *encoding;
3459 int i;
3460
3461 if (zend_parse_parameters_none() == FAILURE) {
3462 return;
3463 }
3464
3465 array_init(return_value);
3466 i = 0;
3467 encodings = mbfl_get_supported_encodings();
3468 while ((encoding = encodings[i++]) != NULL) {
3469 add_next_index_string(return_value, (char *) encoding->name);
3470 }
3471 }
3472 /* }}} */
3473
3474 /* {{{ proto array mb_encoding_aliases(string encoding)
3475 Returns an array of the aliases of a given encoding name */
PHP_FUNCTION(mb_encoding_aliases)3476 PHP_FUNCTION(mb_encoding_aliases)
3477 {
3478 const mbfl_encoding *encoding;
3479 char *name = NULL;
3480 size_t name_len;
3481
3482 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s", &name, &name_len) == FAILURE) {
3483 return;
3484 }
3485
3486 encoding = mbfl_name2encoding(name);
3487 if (!encoding) {
3488 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", name);
3489 RETURN_FALSE;
3490 }
3491
3492 array_init(return_value);
3493 if (encoding->aliases != NULL) {
3494 const char **alias;
3495 for (alias = *encoding->aliases; *alias; ++alias) {
3496 add_next_index_string(return_value, (char *)*alias);
3497 }
3498 }
3499 }
3500 /* }}} */
3501
3502 /* {{{ proto string mb_encode_mimeheader(string str [, string charset [, string transfer-encoding [, string linefeed [, int indent]]]])
3503 Converts the string to MIME "encoded-word" in the format of =?charset?(B|Q)?encoded_string?= */
PHP_FUNCTION(mb_encode_mimeheader)3504 PHP_FUNCTION(mb_encode_mimeheader)
3505 {
3506 enum mbfl_no_encoding charset, transenc;
3507 mbfl_string string, result, *ret;
3508 char *charset_name = NULL;
3509 size_t charset_name_len;
3510 char *trans_enc_name = NULL;
3511 size_t trans_enc_name_len;
3512 char *linefeed = "\r\n";
3513 size_t linefeed_len, string_len;
3514 zend_long indent = 0;
3515
3516 mbfl_string_init(&string);
3517 string.no_language = MBSTRG(language);
3518 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
3519
3520 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|sssl", (char **)&string.val, &string_len, &charset_name, &charset_name_len, &trans_enc_name, &trans_enc_name_len, &linefeed, &linefeed_len, &indent) == FAILURE) {
3521 return;
3522 }
3523
3524 if (ZEND_SIZE_T_UINT_OVFL(string_len)) {
3525 php_error_docref(NULL, E_WARNING, "String length overflows the max allowed length of %u", UINT_MAX);
3526 return;
3527 }
3528
3529 string.len = (uint32_t)string_len;
3530
3531 charset = mbfl_no_encoding_pass;
3532 transenc = mbfl_no_encoding_base64;
3533
3534 if (charset_name != NULL) {
3535 charset = mbfl_name2no_encoding(charset_name);
3536 if (charset == mbfl_no_encoding_invalid) {
3537 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", charset_name);
3538 RETURN_FALSE;
3539 }
3540 } else {
3541 const mbfl_language *lang = mbfl_no2language(MBSTRG(language));
3542 if (lang != NULL) {
3543 charset = lang->mail_charset;
3544 transenc = lang->mail_header_encoding;
3545 }
3546 }
3547
3548 if (trans_enc_name != NULL) {
3549 if (*trans_enc_name == 'B' || *trans_enc_name == 'b') {
3550 transenc = mbfl_no_encoding_base64;
3551 } else if (*trans_enc_name == 'Q' || *trans_enc_name == 'q') {
3552 transenc = mbfl_no_encoding_qprint;
3553 }
3554 }
3555
3556 mbfl_string_init(&result);
3557 ret = mbfl_mime_header_encode(&string, &result, charset, transenc, linefeed, indent);
3558 if (ret != NULL) {
3559 // TODO: avoid reallocation ???
3560 RETVAL_STRINGL((char *)ret->val, ret->len); /* the string is already strdup()'ed */
3561 efree(ret->val);
3562 } else {
3563 RETVAL_FALSE;
3564 }
3565 }
3566 /* }}} */
3567
3568 /* {{{ proto string mb_decode_mimeheader(string string)
3569 Decodes the MIME "encoded-word" in the string */
PHP_FUNCTION(mb_decode_mimeheader)3570 PHP_FUNCTION(mb_decode_mimeheader)
3571 {
3572 mbfl_string string, result, *ret;
3573 size_t string_len;
3574
3575 mbfl_string_init(&string);
3576 string.no_language = MBSTRG(language);
3577 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
3578
3579 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s", (char **)&string.val, &string_len) == FAILURE) {
3580 return;
3581 }
3582
3583 if (ZEND_SIZE_T_UINT_OVFL(string_len)) {
3584 php_error_docref(NULL, E_WARNING, "String length overflows the max allowed length of %u", UINT_MAX);
3585 return;
3586 }
3587
3588 string.len = (uint32_t)string_len;
3589
3590 mbfl_string_init(&result);
3591 ret = mbfl_mime_header_decode(&string, &result, MBSTRG(current_internal_encoding)->no_encoding);
3592 if (ret != NULL) {
3593 // TODO: avoid reallocation ???
3594 RETVAL_STRINGL((char *)ret->val, ret->len); /* the string is already strdup()'ed */
3595 efree(ret->val);
3596 } else {
3597 RETVAL_FALSE;
3598 }
3599 }
3600 /* }}} */
3601
3602 /* {{{ proto string mb_convert_kana(string str [, string option] [, string encoding])
3603 Conversion between full-width character and half-width character (Japanese) */
PHP_FUNCTION(mb_convert_kana)3604 PHP_FUNCTION(mb_convert_kana)
3605 {
3606 int opt, i;
3607 mbfl_string string, result, *ret;
3608 char *optstr = NULL;
3609 size_t optstr_len;
3610 char *encname = NULL;
3611 size_t encname_len, string_len;
3612
3613 mbfl_string_init(&string);
3614 string.no_language = MBSTRG(language);
3615 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
3616
3617 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|ss", (char **)&string.val, &string_len, &optstr, &optstr_len, &encname, &encname_len) == FAILURE) {
3618 return;
3619 }
3620
3621 if (ZEND_SIZE_T_UINT_OVFL(string_len)) {
3622 php_error_docref(NULL, E_WARNING, "String length overflows the max allowed length of %u", UINT_MAX);
3623 return;
3624 }
3625
3626 string.len = (uint32_t)string_len;
3627
3628 /* option */
3629 if (optstr != NULL) {
3630 char *p = optstr;
3631 int n = optstr_len;
3632 i = 0;
3633 opt = 0;
3634 while (i < n) {
3635 i++;
3636 switch (*p++) {
3637 case 'A':
3638 opt |= 0x1;
3639 break;
3640 case 'a':
3641 opt |= 0x10;
3642 break;
3643 case 'R':
3644 opt |= 0x2;
3645 break;
3646 case 'r':
3647 opt |= 0x20;
3648 break;
3649 case 'N':
3650 opt |= 0x4;
3651 break;
3652 case 'n':
3653 opt |= 0x40;
3654 break;
3655 case 'S':
3656 opt |= 0x8;
3657 break;
3658 case 's':
3659 opt |= 0x80;
3660 break;
3661 case 'K':
3662 opt |= 0x100;
3663 break;
3664 case 'k':
3665 opt |= 0x1000;
3666 break;
3667 case 'H':
3668 opt |= 0x200;
3669 break;
3670 case 'h':
3671 opt |= 0x2000;
3672 break;
3673 case 'V':
3674 opt |= 0x800;
3675 break;
3676 case 'C':
3677 opt |= 0x10000;
3678 break;
3679 case 'c':
3680 opt |= 0x20000;
3681 break;
3682 case 'M':
3683 opt |= 0x100000;
3684 break;
3685 case 'm':
3686 opt |= 0x200000;
3687 break;
3688 }
3689 }
3690 } else {
3691 opt = 0x900;
3692 }
3693
3694 /* encoding */
3695 if (encname != NULL) {
3696 string.no_encoding = mbfl_name2no_encoding(encname);
3697 if (string.no_encoding == mbfl_no_encoding_invalid) {
3698 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", encname);
3699 RETURN_FALSE;
3700 }
3701 }
3702
3703 ret = mbfl_ja_jp_hantozen(&string, &result, opt);
3704 if (ret != NULL) {
3705 // TODO: avoid reallocation ???
