1 : /*
2 : +----------------------------------------------------------------------+
3 : | PHP Version 6 |
4 : +----------------------------------------------------------------------+
5 : | Copyright (c) 1997-2009 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: mbstring.c 288613 2009-09-23 15:22:47Z moriyoshi $ */
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/php_smart_str.h"
60 : #include "ext/standard/url.h"
61 : #include "main/php_output.h"
62 : #include "ext/standard/info.h"
63 :
64 : #include "libmbfl/mbfl/mbfl_allocators.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 : #ifdef ZEND_MULTIBYTE
81 : #include "zend_multibyte.h"
82 : #endif /* ZEND_MULTIBYTE */
83 :
84 : #if HAVE_ONIG
85 : #include "php_onig_compat.h"
86 : #include <oniguruma.h>
87 : #undef UChar
88 : #elif HAVE_PCRE || HAVE_BUNDLED_PCRE
89 : #include "ext/pcre/php_pcre.h"
90 : #endif
91 : /* }}} */
92 :
93 : #if HAVE_MBSTRING
94 :
95 : /* {{{ prototypes */
96 : ZEND_DECLARE_MODULE_GLOBALS(mbstring)
97 :
98 : static PHP_GINIT_FUNCTION(mbstring);
99 : static PHP_GSHUTDOWN_FUNCTION(mbstring);
100 :
101 : #ifdef ZEND_MULTIBYTE
102 : static size_t php_mb_oddlen(const unsigned char *string, size_t length, const char *encoding TSRMLS_DC);
103 : static int php_mb_encoding_converter(unsigned char **to, size_t *to_length, const unsigned char *from, size_t from_length, const char *encoding_to, const char *encoding_from TSRMLS_DC);
104 : static char* php_mb_encoding_detector(const unsigned char *arg_string, size_t arg_length, char *arg_list TSRMLS_DC);
105 : static int php_mb_set_zend_encoding(TSRMLS_D);
106 : #endif
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 : int 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_stripos"},
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 : #if HAVE_MBREGEX
208 : {MB_OVERLOAD_REGEX, "ereg", "mb_ereg", "mb_orig_ereg"},
209 : {MB_OVERLOAD_REGEX, "eregi", "mb_eregi", "mb_orig_eregi"},
210 : {MB_OVERLOAD_REGEX, "ereg_replace", "mb_ereg_replace", "mb_orig_ereg_replace"},
211 : {MB_OVERLOAD_REGEX, "eregi_replace", "mb_eregi_replace", "mb_orig_eregi_replace"},
212 : {MB_OVERLOAD_REGEX, "split", "mb_split", "mb_orig_split"},
213 : #endif
214 : {0, NULL, NULL, NULL}
215 : };
216 : /* }}} */
217 :
218 : /* {{{ arginfo */
219 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_language, 0, 0, 0)
220 : ZEND_ARG_INFO(0, language)
221 : ZEND_END_ARG_INFO()
222 :
223 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_internal_encoding, 0, 0, 0)
224 : ZEND_ARG_INFO(0, encoding)
225 : ZEND_END_ARG_INFO()
226 :
227 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_http_input, 0, 0, 0)
228 : ZEND_ARG_INFO(0, type)
229 : ZEND_END_ARG_INFO()
230 :
231 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_http_output, 0, 0, 0)
232 : ZEND_ARG_INFO(0, encoding)
233 : ZEND_END_ARG_INFO()
234 :
235 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_detect_order, 0, 0, 0)
236 : ZEND_ARG_INFO(0, encoding)
237 : ZEND_END_ARG_INFO()
238 :
239 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_substitute_character, 0, 0, 0)
240 : ZEND_ARG_INFO(0, substchar)
241 : ZEND_END_ARG_INFO()
242 :
243 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_preferred_mime_name, 0, 0, 1)
244 : ZEND_ARG_INFO(0, encoding)
245 : ZEND_END_ARG_INFO()
246 :
247 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_parse_str, 0, 0, 1)
248 : ZEND_ARG_INFO(0, encoded_string)
249 : ZEND_ARG_INFO(1, result)
250 : ZEND_END_ARG_INFO()
251 :
252 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_output_handler, 0, 0, 2)
253 : ZEND_ARG_INFO(0, contents)
254 : ZEND_ARG_INFO(0, status)
255 : ZEND_END_ARG_INFO()
256 :
257 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strlen, 0, 0, 1)
258 : ZEND_ARG_INFO(0, str)
259 : ZEND_ARG_INFO(0, encoding)
260 : ZEND_END_ARG_INFO()
261 :
262 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strpos, 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_strrpos, 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_stripos, 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_strripos, 0, 0, 2)
284 : ZEND_ARG_INFO(0, haystack)
285 : ZEND_ARG_INFO(0, needle)
286 : ZEND_ARG_INFO(0, offset)
287 : ZEND_ARG_INFO(0, encoding)
288 : ZEND_END_ARG_INFO()
289 :
290 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strstr, 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_strrchr, 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_stristr, 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_strrichr, 0, 0, 2)
312 : ZEND_ARG_INFO(0, haystack)
313 : ZEND_ARG_INFO(0, needle)
314 : ZEND_ARG_INFO(0, part)
315 : ZEND_ARG_INFO(0, encoding)
316 : ZEND_END_ARG_INFO()
317 :
318 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_substr_count, 0, 0, 2)
319 : ZEND_ARG_INFO(0, haystack)
320 : ZEND_ARG_INFO(0, needle)
321 : ZEND_ARG_INFO(0, encoding)
322 : ZEND_END_ARG_INFO()
323 :
324 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_substr, 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_strcut, 0, 0, 2)
332 : ZEND_ARG_INFO(0, str)
333 : ZEND_ARG_INFO(0, start)
334 : ZEND_ARG_INFO(0, length)
335 : ZEND_ARG_INFO(0, encoding)
336 : ZEND_END_ARG_INFO()
337 :
338 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strwidth, 0, 0, 1)
339 : ZEND_ARG_INFO(0, str)
340 : ZEND_ARG_INFO(0, encoding)
341 : ZEND_END_ARG_INFO()
342 :
343 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strimwidth, 0, 0, 3)
344 : ZEND_ARG_INFO(0, str)
345 : ZEND_ARG_INFO(0, start)
346 : ZEND_ARG_INFO(0, width)
347 : ZEND_ARG_INFO(0, trimmarker)
348 : ZEND_ARG_INFO(0, encoding)
349 : ZEND_END_ARG_INFO()
350 :
351 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_convert_encoding, 0, 0, 2)
352 : ZEND_ARG_INFO(0, str)
353 : ZEND_ARG_INFO(0, to)
354 : ZEND_ARG_INFO(0, from)
355 : ZEND_END_ARG_INFO()
356 :
357 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_convert_case, 0, 0, 2)
358 : ZEND_ARG_INFO(0, sourcestring)
359 : ZEND_ARG_INFO(0, mode)
360 : ZEND_ARG_INFO(0, encoding)
361 : ZEND_END_ARG_INFO()
362 :
363 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strtoupper, 0, 0, 1)
364 : ZEND_ARG_INFO(0, sourcestring)
365 : ZEND_ARG_INFO(0, encoding)
366 : ZEND_END_ARG_INFO()
367 :
368 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_strtolower, 0, 0, 1)
369 : ZEND_ARG_INFO(0, sourcestring)
370 : ZEND_ARG_INFO(0, encoding)
371 : ZEND_END_ARG_INFO()
372 :
373 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_detect_encoding, 0, 0, 1)
374 : ZEND_ARG_INFO(0, str)
375 : ZEND_ARG_INFO(0, encoding_list)
376 : ZEND_ARG_INFO(0, strict)
377 : ZEND_END_ARG_INFO()
378 :
379 : ZEND_BEGIN_ARG_INFO(arginfo_mb_list_encodings, 0)
380 : ZEND_END_ARG_INFO()
381 :
382 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_encoding_aliases, 0, 0, 1)
383 : ZEND_ARG_INFO(0, encoding)
384 : ZEND_END_ARG_INFO()
385 :
386 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_encode_mimeheader, 0, 0, 1)
387 : ZEND_ARG_INFO(0, str)
388 : ZEND_ARG_INFO(0, charset)
389 : ZEND_ARG_INFO(0, transfer)
390 : ZEND_ARG_INFO(0, linefeed)
391 : ZEND_ARG_INFO(0, indent)
392 : ZEND_END_ARG_INFO()
393 :
394 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_decode_mimeheader, 0, 0, 1)
395 : ZEND_ARG_INFO(0, string)
396 : ZEND_END_ARG_INFO()
397 :
398 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_convert_kana, 0, 0, 1)
399 : ZEND_ARG_INFO(0, str)
400 : ZEND_ARG_INFO(0, option)
401 : ZEND_ARG_INFO(0, encoding)
402 : ZEND_END_ARG_INFO()
403 :
404 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_convert_variables, 1, 0, 3)
405 : ZEND_ARG_INFO(0, to)
406 : ZEND_ARG_INFO(0, from)
407 : ZEND_ARG_INFO(1, ...)
408 : ZEND_END_ARG_INFO()
409 :
410 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_encode_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_decode_numericentity, 0, 0, 2)
417 : ZEND_ARG_INFO(0, string)
418 : ZEND_ARG_INFO(0, convmap)
419 : ZEND_ARG_INFO(0, encoding)
420 : ZEND_END_ARG_INFO()
421 :
422 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_send_mail, 0, 0, 3)
423 : ZEND_ARG_INFO(0, to)
424 : ZEND_ARG_INFO(0, subject)
425 : ZEND_ARG_INFO(0, message)
426 : ZEND_ARG_INFO(0, additional_headers)
427 : ZEND_ARG_INFO(0, additional_parameters)
428 : ZEND_END_ARG_INFO()
429 :
430 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_get_info, 0, 0, 0)
431 : ZEND_ARG_INFO(0, type)
432 : ZEND_END_ARG_INFO()
433 :
434 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_check_encoding, 0, 0, 0)
435 : ZEND_ARG_INFO(0, var)
436 : ZEND_ARG_INFO(0, encoding)
437 : ZEND_END_ARG_INFO()
438 :
439 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_regex_encoding, 0, 0, 0)
440 : ZEND_ARG_INFO(0, encoding)
441 : ZEND_END_ARG_INFO()
442 :
443 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg, 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_eregi, 0, 0, 2)
450 : ZEND_ARG_INFO(0, pattern)
451 : ZEND_ARG_INFO(0, string)
452 : ZEND_ARG_INFO(1, registers)
453 : ZEND_END_ARG_INFO()
454 :
455 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_replace, 0, 0, 3)
456 : ZEND_ARG_INFO(0, pattern)
457 : ZEND_ARG_INFO(0, replacement)
458 : ZEND_ARG_INFO(0, string)
459 : ZEND_ARG_INFO(0, option)
460 : ZEND_END_ARG_INFO()
461 :
462 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_eregi_replace, 0, 0, 3)
463 : ZEND_ARG_INFO(0, pattern)
464 : ZEND_ARG_INFO(0, replacement)
465 : ZEND_ARG_INFO(0, string)
466 : ZEND_END_ARG_INFO()
467 :
468 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_split, 0, 0, 2)
469 : ZEND_ARG_INFO(0, pattern)
470 : ZEND_ARG_INFO(0, string)
471 : ZEND_ARG_INFO(0, limit)
472 : ZEND_END_ARG_INFO()
473 :
474 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_match, 0, 0, 2)
475 : ZEND_ARG_INFO(0, pattern)
476 : ZEND_ARG_INFO(0, string)
477 : ZEND_ARG_INFO(0, option)
478 : ZEND_END_ARG_INFO()
479 :
480 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search, 0, 0, 0)
481 : ZEND_ARG_INFO(0, pattern)
482 : ZEND_ARG_INFO(0, option)
483 : ZEND_END_ARG_INFO()
484 :
485 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search_pos, 0, 0, 0)
486 : ZEND_ARG_INFO(0, pattern)
487 : ZEND_ARG_INFO(0, option)
488 : ZEND_END_ARG_INFO()
489 :
490 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search_regs, 0, 0, 0)
491 : ZEND_ARG_INFO(0, pattern)
492 : ZEND_ARG_INFO(0, option)
493 : ZEND_END_ARG_INFO()
494 :
495 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search_init, 0, 0, 1)
496 : ZEND_ARG_INFO(0, string)
497 : ZEND_ARG_INFO(0, pattern)
498 : ZEND_ARG_INFO(0, option)
499 : ZEND_END_ARG_INFO()
500 :
501 : ZEND_BEGIN_ARG_INFO(arginfo_mb_ereg_search_getregs, 0)
502 : ZEND_END_ARG_INFO()
503 :
504 : ZEND_BEGIN_ARG_INFO(arginfo_mb_ereg_search_getpos, 0)
505 : ZEND_END_ARG_INFO()
506 :
507 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_ereg_search_setpos, 0, 0, 1)
508 : ZEND_ARG_INFO(0, position)
509 : ZEND_END_ARG_INFO()
510 :
511 : ZEND_BEGIN_ARG_INFO_EX(arginfo_mb_regex_set_options, 0, 0, 0)
512 : ZEND_ARG_INFO(0, options)
513 : ZEND_END_ARG_INFO()
514 : /* }}} */
515 :
516 : /* {{{ zend_function_entry mbstring_functions[] */
517 : const zend_function_entry mbstring_functions[] = {
518 : PHP_FE(mb_convert_case, arginfo_mb_convert_case)
519 : PHP_FE(mb_strtoupper, arginfo_mb_strtoupper)
520 : PHP_FE(mb_strtolower, arginfo_mb_strtolower)
521 : PHP_FE(mb_language, arginfo_mb_language)
522 : PHP_FE(mb_internal_encoding, arginfo_mb_internal_encoding)
523 : PHP_FE(mb_http_input, arginfo_mb_http_input)
524 : PHP_FE(mb_http_output, arginfo_mb_http_output)
525 : PHP_FE(mb_detect_order, arginfo_mb_detect_order)
526 : PHP_FE(mb_substitute_character, arginfo_mb_substitute_character)
527 : PHP_FE(mb_parse_str, arginfo_mb_parse_str)
528 : PHP_FE(mb_output_handler, arginfo_mb_output_handler)
529 : PHP_FE(mb_preferred_mime_name, arginfo_mb_preferred_mime_name)
530 : PHP_FE(mb_strlen, arginfo_mb_strlen)
531 : PHP_FE(mb_strpos, arginfo_mb_strpos)
532 : PHP_FE(mb_strrpos, arginfo_mb_strrpos)
533 : PHP_FE(mb_stripos, arginfo_mb_stripos)
534 : PHP_FE(mb_strripos, arginfo_mb_strripos)
535 : PHP_FE(mb_strstr, arginfo_mb_strstr)
536 : PHP_FE(mb_strrchr, arginfo_mb_strrchr)
537 : PHP_FE(mb_stristr, arginfo_mb_stristr)
538 : PHP_FE(mb_strrichr, arginfo_mb_strrichr)
539 : PHP_FE(mb_substr_count, arginfo_mb_substr_count)
540 : PHP_FE(mb_substr, arginfo_mb_substr)
541 : PHP_FE(mb_strcut, arginfo_mb_strcut)
542 : PHP_FE(mb_strwidth, arginfo_mb_strwidth)
543 : PHP_FE(mb_strimwidth, arginfo_mb_strimwidth)
544 : PHP_FE(mb_convert_encoding, arginfo_mb_convert_encoding)
545 : PHP_FE(mb_detect_encoding, arginfo_mb_detect_encoding)
546 : PHP_FE(mb_list_encodings, arginfo_mb_list_encodings)
547 : PHP_FE(mb_encoding_aliases, arginfo_mb_encoding_aliases)
548 : PHP_FE(mb_convert_kana, arginfo_mb_convert_kana)
549 : PHP_FE(mb_encode_mimeheader, arginfo_mb_encode_mimeheader)
550 : PHP_FE(mb_decode_mimeheader, arginfo_mb_decode_mimeheader)
551 : PHP_FE(mb_convert_variables, arginfo_mb_convert_variables)
552 : PHP_FE(mb_encode_numericentity, arginfo_mb_encode_numericentity)
553 : PHP_FE(mb_decode_numericentity, arginfo_mb_decode_numericentity)
554 : PHP_FE(mb_send_mail, arginfo_mb_send_mail)
555 : PHP_FE(mb_get_info, arginfo_mb_get_info)
556 : PHP_FE(mb_check_encoding, arginfo_mb_check_encoding)
557 : #if HAVE_MBREGEX
558 : PHP_MBREGEX_FUNCTION_ENTRIES
559 : #endif
560 : { NULL, NULL, NULL }
561 : };
562 : /* }}} */
563 :
564 : /* {{{ zend_module_entry mbstring_module_entry */
565 : zend_module_entry mbstring_module_entry = {
566 : STANDARD_MODULE_HEADER,
567 : "mbstring",
568 : mbstring_functions,
569 : PHP_MINIT(mbstring),
570 : PHP_MSHUTDOWN(mbstring),
571 : PHP_RINIT(mbstring),
572 : PHP_RSHUTDOWN(mbstring),
573 : PHP_MINFO(mbstring),
574 : NO_VERSION_YET,
575 : PHP_MODULE_GLOBALS(mbstring),
576 : PHP_GINIT(mbstring),
577 : PHP_GSHUTDOWN(mbstring),
578 : NULL,
579 : STANDARD_MODULE_PROPERTIES_EX
580 : };
581 : /* }}} */
582 :
583 : /* {{{ static sapi_post_entry php_post_entries[] */
584 : static sapi_post_entry php_post_entries[] = {
585 : { DEFAULT_POST_CONTENT_TYPE, sizeof(DEFAULT_POST_CONTENT_TYPE)-1, sapi_read_standard_form_data, php_std_post_handler },
586 : { MULTIPART_CONTENT_TYPE, sizeof(MULTIPART_CONTENT_TYPE)-1, NULL, rfc1867_post_handler },
587 : { NULL, 0, NULL, NULL }
588 : };
589 : /* }}} */
590 :
591 : #ifdef COMPILE_DL_MBSTRING
592 : ZEND_GET_MODULE(mbstring)
593 : #endif
594 :
595 : /* {{{ allocators */
596 : static void *_php_mb_allocators_malloc(unsigned int sz)
597 71623 : {
598 71623 : return emalloc(sz);
599 : }
600 :
601 : static void *_php_mb_allocators_realloc(void *ptr, unsigned int sz)
602 15844 : {
603 15844 : return erealloc(ptr, sz);
604 : }
605 :
606 : static void *_php_mb_allocators_calloc(unsigned int nelems, unsigned int szelem)
607 6 : {
608 6 : return ecalloc(nelems, szelem);
609 : }
610 :
611 : static void _php_mb_allocators_free(void *ptr)
612 57260 : {
613 57260 : efree(ptr);
614 57260 : }
615 :
616 : static void *_php_mb_allocators_pmalloc(unsigned int sz)
617 0 : {
618 0 : return pemalloc(sz, 1);
619 : }
620 :
621 : static void *_php_mb_allocators_prealloc(void *ptr, unsigned int sz)
622 0 : {
623 0 : return perealloc(ptr, sz, 1);
624 : }
625 :
626 : static void _php_mb_allocators_pfree(void *ptr)
627 0 : {
628 0 : pefree(ptr, 1);
629 0 : }
630 :
631 : static mbfl_allocators _php_mb_allocators = {
632 : _php_mb_allocators_malloc,
633 : _php_mb_allocators_realloc,
634 : _php_mb_allocators_calloc,
635 : _php_mb_allocators_free,
636 : _php_mb_allocators_pmalloc,
637 : _php_mb_allocators_prealloc,
638 : _php_mb_allocators_pfree
639 : };
640 : /* }}} */
641 :
642 : /* {{{ static sapi_post_entry mbstr_post_entries[] */
643 : static sapi_post_entry mbstr_post_entries[] = {
644 : { DEFAULT_POST_CONTENT_TYPE, sizeof(DEFAULT_POST_CONTENT_TYPE)-1, sapi_read_standard_form_data, php_mb_post_handler },
645 : { MULTIPART_CONTENT_TYPE, sizeof(MULTIPART_CONTENT_TYPE)-1, NULL, rfc1867_post_handler },
646 : { NULL, 0, NULL, NULL }
647 : };
648 : /* }}} */
649 :
650 : /* {{{ static int php_mb_parse_encoding_list()
651 : * Return 0 if input contains any illegal encoding, otherwise 1.
652 : * Even if any illegal encoding is detected the result may contain a list
653 : * of parsed encodings.