3706 RETVAL_STRINGL((char *)ret->val, ret->len); /* the string is already strdup()'ed */
3707 efree(ret->val);
3708 } else {
3709 RETVAL_FALSE;
3710 }
3711 }
3712 /* }}} */
3713
3714 #define PHP_MBSTR_STACK_BLOCK_SIZE 32
3715
3716 /* {{{ proto string mb_convert_variables(string to-encoding, mixed from-encoding, mixed vars [, ...])
3717 Converts the string resource in variables to desired encoding */
PHP_FUNCTION(mb_convert_variables)3718 PHP_FUNCTION(mb_convert_variables)
3719 {
3720 zval *args, *stack, *var, *hash_entry, *hash_entry_ptr, *zfrom_enc;
3721 HashTable *target_hash;
3722 mbfl_string string, result, *ret;
3723 const mbfl_encoding *from_encoding, *to_encoding;
3724 mbfl_encoding_detector *identd;
3725 mbfl_buffer_converter *convd;
3726 int n, argc, stack_level, stack_max;
3727 size_t to_enc_len;
3728 size_t elistsz;
3729 const mbfl_encoding **elist;
3730 char *to_enc;
3731 void *ptmp;
3732 int recursion_error = 0;
3733
3734 if (zend_parse_parameters(ZEND_NUM_ARGS(), "sz+", &to_enc, &to_enc_len, &zfrom_enc, &args, &argc) == FAILURE) {
3735 return;
3736 }
3737
3738 /* new encoding */
3739 to_encoding = mbfl_name2encoding(to_enc);
3740 if (!to_encoding) {
3741 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", to_enc);
3742 RETURN_FALSE;
3743 }
3744
3745 /* initialize string */
3746 mbfl_string_init(&string);
3747 mbfl_string_init(&result);
3748 from_encoding = MBSTRG(current_internal_encoding);
3749 string.no_encoding = from_encoding->no_encoding;
3750 string.no_language = MBSTRG(language);
3751
3752 /* pre-conversion encoding */
3753 elist = NULL;
3754 elistsz = 0;
3755 switch (Z_TYPE_P(zfrom_enc)) {
3756 case IS_ARRAY:
3757 php_mb_parse_encoding_array(zfrom_enc, &elist, &elistsz, 0);
3758 break;
3759 default:
3760 convert_to_string_ex(zfrom_enc);
3761 php_mb_parse_encoding_list(Z_STRVAL_P(zfrom_enc), Z_STRLEN_P(zfrom_enc), &elist, &elistsz, 0);
3762 break;
3763 }
3764
3765 if (elistsz <= 0) {
3766 from_encoding = &mbfl_encoding_pass;
3767 } else if (elistsz == 1) {
3768 from_encoding = *elist;
3769 } else {
3770 /* auto detect */
3771 from_encoding = NULL;
3772 stack_max = PHP_MBSTR_STACK_BLOCK_SIZE;
3773 stack = (zval *)safe_emalloc(stack_max, sizeof(zval), 0);
3774 stack_level = 0;
3775 identd = mbfl_encoding_detector_new2(elist, elistsz, MBSTRG(strict_detection));
3776 if (identd != NULL) {
3777 n = 0;
3778 while (n < argc || stack_level > 0) {
3779 if (stack_level <= 0) {
3780 var = &args[n++];
3781 ZVAL_DEREF(var);
3782 SEPARATE_ZVAL_NOREF(var);
3783 if (Z_TYPE_P(var) == IS_ARRAY || Z_TYPE_P(var) == IS_OBJECT) {
3784 target_hash = HASH_OF(var);
3785 if (target_hash != NULL) {
3786 zend_hash_internal_pointer_reset(target_hash);
3787 }
3788 }
3789 } else {
3790 stack_level--;
3791 var = &stack[stack_level];
3792 }
3793 if (Z_TYPE_P(var) == IS_ARRAY || Z_TYPE_P(var) == IS_OBJECT) {
3794 target_hash = HASH_OF(var);
3795 if (target_hash != NULL) {
3796 while ((hash_entry = zend_hash_get_current_data(target_hash)) != NULL) {
3797 if (!Z_IMMUTABLE_P(var)) {
3798 if (++target_hash->u.v.nApplyCount > 1) {
3799 --target_hash->u.v.nApplyCount;
3800 recursion_error = 1;
3801 goto detect_end;
3802 }
3803 }
3804 zend_hash_move_forward(target_hash);
3805 if (Z_TYPE_P(hash_entry) == IS_INDIRECT) {
3806 hash_entry = Z_INDIRECT_P(hash_entry);
3807 }
3808 ZVAL_DEREF(hash_entry);
3809 if (Z_TYPE_P(hash_entry) == IS_ARRAY || Z_TYPE_P(hash_entry) == IS_OBJECT) {
3810 if (stack_level >= stack_max) {
3811 stack_max += PHP_MBSTR_STACK_BLOCK_SIZE;
3812 ptmp = erealloc(stack, sizeof(zval) * stack_max);
3813 stack = (zval *)ptmp;
3814 }
3815 ZVAL_COPY_VALUE(&stack[stack_level], var);
3816 stack_level++;
3817 var = hash_entry;
3818 target_hash = HASH_OF(var);
3819 if (target_hash != NULL) {
3820 zend_hash_internal_pointer_reset(target_hash);
3821 continue;
3822 }
3823 } else if (Z_TYPE_P(hash_entry) == IS_STRING) {
3824 string.val = (unsigned char *)Z_STRVAL_P(hash_entry);
3825 string.len = Z_STRLEN_P(hash_entry);
3826 if (mbfl_encoding_detector_feed(identd, &string)) {
3827 goto detect_end; /* complete detecting */
3828 }
3829 }
3830 }
3831 }
3832 } else if (Z_TYPE_P(var) == IS_STRING) {
3833 string.val = (unsigned char *)Z_STRVAL_P(var);
3834 string.len = Z_STRLEN_P(var);
3835 if (mbfl_encoding_detector_feed(identd, &string)) {
3836 goto detect_end; /* complete detecting */
3837 }
3838 }
3839 }
3840 detect_end:
3841 from_encoding = mbfl_encoding_detector_judge2(identd);
3842 mbfl_encoding_detector_delete(identd);
3843 }
3844 if (recursion_error) {
3845 while(stack_level-- && (var = &stack[stack_level])) {
3846 if (!Z_IMMUTABLE_P(var)) {
3847 if (HASH_OF(var)->u.v.nApplyCount > 1) {
3848 HASH_OF(var)->u.v.nApplyCount--;
3849 }
3850 }
3851 }
3852 efree(stack);
3853 if (elist != NULL) {
3854 efree((void *)elist);
3855 }
3856 php_error_docref(NULL, E_WARNING, "Cannot handle recursive references");
3857 RETURN_FALSE;
3858 }
3859 efree(stack);
3860
3861 if (!from_encoding) {
3862 php_error_docref(NULL, E_WARNING, "Unable to detect encoding");
3863 from_encoding = &mbfl_encoding_pass;
3864 }
3865 }
3866 if (elist != NULL) {
3867 efree((void *)elist);
3868 }
3869 /* create converter */
3870 convd = NULL;
3871 if (from_encoding != &mbfl_encoding_pass) {
3872 convd = mbfl_buffer_converter_new2(from_encoding, to_encoding, 0);
3873 if (convd == NULL) {
3874 php_error_docref(NULL, E_WARNING, "Unable to create converter");