654 : */
655 : static int
656 : php_mb_parse_encoding_list(const char *value, int value_length, enum mbfl_no_encoding **return_list, int *return_size, int persistent TSRMLS_DC)
657 133731 : {
658 133731 : int n, l, size, bauto, ret = 1;
659 : char *p, *p1, *p2, *endp, *tmpstr;
660 : enum mbfl_no_encoding no_encoding;
661 : enum mbfl_no_encoding *src, *entry, *list;
662 :
663 133731 : list = NULL;
664 133731 : if (value == NULL || value_length <= 0) {
665 17006 : if (return_list) {
666 17006 : *return_list = NULL;
667 : }
668 17006 : if (return_size) {
669 17006 : *return_size = 0;
670 : }
671 17006 : return 0;
672 : } else {
673 : enum mbfl_no_encoding *identify_list;
674 : int identify_list_size;
675 :
676 116725 : identify_list = MBSTRG(default_detect_order_list);
677 116725 : identify_list_size = MBSTRG(default_detect_order_list_size);
678 :
679 : /* copy the value string for work */
680 116725 : if (value[0]=='"' && value[value_length-1]=='"' && value_length>2) {
681 0 : tmpstr = (char *)estrndup(value+1, value_length-2);
682 0 : value_length -= 2;
683 : }
684 : else
685 116725 : tmpstr = (char *)estrndup(value, value_length);
686 116725 : if (tmpstr == NULL) {
687 0 : return 0;
688 : }
689 : /* count the number of listed encoding names */
690 116725 : endp = tmpstr + value_length;
691 116725 : n = 1;
692 116725 : p1 = tmpstr;
693 233456 : while ((p2 = php_memnstr(p1, ",", 1, endp)) != NULL) {
694 6 : p1 = p2 + 1;
695 6 : n++;
696 : }
697 116725 : size = n + identify_list_size;
698 : /* make list */
699 116725 : list = (enum mbfl_no_encoding *)pecalloc(size, sizeof(int), persistent);
700 116725 : if (list != NULL) {
701 116725 : entry = list;
702 116725 : n = 0;
703 116725 : bauto = 0;
704 116725 : p1 = tmpstr;
705 : do {
706 116731 : p2 = p = php_memnstr(p1, ",", 1, endp);
707 116731 : if (p == NULL) {
708 116725 : p = endp;
709 : }
710 116731 : *p = '\0';
711 : /* trim spaces */
712 233462 : while (p1 < p && (*p1 == ' ' || *p1 == '\t')) {
713 0 : p1++;
714 : }
715 116731 : p--;
716 233462 : while (p > p1 && (*p == ' ' || *p == '\t')) {
717 0 : *p = '\0';
718 0 : p--;
719 : }
720 : /* convert to the encoding number and check encoding */
721 116731 : if (strcasecmp(p1, "auto") == 0) {
722 1 : if (!bauto) {
723 1 : bauto = 1;
724 1 : l = identify_list_size;
725 1 : src = identify_list;
726 7 : while (l > 0) {
727 5 : *entry++ = *src++;
728 5 : l--;
729 5 : n++;
730 : }
731 : }
732 : } else {
733 116730 : no_encoding = mbfl_name2no_encoding(p1);
734 116730 : if (no_encoding != mbfl_no_encoding_invalid) {
735 116729 : *entry++ = no_encoding;
736 116729 : n++;
737 : } else {
738 1 : ret = 0;
739 : }
740 : }
741 116731 : p1 = p2 + 1;
742 116731 : } while (n < size && p2 != NULL);
743 116725 : if (n > 0) {
744 116724 : if (return_list) {
745 31689 : *return_list = list;
746 : } else {
747 85035 : pefree(list, persistent);
748 : }
749 : } else {
750 1 : pefree(list, persistent);
751 1 : if (return_list) {
752 1 : *return_list = NULL;
753 : }
754 1 : ret = 0;
755 : }
756 116725 : if (return_size) {
757 31690 : *return_size = n;
758 : }
759 : } else {
760 0 : if (return_list) {
761 0 : *return_list = NULL;
762 : }
763 0 : if (return_size) {
764 0 : *return_size = 0;
765 : }
766 0 : ret = 0;
767 : }
768 116725 : efree(tmpstr);
769 : }
770 :
771 116725 : return ret;
772 : }
773 : /* }}} */
774 :
775 : /* {{{ MBSTRING_API php_mb_check_encoding_list */
776 85035 : MBSTRING_API int php_mb_check_encoding_list(const char *encoding_list TSRMLS_DC) {
777 85035 : return php_mb_parse_encoding_list(encoding_list, strlen(encoding_list), NULL, NULL, 0 TSRMLS_CC);
778 : }
779 : /* }}} */
780 :
781 : /* {{{ static int php_mb_parse_encoding_array()
782 : * Return 0 if input contains any illegal encoding, otherwise 1.
783 : * Even if any illegal encoding is detected the result may contain a list
784 : * of parsed encodings.
785 : */
786 : static int
787 : php_mb_parse_encoding_array(zval *array, enum mbfl_no_encoding **return_list, int *return_size, int persistent TSRMLS_DC)
788 2 : {
789 : zval **hash_entry;
790 : HashTable *target_hash;
791 2 : int i, n, l, size, bauto,ret = 1;
792 : enum mbfl_no_encoding no_encoding;
793 : enum mbfl_no_encoding *src, *list, *entry;
794 :
795 2 : list = NULL;
796 2 : if (Z_TYPE_P(array) == IS_ARRAY) {
797 : enum mbfl_no_encoding *identify_list;
798 : int identify_list_size;
799 :
800 2 : identify_list = MBSTRG(default_detect_order_list);
801 2 : identify_list_size = MBSTRG(default_detect_order_list_size);
802 :
803 2 : target_hash = Z_ARRVAL_P(array);
804 2 : zend_hash_internal_pointer_reset(target_hash);
805 2 : i = zend_hash_num_elements(target_hash);
806 2 : size = i + identify_list_size;
807 2 : list = (enum mbfl_no_encoding *)pecalloc(size, sizeof(int), persistent);
808 2 : if (list != NULL) {
809 2 : entry = list;
810 2 : bauto = 0;
811 2 : n = 0;
812 13 : while (i > 0) {
813 9 : if (zend_hash_get_current_data(target_hash, (void **) &hash_entry) == FAILURE) {
814 0 : break;
815 : }
816 9 : convert_to_string_ex(hash_entry);
817 9 : if (strcasecmp(Z_STRVAL_PP(hash_entry), "auto") == 0) {
818 0 : if (!bauto) {
819 0 : bauto = 1;
820 0 : l = identify_list_size;
821 0 : src = identify_list;
822 0 : while (l > 0) {
823 0 : *entry++ = *src++;
824 0 : l--;
825 0 : n++;
826 : }
827 : }
828 : } else {
829 9 : no_encoding = mbfl_name2no_encoding(Z_STRVAL_PP(hash_entry));
830 9 : if (no_encoding != mbfl_no_encoding_invalid) {
831 8 : *entry++ = no_encoding;
832 8 : n++;
833 : } else {
834 1 : ret = 0;
835 : }
836 : }
837 9 : zend_hash_move_forward(target_hash);
838 9 : i--;
839 : }
840 2 : if (n > 0) {
841 2 : if (return_list) {
842 2 : *return_list = list;
843 : } else {
844 0 : pefree(list, persistent);
845 : }
846 : } else {
847 0 : pefree(list, persistent);
848 0 : if (return_list) {
849 0 : *return_list = NULL;
850 : }
851 0 : ret = 0;
852 : }
853 2 : if (return_size) {
854 2 : *return_size = n;
855 : }
856 : } else {
857 0 : if (return_list) {
858 0 : *return_list = NULL;
859 : }
860 0 : if (return_size) {
861 0 : *return_size = 0;
862 : }
863 0 : ret = 0;
864 : }
865 : }
866 :
867 2 : return ret;
868 : }
869 : /* }}} */
870 :
871 : static void *_php_mb_compile_regex(const char *pattern TSRMLS_DC);
872 : static int _php_mb_match_regex(void *opaque, const char *str, size_t str_len);
873 : static void _php_mb_free_regex(void *opaque);
874 :
875 : #if HAVE_ONIG
876 : /* {{{ _php_mb_compile_regex */
877 : static void *_php_mb_compile_regex(const char *pattern TSRMLS_DC)
878 : {
879 : php_mb_regex_t *retval;
880 : OnigErrorInfo err_info;
881 : int err_code;
882 :
883 : if ((err_code = onig_new(&retval,
884 : (const OnigUChar *)pattern,
885 : (const OnigUChar *)pattern + strlen(pattern),
886 : ONIG_OPTION_IGNORECASE | ONIG_OPTION_DONT_CAPTURE_GROUP,
887 : ONIG_ENCODING_ASCII, &OnigSyntaxPerl, &err_info))) {
888 : OnigUChar err_str[ONIG_MAX_ERROR_MESSAGE_LEN];
889 : onig_error_code_to_str(err_str, err_code, err_info);
890 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "%s: %s", pattern, err_str);
891 : retval = NULL;
892 : }
893 : return retval;
894 : }
895 : /* }}} */
896 :
897 : /* {{{ _php_mb_match_regex */
898 : static int _php_mb_match_regex(void *opaque, const char *str, size_t str_len)
899 : {
900 : return onig_search((php_mb_regex_t *)opaque, (const OnigUChar *)str,
901 : (const OnigUChar*)str + str_len, (const OnigUChar *)str,
902 : (const OnigUChar*)str + str_len, NULL, ONIG_OPTION_NONE) >= 0;
903 : }
904 : /* }}} */
905 :
906 : /* {{{ _php_mb_free_regex */
907 : static void _php_mb_free_regex(void *opaque)
908 : {
909 : onig_free((php_mb_regex_t *)opaque);
910 : }
911 : /* }}} */
912 : #elif HAVE_PCRE || HAVE_BUNDLED_PCRE
913 : /* {{{ _php_mb_compile_regex */
914 : static void *_php_mb_compile_regex(const char *pattern TSRMLS_DC)
915 17011 : {
916 : pcre *retval;
917 : const char *err_str;
918 : int err_offset;
919 :
920 17011 : if (!(retval = pcre_compile(pattern,
921 : PCRE_CASELESS, &err_str, &err_offset, NULL))) {
922 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "%s (offset=%d): %s", pattern, err_offset, err_str);
923 : }
924 17011 : return retval;
925 : }
926 : /* }}} */
927 :
928 : /* {{{ _php_mb_match_regex */
929 : static int _php_mb_match_regex(void *opaque, const char *str, size_t str_len)
930 12 : {
931 12 : return pcre_exec((pcre *)opaque, NULL, str, (int)str_len, 0,
932 : 0, NULL, 0) >= 0;
933 : }
934 : /* }}} */
935 :
936 : /* {{{ _php_mb_free_regex */
937 : static void _php_mb_free_regex(void *opaque)
938 17043 : {
939 17043 : pcre_free(opaque);
940 17043 : }
941 : /* }}} */
942 : #endif
943 :
944 : /* {{{ php_mb_nls_get_default_detect_order_list */
945 : static int php_mb_nls_get_default_detect_order_list(enum mbfl_no_language lang, enum mbfl_no_encoding **plist, int* plist_size)
946 17015 : {
947 : size_t i;
948 :
949 17015 : *plist = (enum mbfl_no_encoding *) php_mb_default_identify_list_neut;
950 17015 : *plist_size = sizeof(php_mb_default_identify_list_neut) / sizeof(php_mb_default_identify_list_neut[0]);
951 :
952 153066 : for (i = 0; i < sizeof(php_mb_default_identify_list) / sizeof(php_mb_default_identify_list[0]); i++) {
953 153064 : if (php_mb_default_identify_list[i].lang == lang) {
954 17013 : *plist = (enum mbfl_no_encoding *)php_mb_default_identify_list[i].list;
955 17013 : *plist_size = php_mb_default_identify_list[i].list_size;
956 17013 : return 1;
957 : }
958 : }
959 2 : return 0;
960 : }
961 : /* }}} */
962 :
963 : /* {{{ php.ini directive handler */
964 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_language) */
965 : static PHP_INI_MH(OnUpdate_mbstring_language)
966 17015 : {
967 : enum mbfl_no_language no_language;
968 :
969 17015 : no_language = mbfl_name2no_language(new_value);
970 17015 : if (no_language == mbfl_no_language_invalid) {
971 0 : MBSTRG(language) = mbfl_no_language_neutral;
972 0 : return FAILURE;
973 : }
974 17015 : MBSTRG(language) = no_language;
975 17015 : php_mb_nls_get_default_detect_order_list(no_language, &MBSTRG(default_detect_order_list), &MBSTRG(default_detect_order_list_size));
976 17015 : return SUCCESS;
977 : }
978 : /* }}} */
979 :
980 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_detect_order) */
981 : static PHP_INI_MH(OnUpdate_mbstring_detect_order)
982 17007 : {
983 : enum mbfl_no_encoding *list;
984 : int size;
985 :
986 17007 : if (php_mb_parse_encoding_list(new_value, new_value_length, &list, &size, 1 TSRMLS_CC)) {
987 1 : if (MBSTRG(detect_order_list)) {
988 0 : free(MBSTRG(detect_order_list));
989 : }
990 1 : MBSTRG(detect_order_list) = list;
991 1 : MBSTRG(detect_order_list_size) = size;
992 : } else {
993 17006 : if (MBSTRG(detect_order_list)) {
994 0 : free(MBSTRG(detect_order_list));
995 0 : MBSTRG(detect_order_list) = NULL;
996 : }
997 17006 : return FAILURE;
998 : }
999 :
1000 1 : return SUCCESS;
1001 : }
1002 : /* }}} */
1003 :
1004 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_input) */
1005 : static PHP_INI_MH(OnUpdate_mbstring_http_input)
1006 17007 : {
1007 : enum mbfl_no_encoding *list;
1008 : int size;
1009 :
1010 17007 : if (php_mb_parse_encoding_list(new_value, new_value_length, &list, &size, 1 TSRMLS_CC)) {
1011 17007 : if (MBSTRG(http_input_list)) {
1012 0 : free(MBSTRG(http_input_list));
1013 : }
1014 17007 : MBSTRG(http_input_list) = list;
1015 17007 : MBSTRG(http_input_list_size) = size;
1016 : } else {
1017 0 : if (MBSTRG(http_input_list)) {
1018 0 : free(MBSTRG(http_input_list));
1019 0 : MBSTRG(http_input_list) = NULL;
1020 : }
1021 0 : MBSTRG(http_input_list_size) = 0;
1022 0 : return FAILURE;
1023 : }
1024 :
1025 17007 : return SUCCESS;
1026 : }
1027 : /* }}} */
1028 :
1029 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_output) */
1030 : static PHP_INI_MH(OnUpdate_mbstring_http_output)
1031 17007 : {
1032 : enum mbfl_no_encoding no_encoding;
1033 :
1034 17007 : no_encoding = mbfl_name2no_encoding(new_value);
1035 17007 : if (no_encoding != mbfl_no_encoding_invalid) {
1036 17007 : MBSTRG(http_output_encoding) = no_encoding;
1037 17007 : MBSTRG(current_http_output_encoding) = no_encoding;
1038 : } else {
1039 0 : MBSTRG(http_output_encoding) = mbfl_no_encoding_pass;
1040 0 : MBSTRG(current_http_output_encoding) = mbfl_no_encoding_pass;
1041 0 : if (new_value != NULL && new_value_length > 0) {
1042 0 : return FAILURE;
1043 : }
1044 : }
1045 :
1046 17007 : return SUCCESS;
1047 : }
1048 : /* }}} */
1049 :
1050 : /* {{{ static _php_mb_ini_mbstring_internal_encoding_set */
1051 : int _php_mb_ini_mbstring_internal_encoding_set(const char *new_value, uint new_value_length TSRMLS_DC)
1052 34042 : {
1053 : enum mbfl_no_encoding no_encoding;
1054 34042 : const char *enc_name = NULL;
1055 34042 : uint enc_name_len = 0;
1056 :
1057 34042 : no_encoding = new_value ? mbfl_name2no_encoding(new_value):
1058 : mbfl_no_encoding_invalid;
1059 34042 : if (no_encoding != mbfl_no_encoding_invalid) {
1060 129 : enc_name = new_value;
1061 129 : enc_name_len = new_value_length;
1062 : } else {
1063 33913 : switch (MBSTRG(language)) {
1064 : case mbfl_no_language_uni:
1065 0 : enc_name = "UTF-8";
1066 0 : enc_name_len = sizeof("UTF-8") - 1;
1067 0 : break;
1068 : case mbfl_no_language_japanese:
1069 8 : enc_name = "EUC-JP";
1070 8 : enc_name_len = sizeof("EUC-JP") - 1;
1071 8 : break;
1072 : case mbfl_no_language_korean:
1073 0 : enc_name = "EUC-KR";
1074 0 : enc_name_len = sizeof("EUC-KR") - 1;
1075 0 : break;
1076 : case mbfl_no_language_simplified_chinese:
1077 0 : enc_name = "EUC-CN";
1078 0 : enc_name_len = sizeof("EUC-CN") - 1;
1079 0 : break;
1080 : case mbfl_no_language_traditional_chinese:
1081 0 : enc_name = "EUC-TW";
1082 0 : enc_name_len = sizeof("EUC-TW") - 1;
1083 0 : break;
1084 : case mbfl_no_language_russian:
1085 0 : enc_name = "KOI8-R";
1086 0 : enc_name_len = sizeof("KOI8-R") - 1;
1087 0 : break;
1088 : case mbfl_no_language_german:
1089 0 : enc_name = "ISO-8859-15";
1090 0 : enc_name_len = sizeof("ISO-8859-15") - 1;
1091 0 : break;
1092 : case mbfl_no_language_armenian:
1093 0 : enc_name = "ArmSCII-8";
1094 0 : enc_name_len = sizeof("ArmSCII-8") - 1;
1095 0 : break;
1096 : case mbfl_no_language_turkish:
1097 0 : enc_name = "ISO-8859-9";
1098 0 : enc_name_len = sizeof("ISO-8859-9") - 1;
1099 0 : break;
1100 : default:
1101 33905 : enc_name = "ISO-8859-1";
1102 33905 : enc_name_len = sizeof("ISO-8859-1") - 1;
1103 : break;
1104 : }
1105 33913 : no_encoding = mbfl_name2no_encoding(enc_name);
1106 : }
1107 34042 : MBSTRG(internal_encoding) = no_encoding;
1108 34042 : MBSTRG(current_internal_encoding) = no_encoding;
1109 : #if HAVE_MBREGEX
1110 : {
1111 : const char *enc_name = new_value;
1112 : if (FAILURE == php_mb_regex_set_default_mbctype(enc_name TSRMLS_CC)) {
1113 : /* falls back to EUC-JP if an unknown encoding name is given */
1114 : enc_name = "EUC-JP";
1115 : php_mb_regex_set_default_mbctype(enc_name TSRMLS_CC);
1116 : }
1117 : php_mb_regex_set_mbctype(new_value TSRMLS_CC);
1118 : }
1119 : #endif
1120 34042 : return SUCCESS;
1121 : }
1122 : /* }}} */
1123 :
1124 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_internal_encoding) */
1125 : static PHP_INI_MH(OnUpdate_mbstring_internal_encoding)
1126 17009 : {
1127 17009 : if (stage == PHP_INI_STAGE_STARTUP || stage == PHP_INI_STAGE_SHUTDOWN
1128 : || stage == PHP_INI_STAGE_RUNTIME) {
1129 17008 : return _php_mb_ini_mbstring_internal_encoding_set(new_value, new_value_length TSRMLS_CC);
1130 : } else {
1131 : /* the corresponding mbstring globals needs to be set according to the
1132 : * ini value in the later stage because it never falls back to the
1133 : * default value if 1. no value for mbstring.internal_encoding is given,
1134 : * 2. mbstring.language directive is processed in per-dir or runtime
1135 : * context and 3. call to the handler for mbstring.language is done
1136 : * after mbstring.internal_encoding is handled. */
1137 1 : return SUCCESS;
1138 : }
1139 : }
1140 : /* }}} */
1141 :
1142 : #ifdef ZEND_MULTIBYTE
1143 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_script_encoding) */
1144 : static PHP_INI_MH(OnUpdate_mbstring_script_encoding)
1145 : {
1146 : int *list, size;
1147 :
1148 : if (php_mb_parse_encoding_list(new_value, new_value_length, &list, &size, 1 TSRMLS_CC)) {
1149 : if (MBSTRG(script_encoding_list) != NULL) {
1150 : free(MBSTRG(script_encoding_list));
1151 : }
1152 : MBSTRG(script_encoding_list) = list;
1153 : MBSTRG(script_encoding_list_size) = size;
1154 : } else {
1155 : if (MBSTRG(script_encoding_list) != NULL) {
1156 : free(MBSTRG(script_encoding_list));
1157 : }
1158 : MBSTRG(script_encoding_list) = NULL;
1159 : MBSTRG(script_encoding_list_size) = 0;
1160 : return FAILURE;
1161 : }
1162 :
1163 : return SUCCESS;
1164 : }
1165 : /* }}} */
1166 : #endif /* ZEND_MULTIBYTE */
1167 :
1168 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_substitute_character) */
1169 : static PHP_INI_MH(OnUpdate_mbstring_substitute_character)
1170 17007 : {
1171 : int c;
1172 17007 : char *endptr = NULL;
1173 :
1174 17007 : if (new_value != NULL) {
1175 1 : if (strcasecmp("none", new_value) == 0) {
1176 0 : MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
1177 0 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
1178 1 : } else if (strcasecmp("long", new_value) == 0) {
1179 0 : MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG;
1180 0 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG;
1181 1 : } else if (strcasecmp("entity", new_value) == 0) {
1182 0 : MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY;
1183 0 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY;
1184 : } else {
1185 1 : MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1186 1 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1187 1 : if (new_value_length >0) {
1188 1 : c = strtol(new_value, &endptr, 0);
1189 1 : if (*endptr == '\0') {
1190 1 : MBSTRG(filter_illegal_substchar) = c;
1191 1 : MBSTRG(current_filter_illegal_substchar) = c;
1192 : }
1193 : }
1194 : }
1195 : } else {
1196 17006 : MBSTRG(filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1197 17006 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1198 17006 : MBSTRG(filter_illegal_substchar) = 0x3f; /* '?' */
1199 17006 : MBSTRG(current_filter_illegal_substchar) = 0x3f; /* '?' */
1200 : }
1201 :
1202 17007 : return SUCCESS;