3875 RETURN_FALSE;
3876 }
3877 mbfl_buffer_converter_illegal_mode(convd, MBSTRG(current_filter_illegal_mode));
3878 mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
3879 }
3880
3881 /* convert */
3882 if (convd != NULL) {
3883 stack_max = PHP_MBSTR_STACK_BLOCK_SIZE;
3884 stack = (zval*)safe_emalloc(stack_max, sizeof(zval), 0);
3885 stack_level = 0;
3886 n = 0;
3887 while (n < argc || stack_level > 0) {
3888 if (stack_level <= 0) {
3889 var = &args[n++];
3890 ZVAL_DEREF(var);
3891 SEPARATE_ZVAL_NOREF(var);
3892 if (Z_TYPE_P(var) == IS_ARRAY || Z_TYPE_P(var) == IS_OBJECT) {
3893 target_hash = HASH_OF(var);
3894 if (target_hash != NULL) {
3895 zend_hash_internal_pointer_reset(target_hash);
3896 }
3897 }
3898 } else {
3899 stack_level--;
3900 var = &stack[stack_level];
3901 }
3902 if (Z_TYPE_P(var) == IS_ARRAY || Z_TYPE_P(var) == IS_OBJECT) {
3903 target_hash = HASH_OF(var);
3904 if (target_hash != NULL) {
3905 while ((hash_entry_ptr = zend_hash_get_current_data(target_hash)) != NULL) {
3906 zend_hash_move_forward(target_hash);
3907 if (Z_TYPE_P(hash_entry_ptr) == IS_INDIRECT) {
3908 hash_entry_ptr = Z_INDIRECT_P(hash_entry_ptr);
3909 }
3910 hash_entry = hash_entry_ptr;
3911 ZVAL_DEREF(hash_entry);
3912 if (Z_TYPE_P(hash_entry) == IS_ARRAY || Z_TYPE_P(hash_entry) == IS_OBJECT) {
3913 if (!Z_IMMUTABLE_P(hash_entry)) {
3914 if (++(HASH_OF(hash_entry)->u.v.nApplyCount) > 1) {
3915 --(HASH_OF(hash_entry)->u.v.nApplyCount);
3916 recursion_error = 1;
3917 goto conv_end;
3918 }
3919 }
3920 if (stack_level >= stack_max) {
3921 stack_max += PHP_MBSTR_STACK_BLOCK_SIZE;
3922 ptmp = erealloc(stack, sizeof(zval) * stack_max);
3923 stack = (zval *)ptmp;
3924 }
3925 ZVAL_COPY_VALUE(&stack[stack_level], var);
3926 stack_level++;
3927 var = hash_entry;
3928 SEPARATE_ZVAL(hash_entry);
3929 target_hash = HASH_OF(var);
3930 if (target_hash != NULL) {
3931 zend_hash_internal_pointer_reset(target_hash);
3932 continue;
3933 }
3934 } else if (Z_TYPE_P(hash_entry) == IS_STRING) {
3935 string.val = (unsigned char *)Z_STRVAL_P(hash_entry);
3936 string.len = Z_STRLEN_P(hash_entry);
3937 ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
3938 if (ret != NULL) {
3939 zval_ptr_dtor(hash_entry_ptr);
3940 // TODO: avoid reallocation ???
3941 ZVAL_STRINGL(hash_entry_ptr, (char *)ret->val, ret->len);
3942 efree(ret->val);
3943 }
3944 }
3945 }
3946 }
3947 } else if (Z_TYPE_P(var) == IS_STRING) {
3948 string.val = (unsigned char *)Z_STRVAL_P(var);
3949 string.len = Z_STRLEN_P(var);
3950 ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
3951 if (ret != NULL) {
3952 zval_ptr_dtor(var);
3953 // TODO: avoid reallocation ???
3954 ZVAL_STRINGL(var, (char *)ret->val, ret->len);
3955 efree(ret->val);
3956 }
3957 }
3958 }
3959
3960 conv_end:
3961 MBSTRG(illegalchars) += mbfl_buffer_illegalchars(convd);
3962 mbfl_buffer_converter_delete(convd);
3963
3964 if (recursion_error) {
3965 while(stack_level-- && (var = &stack[stack_level])) {
3966 if (!Z_IMMUTABLE_P(var)) {
3967 if (HASH_OF(var)->u.v.nApplyCount > 1) {
3968 HASH_OF(var)->u.v.nApplyCount--;
3969 }
3970 }
3971 }
3972 efree(stack);
3973 php_error_docref(NULL, E_WARNING, "Cannot handle recursive references");
3974 RETURN_FALSE;
3975 }
3976 efree(stack);
3977 }
3978
3979 if (from_encoding) {
3980 RETURN_STRING(from_encoding->name);
3981 } else {
3982 RETURN_FALSE;
3983 }
3984 }
3985 /* }}} */
3986
3987 /* {{{ HTML numeric entity */
3988 /* {{{ static void php_mb_numericentity_exec() */
3989 static void
php_mb_numericentity_exec(INTERNAL_FUNCTION_PARAMETERS,int type)3990 php_mb_numericentity_exec(INTERNAL_FUNCTION_PARAMETERS, int type)
3991 {
3992 char *str, *encoding = NULL;
3993 size_t str_len, encoding_len;
3994 zval *zconvmap, *hash_entry;
3995 HashTable *target_hash;
3996 int i, *convmap, *mapelm, mapsize=0;
3997 zend_bool is_hex = 0;
3998 mbfl_string string, result, *ret;
3999 enum mbfl_no_encoding no_encoding;
4000
4001 if (zend_parse_parameters(ZEND_NUM_ARGS(), "sz|sb", &str, &str_len, &zconvmap, &encoding, &encoding_len, &is_hex) == FAILURE) {
4002 return;
4003 }
4004
4005 mbfl_string_init(&string);
4006 string.no_language = MBSTRG(language);
4007 string.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
4008 string.val = (unsigned char *)str;
4009 string.len = str_len;
4010
4011 /* encoding */
4012 if (encoding && encoding_len > 0) {
4013 no_encoding = mbfl_name2no_encoding(encoding);
4014 if (no_encoding == mbfl_no_encoding_invalid) {
4015 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", encoding);
4016 RETURN_FALSE;
4017 } else {
4018 string.no_encoding = no_encoding;
4019 }
4020 }
4021
4022 if (type == 0 && is_hex) {
4023 type = 2; /* output in hex format */
4024 }
4025
4026 /* conversion map */
4027 convmap = NULL;
4028 if (Z_TYPE_P(zconvmap) == IS_ARRAY) {
4029 target_hash = Z_ARRVAL_P(zconvmap);
4030 i = zend_hash_num_elements(target_hash);
4031 if (i > 0) {
4032 convmap = (int *)safe_emalloc(i, sizeof(int), 0);
4033 mapelm = convmap;
4034 mapsize = 0;
4035 ZEND_HASH_FOREACH_VAL(target_hash, hash_entry) {
4036 convert_to_long_ex(hash_entry);
4037 *mapelm++ = Z_LVAL_P(hash_entry);
4038 mapsize++;
4039 } ZEND_HASH_FOREACH_END();
4040 }
4041 }
4042 if (convmap == NULL) {
4043 RETURN_FALSE;
4044 }
4045 mapsize /= 4;
4046
4047 ret = mbfl_html_numeric_entity(&string, &result, convmap, mapsize, type);
4048 if (ret != NULL) {
4049 // TODO: avoid reallocation ???
4050 RETVAL_STRINGL((char *)ret->val, ret->len);
4051 efree(ret->val);
4052 } else {
4053 RETVAL_FALSE;
4054 }
4055 efree((void *)convmap);
4056 }
4057 /* }}} */
4058
4059 /* {{{ proto string mb_encode_numericentity(string string, array convmap [, string encoding [, bool is_hex]])
4060 Converts specified characters to HTML numeric entities */
PHP_FUNCTION(mb_encode_numericentity)4061 PHP_FUNCTION(mb_encode_numericentity)
4062 {
4063 php_mb_numericentity_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0);
4064 }
4065 /* }}} */
4066
4067 /* {{{ proto string mb_decode_numericentity(string string, array convmap [, string encoding])
4068 Converts HTML numeric entities to character code */
PHP_FUNCTION(mb_decode_numericentity)4069 PHP_FUNCTION(mb_decode_numericentity)
4070 {
4071 php_mb_numericentity_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, 1);
4072 }
4073 /* }}} */
4074 /* }}} */
4075
4076 /* {{{ proto int mb_send_mail(string to, string subject, string message [, string additional_headers [, string additional_parameters]])
4077 * Sends an email message with MIME scheme
4078 */
4079
4080 #define SKIP_LONG_HEADER_SEP_MBSTRING(str, pos) \
4081 if (str[pos] == '\r' && str[pos + 1] == '\n' && (str[pos + 2] == ' ' || str[pos + 2] == '\t')) { \
4082 pos += 2; \
4083 while (str[pos + 1] == ' ' || str[pos + 1] == '\t') { \
4084 pos++; \
4085 } \
4086 continue; \
4087 }
4088
4089 #define MAIL_ASCIIZ_CHECK_MBSTRING(str, len) \
4090 pp = str; \
4091 ee = pp + len; \
4092 while ((pp = memchr(pp, '\0', (ee - pp)))) { \
4093 *pp = ' '; \
4094 } \
4095
_php_mbstr_parse_mail_headers(HashTable * ht,const char * str,size_t str_len)4096 static int _php_mbstr_parse_mail_headers(HashTable *ht, const char *str, size_t str_len)
4097 {
4098 const char *ps;
4099 size_t icnt;
4100 int state = 0;
4101 int crlf_state = -1;
4102 char *token = NULL;
4103 size_t token_pos = 0;
4104 zend_string *fld_name, *fld_val;
4105
4106 ps = str;
4107 icnt = str_len;
4108 fld_name = fld_val = NULL;
4109
4110 /*
4111 * C o n t e n t - T y p e : t e x t / h t m l \r\n
4112 * ^ ^^^^^^^^^^^^^^^^^^^^^ ^^^ ^^^^^^^^^^^^^^^^^ ^^^^
4113 * state 0 1 2 3
4114 *
4115 * C o n t e n t - T y p e : t e x t / h t m l \r\n
4116 * ^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^
4117 * crlf_state -1 0 1 -1
4118 *
4119 */
4120
4121 while (icnt > 0) {
4122 switch (*ps) {
4123 case ':':
4124 if (crlf_state == 1) {
4125 token_pos++;
4126 }
4127
4128 if (state == 0 || state == 1) {
4129 if(token && token_pos > 0) {
4130 fld_name = zend_string_init(token, token_pos, 0);
4131 }
4132 state = 2;
4133 } else {
4134 token_pos++;
4135 }
4136
4137 crlf_state = 0;
4138 break;
4139
4140 case '\n':
4141 if (crlf_state == -1) {
4142 goto out;
4143 }
4144 crlf_state = -1;
4145 break;
4146
4147 case '\r':
4148 if (crlf_state == 1) {
4149 token_pos++;
4150 } else {
4151 crlf_state = 1;
4152 }
4153 break;
4154
4155 case ' ': case '\t':
4156 if (crlf_state == -1) {
4157 if (state == 3) {
4158 /* continuing from the previous line */
4159 state = 4;
4160 } else {
4161 /* simply skipping this new line */
4162 state = 5;
4163 }
4164 } else {
4165 if (crlf_state == 1) {
4166 token_pos++;
4167 }
4168 if (state == 1 || state == 3) {
4169 token_pos++;
4170 }
4171 }
4172 crlf_state = 0;
4173 break;
4174
4175 default:
4176 switch (state) {
4177 case 0:
4178 token = (char*)ps;
4179 token_pos = 0;
4180 state = 1;
4181 break;
4182
4183 case 2:
4184 if (crlf_state != -1) {
4185 token = (char*)ps;
4186 token_pos = 0;
4187
4188 state = 3;
4189 break;
4190 }
4191 /* break is missing intentionally */
4192
4193 case 3:
4194 if (crlf_state == -1) {
4195 if(token && token_pos > 0) {
4196 fld_val = zend_string_init(token, token_pos, 0);
4197 }
4198
4199 if (fld_name != NULL && fld_val != NULL) {
4200 zval val;
4201 /* FIXME: some locale free implementation is
4202 * really required here,,, */
4203 php_strtoupper(ZSTR_VAL(fld_name), ZSTR_LEN(fld_name));
4204 ZVAL_STR(&val, fld_val);
4205
4206 zend_hash_update(ht, fld_name, &val);
4207
4208 zend_string_release(fld_name);
4209 }
4210
4211 fld_name = fld_val = NULL;
4212 token = (char*)ps;
4213 token_pos = 0;
4214
4215 state = 1;
4216 }
4217 break;
4218
4219 case 4:
4220 token_pos++;
4221 state = 3;
4222 break;
4223 }
4224
4225 if (crlf_state == 1) {
4226 token_pos++;
4227 }
4228
4229 token_pos++;
4230
4231 crlf_state = 0;
4232 break;
4233 }
4234 ps++, icnt--;
4235 }
4236 out:
4237 if (state == 2) {
4238 token = "";
4239 token_pos = 0;
4240
4241 state = 3;
4242 }
4243 if (state == 3) {
4244 if(token && token_pos > 0) {
4245 fld_val = zend_string_init(token, token_pos, 0);
4246 }
4247 if (fld_name != NULL && fld_val != NULL) {
4248 zval val;
4249 /* FIXME: some locale free implementation is
4250 * really required here,,, */
4251 php_strtoupper(ZSTR_VAL(fld_name), ZSTR_LEN(fld_name));
4252 ZVAL_STR(&val, fld_val);
4253
4254 zend_hash_update(ht, fld_name, &val);
4255
4256 zend_string_release(fld_name);
4257 }
4258 }
4259 return state;
4260 }
4261
PHP_FUNCTION(mb_send_mail)4262 PHP_FUNCTION(mb_send_mail)
4263 {
4264 int n;
4265 char *to = NULL;
4266 size_t to_len;
4267 char *message = NULL;
4268 size_t message_len;
4269 char *headers = NULL;
4270 size_t headers_len;
4271 char *subject = NULL;
4272 zend_string *extra_cmd = NULL;
4273 size_t subject_len;
4274 int i;
4275 char *to_r = NULL;
4276 char *force_extra_parameters = INI_STR("mail.force_extra_parameters");
4277 struct {
4278 int cnt_type:1;
4279 int cnt_trans_enc:1;
4280 } suppressed_hdrs = { 0, 0 };
4281
4282 char *message_buf = NULL, *subject_buf = NULL, *p;
4283 mbfl_string orig_str, conv_str;
4284 mbfl_string *pstr; /* pointer to mbfl string for return value */
4285 enum mbfl_no_encoding
4286 tran_cs, /* transfar text charset */
4287 head_enc, /* header transfar encoding */
4288 body_enc; /* body transfar encoding */
4289 mbfl_memory_device device; /* automatic allocateable buffer for additional header */
4290 const mbfl_language *lang;
4291 int err = 0;
4292 HashTable ht_headers;
4293 zval *s;
4294 extern void mbfl_memory_device_unput(mbfl_memory_device *device);
4295 char *pp, *ee;
4296
4297 /* initialize */