1203 : }
1204 : /* }}} */
1205 :
1206 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_encoding_translation) */
1207 : static PHP_INI_MH(OnUpdate_mbstring_encoding_translation)
1208 17007 : {
1209 17007 : if (new_value == NULL) {
1210 0 : return FAILURE;
1211 : }
1212 :
1213 17007 : OnUpdateBool(entry, new_value, new_value_length, mh_arg1, mh_arg2, mh_arg3, stage TSRMLS_CC);
1214 :
1215 17007 : if (MBSTRG(encoding_translation)) {
1216 3 : sapi_unregister_post_entry(php_post_entries TSRMLS_CC);
1217 3 : sapi_register_post_entries(mbstr_post_entries TSRMLS_CC);
1218 : } else {
1219 17004 : sapi_unregister_post_entry(mbstr_post_entries TSRMLS_CC);
1220 17004 : sapi_register_post_entries(php_post_entries TSRMLS_CC);
1221 : }
1222 :
1223 17007 : return SUCCESS;
1224 : }
1225 : /* }}} */
1226 :
1227 : /* {{{ static PHP_INI_MH(OnUpdate_mbstring_http_output_conv_mimetypes */
1228 : static PHP_INI_MH(OnUpdate_mbstring_http_output_conv_mimetypes)
1229 17011 : {
1230 : zval tmp;
1231 17011 : void *re = NULL;
1232 :
1233 17011 : if (!new_value) {
1234 0 : new_value = entry->orig_value;
1235 0 : new_value_length = entry->orig_value_length;
1236 : }
1237 17011 : php_trim(new_value, new_value_length, NULL, 0, &tmp, 3 TSRMLS_CC);
1238 :
1239 17011 : if (Z_STRLEN(tmp) > 0) {
1240 17011 : if (!(re = _php_mb_compile_regex(Z_STRVAL(tmp) TSRMLS_CC))) {
1241 0 : zval_dtor(&tmp);
1242 0 : return FAILURE;
1243 : }
1244 : }
1245 :
1246 17011 : if (MBSTRG(http_output_conv_mimetypes)) {
1247 4 : _php_mb_free_regex(MBSTRG(http_output_conv_mimetypes));
1248 : }
1249 :
1250 17011 : MBSTRG(http_output_conv_mimetypes) = re;
1251 :
1252 17011 : zval_dtor(&tmp);
1253 17011 : return SUCCESS;
1254 : }
1255 : /* }}} */
1256 : /* }}} */
1257 :
1258 : /* {{{ php.ini directive registration */
1259 : PHP_INI_BEGIN()
1260 : PHP_INI_ENTRY("mbstring.language", "neutral", PHP_INI_ALL, OnUpdate_mbstring_language)
1261 : PHP_INI_ENTRY("mbstring.detect_order", NULL, PHP_INI_ALL, OnUpdate_mbstring_detect_order)
1262 : PHP_INI_ENTRY("mbstring.http_input", "pass", PHP_INI_ALL, OnUpdate_mbstring_http_input)
1263 : PHP_INI_ENTRY("mbstring.http_output", "pass", PHP_INI_ALL, OnUpdate_mbstring_http_output)
1264 : PHP_INI_ENTRY("mbstring.internal_encoding", NULL, PHP_INI_ALL, OnUpdate_mbstring_internal_encoding)
1265 : #ifdef ZEND_MULTIBYTE
1266 : PHP_INI_ENTRY("mbstring.script_encoding", NULL, PHP_INI_ALL, OnUpdate_mbstring_script_encoding)
1267 : #endif /* ZEND_MULTIBYTE */
1268 : PHP_INI_ENTRY("mbstring.substitute_character", NULL, PHP_INI_ALL, OnUpdate_mbstring_substitute_character)
1269 : STD_PHP_INI_ENTRY("mbstring.func_overload", "0",
1270 : PHP_INI_SYSTEM, OnUpdateLong, func_overload, zend_mbstring_globals, mbstring_globals)
1271 :
1272 : STD_PHP_INI_BOOLEAN("mbstring.encoding_translation", "0",
1273 : PHP_INI_SYSTEM | PHP_INI_PERDIR,
1274 : OnUpdate_mbstring_encoding_translation,
1275 : encoding_translation, zend_mbstring_globals, mbstring_globals)
1276 : PHP_INI_ENTRY("mbstring.http_output_conv_mimetypes",
1277 : "^(text/|application/xhtml\\+xml)",
1278 : PHP_INI_ALL,
1279 : OnUpdate_mbstring_http_output_conv_mimetypes)
1280 :
1281 : STD_PHP_INI_BOOLEAN("mbstring.strict_detection", "0",
1282 : PHP_INI_ALL,
1283 : OnUpdateLong,
1284 : strict_detection, zend_mbstring_globals, mbstring_globals)
1285 : PHP_INI_END()
1286 : /* }}} */
1287 :
1288 : /* {{{ module global initialize handler */
1289 : static PHP_GINIT_FUNCTION(mbstring)
1290 17007 : {
1291 17007 : mbstring_globals->language = mbfl_no_language_uni;
1292 17007 : mbstring_globals->internal_encoding = mbfl_no_encoding_invalid;
1293 17007 : mbstring_globals->current_internal_encoding = mbstring_globals->internal_encoding;
1294 : #ifdef ZEND_MULTIBYTE
1295 : mbstring_globals->script_encoding_list = NULL;
1296 : mbstring_globals->script_encoding_list_size = 0;
1297 : #endif /* ZEND_MULTIBYTE */
1298 17007 : mbstring_globals->http_output_encoding = mbfl_no_encoding_pass;
1299 17007 : mbstring_globals->current_http_output_encoding = mbfl_no_encoding_pass;
1300 17007 : mbstring_globals->http_input_identify = mbfl_no_encoding_invalid;
1301 17007 : mbstring_globals->http_input_identify_get = mbfl_no_encoding_invalid;
1302 17007 : mbstring_globals->http_input_identify_post = mbfl_no_encoding_invalid;
1303 17007 : mbstring_globals->http_input_identify_cookie = mbfl_no_encoding_invalid;
1304 17007 : mbstring_globals->http_input_identify_string = mbfl_no_encoding_invalid;
1305 17007 : mbstring_globals->http_input_list = NULL;
1306 17007 : mbstring_globals->http_input_list_size = 0;
1307 17007 : mbstring_globals->detect_order_list = NULL;
1308 17007 : mbstring_globals->detect_order_list_size = 0;
1309 17007 : mbstring_globals->current_detect_order_list = NULL;
1310 17007 : mbstring_globals->current_detect_order_list_size = 0;
1311 17007 : mbstring_globals->default_detect_order_list = (enum mbfl_no_encoding *) php_mb_default_identify_list_neut;
1312 17007 : mbstring_globals->default_detect_order_list_size = sizeof(php_mb_default_identify_list_neut) / sizeof(php_mb_default_identify_list_neut[0]);
1313 17007 : mbstring_globals->filter_illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1314 17007 : mbstring_globals->filter_illegal_substchar = 0x3f; /* '?' */
1315 17007 : mbstring_globals->current_filter_illegal_mode = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1316 17007 : mbstring_globals->current_filter_illegal_substchar = 0x3f; /* '?' */
1317 17007 : mbstring_globals->illegalchars = 0;
1318 17007 : mbstring_globals->func_overload = 0;
1319 17007 : mbstring_globals->encoding_translation = 0;
1320 17007 : mbstring_globals->strict_detection = 0;
1321 17007 : mbstring_globals->outconv = NULL;
1322 17007 : mbstring_globals->http_output_conv_mimetypes = NULL;
1323 : #if HAVE_MBREGEX
1324 : mbstring_globals->mb_regex_globals = php_mb_regex_globals_alloc(TSRMLS_C);
1325 : #endif
1326 17007 : }
1327 : /* }}} */
1328 :
1329 : /* {{{ PHP_GSHUTDOWN_FUNCTION */
1330 : static PHP_GSHUTDOWN_FUNCTION(mbstring)
1331 17039 : {
1332 17039 : if (mbstring_globals->http_input_list) {
1333 17039 : free(mbstring_globals->http_input_list);
1334 : }
1335 : #ifdef ZEND_MULTIBYTE
1336 : if (mbstring_globals->script_encoding_list) {
1337 : free(mbstring_globals->script_encoding_list);
1338 : }
1339 : #endif /* ZEND_MULTIBYTE */
1340 17039 : if (mbstring_globals->detect_order_list) {
1341 1 : free(mbstring_globals->detect_order_list);
1342 : }
1343 17039 : if (mbstring_globals->http_output_conv_mimetypes) {
1344 17039 : _php_mb_free_regex(mbstring_globals->http_output_conv_mimetypes);
1345 : }
1346 : #if HAVE_MBREGEX
1347 : php_mb_regex_globals_free(mbstring_globals->mb_regex_globals TSRMLS_CC);
1348 : #endif
1349 17039 : }
1350 : /* }}} */
1351 :
1352 : /* {{{ PHP_MINIT_FUNCTION(mbstring) */
1353 : PHP_MINIT_FUNCTION(mbstring)
1354 17007 : {
1355 17007 : __mbfl_allocators = &_php_mb_allocators;
1356 :
1357 17007 : REGISTER_INI_ENTRIES();
1358 :
1359 : /* This is a global handler. Should not be set in a per-request handler. */
1360 17007 : sapi_register_treat_data(mbstr_treat_data);
1361 :
1362 : /* Post handlers are stored in the thread-local context. */
1363 17007 : if (MBSTRG(encoding_translation)) {
1364 3 : sapi_register_post_entries(mbstr_post_entries TSRMLS_CC);
1365 : }
1366 :
1367 17007 : REGISTER_LONG_CONSTANT("MB_OVERLOAD_MAIL", MB_OVERLOAD_MAIL, CONST_CS | CONST_PERSISTENT);
1368 17007 : REGISTER_LONG_CONSTANT("MB_OVERLOAD_STRING", MB_OVERLOAD_STRING, CONST_CS | CONST_PERSISTENT);
1369 17007 : REGISTER_LONG_CONSTANT("MB_OVERLOAD_REGEX", MB_OVERLOAD_REGEX, CONST_CS | CONST_PERSISTENT);
1370 :
1371 17007 : REGISTER_LONG_CONSTANT("MB_CASE_UPPER", PHP_UNICODE_CASE_UPPER, CONST_CS | CONST_PERSISTENT);
1372 17007 : REGISTER_LONG_CONSTANT("MB_CASE_LOWER", PHP_UNICODE_CASE_LOWER, CONST_CS | CONST_PERSISTENT);
1373 17007 : REGISTER_LONG_CONSTANT("MB_CASE_TITLE", PHP_UNICODE_CASE_TITLE, CONST_CS | CONST_PERSISTENT);
1374 :
1375 : #if HAVE_MBREGEX
1376 : PHP_MINIT(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
1377 : #endif
1378 17007 : return SUCCESS;
1379 : }
1380 : /* }}} */
1381 :
1382 : /* {{{ PHP_MSHUTDOWN_FUNCTION(mbstring) */
1383 : PHP_MSHUTDOWN_FUNCTION(mbstring)
1384 17039 : {
1385 17039 : UNREGISTER_INI_ENTRIES();
1386 :
1387 : #if HAVE_MBREGEX
1388 : PHP_MSHUTDOWN(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
1389 : #endif
1390 :
1391 17039 : return SUCCESS;
1392 : }
1393 : /* }}} */
1394 :
1395 : /* {{{ PHP_RINIT_FUNCTION(mbstring) */
1396 : PHP_RINIT_FUNCTION(mbstring)
1397 16993 : {
1398 : int n;
1399 16993 : enum mbfl_no_encoding *list=NULL, *entry;
1400 : zend_function *func, *orig;
1401 : const struct mb_overload_def *p;
1402 :
1403 16993 : MBSTRG(current_internal_encoding) = MBSTRG(internal_encoding);
1404 16993 : MBSTRG(current_http_output_encoding) = MBSTRG(http_output_encoding);
1405 16993 : MBSTRG(current_filter_illegal_mode) = MBSTRG(filter_illegal_mode);
1406 16993 : MBSTRG(current_filter_illegal_substchar) = MBSTRG(filter_illegal_substchar);
1407 :
1408 16993 : MBSTRG(illegalchars) = 0;
1409 :
1410 16993 : n = 0;
1411 16993 : if (MBSTRG(detect_order_list)) {
1412 1 : list = MBSTRG(detect_order_list);
1413 1 : n = MBSTRG(detect_order_list_size);
1414 : }
1415 16993 : if (n <= 0) {
1416 16992 : list = MBSTRG(default_detect_order_list);
1417 16992 : n = MBSTRG(default_detect_order_list_size);
1418 : }
1419 16993 : entry = (enum mbfl_no_encoding *)safe_emalloc(n, sizeof(int), 0);
1420 16993 : MBSTRG(current_detect_order_list) = entry;
1421 16993 : MBSTRG(current_detect_order_list_size) = n;
1422 67998 : while (n > 0) {
1423 34012 : *entry++ = *list++;
1424 34012 : n--;
1425 : }
1426 :
1427 : /* override original function. */
1428 16993 : if (MBSTRG(func_overload)){
1429 1 : p = &(mb_ovld[0]);
1430 :
1431 15 : while (p->type > 0) {
1432 13 : if ((MBSTRG(func_overload) & p->type) == p->type &&
1433 : zend_hash_find(EG(function_table), p->save_func,
1434 : strlen(p->save_func)+1, (void **)&orig) != SUCCESS) {
1435 :
1436 11 : zend_hash_find(EG(function_table), p->ovld_func, strlen(p->ovld_func)+1 , (void **)&func);
1437 :
1438 11 : if (zend_hash_find(EG(function_table), p->orig_func, strlen(p->orig_func)+1, (void **)&orig) != SUCCESS) {
1439 0 : php_error_docref("ref.mbstring" TSRMLS_CC, E_WARNING, "mbstring couldn't find function %s.", p->orig_func);
1440 0 : return FAILURE;
1441 : } else {
1442 11 : zend_hash_add(EG(function_table), p->save_func, strlen(p->save_func)+1, orig, sizeof(zend_function), NULL);
1443 :
1444 11 : if (zend_hash_update(EG(function_table), p->orig_func, strlen(p->orig_func)+1, func, sizeof(zend_function),
1445 : NULL) == FAILURE) {
1446 0 : php_error_docref("ref.mbstring" TSRMLS_CC, E_WARNING, "mbstring couldn't replace function %s.", p->orig_func);
1447 0 : return FAILURE;
1448 : }
1449 : }
1450 : }
1451 13 : p++;
1452 : }
1453 : }
1454 : #if HAVE_MBREGEX
1455 : PHP_RINIT(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
1456 : #endif
1457 : #ifdef ZEND_MULTIBYTE
1458 : zend_multibyte_set_internal_encoding(mbfl_no_encoding2name(MBSTRG(internal_encoding)) TSRMLS_CC);
1459 : php_mb_set_zend_encoding(TSRMLS_C);
1460 : #endif /* ZEND_MULTIBYTE */
1461 :
1462 16993 : return SUCCESS;
1463 : }
1464 : /* }}} */
1465 :
1466 : /* {{{ PHP_RSHUTDOWN_FUNCTION(mbstring) */
1467 : PHP_RSHUTDOWN_FUNCTION(mbstring)
1468 17025 : {
1469 : const struct mb_overload_def *p;
1470 : zend_function *orig;
1471 :
1472 17025 : if (MBSTRG(current_detect_order_list) != NULL) {
1473 17025 : efree(MBSTRG(current_detect_order_list));
1474 17025 : MBSTRG(current_detect_order_list) = NULL;
1475 17025 : MBSTRG(current_detect_order_list_size) = 0;
1476 : }
1477 17025 : if (MBSTRG(outconv) != NULL) {
1478 0 : MBSTRG(illegalchars) += mbfl_buffer_illegalchars(MBSTRG(outconv));
1479 0 : mbfl_buffer_converter_delete(MBSTRG(outconv));
1480 0 : MBSTRG(outconv) = NULL;
1481 : }
1482 :
1483 : /* clear http input identification. */
1484 17025 : MBSTRG(http_input_identify) = mbfl_no_encoding_invalid;
1485 17025 : MBSTRG(http_input_identify_post) = mbfl_no_encoding_invalid;
1486 17025 : MBSTRG(http_input_identify_get) = mbfl_no_encoding_invalid;
1487 17025 : MBSTRG(http_input_identify_cookie) = mbfl_no_encoding_invalid;
1488 17025 : MBSTRG(http_input_identify_string) = mbfl_no_encoding_invalid;
1489 :
1490 : /* clear overloaded function. */
1491 17025 : if (MBSTRG(func_overload)){
1492 1 : p = &(mb_ovld[0]);
1493 15 : while (p->type > 0) {
1494 13 : if ((MBSTRG(func_overload) & p->type) == p->type &&
1495 : zend_hash_find(EG(function_table), p->save_func,
1496 : strlen(p->save_func)+1, (void **)&orig) == SUCCESS) {
1497 :
1498 11 : zend_hash_update(EG(function_table), p->orig_func, strlen(p->orig_func)+1, orig, sizeof(zend_function), NULL);
1499 11 : zend_hash_del(EG(function_table), p->save_func, strlen(p->save_func)+1);
1500 : }
1501 13 : p++;
1502 : }
1503 : }
1504 :
1505 : #if HAVE_MBREGEX
1506 : PHP_RSHUTDOWN(mb_regex) (INIT_FUNC_ARGS_PASSTHRU);
1507 : #endif
1508 :
1509 17025 : return SUCCESS;
1510 : }
1511 : /* }}} */
1512 :
1513 : /* {{{ PHP_MINFO_FUNCTION(mbstring) */
1514 : PHP_MINFO_FUNCTION(mbstring)
1515 43 : {
1516 43 : php_info_print_table_start();
1517 43 : php_info_print_table_row(2, "Multibyte Support", "enabled");
1518 43 : php_info_print_table_row(2, "Multibyte string engine", "libmbfl");
1519 43 : php_info_print_table_row(2, "HTTP input encoding translation", MBSTRG(encoding_translation) ? "enabled": "disabled");
1520 43 : php_info_print_table_end();
1521 :
1522 43 : php_info_print_table_start();
1523 43 : 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.");
1524 43 : php_info_print_table_end();
1525 :
1526 : #if HAVE_MBREGEX
1527 : PHP_MINFO(mb_regex)(ZEND_MODULE_INFO_FUNC_ARGS_PASSTHRU);
1528 : #endif
1529 :
1530 43 : DISPLAY_INI_ENTRIES();
1531 43 : }
1532 : /* }}} */
1533 :
1534 : /* {{{ proto string mb_language([string language])
1535 : Sets the current language or Returns the current language as a string */
1536 : PHP_FUNCTION(mb_language)
1537 8 : {
1538 8 : char *name = NULL;
1539 8 : int name_len = 0;
1540 :
1541 8 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|s", &name, &name_len) == FAILURE) {
1542 0 : return;
1543 : }
1544 8 : if (name == NULL) {
1545 4 : RETVAL_STRING((char *)mbfl_no_language2name(MBSTRG(language)), 1);
1546 : } else {
1547 4 : if (FAILURE == zend_alter_ini_entry(
1548 : "mbstring.language", sizeof("mbstring.language"),
1549 : name, name_len, PHP_INI_USER, PHP_INI_STAGE_RUNTIME)) {
1550 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown language \"%s\"", name);
1551 0 : RETVAL_FALSE;
1552 : } else {
1553 4 : RETVAL_TRUE;
1554 : }
1555 : }
1556 : }
1557 : /* }}} */
1558 :
1559 : /* {{{ proto string mb_internal_encoding([string encoding])
1560 : Sets the current internal encoding or Returns the current internal encoding as a string */
1561 : PHP_FUNCTION(mb_internal_encoding)
1562 282 : {
1563 282 : char *name = NULL;
1564 : int name_len;
1565 : enum mbfl_no_encoding no_encoding;
1566 :
1567 282 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|s", &name, &name_len) == FAILURE) {
1568 4 : RETURN_FALSE;
1569 : }
1570 278 : if (name == NULL) {
1571 134 : name = (char *)mbfl_no_encoding2name(MBSTRG(current_internal_encoding));
1572 134 : if (name != NULL) {
1573 134 : RETURN_STRING(name, 1);
1574 : } else {
1575 0 : RETURN_FALSE;
1576 : }
1577 : } else {
1578 144 : no_encoding = mbfl_name2no_encoding(name);
1579 144 : if (no_encoding == mbfl_no_encoding_invalid) {
1580 21 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", name);
1581 21 : RETURN_FALSE;
1582 : } else {
1583 123 : MBSTRG(current_internal_encoding) = no_encoding;
1584 : #ifdef ZEND_MULTIBYTE
1585 : /* TODO: make independent from mbstring.encoding_translation? */
1586 : if (MBSTRG(encoding_translation)) {
1587 : zend_multibyte_set_internal_encoding(name TSRMLS_CC);
1588 : }
1589 : #endif /* ZEND_MULTIBYTE */
1590 123 : RETURN_TRUE;
1591 : }
1592 : }
1593 : }
1594 : /* }}} */
1595 :
1596 : /* {{{ proto mixed mb_http_input([string type])
1597 : Returns the input encoding */
1598 : PHP_FUNCTION(mb_http_input)
1599 2 : {
1600 2 : char *typ = NULL;
1601 : int typ_len;
1602 : int retname, n;
1603 : char *name, *list, *temp;
1604 : enum mbfl_no_encoding *entry;
1605 2 : enum mbfl_no_encoding result = mbfl_no_encoding_invalid;
1606 :
1607 2 : retname = 1;
1608 2 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|s", &typ, &typ_len) == FAILURE) {
1609 0 : RETURN_FALSE;
1610 : }
1611 2 : if (typ == NULL) {
1612 1 : result = MBSTRG(http_input_identify);
1613 : } else {
1614 1 : switch (*typ) {
1615 : case 'G':
1616 : case 'g':
1617 0 : result = MBSTRG(http_input_identify_get);
1618 0 : break;
1619 : case 'P':
1620 : case 'p':
1621 0 : result = MBSTRG(http_input_identify_post);
1622 0 : break;
1623 : case 'C':
1624 : case 'c':
1625 0 : result = MBSTRG(http_input_identify_cookie);
1626 0 : break;
1627 : case 'S':
1628 : case 's':
1629 0 : result = MBSTRG(http_input_identify_string);
1630 0 : break;
1631 : case 'I':
1632 : case 'i':
1633 0 : array_init(return_value);
1634 0 : entry = MBSTRG(http_input_list);
1635 0 : n = MBSTRG(http_input_list_size);
1636 0 : while (n > 0) {
1637 0 : name = (char *)mbfl_no_encoding2name(*entry);
1638 0 : if (name) {
1639 0 : add_next_index_string(return_value, name, 1);
1640 : }
1641 0 : entry++;
1642 0 : n--;
1643 : }
1644 0 : retname = 0;
1645 0 : break;
1646 : case 'L':
1647 : case 'l':
1648 1 : entry = MBSTRG(http_input_list);
1649 1 : n = MBSTRG(http_input_list_size);
1650 1 : list = NULL;
1651 3 : while (n > 0) {
1652 1 : name = (char *)mbfl_no_encoding2name(*entry);
1653 1 : if (name) {
1654 1 : if (list) {
1655 0 : temp = list;
1656 0 : spprintf(&list, 0, "%s,%s", temp, name);
1657 0 : efree(temp);
1658 0 : if (!list) {
1659 0 : break;
1660 : }
1661 : } else {
1662 1 : list = estrdup(name);
1663 : }
1664 : }
1665 1 : entry++;
1666 1 : n--;
1667 : }
1668 1 : if (!list) {
1669 0 : RETURN_FALSE;
1670 : }
1671 1 : RETVAL_STRING(list, 0);
1672 1 : retname = 0;
1673 1 : break;
1674 : default:
1675 0 : result = MBSTRG(http_input_identify);
1676 : break;
1677 : }
1678 : }
1679 :
1680 2 : if (retname) {
1681 2 : if (result != mbfl_no_encoding_invalid &&