4298 mbfl_memory_device_init(&device, 0, 0);
4299 mbfl_string_init(&orig_str);
4300 mbfl_string_init(&conv_str);
4301
4302 /* character-set, transfer-encoding */
4303 tran_cs = mbfl_no_encoding_utf8;
4304 head_enc = mbfl_no_encoding_base64;
4305 body_enc = mbfl_no_encoding_base64;
4306 lang = mbfl_no2language(MBSTRG(language));
4307 if (lang != NULL) {
4308 tran_cs = lang->mail_charset;
4309 head_enc = lang->mail_header_encoding;
4310 body_enc = lang->mail_body_encoding;
4311 }
4312
4313 if (zend_parse_parameters(ZEND_NUM_ARGS(), "sss|sS", &to, &to_len, &subject, &subject_len, &message, &message_len, &headers, &headers_len, &extra_cmd) == FAILURE) {
4314 return;
4315 }
4316
4317 /* ASCIIZ check */
4318 MAIL_ASCIIZ_CHECK_MBSTRING(to, to_len);
4319 MAIL_ASCIIZ_CHECK_MBSTRING(subject, subject_len);
4320 MAIL_ASCIIZ_CHECK_MBSTRING(message, message_len);
4321 if (headers) {
4322 MAIL_ASCIIZ_CHECK_MBSTRING(headers, headers_len);
4323 }
4324 if (extra_cmd) {
4325 MAIL_ASCIIZ_CHECK_MBSTRING(ZSTR_VAL(extra_cmd), ZSTR_LEN(extra_cmd));
4326 }
4327
4328 zend_hash_init(&ht_headers, 0, NULL, ZVAL_PTR_DTOR, 0);
4329
4330 if (headers != NULL) {
4331 _php_mbstr_parse_mail_headers(&ht_headers, headers, headers_len);
4332 }
4333
4334 if ((s = zend_hash_str_find(&ht_headers, "CONTENT-TYPE", sizeof("CONTENT-TYPE") - 1))) {
4335 char *tmp;
4336 char *param_name;
4337 char *charset = NULL;
4338
4339 ZEND_ASSERT(Z_TYPE_P(s) == IS_STRING);
4340 p = strchr(Z_STRVAL_P(s), ';');
4341
4342 if (p != NULL) {
4343 /* skipping the padded spaces */
4344 do {
4345 ++p;
4346 } while (*p == ' ' || *p == '\t');
4347
4348 if (*p != '\0') {
4349 if ((param_name = php_strtok_r(p, "= ", &tmp)) != NULL) {
4350 if (strcasecmp(param_name, "charset") == 0) {
4351 enum mbfl_no_encoding _tran_cs = tran_cs;
4352
4353 charset = php_strtok_r(NULL, "= \"", &tmp);
4354 if (charset != NULL) {
4355 _tran_cs = mbfl_name2no_encoding(charset);
4356 }
4357
4358 if (_tran_cs == mbfl_no_encoding_invalid) {
4359 php_error_docref(NULL, E_WARNING, "Unsupported charset \"%s\" - will be regarded as ascii", charset);
4360 _tran_cs = mbfl_no_encoding_ascii;
4361 }
4362 tran_cs = _tran_cs;
4363 }
4364 }
4365 }
4366 }
4367 suppressed_hdrs.cnt_type = 1;
4368 }
4369
4370 if ((s = zend_hash_str_find(&ht_headers, "CONTENT-TRANSFER-ENCODING", sizeof("CONTENT-TRANSFER-ENCODING") - 1))) {
4371 enum mbfl_no_encoding _body_enc;
4372
4373 ZEND_ASSERT(Z_TYPE_P(s) == IS_STRING);
4374 _body_enc = mbfl_name2no_encoding(Z_STRVAL_P(s));
4375 switch (_body_enc) {
4376 case mbfl_no_encoding_base64:
4377 case mbfl_no_encoding_7bit:
4378 case mbfl_no_encoding_8bit:
4379 body_enc = _body_enc;
4380 break;
4381
4382 default:
4383 php_error_docref(NULL, E_WARNING, "Unsupported transfer encoding \"%s\" - will be regarded as 8bit", Z_STRVAL_P(s));
4384 body_enc = mbfl_no_encoding_8bit;
4385 break;
4386 }
4387 suppressed_hdrs.cnt_trans_enc = 1;
4388 }
4389
4390 /* To: */
4391 if (to != NULL) {
4392 if (to_len > 0) {
4393 to_r = estrndup(to, to_len);
4394 for (; to_len; to_len--) {
4395 if (!isspace((unsigned char) to_r[to_len - 1])) {
4396 break;
4397 }
4398 to_r[to_len - 1] = '\0';
4399 }
4400 for (i = 0; to_r[i]; i++) {
4401 if (iscntrl((unsigned char) to_r[i])) {
4402 /* According to RFC 822, section 3.1.1 long headers may be separated into
4403 * parts using CRLF followed at least one linear-white-space character ('\t' or ' ').
4404 * To prevent these separators from being replaced with a space, we use the
4405 * SKIP_LONG_HEADER_SEP_MBSTRING to skip over them.
4406 */
4407 SKIP_LONG_HEADER_SEP_MBSTRING(to_r, i);
4408 to_r[i] = ' ';
4409 }
4410 }
4411 } else {
4412 to_r = to;
4413 }
4414 } else {
4415 php_error_docref(NULL, E_WARNING, "Missing To: field");
4416 err = 1;
4417 }
4418
4419 /* Subject: */
4420 if (subject != NULL) {
4421 orig_str.no_language = MBSTRG(language);
4422 orig_str.val = (unsigned char *)subject;
4423 orig_str.len = subject_len;
4424 orig_str.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
4425 if (orig_str.no_encoding == mbfl_no_encoding_invalid || orig_str.no_encoding == mbfl_no_encoding_pass) {
4426 const mbfl_encoding *encoding = mbfl_identify_encoding2(&orig_str, MBSTRG(current_detect_order_list), MBSTRG(current_detect_order_list_size), MBSTRG(strict_detection));
4427 orig_str.no_encoding = encoding ? encoding->no_encoding: mbfl_no_encoding_invalid;
4428 }
4429 pstr = mbfl_mime_header_encode(&orig_str, &conv_str, tran_cs, head_enc, "\n", sizeof("Subject: [PHP-jp nnnnnnnn]"));
4430 if (pstr != NULL) {
4431 subject_buf = subject = (char *)pstr->val;
4432 }
4433 } else {
4434 php_error_docref(NULL, E_WARNING, "Missing Subject: field");
4435 err = 1;
4436 }
4437
4438 /* message body */
4439 if (message != NULL) {
4440 orig_str.no_language = MBSTRG(language);
4441 orig_str.val = (unsigned char *)message;
4442 orig_str.len = (unsigned int)message_len;
4443 orig_str.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
4444
4445 if (orig_str.no_encoding == mbfl_no_encoding_invalid || orig_str.no_encoding == mbfl_no_encoding_pass) {
4446 const mbfl_encoding *encoding = mbfl_identify_encoding2(&orig_str, MBSTRG(current_detect_order_list), MBSTRG(current_detect_order_list_size), MBSTRG(strict_detection));
4447 orig_str.no_encoding = encoding ? encoding->no_encoding: mbfl_no_encoding_invalid;
4448 }
4449
4450 pstr = NULL;
4451 {
4452 mbfl_string tmpstr;
4453
4454 if (mbfl_convert_encoding(&orig_str, &tmpstr, tran_cs) != NULL) {
4455 tmpstr.no_encoding=mbfl_no_encoding_8bit;
4456 pstr = mbfl_convert_encoding(&tmpstr, &conv_str, body_enc);
4457 efree(tmpstr.val);
4458 }
4459 }
4460 if (pstr != NULL) {
4461 message_buf = message = (char *)pstr->val;
4462 }
4463 } else {
4464 /* this is not really an error, so it is allowed. */
4465 php_error_docref(NULL, E_WARNING, "Empty message body");
4466 message = NULL;
4467 }
4468
4469 /* other headers */
4470 #define PHP_MBSTR_MAIL_MIME_HEADER1 "MIME-Version: 1.0"
4471 #define PHP_MBSTR_MAIL_MIME_HEADER2 "Content-Type: text/plain"
4472 #define PHP_MBSTR_MAIL_MIME_HEADER3 "; charset="
4473 #define PHP_MBSTR_MAIL_MIME_HEADER4 "Content-Transfer-Encoding: "