1682 : (name = (char *)mbfl_no_encoding2name(result)) != NULL) {
1683 1 : RETVAL_STRING(name, 1);
1684 : } else {
1685 0 : RETVAL_FALSE;
1686 : }
1687 : }
1688 : }
1689 : /* }}} */
1690 :
1691 : /* {{{ proto string mb_http_output([string encoding])
1692 : Sets the current output_encoding or returns the current output_encoding as a string */
1693 : PHP_FUNCTION(mb_http_output)
1694 31 : {
1695 31 : char *name = NULL;
1696 : int name_len;
1697 : enum mbfl_no_encoding no_encoding;
1698 :
1699 31 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|s", (char **)&name, &name_len) == FAILURE) {
1700 2 : RETURN_FALSE;
1701 : }
1702 :
1703 29 : if (name == NULL) {
1704 9 : name = (char *)mbfl_no_encoding2name(MBSTRG(current_http_output_encoding));
1705 9 : if (name != NULL) {
1706 9 : RETURN_STRING(name, 1);
1707 : } else {
1708 0 : RETURN_FALSE;
1709 : }
1710 : } else {
1711 20 : no_encoding = mbfl_name2no_encoding(name);
1712 20 : if (no_encoding == mbfl_no_encoding_invalid) {
1713 1 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", name);
1714 1 : RETURN_FALSE;
1715 : } else {
1716 19 : MBSTRG(current_http_output_encoding) = no_encoding;
1717 19 : RETURN_TRUE;
1718 : }
1719 : }
1720 : }
1721 : /* }}} */
1722 :
1723 : /* {{{ proto bool|array mb_detect_order([mixed encoding-list])
1724 : Sets the current detect_order or Return the current detect_order as a array */
1725 : PHP_FUNCTION(mb_detect_order)
1726 10 : {
1727 10 : zval **arg1 = NULL;
1728 : int n, size;
1729 : enum mbfl_no_encoding *list, *entry;
1730 : char *name;
1731 :
1732 10 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|Z", &arg1) == FAILURE) {
1733 0 : return;
1734 : }
1735 :
1736 10 : if (!arg1) {
1737 5 : array_init(return_value);
1738 5 : entry = MBSTRG(current_detect_order_list);
1739 5 : n = MBSTRG(current_detect_order_list_size);
1740 31 : while (n > 0) {
1741 21 : name = (char *)mbfl_no_encoding2name(*entry);
1742 21 : if (name) {
1743 21 : add_next_index_string(return_value, name, 1);
1744 : }
1745 21 : entry++;
1746 21 : n--;
1747 : }
1748 : } else {
1749 5 : list = NULL;
1750 5 : size = 0;
1751 5 : switch (Z_TYPE_PP(arg1)) {
1752 : case IS_ARRAY:
1753 2 : if (!php_mb_parse_encoding_array(*arg1, &list, &size, 0 TSRMLS_CC)) {
1754 1 : if (list) {
1755 1 : efree(list);
1756 : }
1757 1 : RETURN_FALSE;
1758 : }
1759 1 : break;
1760 : default:
1761 3 : convert_to_string_ex(arg1);
1762 3 : if (!php_mb_parse_encoding_list(Z_STRVAL_PP(arg1), Z_STRLEN_PP(arg1), &list, &size, 0 TSRMLS_CC)) {
1763 1 : if (list) {
1764 0 : efree(list);
1765 : }
1766 1 : RETURN_FALSE;
1767 : }
1768 : break;
1769 : }
1770 :
1771 3 : if (list == NULL) {
1772 0 : RETURN_FALSE;
1773 : }
1774 :
1775 3 : if (MBSTRG(current_detect_order_list)) {
1776 3 : efree(MBSTRG(current_detect_order_list));
1777 : }
1778 3 : MBSTRG(current_detect_order_list) = list;
1779 3 : MBSTRG(current_detect_order_list_size) = size;
1780 3 : RETURN_TRUE;
1781 : }
1782 : }
1783 : /* }}} */
1784 :
1785 : /* {{{ proto mixed mb_substitute_character([mixed substchar])
1786 : Sets the current substitute_character or returns the current substitute_character */
1787 : PHP_FUNCTION(mb_substitute_character)
1788 50 : {
1789 50 : zval **arg1 = NULL;
1790 :
1791 50 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|Z", &arg1) == FAILURE) {
1792 1 : return;
1793 : }
1794 :
1795 49 : if (!arg1) {
1796 8 : if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
1797 0 : RETURN_STRING("none", 1);
1798 8 : } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG) {
1799 0 : RETURN_STRING("long", 1);
1800 8 : } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY) {
1801 0 : RETURN_STRING("entity", 1);
1802 : } else {
1803 8 : RETURN_LONG(MBSTRG(current_filter_illegal_substchar));
1804 : }
1805 : } else {
1806 41 : RETVAL_TRUE;
1807 :
1808 41 : switch (Z_TYPE_PP(arg1)) {
1809 : case IS_STRING:
1810 0 : if (strncasecmp("none", Z_STRVAL_PP(arg1), Z_STRLEN_PP(arg1)) == 0) {
1811 0 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE;
1812 0 : } else if (strncasecmp("long", Z_STRVAL_PP(arg1), Z_STRLEN_PP(arg1)) == 0) {
1813 0 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG;
1814 0 : } else if (strncasecmp("entity", Z_STRVAL_PP(arg1), Z_STRLEN_PP(arg1)) == 0) {
1815 0 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY;
1816 : } else {
1817 0 : convert_to_long_ex(arg1);
1818 :
1819 0 : if (Z_LVAL_PP(arg1) < 0xffff && Z_LVAL_PP(arg1) > 0x0) {
1820 0 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1821 0 : MBSTRG(current_filter_illegal_substchar) = Z_LVAL_PP(arg1);
1822 : } else {
1823 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown character.");
1824 0 : RETURN_FALSE;
1825 : }
1826 : }
1827 0 : break;
1828 : default:
1829 41 : convert_to_long_ex(arg1);
1830 56 : if (Z_LVAL_PP(arg1) < 0xffff && Z_LVAL_PP(arg1) > 0x0) {
1831 15 : MBSTRG(current_filter_illegal_mode) = MBFL_OUTPUTFILTER_ILLEGAL_MODE_CHAR;
1832 15 : MBSTRG(current_filter_illegal_substchar) = Z_LVAL_PP(arg1);
1833 : } else {
1834 26 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown character.");
1835 26 : RETURN_FALSE;
1836 : }
1837 : break;
1838 : }
1839 : }
1840 : }
1841 : /* }}} */
1842 :
1843 : /* {{{ proto string mb_preferred_mime_name(string encoding)
1844 : Return the preferred MIME name (charset) as a string */
1845 : PHP_FUNCTION(mb_preferred_mime_name)
1846 10 : {
1847 : enum mbfl_no_encoding no_encoding;
1848 10 : char *name = NULL;
1849 : int name_len;
1850 :
1851 10 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s", &name, &name_len) == FAILURE) {
1852 0 : return;
1853 : } else {
1854 10 : no_encoding = mbfl_name2no_encoding(name);
1855 10 : if (no_encoding == mbfl_no_encoding_invalid) {
1856 1 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", name);
1857 1 : RETVAL_FALSE;
1858 : } else {
1859 9 : const char *preferred_name = mbfl_no2preferred_mime_name(no_encoding);
1860 9 : if (preferred_name == NULL || *preferred_name == '\0') {
1861 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "No MIME preferred name corresponding to \"%s\"", name);
1862 0 : RETVAL_FALSE;
1863 : } else {
1864 9 : RETVAL_STRING((char *)preferred_name, 1);
1865 : }
1866 : }
1867 : }
1868 : }
1869 : /* }}} */
1870 :
1871 : #define IS_SJIS1(c) ((((c)>=0x81 && (c)<=0x9f) || ((c)>=0xe0 && (c)<=0xf5)) ? 1 : 0)
1872 : #define IS_SJIS2(c) ((((c)>=0x40 && (c)<=0x7e) || ((c)>=0x80 && (c)<=0xfc)) ? 1 : 0)
1873 :
1874 : /* {{{ proto bool mb_parse_str(string encoded_string [, array result])
1875 : Parses GET/POST/COOKIE data and sets global variables */
1876 : PHP_FUNCTION(mb_parse_str)
1877 17 : {
1878 17 : zval *track_vars_array = NULL;
1879 17 : char *encstr = NULL;
1880 : int encstr_len;
1881 : php_mb_encoding_handler_info_t info;
1882 : enum mbfl_no_encoding detected;
1883 :
1884 17 : track_vars_array = NULL;
1885 17 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|z", &encstr, &encstr_len, &track_vars_array) == FAILURE) {
1886 0 : return;
1887 : }
1888 :
1889 : /* Clear out the array */
1890 17 : if (track_vars_array != NULL) {
1891 9 : zval_dtor(track_vars_array);
1892 9 : array_init(track_vars_array);
1893 : }
1894 :
1895 17 : encstr = estrndup(encstr, encstr_len);
1896 :
1897 17 : info.data_type = PARSE_STRING;
1898 17 : info.separator = PG(arg_separator).input;
1899 17 : info.report_errors = 1;
1900 17 : info.to_encoding = MBSTRG(current_internal_encoding);
1901 17 : info.to_language = MBSTRG(language);
1902 17 : info.from_encodings = MBSTRG(http_input_list);
1903 17 : info.num_from_encodings = MBSTRG(http_input_list_size);
1904 17 : info.from_language = MBSTRG(language);
1905 :
1906 17 : detected = _php_mb_encoding_handler_ex(&info, track_vars_array, encstr TSRMLS_CC);
1907 :
1908 17 : MBSTRG(http_input_identify) = detected;
1909 :
1910 17 : RETVAL_BOOL(detected != mbfl_no_encoding_invalid);
1911 :
1912 17 : if (encstr != NULL) efree(encstr);
1913 : }
1914 : /* }}} */
1915 :
1916 : /* {{{ proto string mb_output_handler(string contents, int status)
1917 : Returns string in output buffer converted to the http_output encoding */
1918 : PHP_FUNCTION(mb_output_handler)
1919 16 : {
1920 : char *arg_string;
1921 : int arg_string_len;
1922 : long arg_status;
1923 : mbfl_string string, result;
1924 : const char *charset;
1925 : char *p;
1926 : enum mbfl_no_encoding encoding;
1927 : int last_feed, len;
1928 16 : unsigned char send_text_mimetype = 0;
1929 16 : char *s, *mimetype = NULL;
1930 :
1931 16 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "sl", &arg_string, &arg_string_len, &arg_status) == FAILURE) {
1932 0 : return;
1933 : }
1934 :
1935 16 : encoding = MBSTRG(current_http_output_encoding);
1936 :
1937 : /* start phase only */
1938 16 : if ((arg_status & PHP_OUTPUT_HANDLER_START) != 0) {
1939 : /* delete the converter just in case. */
1940 16 : if (MBSTRG(outconv)) {
1941 0 : MBSTRG(illegalchars) += mbfl_buffer_illegalchars(MBSTRG(outconv));
1942 0 : mbfl_buffer_converter_delete(MBSTRG(outconv));
1943 0 : MBSTRG(outconv) = NULL;
1944 : }
1945 16 : if (encoding == mbfl_no_encoding_pass) {
1946 0 : RETURN_STRINGL(arg_string, arg_string_len, 1);
1947 : }
1948 :
1949 : /* analyze mime type */
1950 24 : if (SG(sapi_headers).mimetype &&
1951 : _php_mb_match_regex(
1952 : MBSTRG(http_output_conv_mimetypes),
1953 : SG(sapi_headers).mimetype,
1954 : strlen(SG(sapi_headers).mimetype))) {
1955 8 : if ((s = strchr(SG(sapi_headers).mimetype,';')) == NULL){
1956 2 : mimetype = estrdup(SG(sapi_headers).mimetype);
1957 : } else {
1958 6 : mimetype = estrndup(SG(sapi_headers).mimetype,s-SG(sapi_headers).mimetype);
1959 : }
1960 8 : send_text_mimetype = 1;
1961 8 : } else if (SG(sapi_headers).send_default_content_type) {
1962 4 : mimetype = SG(default_mimetype) ? SG(default_mimetype) : SAPI_DEFAULT_MIMETYPE;
1963 : }
1964 :
1965 : /* if content-type is not yet set, set it and activate the converter */
1966 16 : if (SG(sapi_headers).send_default_content_type || send_text_mimetype) {
1967 12 : charset = mbfl_no2preferred_mime_name(encoding);
1968 12 : if (charset) {
1969 12 : len = spprintf( &p, 0, "Content-Type: %s; charset=%s", mimetype, charset );
1970 12 : if (sapi_add_header(p, len, 0) != FAILURE) {
1971 12 : SG(sapi_headers).send_default_content_type = 0;
1972 : }
1973 : }
1974 : /* activate the converter */
1975 12 : MBSTRG(outconv) = mbfl_buffer_converter_new(MBSTRG(current_internal_encoding), encoding, 0);
1976 12 : if (send_text_mimetype){
1977 8 : efree(mimetype);
1978 : }
1979 : }
1980 : }
1981 :
1982 : /* just return if the converter is not activated. */
1983 16 : if (MBSTRG(outconv) == NULL) {
1984 4 : RETURN_STRINGL(arg_string, arg_string_len, 1);
1985 : }
1986 :
1987 : /* flag */
1988 12 : last_feed = ((arg_status & PHP_OUTPUT_HANDLER_END) != 0);
1989 : /* mode */
1990 12 : mbfl_buffer_converter_illegal_mode(MBSTRG(outconv), MBSTRG(current_filter_illegal_mode));
1991 12 : mbfl_buffer_converter_illegal_substchar(MBSTRG(outconv), MBSTRG(current_filter_illegal_substchar));
1992 :
1993 : /* feed the string */
1994 12 : mbfl_string_init(&string);
1995 12 : string.no_language = MBSTRG(language);
1996 12 : string.no_encoding = MBSTRG(current_internal_encoding);
1997 12 : string.val = (unsigned char *)arg_string;
1998 12 : string.len = arg_string_len;
1999 12 : mbfl_buffer_converter_feed(MBSTRG(outconv), &string);
2000 12 : if (last_feed) {
2001 12 : mbfl_buffer_converter_flush(MBSTRG(outconv));
2002 : }
2003 : /* get the converter output, and return it */
2004 12 : mbfl_buffer_converter_result(MBSTRG(outconv), &result);
2005 12 : RETVAL_STRINGL((char *)result.val, result.len, 0); /* the string is already strdup()'ed */
2006 :
2007 : /* delete the converter if it is the last feed. */
2008 12 : if (last_feed) {
2009 12 : MBSTRG(illegalchars) += mbfl_buffer_illegalchars(MBSTRG(outconv));
2010 12 : mbfl_buffer_converter_delete(MBSTRG(outconv));
2011 12 : MBSTRG(outconv) = NULL;
2012 : }
2013 : }
2014 : /* }}} */
2015 :
2016 : /* {{{ proto int mb_strlen(string str [, string encoding])
2017 : Get character numbers of a string */
2018 : PHP_FUNCTION(mb_strlen)
2019 217 : {
2020 : int n;
2021 : mbfl_string string;
2022 217 : char *enc_name = NULL;
2023 : int enc_name_len;
2024 :
2025 217 : mbfl_string_init(&string);
2026 :
2027 217 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|s", (char **)&string.val, &string.len, &enc_name, &enc_name_len) == FAILURE) {
2028 4 : RETURN_FALSE;
2029 : }
2030 :
2031 213 : string.no_language = MBSTRG(language);
2032 213 : if (enc_name == NULL) {
2033 1 : string.no_encoding = MBSTRG(current_internal_encoding);
2034 : } else {
2035 212 : string.no_encoding = mbfl_name2no_encoding(enc_name);
2036 212 : if (string.no_encoding == mbfl_no_encoding_invalid) {
2037 20 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2038 20 : RETURN_FALSE;
2039 : }
2040 : }
2041 :
2042 193 : n = mbfl_strlen(&string);
2043 193 : if (n >= 0) {
2044 193 : RETVAL_LONG(n);
2045 : } else {
2046 0 : RETVAL_FALSE;
2047 : }
2048 : }
2049 : /* }}} */
2050 :
2051 : /* {{{ proto int mb_strpos(string haystack, string needle [, int offset [, string encoding]])
2052 : Find position of first occurrence of a string within another */
2053 : PHP_FUNCTION(mb_strpos)
2054 138 : {
2055 138 : int n, reverse = 0;
2056 : long offset;
2057 : mbfl_string haystack, needle;
2058 138 : char *enc_name = NULL;
2059 : int enc_name_len;
2060 :
2061 138 : mbfl_string_init(&haystack);
2062 138 : mbfl_string_init(&needle);
2063 138 : haystack.no_language = MBSTRG(language);
2064 138 : haystack.no_encoding = MBSTRG(current_internal_encoding);
2065 138 : needle.no_language = MBSTRG(language);
2066 138 : needle.no_encoding = MBSTRG(current_internal_encoding);
2067 138 : offset = 0;
2068 :
2069 138 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|ls", (char **)&haystack.val, &haystack.len, (char **)&needle.val, &needle.len, &offset, &enc_name, &enc_name_len) == FAILURE) {
2070 12 : RETURN_FALSE;
2071 : }
2072 :
2073 126 : if (enc_name != NULL) {
2074 101 : haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2075 101 : if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2076 20 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2077 20 : RETURN_FALSE;
2078 : }
2079 : }
2080 :
2081 106 : if (offset < 0 || offset > mbfl_strlen(&haystack)) {
2082 24 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Offset not contained in string");
2083 24 : RETURN_FALSE;
2084 : }
2085 82 : if (needle.len == 0) {
2086 8 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty delimiter");
2087 8 : RETURN_FALSE;
2088 : }
2089 :
2090 74 : n = mbfl_strpos(&haystack, &needle, offset, reverse);
2091 74 : if (n >= 0) {
2092 31 : RETVAL_LONG(n);
2093 : } else {
2094 43 : switch (-n) {
2095 : case 1:
2096 43 : break;
2097 : case 2:
2098 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Needle has not positive length");
2099 0 : break;
2100 : case 4:
2101 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding or conversion error");
2102 0 : break;
2103 : case 8:
2104 0 : php_error_docref(NULL TSRMLS_CC, E_NOTICE, "Argument is empty");
2105 0 : break;
2106 : default:
2107 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown error in mb_strpos");
2108 : break;
2109 : }
2110 43 : RETVAL_FALSE;
2111 : }
2112 : }
2113 : /* }}} */
2114 :
2115 : /* {{{ proto int mb_strrpos(string haystack, string needle [, int offset [, string encoding]])
2116 : Find position of last occurrence of a string within another */
2117 : PHP_FUNCTION(mb_strrpos)
2118 146 : {
2119 : int n;
2120 : mbfl_string haystack, needle;
2121 146 : char *enc_name = NULL;
2122 : int enc_name_len;
2123 146 : zval **zoffset = NULL;
2124 146 : long offset = 0, str_flg;
2125 146 : char *enc_name2 = NULL;
2126 : int enc_name_len2;
2127 :
2128 146 : mbfl_string_init(&haystack);
2129 146 : mbfl_string_init(&needle);
2130 146 : haystack.no_language = MBSTRG(language);
2131 146 : haystack.no_encoding = MBSTRG(current_internal_encoding);
2132 146 : needle.no_language = MBSTRG(language);
2133 146 : needle.no_encoding = MBSTRG(current_internal_encoding);
2134 :
2135 146 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|Zs", (char **)&haystack.val, &haystack.len, (char **)&needle.val, &needle.len, &zoffset, &enc_name, &enc_name_len) == FAILURE) {
2136 5 : RETURN_FALSE;
2137 : }
2138 :
2139 141 : if (zoffset) {
2140 138 : if (Z_TYPE_PP(zoffset) == IS_STRING) {
2141 0 : enc_name2 = Z_STRVAL_PP(zoffset);
2142 0 : enc_name_len2 = Z_STRLEN_PP(zoffset);
2143 0 : str_flg = 1;
2144 :
2145 0 : if (enc_name2 != NULL) {
2146 0 : switch (*enc_name2) {
2147 : case '0':
2148 : case '1':
2149 : case '2':
2150 : case '3':
2151 : case '4':
2152 : case '5':
2153 : case '6':
2154 : case '7':
2155 : case '8':
2156 : case '9':
2157 : case ' ':
2158 : case '-':
2159 : case '.':
2160 0 : break;
2161 : default :
2162 0 : str_flg = 0;
2163 : break;
2164 : }
2165 : }
2166 :
2167 0 : if (str_flg) {
2168 0 : convert_to_long_ex(zoffset);
2169 0 : offset = Z_LVAL_PP(zoffset);
2170 : } else {
2171 0 : enc_name = enc_name2;
2172 0 : enc_name_len = enc_name_len2;
2173 : }
2174 : } else {
2175 138 : convert_to_long_ex(zoffset);
2176 138 : offset = Z_LVAL_PP(zoffset);
2177 : }
2178 : }
2179 :
2180 141 : if (enc_name != NULL) {
2181 109 : haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2182 109 : if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2183 20 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2184 20 : RETURN_FALSE;
2185 : }
2186 : }
2187 :
2188 121 : if (haystack.len <= 0) {
2189 8 : RETURN_FALSE;
2190 : }
2191 113 : if (needle.len <= 0) {
2192 8 : RETURN_FALSE;
2193 : }
2194 :
2195 : {
2196 105 : int haystack_char_len = mbfl_strlen(&haystack);
2197 105 : if ((offset > 0 && offset > haystack_char_len) ||
2198 : (offset < 0 && -offset > haystack_char_len)) {
2199 24 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Offset is greater than the length of haystack string");
2200 24 : RETURN_FALSE;
2201 : }
2202 : }
2203 :
2204 81 : n = mbfl_strpos(&haystack, &needle, offset, 1);
2205 81 : if (n >= 0) {
2206 49 : RETVAL_LONG(n);
2207 : } else {
2208 32 : RETVAL_FALSE;
2209 : }
2210 : }
2211 : /* }}} */
2212 :
2213 : /* {{{ proto int mb_stripos(string haystack, string needle [, int offset [, string encoding]])
2214 : Finds position of first occurrence of a string within another, case insensitive */