4474 if (headers != NULL) {
4475 p = headers;
4476 n = headers_len;
4477 mbfl_memory_device_strncat(&device, p, n);
4478 if (n > 0 && p[n - 1] != '\n') {
4479 mbfl_memory_device_strncat(&device, "\n", 1);
4480 }
4481 }
4482
4483 if (!zend_hash_str_exists(&ht_headers, "MIME-VERSION", sizeof("MIME-VERSION") - 1)) {
4484 mbfl_memory_device_strncat(&device, PHP_MBSTR_MAIL_MIME_HEADER1, sizeof(PHP_MBSTR_MAIL_MIME_HEADER1) - 1);
4485 mbfl_memory_device_strncat(&device, "\n", 1);
4486 }
4487
4488 if (!suppressed_hdrs.cnt_type) {
4489 mbfl_memory_device_strncat(&device, PHP_MBSTR_MAIL_MIME_HEADER2, sizeof(PHP_MBSTR_MAIL_MIME_HEADER2) - 1);
4490
4491 p = (char *)mbfl_no2preferred_mime_name(tran_cs);
4492 if (p != NULL) {
4493 mbfl_memory_device_strncat(&device, PHP_MBSTR_MAIL_MIME_HEADER3, sizeof(PHP_MBSTR_MAIL_MIME_HEADER3) - 1);
4494 mbfl_memory_device_strcat(&device, p);
4495 }
4496 mbfl_memory_device_strncat(&device, "\n", 1);
4497 }
4498 if (!suppressed_hdrs.cnt_trans_enc) {
4499 mbfl_memory_device_strncat(&device, PHP_MBSTR_MAIL_MIME_HEADER4, sizeof(PHP_MBSTR_MAIL_MIME_HEADER4) - 1);
4500 p = (char *)mbfl_no2preferred_mime_name(body_enc);
4501 if (p == NULL) {
4502 p = "7bit";
4503 }
4504 mbfl_memory_device_strcat(&device, p);
4505 mbfl_memory_device_strncat(&device, "\n", 1);
4506 }
4507
4508 mbfl_memory_device_unput(&device);
4509 mbfl_memory_device_output('\0', &device);
4510 headers = (char *)device.buffer;
4511
4512 if (force_extra_parameters) {
4513 extra_cmd = php_escape_shell_cmd(force_extra_parameters);
4514 } else if (extra_cmd) {
4515 extra_cmd = php_escape_shell_cmd(ZSTR_VAL(extra_cmd));
4516 }
4517
4518 if (!err && php_mail(to_r, subject, message, headers, extra_cmd ? ZSTR_VAL(extra_cmd) : NULL)) {
4519 RETVAL_TRUE;
4520 } else {
4521 RETVAL_FALSE;
4522 }
4523
4524 if (extra_cmd) {
4525 zend_string_release(extra_cmd);
4526 }
4527
4528 if (to_r != to) {
4529 efree(to_r);
4530 }
4531 if (subject_buf) {
4532 efree((void *)subject_buf);
4533 }
4534 if (message_buf) {
4535 efree((void *)message_buf);
4536 }
4537 mbfl_memory_device_clear(&device);
4538 zend_hash_destroy(&ht_headers);
4539 }
4540
4541 #undef SKIP_LONG_HEADER_SEP_MBSTRING
4542 #undef MAIL_ASCIIZ_CHECK_MBSTRING
4543 #undef PHP_MBSTR_MAIL_MIME_HEADER1
4544 #undef PHP_MBSTR_MAIL_MIME_HEADER2
4545 #undef PHP_MBSTR_MAIL_MIME_HEADER3
4546 #undef PHP_MBSTR_MAIL_MIME_HEADER4
4547 /* }}} */
4548
4549 /* {{{ proto mixed mb_get_info([string type])
4550 Returns the current settings of mbstring */
PHP_FUNCTION(mb_get_info)4551 PHP_FUNCTION(mb_get_info)
4552 {
4553 char *typ = NULL;
4554 size_t typ_len;
4555 size_t n;
4556 char *name;
4557 const struct mb_overload_def *over_func;
4558 zval row1, row2;
4559 const mbfl_language *lang = mbfl_no2language(MBSTRG(language));
4560 const mbfl_encoding **entry;
4561
4562 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|s", &typ, &typ_len) == FAILURE) {
4563 return;
4564 }
4565
4566 if (!typ || !strcasecmp("all", typ)) {
4567 array_init(return_value);
4568 if (MBSTRG(current_internal_encoding)) {
4569 add_assoc_string(return_value, "internal_encoding", (char *)MBSTRG(current_internal_encoding)->name);
4570 }
4571 if (MBSTRG(http_input_identify)) {
4572 add_assoc_string(return_value, "http_input", (char *)MBSTRG(http_input_identify)->name);
4573 }
4574 if (MBSTRG(current_http_output_encoding)) {
4575 add_assoc_string(return_value, "http_output", (char *)MBSTRG(current_http_output_encoding)->name);
4576 }
4577 if ((name = (char *)zend_ini_string("mbstring.http_output_conv_mimetypes", sizeof("mbstring.http_output_conv_mimetypes") - 1, 0)) != NULL) {
4578 add_assoc_string(return_value, "http_output_conv_mimetypes", name);
4579 }
4580 add_assoc_long(return_value, "func_overload", MBSTRG(func_overload));
4581 if (MBSTRG(func_overload)){
4582 over_func = &(mb_ovld[0]);
4583 array_init(&row1);
4584 while (over_func->type > 0) {
4585 if ((MBSTRG(func_overload) & over_func->type) == over_func->type ) {
4586 add_assoc_string(&row1, over_func->orig_func, over_func->ovld_func);
4587 }
4588 over_func++;
4589 }
4590 add_assoc_zval(return_value, "func_overload_list", &row1);
4591 } else {
4592 add_assoc_string(return_value, "func_overload_list", "no overload");
4593 }
4594 if (lang != NULL) {
4595 if ((name = (char *)mbfl_no_encoding2name(lang->mail_charset)) != NULL) {
4596 add_assoc_string(return_value, "mail_charset", name);
4597 }
4598 if ((name = (char *)mbfl_no_encoding2name(lang->mail_header_encoding)) != NULL) {
4599 add_assoc_string(return_value, "mail_header_encoding", name);
4600 }
4601 if ((name = (char *)mbfl_no_encoding2name(lang->mail_body_encoding)) != NULL) {
4602 add_assoc_string(return_value, "mail_body_encoding", name);
4603 }
4604 }
4605 add_assoc_long(return_value, "illegal_chars", MBSTRG(illegalchars));
4606 if (MBSTRG(encoding_translation)) {
4607 add_assoc_string(return_value, "encoding_translation", "On");
4608 } else {
4609 add_assoc_string(return_value, "encoding_translation", "Off");
4610 }
4611 if ((name = (char *)mbfl_no_language2name(MBSTRG(language))) != NULL) {
4612 add_assoc_string(return_value, "language", name);
4613 }
4614 n = MBSTRG(current_detect_order_list_size);
4615 entry = MBSTRG(current_detect_order_list);
4616 if (n > 0) {
4617 size_t i;
4618 array_init(&row2);
4619 for (i = 0; i < n; i++) {
4620 add_next_index_string(&row2, (*entry)->name);
4621 entry++;
4622 }
4623 add_assoc_zval(return_value, "detect_order", &row2);
4624 }
4625 if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
4626 add_assoc_string(return_value, "substitute_character", "none");
4627 } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG) {
4628 add_assoc_string(return_value, "substitute_character", "long");
4629 } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY) {
4630 add_assoc_string(return_value, "substitute_character", "entity");
4631 } else {
4632 add_assoc_long(return_value, "substitute_character", MBSTRG(current_filter_illegal_substchar));
4633 }
4634 if (MBSTRG(strict_detection)) {
4635 add_assoc_string(return_value, "strict_detection", "On");
4636 } else {
4637 add_assoc_string(return_value, "strict_detection", "Off");