2215 : PHP_FUNCTION(mb_stripos)
2216 220 : {
2217 : int n;
2218 : long offset;
2219 : mbfl_string haystack, needle;
2220 220 : char *from_encoding = (char*)mbfl_no2preferred_mime_name(MBSTRG(current_internal_encoding));
2221 : int from_encoding_len;
2222 220 : n = -1;
2223 220 : offset = 0;
2224 :
2225 220 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|ls", (char **)&haystack.val, (int *)&haystack.len, (char **)&needle.val, (int *)&needle.len, &offset, &from_encoding, &from_encoding_len) == FAILURE) {
2226 12 : RETURN_FALSE;
2227 : }
2228 208 : if (needle.len == 0) {
2229 8 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty delimiter");
2230 8 : RETURN_FALSE;
2231 : }
2232 200 : n = php_mb_stripos(0, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, offset, from_encoding TSRMLS_CC);
2233 :
2234 200 : if (n >= 0) {
2235 99 : RETVAL_LONG(n);
2236 : } else {
2237 101 : RETVAL_FALSE;
2238 : }
2239 : }
2240 : /* }}} */
2241 :
2242 : /* {{{ proto int mb_strripos(string haystack, string needle [, int offset [, string encoding]])
2243 : Finds position of last occurrence of a string within another, case insensitive */
2244 : PHP_FUNCTION(mb_strripos)
2245 220 : {
2246 : int n;
2247 : long offset;
2248 : mbfl_string haystack, needle;
2249 220 : const char *from_encoding = mbfl_no2preferred_mime_name(MBSTRG(current_internal_encoding));
2250 : int from_encoding_len;
2251 220 : n = -1;
2252 220 : offset = 0;
2253 :
2254 220 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|ls", (char **)&haystack.val, (int *)&haystack.len, (char **)&needle.val, (int *)&needle.len, &offset, &from_encoding, &from_encoding_len) == FAILURE) {
2255 12 : RETURN_FALSE;
2256 : }
2257 :
2258 208 : n = php_mb_stripos(1, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, offset, from_encoding TSRMLS_CC);
2259 :
2260 208 : if (n >= 0) {
2261 80 : RETVAL_LONG(n);
2262 : } else {
2263 128 : RETVAL_FALSE;
2264 : }
2265 : }
2266 : /* }}} */
2267 :
2268 : /* {{{ proto string mb_strstr(string haystack, string needle[, bool part[, string encoding]])
2269 : Finds first occurrence of a string within another */
2270 : PHP_FUNCTION(mb_strstr)
2271 137 : {
2272 : int n, len, mblen;
2273 137 : mbfl_string haystack, needle, result, *ret = NULL;
2274 137 : char *enc_name = NULL;
2275 : int enc_name_len;
2276 137 : zend_bool part = 0;
2277 :
2278 137 : mbfl_string_init(&haystack);
2279 137 : mbfl_string_init(&needle);
2280 137 : haystack.no_language = MBSTRG(language);
2281 137 : haystack.no_encoding = MBSTRG(current_internal_encoding);
2282 137 : needle.no_language = MBSTRG(language);
2283 137 : needle.no_encoding = MBSTRG(current_internal_encoding);
2284 :
2285 137 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|bs", (char **)&haystack.val, (int *)&haystack.len, (char **)&needle.val, (int *)&needle.len, &part, &enc_name, &enc_name_len) == FAILURE) {
2286 27 : RETURN_FALSE;
2287 : }
2288 :
2289 110 : if (enc_name != NULL) {
2290 88 : haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2291 88 : if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2292 21 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2293 21 : RETURN_FALSE;
2294 : }
2295 : }
2296 :
2297 89 : if (needle.len <= 0) {
2298 8 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty delimiter");
2299 8 : RETURN_FALSE;
2300 : }
2301 81 : n = mbfl_strpos(&haystack, &needle, 0, 0);
2302 81 : if (n >= 0) {
2303 43 : mblen = mbfl_strlen(&haystack);
2304 43 : if (part) {
2305 21 : ret = mbfl_substr(&haystack, &result, 0, n);
2306 21 : if (ret != NULL) {
2307 21 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
2308 : } else {
2309 0 : RETVAL_FALSE;
2310 : }
2311 : } else {
2312 22 : len = (mblen - n);
2313 22 : ret = mbfl_substr(&haystack, &result, n, len);
2314 22 : if (ret != NULL) {
2315 22 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
2316 : } else {
2317 0 : RETVAL_FALSE;
2318 : }
2319 : }
2320 : } else {
2321 38 : RETVAL_FALSE;
2322 : }
2323 : }
2324 : /* }}} */
2325 :
2326 : /* {{{ proto string mb_strrchr(string haystack, string needle[, bool part[, string encoding]])
2327 : Finds the last occurrence of a character in a string within another */
2328 : PHP_FUNCTION(mb_strrchr)
2329 129 : {
2330 : int n, len, mblen;
2331 129 : mbfl_string haystack, needle, result, *ret = NULL;
2332 129 : char *enc_name = NULL;
2333 : int enc_name_len;
2334 129 : zend_bool part = 0;
2335 :
2336 129 : mbfl_string_init(&haystack);
2337 129 : mbfl_string_init(&needle);
2338 129 : haystack.no_language = MBSTRG(language);
2339 129 : haystack.no_encoding = MBSTRG(current_internal_encoding);
2340 129 : needle.no_language = MBSTRG(language);
2341 129 : needle.no_encoding = MBSTRG(current_internal_encoding);
2342 :
2343 129 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|bs", (char **)&haystack.val, &haystack.len, (char **)&needle.val, &needle.len, &part, &enc_name, &enc_name_len) == FAILURE) {
2344 27 : RETURN_FALSE;
2345 : }
2346 :
2347 102 : if (enc_name != NULL) {
2348 86 : haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2349 86 : if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2350 21 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2351 21 : RETURN_FALSE;
2352 : }
2353 : }
2354 :
2355 81 : if (haystack.len <= 0) {
2356 8 : RETURN_FALSE;
2357 : }
2358 73 : if (needle.len <= 0) {
2359 8 : RETURN_FALSE;
2360 : }
2361 65 : n = mbfl_strpos(&haystack, &needle, 0, 1);
2362 65 : if (n >= 0) {
2363 35 : mblen = mbfl_strlen(&haystack);
2364 35 : if (part) {
2365 18 : ret = mbfl_substr(&haystack, &result, 0, n);
2366 18 : if (ret != NULL) {
2367 18 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
2368 : } else {
2369 0 : RETVAL_FALSE;
2370 : }
2371 : } else {
2372 17 : len = (mblen - n);
2373 17 : ret = mbfl_substr(&haystack, &result, n, len);
2374 17 : if (ret != NULL) {
2375 17 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
2376 : } else {
2377 0 : RETVAL_FALSE;
2378 : }
2379 : }
2380 : } else {
2381 30 : RETVAL_FALSE;
2382 : }
2383 : }
2384 : /* }}} */
2385 :
2386 : /* {{{ proto string mb_stristr(string haystack, string needle[, bool part[, string encoding]])
2387 : Finds first occurrence of a string within another, case insensitive */
2388 : PHP_FUNCTION(mb_stristr)
2389 131 : {
2390 131 : zend_bool part = 0;
2391 : unsigned int from_encoding_len, len, mblen;
2392 : int n;
2393 131 : mbfl_string haystack, needle, result, *ret = NULL;
2394 131 : const char *from_encoding = mbfl_no2preferred_mime_name(MBSTRG(current_internal_encoding));
2395 131 : mbfl_string_init(&haystack);
2396 131 : mbfl_string_init(&needle);
2397 131 : haystack.no_language = MBSTRG(language);
2398 131 : haystack.no_encoding = MBSTRG(current_internal_encoding);
2399 131 : needle.no_language = MBSTRG(language);
2400 131 : needle.no_encoding = MBSTRG(current_internal_encoding);
2401 :
2402 :
2403 131 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|bs", (char **)&haystack.val, &haystack.len, (char **)&needle.val, &needle.len, &part, &from_encoding, &from_encoding_len) == FAILURE) {
2404 27 : RETURN_FALSE;
2405 : }
2406 :
2407 104 : if (!needle.len) {
2408 8 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty delimiter");
2409 8 : RETURN_FALSE;
2410 : }
2411 :
2412 96 : haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(from_encoding);
2413 96 : if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2414 21 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", from_encoding);
2415 21 : RETURN_FALSE;
2416 : }
2417 :
2418 75 : n = php_mb_stripos(0, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, 0, from_encoding TSRMLS_CC);
2419 :
2420 75 : if (n <0) {
2421 34 : RETURN_FALSE;
2422 : }
2423 :
2424 41 : mblen = mbfl_strlen(&haystack);
2425 :
2426 41 : if (part) {
2427 22 : ret = mbfl_substr(&haystack, &result, 0, n);
2428 22 : if (ret != NULL) {
2429 22 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
2430 : } else {
2431 0 : RETVAL_FALSE;
2432 : }
2433 : } else {
2434 19 : len = (mblen - n);
2435 19 : ret = mbfl_substr(&haystack, &result, n, len);
2436 19 : if (ret != NULL) {
2437 19 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
2438 : } else {
2439 0 : RETVAL_FALSE;
2440 : }
2441 : }
2442 : }
2443 : /* }}} */
2444 :
2445 : /* {{{ proto string mb_strrichr(string haystack, string needle[, bool part[, string encoding]])
2446 : Finds the last occurrence of a character in a string within another, case insensitive */
2447 : PHP_FUNCTION(mb_strrichr)
2448 131 : {
2449 131 : zend_bool part = 0;
2450 : int n, from_encoding_len, len, mblen;
2451 131 : mbfl_string haystack, needle, result, *ret = NULL;
2452 131 : char *from_encoding = (char*)mbfl_no2preferred_mime_name(MBSTRG(current_internal_encoding));
2453 131 : mbfl_string_init(&haystack);
2454 131 : mbfl_string_init(&needle);
2455 131 : haystack.no_language = MBSTRG(language);
2456 131 : haystack.no_encoding = MBSTRG(current_internal_encoding);
2457 131 : needle.no_language = MBSTRG(language);
2458 131 : needle.no_encoding = MBSTRG(current_internal_encoding);
2459 :
2460 :
2461 131 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|bs", (char **)&haystack.val, &haystack.len, (char **)&needle.val, &needle.len, &part, &from_encoding, &from_encoding_len) == FAILURE) {
2462 27 : RETURN_FALSE;
2463 : }
2464 :
2465 104 : haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(from_encoding);
2466 104 : if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2467 21 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", from_encoding);
2468 21 : RETURN_FALSE;
2469 : }
2470 :
2471 83 : n = php_mb_stripos(1, (char *)haystack.val, haystack.len, (char *)needle.val, needle.len, 0, from_encoding TSRMLS_CC);
2472 :
2473 83 : if (n <0) {
2474 42 : RETURN_FALSE;
2475 : }
2476 :
2477 41 : mblen = mbfl_strlen(&haystack);
2478 :
2479 41 : if (part) {
2480 22 : ret = mbfl_substr(&haystack, &result, 0, n);
2481 22 : if (ret != NULL) {
2482 22 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
2483 : } else {
2484 0 : RETVAL_FALSE;
2485 : }
2486 : } else {
2487 19 : len = (mblen - n);
2488 19 : ret = mbfl_substr(&haystack, &result, n, len);
2489 19 : if (ret != NULL) {
2490 19 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
2491 : } else {
2492 0 : RETVAL_FALSE;
2493 : }
2494 : }
2495 : }
2496 : /* }}} */
2497 :
2498 : /* {{{ proto int mb_substr_count(string haystack, string needle [, string encoding])
2499 : Count the number of substring occurrences */
2500 : PHP_FUNCTION(mb_substr_count)
2501 90 : {
2502 : int n;
2503 : mbfl_string haystack, needle;
2504 90 : char *enc_name = NULL;
2505 : int enc_name_len;
2506 :
2507 90 : mbfl_string_init(&haystack);
2508 90 : mbfl_string_init(&needle);
2509 90 : haystack.no_language = MBSTRG(language);
2510 90 : haystack.no_encoding = MBSTRG(current_internal_encoding);
2511 90 : needle.no_language = MBSTRG(language);
2512 90 : needle.no_encoding = MBSTRG(current_internal_encoding);
2513 :
2514 90 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|s", (char **)&haystack.val, &haystack.len, (char **)&needle.val, &needle.len, &enc_name, &enc_name_len) == FAILURE) {
2515 5 : return;
2516 : }
2517 :
2518 85 : if (enc_name != NULL) {
2519 25 : haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(enc_name);
2520 25 : if (haystack.no_encoding == mbfl_no_encoding_invalid) {
2521 20 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2522 20 : RETURN_FALSE;
2523 : }
2524 : }
2525 :
2526 65 : if (needle.len <= 0) {
2527 8 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty substring");
2528 8 : RETURN_FALSE;
2529 : }
2530 :
2531 57 : n = mbfl_substr_count(&haystack, &needle);
2532 57 : if (n >= 0) {
2533 57 : RETVAL_LONG(n);
2534 : } else {
2535 0 : RETVAL_FALSE;
2536 : }
2537 : }
2538 : /* }}} */
2539 :
2540 : /* {{{ proto string mb_substr(string str, int start [, int length [, string encoding]])
2541 : Returns part of a string */
2542 : PHP_FUNCTION(mb_substr)
2543 271 : {
2544 271 : size_t argc = ZEND_NUM_ARGS();
2545 : char *str, *encoding;
2546 : long from, len;
2547 : int mblen, str_len, encoding_len;
2548 : mbfl_string string, result, *ret;
2549 :
2550 271 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "sl|ls", &str, &str_len, &from, &len, &encoding, &encoding_len) == FAILURE) {
2551 4 : return;
2552 : }
2553 :
2554 267 : mbfl_string_init(&string);
2555 267 : string.no_language = MBSTRG(language);
2556 267 : string.no_encoding = MBSTRG(current_internal_encoding);
2557 :
2558 267 : if (argc == 4) {
2559 190 : string.no_encoding = mbfl_name2no_encoding(encoding);
2560 190 : if (string.no_encoding == mbfl_no_encoding_invalid) {
2561 20 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", encoding);
2562 20 : RETURN_FALSE;
2563 : }
2564 : }
2565 :
2566 247 : string.val = (unsigned char *)str;
2567 247 : string.len = str_len;
2568 :
2569 247 : if (argc < 3) {
2570 1 : len = str_len;
2571 : }
2572 :
2573 : /* measures length */
2574 247 : mblen = 0;
2575 247 : if (from < 0 || len < 0) {
2576 48 : mblen = mbfl_strlen(&string);
2577 : }
2578 :
2579 : /* if "from" position is negative, count start position from the end
2580 : * of the string
2581 : */
2582 247 : if (from < 0) {
2583 24 : from = mblen + from;
2584 24 : if (from < 0) {
2585 16 : from = 0;
2586 : }
2587 : }
2588 :
2589 : /* if "length" position is negative, set it to the length
2590 : * needed to stop that many chars from the end of the string
2591 : */
2592 247 : if (len < 0) {
2593 24 : len = (mblen - from) + len;
2594 24 : if (len < 0) {
2595 16 : len = 0;
2596 : }
2597 : }
2598 :
2599 247 : if (((MBSTRG(func_overload) & MB_OVERLOAD_STRING) == MB_OVERLOAD_STRING)
2600 : && (from >= mbfl_strlen(&string))) {
2601 0 : RETURN_FALSE;
2602 : }
2603 :
2604 247 : ret = mbfl_substr(&string, &result, from, len);
2605 247 : if (NULL == ret) {
2606 0 : RETURN_FALSE;
2607 : }
2608 :
2609 247 : RETURN_STRINGL((char *)ret->val, ret->len, 0); /* the string is already strdup()'ed */
2610 : }
2611 : /* }}} */
2612 :
2613 : /* {{{ proto string mb_strcut(string str, int start [, int length [, string encoding]])
2614 : Returns part of a string */
2615 : PHP_FUNCTION(mb_strcut)
2616 6 : {
2617 6 : size_t argc = ZEND_NUM_ARGS();
2618 : char *encoding;
2619 : long from, len;
2620 : int encoding_len;
2621 : mbfl_string string, result, *ret;
2622 :
2623 6 : mbfl_string_init(&string);
2624 6 : string.no_language = MBSTRG(language);
2625 6 : string.no_encoding = MBSTRG(current_internal_encoding);
2626 :
2627 6 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "sl|ls", (char **)&string.val, (int **)&string.len, &from, &len, &encoding, &encoding_len) == FAILURE) {
2628 0 : return;
2629 : }
2630 :
2631 6 : if (argc == 4) {
2632 6 : string.no_encoding = mbfl_name2no_encoding(encoding);
2633 6 : if (string.no_encoding == mbfl_no_encoding_invalid) {
2634 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", encoding);
2635 0 : RETURN_FALSE;
2636 : }
2637 : }
2638 :
2639 6 : if (argc < 3) {
2640 0 : len = string.len;
2641 : }
2642 :
2643 : /* if "from" position is negative, count start position from the end
2644 : * of the string
2645 : */
2646 6 : if (from < 0) {
2647 0 : from = string.len + from;
2648 0 : if (from < 0) {
2649 0 : from = 0;
2650 : }
2651 : }
2652 :
2653 : /* if "length" position is negative, set it to the length
2654 : * needed to stop that many chars from the end of the string
2655 : */
2656 6 : if (len < 0) {
2657 0 : len = (string.len - from) + len;
2658 0 : if (len < 0) {
2659 0 : len = 0;
2660 : }
2661 : }
2662 :
2663 6 : if ((unsigned int)from > string.len) {
2664 2 : RETURN_FALSE;
2665 : }
2666 :
2667 4 : ret = mbfl_strcut(&string, &result, from, len);
2668 4 : if (ret == NULL) {
2669 0 : RETURN_FALSE;
2670 : }
2671 :
2672 4 : RETURN_STRINGL((char *)ret->val, ret->len, 0); /* the string is already strdup()'ed */
2673 : }
2674 : /* }}} */
2675 :
2676 : /* {{{ proto int mb_strwidth(string str [, string encoding])
2677 : Gets terminal width of a string */
2678 : PHP_FUNCTION(mb_strwidth)
2679 8256 : {
2680 : int n;
2681 : mbfl_string string;
2682 8256 : char *enc_name = NULL;
2683 : int enc_name_len;
2684 :
2685 8256 : mbfl_string_init(&string);
2686 :
2687 8256 : string.no_language = MBSTRG(language);
2688 8256 : string.no_encoding = MBSTRG(current_internal_encoding);
2689 :
2690 8256 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|s", (char **)&string.val, &string.len, &enc_name, &enc_name_len) == FAILURE) {
2691 0 : return;
2692 : }
2693 :
2694 8256 : if (enc_name != NULL) {
2695 8256 : string.no_encoding = mbfl_name2no_encoding(enc_name);
2696 8256 : if (string.no_encoding == mbfl_no_encoding_invalid) {
2697 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", enc_name);
2698 0 : RETURN_FALSE;
2699 : }
2700 : }
2701 :
2702 8256 : n = mbfl_strwidth(&string);
2703 8256 : if (n >= 0) {
2704 8256 : RETVAL_LONG(n);
2705 : } else {
2706 0 : RETVAL_FALSE;
2707 : }
2708 : }
2709 : /* }}} */
2710 :
2711 : /* {{{ proto string mb_strimwidth(string str, int start, int width [, string trimmarker [, string encoding]])
2712 : Trim the string in terminal width */
2713 : PHP_FUNCTION(mb_strimwidth)
2714 2 : {
2715 : char *str, *trimmarker, *encoding;
2716 : long from, width;
2717 : int str_len, trimmarker_len, encoding_len;
2718 : mbfl_string string, result, marker, *ret;
2719 :
2720 2 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "sll|ss", &str, &str_len, &from, &width, &trimmarker, &trimmarker_len, &encoding, &encoding_len) == FAILURE) {
2721 0 : return;
2722 : }
2723 :
2724 2 : mbfl_string_init(&string);
2725 2 : mbfl_string_init(&marker);
2726 2 : string.no_language = MBSTRG(language);
2727 2 : string.no_encoding = MBSTRG(current_internal_encoding);
2728 2 : marker.no_language = MBSTRG(language);
2729 2 : marker.no_encoding = MBSTRG(current_internal_encoding);
2730 2 : marker.val = NULL;
2731 2 : marker.len = 0;
2732 :
2733 2 : if (ZEND_NUM_ARGS() == 5) {
2734 2 : string.no_encoding = marker.no_encoding = mbfl_name2no_encoding(encoding);
2735 2 : if (string.no_encoding == mbfl_no_encoding_invalid) {
2736 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", encoding);
2737 0 : RETURN_FALSE;
2738 : }
2739 : }
2740 :
2741 2 : string.val = (unsigned char *)str;
2742 2 : string.len = str_len;