4638 }
4639 } else if (!strcasecmp("internal_encoding", typ)) {
4640 if (MBSTRG(current_internal_encoding)) {
4641 RETVAL_STRING((char *)MBSTRG(current_internal_encoding)->name);
4642 }
4643 } else if (!strcasecmp("http_input", typ)) {
4644 if (MBSTRG(http_input_identify)) {
4645 RETVAL_STRING((char *)MBSTRG(http_input_identify)->name);
4646 }
4647 } else if (!strcasecmp("http_output", typ)) {
4648 if (MBSTRG(current_http_output_encoding)) {
4649 RETVAL_STRING((char *)MBSTRG(current_http_output_encoding)->name);
4650 }
4651 } else if (!strcasecmp("http_output_conv_mimetypes", typ)) {
4652 if ((name = (char *)zend_ini_string("mbstring.http_output_conv_mimetypes", sizeof("mbstring.http_output_conv_mimetypes") - 1, 0)) != NULL) {
4653 RETVAL_STRING(name);
4654 }
4655 } else if (!strcasecmp("func_overload", typ)) {
4656 RETVAL_LONG(MBSTRG(func_overload));
4657 } else if (!strcasecmp("func_overload_list", typ)) {
4658 if (MBSTRG(func_overload)){
4659 over_func = &(mb_ovld[0]);
4660 array_init(return_value);
4661 while (over_func->type > 0) {
4662 if ((MBSTRG(func_overload) & over_func->type) == over_func->type ) {
4663 add_assoc_string(return_value, over_func->orig_func, over_func->ovld_func);
4664 }
4665 over_func++;
4666 }
4667 } else {
4668 RETVAL_STRING("no overload");
4669 }
4670 } else if (!strcasecmp("mail_charset", typ)) {
4671 if (lang != NULL && (name = (char *)mbfl_no_encoding2name(lang->mail_charset)) != NULL) {
4672 RETVAL_STRING(name);
4673 }
4674 } else if (!strcasecmp("mail_header_encoding", typ)) {
4675 if (lang != NULL && (name = (char *)mbfl_no_encoding2name(lang->mail_header_encoding)) != NULL) {
4676 RETVAL_STRING(name);
4677 }
4678 } else if (!strcasecmp("mail_body_encoding", typ)) {
4679 if (lang != NULL && (name = (char *)mbfl_no_encoding2name(lang->mail_body_encoding)) != NULL) {
4680 RETVAL_STRING(name);
4681 }
4682 } else if (!strcasecmp("illegal_chars", typ)) {
4683 RETVAL_LONG(MBSTRG(illegalchars));
4684 } else if (!strcasecmp("encoding_translation", typ)) {
4685 if (MBSTRG(encoding_translation)) {
4686 RETVAL_STRING("On");
4687 } else {
4688 RETVAL_STRING("Off");
4689 }
4690 } else if (!strcasecmp("language", typ)) {
4691 if ((name = (char *)mbfl_no_language2name(MBSTRG(language))) != NULL) {
4692 RETVAL_STRING(name);
4693 }
4694 } else if (!strcasecmp("detect_order", typ)) {
4695 n = MBSTRG(current_detect_order_list_size);
4696 entry = MBSTRG(current_detect_order_list);
4697 if (n > 0) {
4698 size_t i;
4699 array_init(return_value);
4700 for (i = 0; i < n; i++) {
4701 add_next_index_string(return_value, (*entry)->name);
4702 entry++;
4703 }
4704 }
4705 } else if (!strcasecmp("substitute_character", typ)) {
4706 if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
4707 RETVAL_STRING("none");
4708 } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG) {
4709 RETVAL_STRING("long");
4710 } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY) {
4711 RETVAL_STRING("entity");
4712 } else {
4713 RETVAL_LONG(MBSTRG(current_filter_illegal_substchar));
4714 }
4715 } else if (!strcasecmp("strict_detection", typ)) {
4716 if (MBSTRG(strict_detection)) {
4717 RETVAL_STRING("On");
4718 } else {
4719 RETVAL_STRING("Off");
4720 }
4721 } else {
4722 RETURN_FALSE;
4723 }
4724 }
4725 /* }}} */
4726
php_mb_check_encoding(const char * input,size_t length,const char * enc)4727 MBSTRING_API int php_mb_check_encoding(const char *input, size_t length, const char *enc)
4728 {
4729 const mbfl_encoding *encoding = MBSTRG(current_internal_encoding);
4730 mbfl_buffer_converter *convd;
4731 mbfl_string string, result, *ret = NULL;
4732 long illegalchars = 0;
4733
4734 if (input == NULL) {
4735 return MBSTRG(illegalchars) == 0;
4736 }
4737
4738 if (enc != NULL) {
4739 encoding = mbfl_name2encoding(enc);
4740 if (!encoding || encoding == &mbfl_encoding_pass) {
4741 php_error_docref(NULL, E_WARNING, "Invalid encoding \"%s\"", enc);
4742 return 0;
4743 }
4744 }
4745
4746 convd = mbfl_buffer_converter_new2(encoding, encoding, 0);
4747
4748 if (convd == NULL) {
4749 php_error_docref(NULL, E_WARNING, "Unable to create converter");
4750 return 0;
4751 }
4752
4753 mbfl_buffer_converter_illegal_mode(convd, MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE);
4754 mbfl_buffer_converter_illegal_substchar(convd, 0);
4755
4756 /* initialize string */
4757 mbfl_string_init_set(&string, mbfl_no_language_neutral, encoding->no_encoding);
4758 mbfl_string_init(&result);
4759
4760 string.val = (unsigned char *) input;
4761 string.len = length;
4762
4763 ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
4764 illegalchars = mbfl_buffer_illegalchars(convd);
4765 mbfl_buffer_converter_delete(convd);
4766
4767 if (ret != NULL) {
4768 if (illegalchars == 0 && string.len == result.len && memcmp(string.val, result.val, string.len) == 0) {
4769 mbfl_string_clear(&result);
4770 return 1;
4771 }
4772
4773 mbfl_string_clear(&result);
4774 }
4775
4776 return 0;
4777 }
4778
4779 /* {{{ proto bool mb_check_encoding([string var[, string encoding]])
4780 Check if the string is valid for the specified encoding */
PHP_FUNCTION(mb_check_encoding)4781 PHP_FUNCTION(mb_check_encoding)
4782 {
4783 char *var = NULL;
4784 size_t var_len;
4785 char *enc = NULL;
4786 size_t enc_len;
4787
4788 if (zend_parse_parameters(ZEND_NUM_ARGS(), "|ss", &var, &var_len, &enc, &enc_len) == FAILURE) {
4789 return;
4790 }
4791
4792 RETVAL_FALSE;
4793
4794 if (php_mb_check_encoding(var, var_len, enc)) {
4795 RETVAL_TRUE;
4796 }
4797 }
4798 /* }}} */
4799
4800 /* {{{ php_mb_populate_current_detect_order_list */
php_mb_populate_current_detect_order_list(void)4801 static void php_mb_populate_current_detect_order_list(void)
4802 {
4803 const mbfl_encoding **entry = 0;
4804 size_t nentries;
4805
4806 if (MBSTRG(current_detect_order_list)) {
4807 return;
4808 }
4809
4810 if (MBSTRG(detect_order_list) && MBSTRG(detect_order_list_size)) {
4811 nentries = MBSTRG(detect_order_list_size);
4812 entry = (const mbfl_encoding **)safe_emalloc(nentries, sizeof(mbfl_encoding*), 0);
4813 memcpy(entry, MBSTRG(detect_order_list), sizeof(mbfl_encoding*) * nentries);
4814 } else {