2743 :
2744 2 : if (from < 0 || from > str_len) {
2745 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Start position is out of range");
2746 0 : RETURN_FALSE;
2747 : }
2748 :
2749 2 : if (width < 0) {
2750 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Width is negative value");
2751 0 : RETURN_FALSE;
2752 : }
2753 :
2754 2 : if (ZEND_NUM_ARGS() >= 4) {
2755 2 : marker.val = (unsigned char *)trimmarker;
2756 2 : marker.len = trimmarker_len;
2757 : }
2758 :
2759 2 : ret = mbfl_strimwidth(&string, &marker, &result, from, width);
2760 :
2761 2 : if (ret == NULL) {
2762 0 : RETURN_FALSE;
2763 : }
2764 :
2765 2 : RETVAL_STRINGL((char *)ret->val, ret->len, 0); /* the string is already strdup()'ed */
2766 : }
2767 : /* }}} */
2768 :
2769 : /* {{{ MBSTRING_API char *php_mb_convert_encoding() */
2770 : 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 TSRMLS_DC)
2771 14687 : {
2772 : mbfl_string string, result, *ret;
2773 : enum mbfl_no_encoding from_encoding, to_encoding;
2774 : mbfl_buffer_converter *convd;
2775 : int size, *list;
2776 14687 : char *output=NULL;
2777 :
2778 14687 : if (output_len) {
2779 14687 : *output_len = 0;
2780 : }
2781 14687 : if (!input) {
2782 0 : return NULL;
2783 : }
2784 : /* new encoding */
2785 29354 : if (_to_encoding && strlen(_to_encoding)) {
2786 14679 : to_encoding = mbfl_name2no_encoding(_to_encoding);
2787 14679 : if (to_encoding == mbfl_no_encoding_invalid) {
2788 12 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", _to_encoding);
2789 12 : return NULL;
2790 : }
2791 : } else {
2792 8 : to_encoding = MBSTRG(current_internal_encoding);
2793 : }
2794 :
2795 : /* initialize string */
2796 14675 : mbfl_string_init(&string);
2797 14675 : mbfl_string_init(&result);
2798 14675 : from_encoding = MBSTRG(current_internal_encoding);
2799 14675 : string.no_encoding = from_encoding;
2800 14675 : string.no_language = MBSTRG(language);
2801 14675 : string.val = (unsigned char *)input;
2802 14675 : string.len = length;
2803 :
2804 : /* pre-conversion encoding */
2805 14675 : if (_from_encodings) {
2806 14675 : list = NULL;
2807 14675 : size = 0;
2808 14675 : php_mb_parse_encoding_list(_from_encodings, strlen(_from_encodings), &list, &size, 0 TSRMLS_CC);
2809 14675 : if (size == 1) {
2810 14675 : from_encoding = *list;
2811 14675 : string.no_encoding = from_encoding;
2812 0 : } else if (size > 1) {
2813 : /* auto detect */
2814 0 : from_encoding = mbfl_identify_encoding_no(&string, list, size, MBSTRG(strict_detection));
2815 0 : if (from_encoding != mbfl_no_encoding_invalid) {
2816 0 : string.no_encoding = from_encoding;
2817 : } else {
2818 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unable to detect character encoding");
2819 0 : from_encoding = mbfl_no_encoding_pass;
2820 0 : to_encoding = from_encoding;
2821 0 : string.no_encoding = from_encoding;
2822 : }
2823 : } else {
2824 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Illegal character encoding specified");
2825 : }
2826 14675 : if (list != NULL) {
2827 14675 : efree((void *)list);
2828 : }
2829 : }
2830 :
2831 : /* initialize converter */
2832 14675 : convd = mbfl_buffer_converter_new(from_encoding, to_encoding, string.len);
2833 14675 : if (convd == NULL) {
2834 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unable to create character encoding converter");
2835 0 : return NULL;
2836 : }
2837 14675 : mbfl_buffer_converter_illegal_mode(convd, MBSTRG(current_filter_illegal_mode));
2838 14675 : mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
2839 :
2840 : /* do it */
2841 14675 : ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
2842 14675 : if (ret) {
2843 14675 : if (output_len) {
2844 14675 : *output_len = ret->len;
2845 : }
2846 14675 : output = (char *)ret->val;
2847 : }
2848 :
2849 14675 : MBSTRG(illegalchars) += mbfl_buffer_illegalchars(convd);
2850 14675 : mbfl_buffer_converter_delete(convd);
2851 14675 : return output;
2852 : }
2853 : /* }}} */
2854 :
2855 : /* {{{ proto string mb_convert_encoding(string str, string to-encoding [, mixed from-encoding])
2856 : Returns converted string in desired encoding */
2857 : PHP_FUNCTION(mb_convert_encoding)
2858 12506 : {
2859 : char *arg_str, *arg_new;
2860 : int str_len, new_len;
2861 : zval *arg_old;
2862 : int i;
2863 : size_t size, l, n;
2864 12506 : char *_from_encodings = NULL, *ret, *s_free = NULL;
2865 :
2866 : zval **hash_entry;
2867 : HashTable *target_hash;
2868 :
2869 12506 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|z", &arg_str, &str_len, &arg_new, &new_len, &arg_old) == FAILURE) {
2870 1 : return;
2871 : }
2872 :
2873 12505 : if (ZEND_NUM_ARGS() == 3) {
2874 12505 : switch (Z_TYPE_P(arg_old)) {
2875 : case IS_ARRAY:
2876 0 : target_hash = Z_ARRVAL_P(arg_old);
2877 0 : zend_hash_internal_pointer_reset(target_hash);
2878 0 : i = zend_hash_num_elements(target_hash);
2879 0 : _from_encodings = NULL;
2880 :
2881 0 : while (i > 0) {
2882 0 : if (zend_hash_get_current_data(target_hash, (void **) &hash_entry) == FAILURE) {
2883 0 : break;
2884 : }
2885 :
2886 0 : convert_to_string_ex(hash_entry);
2887 :
2888 0 : if ( _from_encodings) {
2889 0 : l = strlen(_from_encodings);
2890 0 : n = strlen(Z_STRVAL_PP(hash_entry));
2891 0 : _from_encodings = erealloc(_from_encodings, l+n+2);
2892 0 : strcpy(_from_encodings+l, ",");
2893 0 : strcpy(_from_encodings+l+1, Z_STRVAL_PP(hash_entry));
2894 : } else {
2895 0 : _from_encodings = estrdup(Z_STRVAL_PP(hash_entry));
2896 : }
2897 :
2898 0 : zend_hash_move_forward(target_hash);
2899 0 : i--;
2900 : }
2901 :
2902 0 : if (_from_encodings != NULL && !strlen(_from_encodings)) {
2903 0 : efree(_from_encodings);
2904 0 : _from_encodings = NULL;
2905 : }
2906 0 : s_free = _from_encodings;
2907 0 : break;
2908 : default:
2909 12505 : convert_to_string(arg_old);
2910 12505 : _from_encodings = Z_STRVAL_P(arg_old);
2911 : break;
2912 : }
2913 : }
2914 :
2915 : /* new encoding */
2916 12505 : ret = php_mb_convert_encoding(arg_str, str_len, arg_new, _from_encodings, &size TSRMLS_CC);
2917 12505 : if (ret != NULL) {
2918 12493 : RETVAL_STRINGL(ret, size, 0); /* the string is already strdup()'ed */
2919 : } else {
2920 12 : RETVAL_FALSE;
2921 : }
2922 :
2923 12505 : if ( s_free) {
2924 0 : efree(s_free);
2925 : }
2926 : }
2927 : /* }}} */
2928 :
2929 : /* {{{ proto string mb_convert_case(string sourcestring, int mode [, string encoding])
2930 : Returns a case-folded version of sourcestring */
2931 : PHP_FUNCTION(mb_convert_case)
2932 0 : {
2933 0 : char *str, *from_encoding = (char*)mbfl_no2preferred_mime_name(MBSTRG(current_internal_encoding));
2934 : int str_len, from_encoding_len;
2935 0 : long case_mode = 0;
2936 : char *newstr;
2937 : size_t ret_len;
2938 :
2939 0 : RETVAL_FALSE;
2940 0 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "sl|s!", &str, &str_len,
2941 : &case_mode, &from_encoding, &from_encoding_len) == FAILURE)
2942 0 : RETURN_FALSE;
2943 :
2944 0 : newstr = php_unicode_convert_case(case_mode, str, (size_t) str_len, &ret_len, from_encoding TSRMLS_CC);
2945 :
2946 0 : if (newstr) {
2947 0 : RETVAL_STRINGL(newstr, ret_len, 0);
2948 : }
2949 : }
2950 : /* }}} */
2951 :
2952 : /* {{{ proto string mb_strtoupper(string sourcestring [, string encoding])
2953 : * Returns a uppercased version of sourcestring
2954 : */
2955 : PHP_FUNCTION(mb_strtoupper)
2956 63 : {
2957 63 : char *str, *from_encoding = (char*)mbfl_no2preferred_mime_name(MBSTRG(current_internal_encoding));
2958 : int str_len, from_encoding_len;
2959 : char *newstr;
2960 : size_t ret_len;
2961 :
2962 63 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|s!", &str, &str_len,
2963 : &from_encoding, &from_encoding_len) == FAILURE) {
2964 4 : return;
2965 : }
2966 59 : newstr = php_unicode_convert_case(PHP_UNICODE_CASE_UPPER, str, (size_t) str_len, &ret_len, from_encoding TSRMLS_CC);
2967 :
2968 59 : if (newstr) {
2969 34 : RETURN_STRINGL(newstr, ret_len, 0);
2970 : }
2971 25 : RETURN_FALSE;
2972 : }
2973 : /* }}} */
2974 :
2975 : /* {{{ proto string mb_strtolower(string sourcestring [, string encoding])
2976 : * Returns a lowercased version of sourcestring
2977 : */
2978 : PHP_FUNCTION(mb_strtolower)
2979 63 : {
2980 63 : char *str, *from_encoding = (char*)mbfl_no2preferred_mime_name(MBSTRG(current_internal_encoding));
2981 : int str_len, from_encoding_len;
2982 : char *newstr;
2983 : size_t ret_len;
2984 :
2985 63 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|s!", &str, &str_len,
2986 : &from_encoding, &from_encoding_len) == FAILURE) {
2987 4 : return;
2988 : }
2989 59 : newstr = php_unicode_convert_case(PHP_UNICODE_CASE_LOWER, str, (size_t) str_len, &ret_len, from_encoding TSRMLS_CC);
2990 :
2991 59 : if (newstr) {
2992 34 : RETURN_STRINGL(newstr, ret_len, 0);
2993 : }
2994 25 : RETURN_FALSE;
2995 : }
2996 : /* }}} */
2997 :
2998 : /* {{{ proto string mb_detect_encoding(string str [, mixed encoding_list [, bool strict]])
2999 : Encodings of the given string is returned (as a string) */
3000 : PHP_FUNCTION(mb_detect_encoding)
3001 4 : {
3002 : char *str;
3003 : int str_len;
3004 4 : zend_bool strict=0;
3005 : zval *encoding_list;
3006 :
3007 : mbfl_string string;
3008 : const char *ret;
3009 : enum mbfl_no_encoding *elist;
3010 : int size, *list;
3011 :
3012 4 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|zb", &str, &str_len, &encoding_list, &strict) == FAILURE) {
3013 0 : return;
3014 : }
3015 :
3016 : /* make encoding list */
3017 4 : list = NULL;
3018 4 : size = 0;
3019 4 : if (ZEND_NUM_ARGS() >= 2 && Z_STRVAL_P(encoding_list)) {
3020 4 : switch (Z_TYPE_P(encoding_list)) {
3021 : case IS_ARRAY:
3022 0 : if (!php_mb_parse_encoding_array(encoding_list, &list, &size, 0 TSRMLS_CC)) {
3023 0 : if (list) {
3024 0 : efree(list);
3025 0 : list = NULL;
3026 0 : size = 0;
3027 : }
3028 : }
3029 0 : break;
3030 : default:
3031 4 : convert_to_string(encoding_list);
3032 4 : if (!php_mb_parse_encoding_list(Z_STRVAL_P(encoding_list), Z_STRLEN_P(encoding_list), &list, &size, 0 TSRMLS_CC)) {
3033 0 : if (list) {
3034 0 : efree(list);
3035 0 : list = NULL;
3036 0 : size = 0;
3037 : }
3038 : }
3039 : break;
3040 : }
3041 4 : if (size <= 0) {
3042 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Illegal argument");
3043 : }
3044 : }
3045 :
3046 4 : if (ZEND_NUM_ARGS() < 3) {
3047 0 : strict = (zend_bool)MBSTRG(strict_detection);
3048 : }
3049 :
3050 8 : if (size > 0 && list != NULL) {
3051 4 : elist = list;
3052 : } else {
3053 0 : elist = MBSTRG(current_detect_order_list);
3054 0 : size = MBSTRG(current_detect_order_list_size);
3055 : }
3056 :
3057 4 : mbfl_string_init(&string);
3058 4 : string.no_language = MBSTRG(language);
3059 4 : string.val = (unsigned char *)str;
3060 4 : string.len = str_len;
3061 4 : ret = mbfl_identify_encoding_name(&string, elist, size, strict);
3062 :
3063 4 : if (list != NULL) {
3064 4 : efree((void *)list);
3065 : }
3066 :
3067 4 : if (ret == NULL) {
3068 2 : RETURN_FALSE;
3069 : }
3070 :
3071 2 : RETVAL_STRING((char *)ret, 1);
3072 : }
3073 : /* }}} */
3074 :
3075 : /* {{{ proto mixed mb_list_encodings()
3076 : Returns an array of all supported entity encodings */
3077 : PHP_FUNCTION(mb_list_encodings)
3078 4 : {
3079 : const mbfl_encoding **encodings;
3080 : const mbfl_encoding *encoding;
3081 : int i;
3082 :
3083 4 : array_init(return_value);
3084 4 : i = 0;
3085 4 : encodings = mbfl_get_supported_encodings();
3086 276 : while ((encoding = encodings[i++]) != NULL) {
3087 268 : add_next_index_string(return_value, (char *) encoding->name, 1);
3088 : }
3089 4 : }
3090 : /* }}} */
3091 :
3092 : /* {{{ proto array mb_encoding_aliases(string encoding)
3093 : Returns an array of the aliases of a given encoding name */
3094 : PHP_FUNCTION(mb_encoding_aliases)
3095 4 : {
3096 : const mbfl_encoding *encoding;
3097 4 : char *name = NULL;
3098 : int name_len;
3099 :
3100 4 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s", &name, &name_len) == FAILURE) {
3101 1 : RETURN_FALSE;
3102 : }
3103 :
3104 3 : encoding = mbfl_name2encoding(name);
3105 3 : if (!encoding) {
3106 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", name);
3107 0 : RETURN_FALSE;
3108 : }
3109 :
3110 3 : array_init(return_value);
3111 3 : if (encoding->aliases != NULL) {
3112 : const char **alias;
3113 13 : for (alias = *encoding->aliases; *alias; ++alias) {
3114 11 : add_next_index_string(return_value, (char *)*alias, 1);
3115 : }
3116 : }
3117 : }
3118 : /* }}} */
3119 :
3120 : /* {{{ proto string mb_encode_mimeheader(string str [, string charset [, string transfer-encoding [, string linefeed [, int indent]]]])
3121 : Converts the string to MIME "encoded-word" in the format of =?charset?(B|Q)?encoded_string?= */
3122 : PHP_FUNCTION(mb_encode_mimeheader)
3123 341 : {
3124 : enum mbfl_no_encoding charset, transenc;
3125 : mbfl_string string, result, *ret;
3126 341 : char *charset_name = NULL;
3127 : int charset_name_len;
3128 341 : char *trans_enc_name = NULL;
3129 : int trans_enc_name_len;
3130 341 : char *linefeed = "\r\n";
3131 : int linefeed_len;
3132 341 : long indent = 0;
3133 :
3134 341 : mbfl_string_init(&string);
3135 341 : string.no_language = MBSTRG(language);
3136 341 : string.no_encoding = MBSTRG(current_internal_encoding);
3137 :
3138 341 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|sssl", (char **)&string.val, &string.len, &charset_name, &charset_name_len, &trans_enc_name, &trans_enc_name_len, &linefeed, &linefeed_len, &indent) == FAILURE) {
3139 11 : return;
3140 : }
3141 :
3142 330 : charset = mbfl_no_encoding_pass;
3143 330 : transenc = mbfl_no_encoding_base64;
3144 :
3145 330 : if (charset_name != NULL) {
3146 330 : charset = mbfl_name2no_encoding(charset_name);
3147 330 : if (charset == mbfl_no_encoding_invalid) {
3148 19 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", charset_name);
3149 19 : RETURN_FALSE;
3150 : }
3151 : } else {
3152 0 : const mbfl_language *lang = mbfl_no2language(MBSTRG(language));
3153 0 : if (lang != NULL) {
3154 0 : charset = lang->mail_charset;
3155 0 : transenc = lang->mail_header_encoding;
3156 : }
3157 : }
3158 :
3159 311 : if (trans_enc_name != NULL) {
3160 490 : if (*trans_enc_name == 'B' || *trans_enc_name == 'b') {
3161 179 : transenc = mbfl_no_encoding_base64;
3162 132 : } else if (*trans_enc_name == 'Q' || *trans_enc_name == 'q') {
3163 109 : transenc = mbfl_no_encoding_qprint;
3164 : }
3165 : }
3166 :
3167 311 : mbfl_string_init(&result);
3168 311 : ret = mbfl_mime_header_encode(&string, &result, charset, transenc, linefeed, indent);
3169 311 : if (ret != NULL) {
3170 311 : RETVAL_STRINGL((char *)ret->val, ret->len, 0) /* the string is already strdup()'ed */
3171 : } else {
3172 0 : RETVAL_FALSE;
3173 : }
3174 : }
3175 : /* }}} */
3176 :
3177 : /* {{{ proto string mb_decode_mimeheader(string string)
3178 : Decodes the MIME "encoded-word" in the string */
3179 : PHP_FUNCTION(mb_decode_mimeheader)
3180 39 : {
3181 : mbfl_string string, result, *ret;
3182 :
3183 39 : mbfl_string_init(&string);
3184 39 : string.no_language = MBSTRG(language);
3185 39 : string.no_encoding = MBSTRG(current_internal_encoding);
3186 :
3187 39 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s", (char **)&string.val, &string.len) == FAILURE) {
3188 8 : return;
3189 : }
3190 :
3191 31 : mbfl_string_init(&result);
3192 31 : ret = mbfl_mime_header_decode(&string, &result, MBSTRG(current_internal_encoding));
3193 31 : if (ret != NULL) {
3194 31 : RETVAL_STRINGL((char *)ret->val, ret->len, 0) /* the string is already strdup()'ed */
3195 : } else {
3196 0 : RETVAL_FALSE;
3197 : }
3198 : }
3199 : /* }}} */
3200 :
3201 : /* {{{ proto string mb_convert_kana(string str [, string option] [, string encoding])
3202 : Conversion between full-width character and half-width character (Japanese) */
3203 : PHP_FUNCTION(mb_convert_kana)
3204 10 : {
3205 : int opt, i;
3206 : mbfl_string string, result, *ret;
3207 10 : char *optstr = NULL;
3208 : int optstr_len;
3209 10 : char *encname = NULL;
3210 : int encname_len;
3211 :
3212 10 : mbfl_string_init(&string);
3213 10 : string.no_language = MBSTRG(language);
3214 10 : string.no_encoding = MBSTRG(current_internal_encoding);
3215 :
3216 10 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|ss", (char **)&string.val, &string.len, &optstr, &optstr_len, &encname, &encname_len) == FAILURE) {
3217 0 : return;
3218 : }
3219 :
3220 : /* option */
3221 10 : if (optstr != NULL) {
3222 10 : char *p = optstr;
3223 10 : int n = optstr_len;
3224 10 : i = 0;
3225 10 : opt = 0;
3226 60 : while (i < n) {
3227 40 : i++;
3228 40 : switch (*p++) {
3229 : case 'A':
3230 5 : opt |= 0x1;
3231 5 : break;
3232 : case 'a':
3233 5 : opt |= 0x10;
3234 5 : break;
3235 : case 'R':
3236 0 : opt |= 0x2;
3237 0 : break;
3238 : case 'r':
3239 0 : opt |= 0x20;
3240 0 : break;
3241 : case 'N':
3242 0 : opt |= 0x4;
3243 0 : break;
3244 : case 'n':
3245 0 : opt |= 0x40;
3246 0 : break;
3247 : case 'S':
3248 0 : opt |= 0x8;
3249 0 : break;
3250 : case 's':
3251 0 : opt |= 0x80;
3252 0 : break;
3253 : case 'K':
3254 5 : opt |= 0x100;
3255 5 : break;
3256 : case 'k':
3257 5 : opt |= 0x1000;
3258 5 : break;
3259 : case 'H':
3260 5 : opt |= 0x200;
3261 5 : break;
3262 : case 'h':
3263 5 : opt |= 0x2000;
3264 5 : break;
3265 : case 'V':
3266 0 : opt |= 0x800;
3267 0 : break;
3268 : case 'C':
3269 0 : opt |= 0x10000;
3270 0 : break;
3271 : case 'c':
3272 0 : opt |= 0x20000;
3273 0 : break;
3274 : case 'M':
3275 0 : opt |= 0x100000;
3276 0 : break;
3277 : case 'm':
3278 0 : opt |= 0x200000;
3279 : break;
3280 : }
3281 : }
3282 : } else {
3283 0 : opt = 0x900;
3284 : }
3285 :
3286 : /* encoding */
3287 10 : if (encname != NULL) {
3288 10 : string.no_encoding = mbfl_name2no_encoding(encname);
3289 10 : if (string.no_encoding == mbfl_no_encoding_invalid) {
3290 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", encname);
3291 0 : RETURN_FALSE;
3292 : }
3293 : }
3294 :
3295 10 : ret = mbfl_ja_jp_hantozen(&string, &result, opt);
3296 10 : if (ret != NULL) {
3297 10 : RETVAL_STRINGL((char *)ret->val, ret->len, 0); /* the string is already strdup()'ed */
3298 : } else {
3299 0 : RETVAL_FALSE;
3300 : }
3301 : }
3302 : /* }}} */
3303 :
3304 : #define PHP_MBSTR_STACK_BLOCK_SIZE 32
3305 :
3306 : /* {{{ proto string mb_convert_variables(string to-encoding, mixed from-encoding, mixed vars [, ...])