4815 const enum mbfl_no_encoding *src = MBSTRG(default_detect_order_list);
4816 size_t i;
4817 nentries = MBSTRG(default_detect_order_list_size);
4818 entry = (const mbfl_encoding **)safe_emalloc(nentries, sizeof(mbfl_encoding*), 0);
4819 for (i = 0; i < nentries; i++) {
4820 entry[i] = mbfl_no2encoding(src[i]);
4821 }
4822 }
4823 MBSTRG(current_detect_order_list) = entry;
4824 MBSTRG(current_detect_order_list_size) = nentries;
4825 }
4826 /* }}} */
4827
4828 /* {{{ static int php_mb_encoding_translation() */
php_mb_encoding_translation(void)4829 static int php_mb_encoding_translation(void)
4830 {
4831 return MBSTRG(encoding_translation);
4832 }
4833 /* }}} */
4834
4835 /* {{{ MBSTRING_API size_t php_mb_mbchar_bytes_ex() */
php_mb_mbchar_bytes_ex(const char * s,const mbfl_encoding * enc)4836 MBSTRING_API size_t php_mb_mbchar_bytes_ex(const char *s, const mbfl_encoding *enc)
4837 {
4838 if (enc != NULL) {
4839 if (enc->flag & MBFL_ENCTYPE_MBCS) {
4840 if (enc->mblen_table != NULL) {
4841 if (s != NULL) return enc->mblen_table[*(unsigned char *)s];
4842 }
4843 } else if (enc->flag & (MBFL_ENCTYPE_WCS2BE | MBFL_ENCTYPE_WCS2LE)) {
4844 return 2;
4845 } else if (enc->flag & (MBFL_ENCTYPE_WCS4BE | MBFL_ENCTYPE_WCS4LE)) {
4846 return 4;
4847 }
4848 }
4849 return 1;
4850 }
4851 /* }}} */
4852
4853 /* {{{ MBSTRING_API size_t php_mb_mbchar_bytes() */
php_mb_mbchar_bytes(const char * s)4854 MBSTRING_API size_t php_mb_mbchar_bytes(const char *s)
4855 {
4856 return php_mb_mbchar_bytes_ex(s, MBSTRG(internal_encoding));
4857 }
4858 /* }}} */
4859
4860 /* {{{ MBSTRING_API char *php_mb_safe_strrchr_ex() */
php_mb_safe_strrchr_ex(const char * s,unsigned int c,size_t nbytes,const mbfl_encoding * enc)4861 MBSTRING_API char *php_mb_safe_strrchr_ex(const char *s, unsigned int c, size_t nbytes, const mbfl_encoding *enc)
4862 {
4863 register const char *p = s;
4864 char *last=NULL;
4865
4866 if (nbytes == (size_t)-1) {
4867 size_t nb = 0;
4868
4869 while (*p != '\0') {
4870 if (nb == 0) {
4871 if ((unsigned char)*p == (unsigned char)c) {
4872 last = (char *)p;
4873 }
4874 nb = php_mb_mbchar_bytes_ex(p, enc);
4875 if (nb == 0) {
4876 return NULL; /* something is going wrong! */
4877 }
4878 }
4879 --nb;
4880 ++p;
4881 }
4882 } else {
4883 register size_t bcnt = nbytes;
4884 register size_t nbytes_char;
4885 while (bcnt > 0) {
4886 if ((unsigned char)*p == (unsigned char)c) {
4887 last = (char *)p;
4888 }
4889 nbytes_char = php_mb_mbchar_bytes_ex(p, enc);
4890 if (bcnt < nbytes_char) {
4891 return NULL;
4892 }
4893 p += nbytes_char;
4894 bcnt -= nbytes_char;
4895 }
4896 }
4897 return last;
4898 }
4899 /* }}} */
4900
4901 /* {{{ MBSTRING_API char *php_mb_safe_strrchr() */
php_mb_safe_strrchr(const char * s,unsigned int c,size_t nbytes)4902 MBSTRING_API char *php_mb_safe_strrchr(const char *s, unsigned int c, size_t nbytes)
4903 {
4904 return php_mb_safe_strrchr_ex(s, c, nbytes, MBSTRG(internal_encoding));
4905 }
4906 /* }}} */
4907
4908 /* {{{ MBSTRING_API int php_mb_stripos()
4909 */
php_mb_stripos(int mode,const char * old_haystack,unsigned int old_haystack_len,const char * old_needle,unsigned int old_needle_len,long offset,const char * from_encoding)4910 MBSTRING_API int php_mb_stripos(int mode, const char *old_haystack, unsigned int old_haystack_len, const char *old_needle, unsigned int old_needle_len, long offset, const char *from_encoding)
4911 {
4912 int n;
4913 mbfl_string haystack, needle;
4914 n = -1;
4915
4916 mbfl_string_init(&haystack);
4917 mbfl_string_init(&needle);
4918 haystack.no_language = MBSTRG(language);
4919 haystack.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
4920 needle.no_language = MBSTRG(language);
4921 needle.no_encoding = MBSTRG(current_internal_encoding)->no_encoding;
4922
4923 do {
4924 size_t len = 0;
4925 haystack.val = (unsigned char *)php_unicode_convert_case(PHP_UNICODE_CASE_UPPER, (char *)old_haystack, old_haystack_len, &len, from_encoding);
4926 haystack.len = len;
4927
4928 if (!haystack.val) {
4929 break;
4930 }
4931
4932 if (haystack.len <= 0) {
4933 break;
4934 }
4935
4936 needle.val = (unsigned char *)php_unicode_convert_case(PHP_UNICODE_CASE_UPPER, (char *)old_needle, old_needle_len, &len, from_encoding);
4937 needle.len = len;
4938
4939 if (!needle.val) {
4940 break;
4941 }
4942
4943 if (needle.len <= 0) {
4944 break;
4945 }
4946
4947 haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(from_encoding);
4948 if (haystack.no_encoding == mbfl_no_encoding_invalid) {
4949 php_error_docref(NULL, E_WARNING, "Unknown encoding \"%s\"", from_encoding);
4950 break;
4951 }
4952
4953 {
4954 int haystack_char_len = mbfl_strlen(&haystack);
4955
4956 if (mode) {
4957 if ((offset > 0 && offset > haystack_char_len) ||
4958 (offset < 0 && -offset > haystack_char_len)) {
4959 php_error_docref(NULL, E_WARNING, "Offset is greater than the length of haystack string");
4960 break;
4961 }
4962 } else {
4963 if (offset < 0) {
4964 offset += (long)haystack_char_len;
4965 }
4966 if (offset < 0 || offset > haystack_char_len) {
4967 php_error_docref(NULL, E_WARNING, "Offset not contained in string");
4968 break;
4969 }
4970 }
4971 }
4972
4973 n = mbfl_strpos(&haystack, &needle, offset, mode);
4974 } while(0);
4975
4976 if (haystack.val) {
4977 efree(haystack.val);
4978 }
4979
4980 if (needle.val) {
4981 efree(needle.val);
4982 }
4983
4984 return n;
4985 }
4986 /* }}} */
4987
php_mb_gpc_get_detect_order(const zend_encoding *** list,size_t * list_size)4988 static void php_mb_gpc_get_detect_order(const zend_encoding ***list, size_t *list_size) /* {{{ */
4989 {
4990 *list = (const zend_encoding **)MBSTRG(http_input_list);
4991 *list_size = MBSTRG(http_input_list_size);
4992 }
4993 /* }}} */
4994
php_mb_gpc_set_input_encoding(const zend_encoding * encoding)4995 static void php_mb_gpc_set_input_encoding(const zend_encoding *encoding) /* {{{ */
4996 {
4997 MBSTRG(http_input_identify) = (const mbfl_encoding*)encoding;
4998 }
4999 /* }}} */
5000
5001 #endif /* HAVE_MBSTRING */
5002
5003 /*
5004 * Local variables:
5005 * tab-width: 4
5006 * c-basic-offset: 4
5007 * End:
5008 * vim600: fdm=marker
5009 * vim: noet sw=4 ts=4
5010 */
5011