3307 : Converts the string resource in variables to desired encoding */
3308 : PHP_FUNCTION(mb_convert_variables)
3309 0 : {
3310 : zval ***args, ***stack, **var, **hash_entry, **zfrom_enc;
3311 : HashTable *target_hash;
3312 : mbfl_string string, result, *ret;
3313 : enum mbfl_no_encoding from_encoding, to_encoding;
3314 : mbfl_encoding_detector *identd;
3315 : mbfl_buffer_converter *convd;
3316 : int n, to_enc_len, argc, stack_level, stack_max, elistsz;
3317 : enum mbfl_no_encoding *elist;
3318 : char *name, *to_enc;
3319 : void *ptmp;
3320 :
3321 0 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "sZ+", &to_enc, &to_enc_len, &zfrom_enc, &args, &argc) == FAILURE) {
3322 0 : return;
3323 : }
3324 :
3325 : /* new encoding */
3326 0 : to_encoding = mbfl_name2no_encoding(to_enc);
3327 0 : if (to_encoding == mbfl_no_encoding_invalid) {
3328 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", to_enc);
3329 0 : efree(args);
3330 0 : RETURN_FALSE;
3331 : }
3332 :
3333 : /* initialize string */
3334 0 : mbfl_string_init(&string);
3335 0 : mbfl_string_init(&result);
3336 0 : from_encoding = MBSTRG(current_internal_encoding);
3337 0 : string.no_encoding = from_encoding;
3338 0 : string.no_language = MBSTRG(language);
3339 :
3340 : /* pre-conversion encoding */
3341 0 : elist = NULL;
3342 0 : elistsz = 0;
3343 0 : switch (Z_TYPE_PP(zfrom_enc)) {
3344 : case IS_ARRAY:
3345 0 : php_mb_parse_encoding_array(*zfrom_enc, &elist, &elistsz, 0 TSRMLS_CC);
3346 0 : break;
3347 : default:
3348 0 : convert_to_string_ex(zfrom_enc);
3349 0 : php_mb_parse_encoding_list(Z_STRVAL_PP(zfrom_enc), Z_STRLEN_PP(zfrom_enc), &elist, &elistsz, 0 TSRMLS_CC);
3350 : break;
3351 : }
3352 0 : if (elistsz <= 0) {
3353 0 : from_encoding = mbfl_no_encoding_pass;
3354 0 : } else if (elistsz == 1) {
3355 0 : from_encoding = *elist;
3356 : } else {
3357 : /* auto detect */
3358 0 : from_encoding = mbfl_no_encoding_invalid;
3359 0 : stack_max = PHP_MBSTR_STACK_BLOCK_SIZE;
3360 0 : stack = (zval ***)safe_emalloc(stack_max, sizeof(zval **), 0);
3361 0 : stack_level = 0;
3362 0 : identd = mbfl_encoding_detector_new(elist, elistsz, MBSTRG(strict_detection));
3363 0 : if (identd != NULL) {
3364 0 : n = 0;
3365 0 : while (n < argc || stack_level > 0) {
3366 0 : if (stack_level <= 0) {
3367 0 : var = args[n++];
3368 0 : if (Z_TYPE_PP(var) == IS_ARRAY || Z_TYPE_PP(var) == IS_OBJECT) {
3369 0 : target_hash = HASH_OF(*var);
3370 0 : if (target_hash != NULL) {
3371 0 : zend_hash_internal_pointer_reset(target_hash);
3372 : }
3373 : }
3374 : } else {
3375 0 : stack_level--;
3376 0 : var = stack[stack_level];
3377 : }
3378 0 : if (Z_TYPE_PP(var) == IS_ARRAY || Z_TYPE_PP(var) == IS_OBJECT) {
3379 0 : target_hash = HASH_OF(*var);
3380 0 : if (target_hash != NULL) {
3381 0 : while (zend_hash_get_current_data(target_hash, (void **) &hash_entry) != FAILURE) {
3382 0 : zend_hash_move_forward(target_hash);
3383 0 : if (Z_TYPE_PP(hash_entry) == IS_ARRAY || Z_TYPE_PP(hash_entry) == IS_OBJECT) {
3384 0 : if (stack_level >= stack_max) {
3385 0 : stack_max += PHP_MBSTR_STACK_BLOCK_SIZE;
3386 0 : ptmp = erealloc(stack, sizeof(zval **)*stack_max);
3387 0 : stack = (zval ***)ptmp;
3388 : }
3389 0 : stack[stack_level] = var;
3390 0 : stack_level++;
3391 0 : var = hash_entry;
3392 0 : target_hash = HASH_OF(*var);
3393 0 : if (target_hash != NULL) {
3394 0 : zend_hash_internal_pointer_reset(target_hash);
3395 0 : continue;
3396 : }
3397 0 : } else if (Z_TYPE_PP(hash_entry) == IS_STRING) {
3398 0 : string.val = (unsigned char *)Z_STRVAL_PP(hash_entry);
3399 0 : string.len = Z_STRLEN_PP(hash_entry);
3400 0 : if (mbfl_encoding_detector_feed(identd, &string)) {
3401 0 : goto detect_end; /* complete detecting */
3402 : }
3403 : }
3404 : }
3405 : }
3406 0 : } else if (Z_TYPE_PP(var) == IS_STRING) {
3407 0 : string.val = (unsigned char *)Z_STRVAL_PP(var);
3408 0 : string.len = Z_STRLEN_PP(var);
3409 0 : if (mbfl_encoding_detector_feed(identd, &string)) {
3410 0 : goto detect_end; /* complete detecting */
3411 : }
3412 : }
3413 : }
3414 0 : detect_end:
3415 0 : from_encoding = mbfl_encoding_detector_judge(identd);
3416 0 : mbfl_encoding_detector_delete(identd);
3417 : }
3418 0 : efree(stack);
3419 :
3420 0 : if (from_encoding == mbfl_no_encoding_invalid) {
3421 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unable to detect encoding");
3422 0 : from_encoding = mbfl_no_encoding_pass;
3423 : }
3424 : }
3425 0 : if (elist != NULL) {
3426 0 : efree((void *)elist);
3427 : }
3428 : /* create converter */
3429 0 : convd = NULL;
3430 0 : if (from_encoding != mbfl_no_encoding_pass) {
3431 0 : convd = mbfl_buffer_converter_new(from_encoding, to_encoding, 0);
3432 0 : if (convd == NULL) {
3433 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unable to create converter");
3434 0 : RETURN_FALSE;
3435 : }
3436 0 : mbfl_buffer_converter_illegal_mode(convd, MBSTRG(current_filter_illegal_mode));
3437 0 : mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
3438 : }
3439 :
3440 : /* convert */
3441 0 : if (convd != NULL) {
3442 0 : stack_max = PHP_MBSTR_STACK_BLOCK_SIZE;
3443 0 : stack = (zval ***)safe_emalloc(stack_max, sizeof(zval **), 0);
3444 0 : stack_level = 0;
3445 0 : n = 0;
3446 0 : while (n < argc || stack_level > 0) {
3447 0 : if (stack_level <= 0) {
3448 0 : var = args[n++];
3449 0 : if (Z_TYPE_PP(var) == IS_ARRAY || Z_TYPE_PP(var) == IS_OBJECT) {
3450 0 : target_hash = HASH_OF(*var);
3451 0 : if (target_hash != NULL) {
3452 0 : zend_hash_internal_pointer_reset(target_hash);
3453 : }
3454 : }
3455 : } else {
3456 0 : stack_level--;
3457 0 : var = stack[stack_level];
3458 : }
3459 0 : if (Z_TYPE_PP(var) == IS_ARRAY || Z_TYPE_PP(var) == IS_OBJECT) {
3460 0 : target_hash = HASH_OF(*var);
3461 0 : if (target_hash != NULL) {
3462 0 : while (zend_hash_get_current_data(target_hash, (void **) &hash_entry) != FAILURE) {
3463 0 : zend_hash_move_forward(target_hash);
3464 0 : if (Z_TYPE_PP(hash_entry) == IS_ARRAY || Z_TYPE_PP(hash_entry) == IS_OBJECT) {
3465 0 : if (stack_level >= stack_max) {
3466 0 : stack_max += PHP_MBSTR_STACK_BLOCK_SIZE;
3467 0 : ptmp = erealloc(stack, sizeof(zval **)*stack_max);
3468 0 : stack = (zval ***)ptmp;
3469 : }
3470 0 : stack[stack_level] = var;
3471 0 : stack_level++;
3472 0 : var = hash_entry;
3473 0 : SEPARATE_ZVAL(hash_entry);
3474 0 : target_hash = HASH_OF(*var);
3475 0 : if (target_hash != NULL) {
3476 0 : zend_hash_internal_pointer_reset(target_hash);
3477 0 : continue;
3478 : }
3479 0 : } else if (Z_TYPE_PP(hash_entry) == IS_STRING) {
3480 0 : string.val = (unsigned char *)Z_STRVAL_PP(hash_entry);
3481 0 : string.len = Z_STRLEN_PP(hash_entry);
3482 0 : ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
3483 0 : if (ret != NULL) {
3484 0 : if (Z_REFCOUNT_PP(hash_entry) > 1) {
3485 0 : Z_DELREF_PP(hash_entry);
3486 0 : MAKE_STD_ZVAL(*hash_entry);
3487 : } else {
3488 0 : zval_dtor(*hash_entry);
3489 : }
3490 0 : ZVAL_STRINGL(*hash_entry, (char *)ret->val, ret->len, 0);
3491 : }
3492 : }
3493 : }
3494 : }
3495 0 : } else if (Z_TYPE_PP(var) == IS_STRING) {
3496 0 : string.val = (unsigned char *)Z_STRVAL_PP(var);
3497 0 : string.len = Z_STRLEN_PP(var);
3498 0 : ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
3499 0 : if (ret != NULL) {
3500 0 : zval_dtor(*var);
3501 0 : ZVAL_STRINGL(*var, (char *)ret->val, ret->len, 0);
3502 : }
3503 : }
3504 : }
3505 0 : efree(stack);
3506 :
3507 0 : MBSTRG(illegalchars) += mbfl_buffer_illegalchars(convd);
3508 0 : mbfl_buffer_converter_delete(convd);
3509 : }
3510 :
3511 0 : efree(args);
3512 :
3513 0 : name = (char *)mbfl_no_encoding2name(from_encoding);
3514 0 : if (name != NULL) {
3515 0 : RETURN_STRING(name, 1);
3516 : } else {
3517 0 : RETURN_FALSE;
3518 : }
3519 : }
3520 : /* }}} */
3521 :
3522 : /* {{{ HTML numeric entity */
3523 : /* {{{ static void php_mb_numericentity_exec() */
3524 : static void
3525 : php_mb_numericentity_exec(INTERNAL_FUNCTION_PARAMETERS, int type)
3526 6 : {
3527 : char *str, *encoding;
3528 : int str_len, encoding_len;
3529 : zval *zconvmap, **hash_entry;
3530 : HashTable *target_hash;
3531 6 : size_t argc = ZEND_NUM_ARGS();
3532 6 : int i, *convmap, *mapelm, mapsize=0;
3533 : mbfl_string string, result, *ret;
3534 : enum mbfl_no_encoding no_encoding;
3535 :
3536 6 : if (zend_parse_parameters(argc TSRMLS_CC, "szs", &str, &str_len, &zconvmap, &encoding, &encoding_len) == FAILURE) {
3537 0 : return;
3538 : }
3539 :
3540 6 : mbfl_string_init(&string);
3541 6 : string.no_language = MBSTRG(language);
3542 6 : string.no_encoding = MBSTRG(current_internal_encoding);
3543 6 : string.val = (unsigned char *)str;
3544 6 : string.len = str_len;
3545 :
3546 : /* encoding */
3547 6 : if (argc == 3) {
3548 6 : no_encoding = mbfl_name2no_encoding(encoding);
3549 6 : if (no_encoding == mbfl_no_encoding_invalid) {
3550 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", encoding);
3551 0 : RETURN_FALSE;
3552 : } else {
3553 6 : string.no_encoding = no_encoding;
3554 : }
3555 : }
3556 :
3557 : /* conversion map */
3558 6 : convmap = NULL;
3559 6 : if (Z_TYPE_P(zconvmap) == IS_ARRAY) {
3560 6 : target_hash = Z_ARRVAL_P(zconvmap);
3561 6 : zend_hash_internal_pointer_reset(target_hash);
3562 6 : i = zend_hash_num_elements(target_hash);
3563 6 : if (i > 0) {
3564 6 : convmap = (int *)safe_emalloc(i, sizeof(int), 0);
3565 6 : mapelm = convmap;
3566 6 : mapsize = 0;
3567 36 : while (i > 0) {
3568 24 : if (zend_hash_get_current_data(target_hash, (void **) &hash_entry) == FAILURE) {
3569 0 : break;
3570 : }
3571 24 : convert_to_long_ex(hash_entry);
3572 24 : *mapelm++ = Z_LVAL_PP(hash_entry);
3573 24 : mapsize++;
3574 24 : i--;
3575 24 : zend_hash_move_forward(target_hash);
3576 : }
3577 : }
3578 : }
3579 6 : if (convmap == NULL) {
3580 0 : RETURN_FALSE;
3581 : }
3582 6 : mapsize /= 4;
3583 :
3584 6 : ret = mbfl_html_numeric_entity(&string, &result, convmap, mapsize, type);
3585 6 : if (ret != NULL) {
3586 6 : RETVAL_STRINGL((char *)ret->val, ret->len, 0);
3587 : } else {
3588 0 : RETVAL_FALSE;
3589 : }
3590 6 : efree((void *)convmap);
3591 : }
3592 : /* }}} */
3593 :
3594 : /* {{{ proto string mb_encode_numericentity(string string, array convmap [, string encoding])
3595 : Converts specified characters to HTML numeric entities */
3596 : PHP_FUNCTION(mb_encode_numericentity)
3597 3 : {
3598 3 : php_mb_numericentity_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0);
3599 3 : }
3600 : /* }}} */
3601 :
3602 : /* {{{ proto string mb_decode_numericentity(string string, array convmap [, string encoding])
3603 : Converts HTML numeric entities to character code */
3604 : PHP_FUNCTION(mb_decode_numericentity)
3605 3 : {
3606 3 : php_mb_numericentity_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, 1);
3607 3 : }
3608 : /* }}} */
3609 : /* }}} */
3610 :
3611 : /* {{{ proto int mb_send_mail(string to, string subject, string message [, string additional_headers [, string additional_parameters]])
3612 : * Sends an email message with MIME scheme
3613 : */
3614 :
3615 : #define SKIP_LONG_HEADER_SEP_MBSTRING(str, pos) \
3616 : if (str[pos] == '\r' && str[pos + 1] == '\n' && (str[pos + 2] == ' ' || str[pos + 2] == '\t')) { \
3617 : pos += 2; \
3618 : while (str[pos + 1] == ' ' || str[pos + 1] == '\t') { \
3619 : pos++; \
3620 : } \
3621 : continue; \
3622 : }
3623 :
3624 : #define MAIL_ASCIIZ_CHECK_MBSTRING(str, len) \
3625 : pp = str; \
3626 : ee = pp + len; \
3627 : while ((pp = memchr(pp, '\0', (ee - pp)))) { \
3628 : *pp = ' '; \
3629 : } \
3630 :
3631 : #define APPEND_ONE_CHAR(ch) do { \
3632 : if (token.a > 0) { \
3633 : smart_str_appendc(&token, ch); \
3634 : } else {\
3635 : token.len++; \
3636 : } \
3637 : } while (0)
3638 :
3639 : #define SEPARATE_SMART_STR(str) do {\
3640 : if ((str)->a == 0) { \
3641 : char *tmp_ptr; \
3642 : (str)->a = 1; \
3643 : while ((str)->a < (str)->len) { \
3644 : (str)->a <<= 1; \
3645 : } \
3646 : tmp_ptr = emalloc((str)->a + 1); \
3647 : memcpy(tmp_ptr, (str)->c, (str)->len); \
3648 : (str)->c = tmp_ptr; \
3649 : } \
3650 : } while (0)
3651 :
3652 : static void my_smart_str_dtor(smart_str *s)
3653 0 : {
3654 0 : if (s->a > 0) {
3655 0 : smart_str_free(s);
3656 : }
3657 0 : }
3658 :
3659 : static int _php_mbstr_parse_mail_headers(HashTable *ht, const char *str, size_t str_len)
3660 0 : {
3661 : const char *ps;
3662 : size_t icnt;
3663 0 : int state = 0;
3664 0 : int crlf_state = -1;
3665 :
3666 0 : smart_str token = { 0, 0, 0 };
3667 0 : smart_str fld_name = { 0, 0, 0 }, fld_val = { 0, 0, 0 };
3668 :
3669 0 : ps = str;
3670 0 : icnt = str_len;
3671 :
3672 : /*
3673 : * C o n t e n t - T y p e : t e x t / h t m l \r\n
3674 : * ^ ^^^^^^^^^^^^^^^^^^^^^ ^^^ ^^^^^^^^^^^^^^^^^ ^^^^
3675 : * state 0 1 2 3
3676 : *
3677 : * C o n t e n t - T y p e : t e x t / h t m l \r\n
3678 : * ^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^
3679 : * crlf_state -1 0 1 -1
3680 : *
3681 : */
3682 :
3683 0 : while (icnt > 0) {
3684 0 : switch (*ps) {
3685 : case ':':
3686 0 : if (crlf_state == 1) {
3687 0 : APPEND_ONE_CHAR('\r');
3688 : }
3689 :
3690 0 : if (state == 0 || state == 1) {
3691 0 : fld_name = token;
3692 :
3693 0 : state = 2;
3694 : } else {
3695 0 : APPEND_ONE_CHAR(*ps);
3696 : }
3697 :
3698 0 : crlf_state = 0;
3699 0 : break;
3700 :
3701 : case '\n':
3702 0 : if (crlf_state == -1) {
3703 0 : goto out;
3704 : }
3705 0 : crlf_state = -1;
3706 0 : break;
3707 :
3708 : case '\r':
3709 0 : if (crlf_state == 1) {
3710 0 : APPEND_ONE_CHAR('\r');
3711 : } else {
3712 0 : crlf_state = 1;
3713 : }
3714 0 : break;
3715 :
3716 : case ' ': case '\t':
3717 0 : if (crlf_state == -1) {
3718 0 : if (state == 3) {
3719 : /* continuing from the previous line */
3720 0 : SEPARATE_SMART_STR(&token);
3721 0 : state = 4;
3722 : } else {
3723 : /* simply skipping this new line */
3724 0 : state = 5;
3725 : }
3726 : } else {
3727 0 : if (crlf_state == 1) {
3728 0 : APPEND_ONE_CHAR('\r');
3729 : }
3730 0 : if (state == 1 || state == 3) {
3731 0 : APPEND_ONE_CHAR(*ps);
3732 : }
3733 : }
3734 0 : crlf_state = 0;
3735 0 : break;
3736 :
3737 : default:
3738 0 : switch (state) {
3739 : case 0:
3740 0 : token.c = (char *)ps;
3741 0 : token.len = 0;
3742 0 : token.a = 0;
3743 0 : state = 1;
3744 0 : break;
3745 :
3746 : case 2:
3747 0 : if (crlf_state != -1) {
3748 0 : token.c = (char *)ps;
3749 0 : token.len = 0;
3750 0 : token.a = 0;
3751 :
3752 0 : state = 3;
3753 0 : break;
3754 : }
3755 : /* break is missing intentionally */
3756 :
3757 : case 3:
3758 0 : if (crlf_state == -1) {
3759 0 : fld_val = token;
3760 :
3761 0 : if (fld_name.c != NULL && fld_val.c != NULL) {
3762 : char *dummy;
3763 :
3764 : /* FIXME: some locale free implementation is
3765 : * really required here,,, */
3766 0 : SEPARATE_SMART_STR(&fld_name);
3767 0 : php_strtoupper(fld_name.c, fld_name.len);
3768 :
3769 0 : zend_hash_update(ht, (char *)fld_name.c, fld_name.len, &fld_val, sizeof(smart_str), (void **)&dummy);
3770 :
3771 0 : my_smart_str_dtor(&fld_name);
3772 : }
3773 :
3774 0 : memset(&fld_name, 0, sizeof(smart_str));
3775 0 : memset(&fld_val, 0, sizeof(smart_str));
3776 :
3777 0 : token.c = (char *)ps;
3778 0 : token.len = 0;
3779 0 : token.a = 0;
3780 :
3781 0 : state = 1;
3782 : }
3783 0 : break;
3784 :
3785 : case 4:
3786 0 : APPEND_ONE_CHAR(' ');
3787 0 : state = 3;
3788 : break;
3789 : }
3790 :
3791 0 : if (crlf_state == 1) {
3792 0 : APPEND_ONE_CHAR('\r');
3793 : }
3794 :
3795 0 : APPEND_ONE_CHAR(*ps);
3796 :
3797 0 : crlf_state = 0;
3798 : break;
3799 : }
3800 0 : ps++, icnt--;
3801 : }
3802 0 : out:
3803 0 : if (state == 2) {
3804 0 : token.c = "";
3805 0 : token.len = 0;
3806 0 : token.a = 0;
3807 :
3808 0 : state = 3;
3809 : }
3810 0 : if (state == 3) {
3811 0 : fld_val = token;
3812 :
3813 0 : if (fld_name.c != NULL && fld_val.c != NULL) {
3814 : void *dummy;
3815 :
3816 : /* FIXME: some locale free implementation is
3817 : * really required here,,, */
3818 0 : SEPARATE_SMART_STR(&fld_name);
3819 0 : php_strtoupper(fld_name.c, fld_name.len);
3820 :
3821 0 : zend_hash_update(ht, (char *)fld_name.c, fld_name.len, &fld_val, sizeof(smart_str), (void **)&dummy);
3822 :
3823 0 : my_smart_str_dtor(&fld_name);
3824 : }
3825 : }
3826 0 : return state;
3827 : }
3828 :
3829 : PHP_FUNCTION(mb_send_mail)
3830 4 : {
3831 : int n;
3832 4 : char *to = NULL;
3833 : int to_len;
3834 4 : char *message = NULL;
3835 : int message_len;
3836 4 : char *headers = NULL;
3837 : int headers_len;
3838 4 : char *subject = NULL;
3839 : int subject_len;
3840 4 : char *extra_cmd = NULL;
3841 : int extra_cmd_len;
3842 : int i;
3843 4 : char *to_r = NULL;
3844 4 : char *force_extra_parameters = INI_STR("mail.force_extra_parameters");
3845 : struct {
3846 : int cnt_type:1;
3847 : int cnt_trans_enc:1;
3848 4 : } suppressed_hdrs = { 0, 0 };
3849 :
3850 4 : char *message_buf = NULL, *subject_buf = NULL, *p;
3851 : mbfl_string orig_str, conv_str;
3852 : mbfl_string *pstr; /* pointer to mbfl string for return value */
3853 : enum mbfl_no_encoding
3854 : tran_cs, /* transfar text charset */
3855 : head_enc, /* header transfar encoding */
3856 : body_enc; /* body transfar encoding */
3857 : mbfl_memory_device device; /* automatic allocateable buffer for additional header */
3858 : const mbfl_language *lang;
3859 4 : int err = 0;
3860 : HashTable ht_headers;
3861 : smart_str *s;
3862 : extern void mbfl_memory_device_unput(mbfl_memory_device *device);
3863 : char *pp, *ee;
3864 :
3865 : /* initialize */
3866 4 : mbfl_memory_device_init(&device, 0, 0);
3867 4 : mbfl_string_init(&orig_str);
3868 4 : mbfl_string_init(&conv_str);
3869 :
3870 : /* character-set, transfer-encoding */
3871 4 : tran_cs = mbfl_no_encoding_utf8;
3872 4 : head_enc = mbfl_no_encoding_base64;
3873 4 : body_enc = mbfl_no_encoding_base64;
3874 4 : lang = mbfl_no2language(MBSTRG(language));
3875 4 : if (lang != NULL) {
3876 4 : tran_cs = lang->mail_charset;
3877 4 : head_enc = lang->mail_header_encoding;
3878 4 : body_enc = lang->mail_body_encoding;
3879 : }
3880 :
3881 4 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "sss|ss", &to, &to_len, &subject, &subject_len, &message, &message_len, &headers, &headers_len, &extra_cmd, &extra_cmd_len) == FAILURE) {
3882 0 : return;
3883 : }
3884 :
3885 : /* ASCIIZ check */
3886 4 : MAIL_ASCIIZ_CHECK_MBSTRING(to, to_len);
3887 4 : MAIL_ASCIIZ_CHECK_MBSTRING(subject, subject_len);
3888 4 : MAIL_ASCIIZ_CHECK_MBSTRING(message, message_len);
3889 4 : if (headers) {
3890 0 : MAIL_ASCIIZ_CHECK_MBSTRING(headers, headers_len);
3891 : }
3892 4 : if (extra_cmd) {
3893 0 : MAIL_ASCIIZ_CHECK_MBSTRING(extra_cmd, extra_cmd_len);
3894 : }
3895 :
3896 4 : zend_hash_init(&ht_headers, 0, NULL, (dtor_func_t) my_smart_str_dtor, 0);
3897 :
3898 4 : if (headers != NULL) {
3899 0 : _php_mbstr_parse_mail_headers(&ht_headers, headers, headers_len);
3900 : }
3901 :
3902 4 : if (zend_hash_find(&ht_headers, "CONTENT-TYPE", sizeof("CONTENT-TYPE") - 1, (void **)&s) == SUCCESS) {
3903 : char *tmp;
3904 : char *param_name;
3905 0 : char *charset = NULL;
3906 :
3907 0 : SEPARATE_SMART_STR(s);
3908 0 : smart_str_0(s);
3909 :
3910 0 : p = strchr(s->c, ';');
3911 :
3912 0 : if (p != NULL) {
3913 : /* skipping the padded spaces */
3914 : do {
3915 0 : ++p;
3916 0 : } while (*p == ' ' || *p == '\t');
3917 :
3918 0 : if (*p != '\0') {
3919 0 : if ((param_name = php_strtok_r(p, "= ", &tmp)) != NULL) {
3920 0 : if (strcasecmp(param_name, "charset") == 0) {
3921 0 : enum mbfl_no_encoding _tran_cs = tran_cs;
3922 :
3923 0 : charset = php_strtok_r(NULL, "= \"", &tmp);
3924 0 : if (charset != NULL) {
3925 0 : _tran_cs = mbfl_name2no_encoding(charset);
3926 : }
3927 :
3928 0 : if (_tran_cs == mbfl_no_encoding_invalid) {
3929 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unsupported charset \"%s\" - will be regarded as ascii", charset);
3930 0 : _tran_cs = mbfl_no_encoding_ascii;
3931 : }
3932 0 : tran_cs = _tran_cs;
3933 : }
3934 : }
3935 : }
3936 : }
3937 0 : suppressed_hdrs.cnt_type = 1;
3938 : }
3939 :
3940 4 : if (zend_hash_find(&ht_headers, "CONTENT-TRANSFER-ENCODING", sizeof("CONTENT-TRANSFER-ENCODING") - 1, (void **)&s) == SUCCESS) {
3941 : enum mbfl_no_encoding _body_enc;
3942 0 : SEPARATE_SMART_STR(s);
3943 0 : smart_str_0(s);
3944 :
3945 0 : _body_enc = mbfl_name2no_encoding(s->c);
3946 0 : switch (_body_enc) {
3947 : case mbfl_no_encoding_base64:
3948 : case mbfl_no_encoding_7bit:
3949 : case mbfl_no_encoding_8bit:
3950 0 : body_enc = _body_enc;
3951 0 : break;
3952 :
3953 : default:
3954 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unsupported transfer encoding \"%s\" - will be regarded as 8bit", s->c);
3955 0 : body_enc = mbfl_no_encoding_8bit;
3956 : break;
3957 : }
3958 0 : suppressed_hdrs.cnt_trans_enc = 1;
3959 : }
3960 :
3961 : /* To: */
3962 4 : if (to != NULL) {
3963 4 : if (to_len > 0) {
3964 4 : to_r = estrndup(to, to_len);
3965 4 : for (; to_len; to_len--) {
3966 4 : if (!isspace((unsigned char) to_r[to_len - 1])) {
3967 4 : break;
3968 : }
3969 0 : to_r[to_len - 1] = '\0';
3970 : }
3971 80 : for (i = 0; to_r[i]; i++) {
3972 76 : if (iscntrl((unsigned char) to_r[i])) {
3973 : /* According to RFC 822, section 3.1.1 long headers may be separated into
3974 : * parts using CRLF followed at least one linear-white-space character ('\t' or ' ').
3975 : * To prevent these separators from being replaced with a space, we use the
3976 : * SKIP_LONG_HEADER_SEP_MBSTRING to skip over them.
3977 : */
3978 0 : SKIP_LONG_HEADER_SEP_MBSTRING(to_r, i);
3979 0 : to_r[i] = ' ';
3980 : }
3981 : }
3982 : } else {
3983 0 : to_r = to;
3984 : }
3985 : } else {
3986 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Missing To: field");
3987 0 : err = 1;
3988 : }
3989 :
3990 : /* Subject: */
3991 8 : if (subject != NULL && subject_len >= 0) {
3992 4 : orig_str.no_language = MBSTRG(language);
3993 4 : orig_str.val = (unsigned char *)subject;
3994 4 : orig_str.len = subject_len;
3995 4 : orig_str.no_encoding = MBSTRG(current_internal_encoding);
3996 4 : if (orig_str.no_encoding == mbfl_no_encoding_invalid
3997 : || orig_str.no_encoding == mbfl_no_encoding_pass) {
3998 1 : orig_str.no_encoding = mbfl_identify_encoding_no(&orig_str, MBSTRG(current_detect_order_list), MBSTRG(current_detect_order_list_size), MBSTRG(strict_detection));
3999 : }
4000 4 : pstr = mbfl_mime_header_encode(&orig_str, &conv_str, tran_cs, head_enc, "\n", sizeof("Subject: [PHP-jp nnnnnnnn]"));
4001 4 : if (pstr != NULL) {
4002 4 : subject_buf = subject = (char *)pstr->val;
4003 : }
4004 : } else {
4005 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Missing Subject: field");
4006 0 : err = 1;
4007 : }
4008 :
4009 : /* message body */
4010 4 : if (message != NULL) {
4011 4 : orig_str.no_language = MBSTRG(language);
4012 4 : orig_str.val = (unsigned char *)message;
4013 4 : orig_str.len = (unsigned int)message_len;
4014 4 : orig_str.no_encoding = MBSTRG(current_internal_encoding);
4015 :
4016 4 : if (orig_str.no_encoding == mbfl_no_encoding_invalid
4017 : || orig_str.no_encoding == mbfl_no_encoding_pass) {
4018 1 : orig_str.no_encoding = mbfl_identify_encoding_no(&orig_str, MBSTRG(current_detect_order_list), MBSTRG(current_detect_order_list_size), MBSTRG(strict_detection));
4019 : }
4020 :
4021 4 : pstr = NULL;
4022 : {
4023 : mbfl_string tmpstr;
4024 :
4025 4 : if (mbfl_convert_encoding(&orig_str, &tmpstr, tran_cs) != NULL) {
4026 4 : tmpstr.no_encoding=mbfl_no_encoding_8bit;
4027 4 : pstr = mbfl_convert_encoding(&tmpstr, &conv_str, body_enc);
4028 4 : efree(tmpstr.val);
4029 : }
4030 : }
4031 4 : if (pstr != NULL) {
4032 4 : message_buf = message = (char *)pstr->val;
4033 : }
4034 : } else {
4035 : /* this is not really an error, so it is allowed. */
4036 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty message body");
4037 0 : message = NULL;
4038 : }
4039 :
4040 : /* other headers */
4041 : #define PHP_MBSTR_MAIL_MIME_HEADER1 "Mime-Version: 1.0"
4042 : #define PHP_MBSTR_MAIL_MIME_HEADER2 "Content-Type: text/plain"
4043 : #define PHP_MBSTR_MAIL_MIME_HEADER3 "; charset="
4044 : #define PHP_MBSTR_MAIL_MIME_HEADER4 "Content-Transfer-Encoding: "
4045 4 : if (headers != NULL) {
4046 0 : p = headers;
4047 0 : n = headers_len;
4048 0 : mbfl_memory_device_strncat(&device, p, n);
4049 0 : if (n > 0 && p[n - 1] != '\n') {
4050 0 : mbfl_memory_device_strncat(&device, "\n", 1);
4051 : }
4052 : }
4053 :
4054 4 : mbfl_memory_device_strncat(&device, PHP_MBSTR_MAIL_MIME_HEADER1, sizeof(PHP_MBSTR_MAIL_MIME_HEADER1) - 1);
4055 4 : mbfl_memory_device_strncat(&device, "\n", 1);
4056 :
4057 4 : if (!suppressed_hdrs.cnt_type) {
4058 4 : mbfl_memory_device_strncat(&device, PHP_MBSTR_MAIL_MIME_HEADER2, sizeof(PHP_MBSTR_MAIL_MIME_HEADER2) - 1);
4059 :
4060 4 : p = (char *)mbfl_no2preferred_mime_name(tran_cs);
4061 4 : if (p != NULL) {
4062 4 : mbfl_memory_device_strncat(&device, PHP_MBSTR_MAIL_MIME_HEADER3, sizeof(PHP_MBSTR_MAIL_MIME_HEADER3) - 1);
4063 4 : mbfl_memory_device_strcat(&device, p);
4064 : }
4065 4 : mbfl_memory_device_strncat(&device, "\n", 1);
4066 : }
4067 4 : if (!suppressed_hdrs.cnt_trans_enc) {
4068 4 : mbfl_memory_device_strncat(&device, PHP_MBSTR_MAIL_MIME_HEADER4, sizeof(PHP_MBSTR_MAIL_MIME_HEADER4) - 1);
4069 4 : p = (char *)mbfl_no2preferred_mime_name(body_enc);
4070 4 : if (p == NULL) {
4071 0 : p = "7bit";
4072 : }
4073 4 : mbfl_memory_device_strcat(&device, p);
4074 4 : mbfl_memory_device_strncat(&device, "\n", 1);
4075 : }
4076 :
4077 4 : mbfl_memory_device_unput(&device);
4078 4 : mbfl_memory_device_output('\0', &device);
4079 4 : headers = (char *)device.buffer;
4080 :
4081 4 : if (force_extra_parameters) {
4082 0 : extra_cmd = php_escape_shell_cmd(force_extra_parameters);
4083 4 : } else if (extra_cmd) {
4084 0 : extra_cmd = php_escape_shell_cmd(extra_cmd);
4085 : }
4086 :
4087 8 : if (!err && php_mail(to_r, subject, message, headers, extra_cmd TSRMLS_CC)) {
4088 4 : RETVAL_TRUE;
4089 : } else {
4090 0 : RETVAL_FALSE;
4091 : }
4092 :
4093 4 : if (extra_cmd) {
4094 0 : efree(extra_cmd);
4095 : }
4096 4 : if (to_r != to) {
4097 4 : efree(to_r);
4098 : }
4099 4 : if (subject_buf) {
4100 4 : efree((void *)subject_buf);
4101 : }
4102 4 : if (message_buf) {
4103 4 : efree((void *)message_buf);
4104 : }
4105 4 : mbfl_memory_device_clear(&device);
4106 4 : zend_hash_destroy(&ht_headers);
4107 : }
4108 :
4109 : #undef SKIP_LONG_HEADER_SEP_MBSTRING
4110 : #undef MAIL_ASCIIZ_CHECK_MBSTRING
4111 : #undef APPEND_ONE_CHAR
4112 : #undef SEPARATE_SMART_STR
4113 : #undef PHP_MBSTR_MAIL_MIME_HEADER1
4114 : #undef PHP_MBSTR_MAIL_MIME_HEADER2
4115 : #undef PHP_MBSTR_MAIL_MIME_HEADER3
4116 : #undef PHP_MBSTR_MAIL_MIME_HEADER4
4117 : /* }}} */
4118 :
4119 : /* {{{ proto mixed mb_get_info([string type])
4120 : Returns the current settings of mbstring */
4121 : PHP_FUNCTION(mb_get_info)
4122 16 : {
4123 16 : char *typ = NULL;
4124 : int typ_len, n;
4125 : char *name;
4126 : const struct mb_overload_def *over_func;
4127 : zval *row1, *row2;
4128 16 : const mbfl_language *lang = mbfl_no2language(MBSTRG(language));
4129 : enum mbfl_no_encoding *entry;
4130 : #ifdef ZEND_MULTIBYTE
4131 : zval *row3;
4132 : #endif /* ZEND_MULTIBYTE */
4133 :
4134 16 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|s", &typ, &typ_len) == FAILURE) {
4135 0 : RETURN_FALSE;
4136 : }
4137 :
4138 17 : if (!typ || !strcasecmp("all", typ)) {
4139 1 : array_init(return_value);
4140 1 : if ((name = (char *)mbfl_no_encoding2name(MBSTRG(current_internal_encoding))) != NULL) {
4141 1 : add_assoc_string(return_value, "internal_encoding", name, 1);
4142 : }
4143 1 : if ((name = (char *)mbfl_no_encoding2name(MBSTRG(http_input_identify))) != NULL) {
4144 1 : add_assoc_string(return_value, "http_input", name, 1);
4145 : }
4146 1 : if ((name = (char *)mbfl_no_encoding2name(MBSTRG(current_http_output_encoding))) != NULL) {
4147 1 : add_assoc_string(return_value, "http_output", name, 1);
4148 : }
4149 1 : if ((name = (char *)zend_ini_string("mbstring.http_output_conv_mimetypes", sizeof("mbstring.http_output_conv_mimetypes"), 0)) != NULL) {
4150 1 : add_assoc_string(return_value, "http_output_conv_mimetypes", name, 1);
4151 : }
4152 1 : add_assoc_long(return_value, "func_overload", MBSTRG(func_overload));
4153 1 : if (MBSTRG(func_overload)){
4154 1 : over_func = &(mb_ovld[0]);
4155 1 : MAKE_STD_ZVAL(row1);
4156 1 : array_init(row1);
4157 15 : while (over_func->type > 0) {
4158 13 : if ((MBSTRG(func_overload) & over_func->type) == over_func->type ) {
4159 12 : add_assoc_string(row1, over_func->orig_func, over_func->ovld_func, 1);
4160 : }
4161 13 : over_func++;
4162 : }
4163 1 : add_assoc_zval(return_value, "func_overload_list", row1);
4164 : } else {
4165 0 : add_assoc_string(return_value, "func_overload_list", "no overload", 1);
4166 : }
4167 1 : if (lang != NULL) {
4168 1 : if ((name = (char *)mbfl_no_encoding2name(lang->mail_charset)) != NULL) {
4169 1 : add_assoc_string(return_value, "mail_charset", name, 1);
4170 : }
4171 1 : if ((name = (char *)mbfl_no_encoding2name(lang->mail_header_encoding)) != NULL) {
4172 1 : add_assoc_string(return_value, "mail_header_encoding", name, 1);
4173 : }
4174 1 : if ((name = (char *)mbfl_no_encoding2name(lang->mail_body_encoding)) != NULL) {
4175 1 : add_assoc_string(return_value, "mail_body_encoding", name, 1);
4176 : }
4177 : }
4178 1 : add_assoc_long(return_value, "illegal_chars", MBSTRG(illegalchars));
4179 1 : if (MBSTRG(encoding_translation)) {
4180 1 : add_assoc_string(return_value, "encoding_translation", "On", 1);
4181 : } else {
4182 0 : add_assoc_string(return_value, "encoding_translation", "Off", 1);
4183 : }
4184 1 : if ((name = (char *)mbfl_no_language2name(MBSTRG(language))) != NULL) {
4185 1 : add_assoc_string(return_value, "language", name, 1);
4186 : }
4187 1 : n = MBSTRG(current_detect_order_list_size);
4188 1 : entry = MBSTRG(current_detect_order_list);
4189 1 : if(n > 0) {
4190 1 : MAKE_STD_ZVAL(row2);
4191 1 : array_init(row2);
4192 6 : while (n > 0) {
4193 4 : if ((name = (char *)mbfl_no_encoding2name(*entry)) != NULL) {
4194 4 : add_next_index_string(row2, name, 1);
4195 : }
4196 4 : entry++;
4197 4 : n--;
4198 : }
4199 1 : add_assoc_zval(return_value, "detect_order", row2);
4200 : }
4201 1 : if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
4202 0 : add_assoc_string(return_value, "substitute_character", "none", 1);
4203 1 : } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG) {
4204 0 : add_assoc_string(return_value, "substitute_character", "long", 1);
4205 1 : } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY) {
4206 0 : add_assoc_string(return_value, "substitute_character", "entity", 1);
4207 : } else {
4208 1 : add_assoc_long(return_value, "substitute_character", MBSTRG(current_filter_illegal_substchar));
4209 : }
4210 1 : if (MBSTRG(strict_detection)) {
4211 1 : add_assoc_string(return_value, "strict_detection", "On", 1);
4212 : } else {
4213 0 : add_assoc_string(return_value, "strict_detection", "Off", 1);
4214 : }
4215 : #ifdef ZEND_MULTIBYTE
4216 : entry = MBSTRG(script_encoding_list);
4217 : n = MBSTRG(script_encoding_list_size);
4218 : if(n > 0) {
4219 : MAKE_STD_ZVAL(row3);
4220 : array_init(row3);
4221 : while (n > 0) {
4222 : if ((name = (char *)mbfl_no_encoding2name(*entry)) != NULL) {
4223 : add_next_index_string(row3, name, 1);
4224 : }
4225 : entry++;
4226 : n--;
4227 : }
4228 : add_assoc_zval(return_value, "script_encoding", row3);
4229 : }
4230 : #endif /* ZEND_MULTIBYTE */
4231 15 : } else if (!strcasecmp("internal_encoding", typ)) {
4232 1 : if ((name = (char *)mbfl_no_encoding2name(MBSTRG(current_internal_encoding))) != NULL) {
4233 1 : RETVAL_STRING(name, 1);
4234 : }
4235 14 : } else if (!strcasecmp("http_input", typ)) {
4236 1 : if ((name = (char *)mbfl_no_encoding2name(MBSTRG(http_input_identify))) != NULL) {
4237 1 : RETVAL_STRING(name, 1);
4238 : }
4239 13 : } else if (!strcasecmp("http_output", typ)) {
4240 1 : if ((name = (char *)mbfl_no_encoding2name(MBSTRG(current_http_output_encoding))) != NULL) {
4241 1 : RETVAL_STRING(name, 1);
4242 : }
4243 12 : } else if (!strcasecmp("http_output_conv_mimetypes", typ)) {
4244 1 : if ((name = (char *)zend_ini_string("mbstring.http_output_conv_mimetypes", sizeof("mbstring.http_output_conv_mimetypes"), 0)) != NULL) {
4245 1 : RETVAL_STRING(name, 1);
4246 : }
4247 11 : } else if (!strcasecmp("func_overload", typ)) {
4248 1 : RETVAL_LONG(MBSTRG(func_overload));
4249 10 : } else if (!strcasecmp("func_overload_list", typ)) {
4250 1 : if (MBSTRG(func_overload)){
4251 1 : over_func = &(mb_ovld[0]);
4252 1 : array_init(return_value);
4253 15 : while (over_func->type > 0) {
4254 13 : if ((MBSTRG(func_overload) & over_func->type) == over_func->type ) {
4255 12 : add_assoc_string(return_value, over_func->orig_func, over_func->ovld_func, 1);
4256 : }
4257 13 : over_func++;
4258 : }
4259 : } else {
4260 0 : RETVAL_STRING("no overload", 1);
4261 : }
4262 9 : } else if (!strcasecmp("mail_charset", typ)) {
4263 1 : if (lang != NULL && (name = (char *)mbfl_no_encoding2name(lang->mail_charset)) != NULL) {
4264 1 : RETVAL_STRING(name, 1);
4265 : }
4266 8 : } else if (!strcasecmp("mail_header_encoding", typ)) {
4267 1 : if (lang != NULL && (name = (char *)mbfl_no_encoding2name(lang->mail_header_encoding)) != NULL) {
4268 1 : RETVAL_STRING(name, 1);
4269 : }
4270 7 : } else if (!strcasecmp("mail_body_encoding", typ)) {
4271 1 : if (lang != NULL && (name = (char *)mbfl_no_encoding2name(lang->mail_body_encoding)) != NULL) {
4272 1 : RETVAL_STRING(name, 1);
4273 : }
4274 6 : } else if (!strcasecmp("illegal_chars", typ)) {
4275 1 : RETVAL_LONG(MBSTRG(illegalchars));
4276 5 : } else if (!strcasecmp("encoding_translation", typ)) {
4277 1 : if (MBSTRG(encoding_translation)) {
4278 1 : RETVAL_STRING("On", 1);
4279 : } else {
4280 0 : RETVAL_STRING("Off", 1);
4281 : }
4282 4 : } else if (!strcasecmp("language", typ)) {
4283 1 : if ((name = (char *)mbfl_no_language2name(MBSTRG(language))) != NULL) {
4284 1 : RETVAL_STRING(name, 1);
4285 : }
4286 3 : } else if (!strcasecmp("detect_order", typ)) {
4287 1 : n = MBSTRG(current_detect_order_list_size);
4288 1 : entry = MBSTRG(current_detect_order_list);
4289 1 : if(n > 0) {
4290 1 : array_init(return_value);
4291 6 : while (n > 0) {
4292 4 : name = (char *)mbfl_no_encoding2name(*entry);
4293 4 : if (name) {
4294 4 : add_next_index_string(return_value, name, 1);
4295 : }
4296 4 : entry++;
4297 4 : n--;
4298 : }
4299 : }
4300 2 : } else if (!strcasecmp("substitute_character", typ)) {
4301 1 : if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE) {
4302 0 : RETVAL_STRING("none", 1);
4303 1 : } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_LONG) {
4304 0 : RETVAL_STRING("long", 1);
4305 1 : } else if (MBSTRG(current_filter_illegal_mode) == MBFL_OUTPUTFILTER_ILLEGAL_MODE_ENTITY) {
4306 0 : RETVAL_STRING("entity", 1);
4307 : } else {
4308 1 : RETVAL_LONG(MBSTRG(current_filter_illegal_substchar));
4309 : }
4310 1 : } else if (!strcasecmp("strict_detection", typ)) {
4311 1 : if (MBSTRG(strict_detection)) {
4312 1 : RETVAL_STRING("On", 1);
4313 : } else {
4314 0 : RETVAL_STRING("Off", 1);
4315 : }
4316 : } else {
4317 : #ifdef ZEND_MULTIBYTE
4318 : if (!strcasecmp("script_encoding", typ)) {
4319 : entry = MBSTRG(script_encoding_list);
4320 : n = MBSTRG(script_encoding_list_size);
4321 : if(n > 0) {
4322 : array_init(return_value);
4323 : while (n > 0) {
4324 : name = (char *)mbfl_no_encoding2name(*entry);
4325 : if (name) {
4326 : add_next_index_string(return_value, name, 1);
4327 : }
4328 : entry++;
4329 : n--;
4330 : }
4331 : }
4332 : return;
4333 : }
4334 : #endif /* ZEND_MULTIBYTE */
4335 0 : RETURN_FALSE;
4336 : }
4337 : }
4338 : /* }}} */
4339 :
4340 : /* {{{ proto bool mb_check_encoding([string var[, string encoding]])
4341 : Check if the string is valid for the specified encoding */
4342 : PHP_FUNCTION(mb_check_encoding)
4343 513 : {
4344 513 : char *var = NULL;
4345 : int var_len;
4346 513 : char *enc = NULL;
4347 : int enc_len;
4348 : mbfl_buffer_converter *convd;
4349 513 : enum mbfl_no_encoding no_encoding = MBSTRG(current_internal_encoding);
4350 513 : mbfl_string string, result, *ret = NULL;
4351 513 : long illegalchars = 0;
4352 :
4353 513 : if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|ss", &var, &var_len, &enc, &enc_len) == FAILURE) {
4354 0 : RETURN_FALSE;
4355 : }
4356 :
4357 513 : if (var == NULL) {
4358 0 : RETURN_BOOL(MBSTRG(illegalchars) == 0);
4359 : }
4360 :
4361 513 : if (enc != NULL) {
4362 513 : no_encoding = mbfl_name2no_encoding(enc);
4363 513 : if (no_encoding == mbfl_no_encoding_invalid || no_encoding == mbfl_no_encoding_pass) {
4364 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Invalid encoding \"%s\"", enc);
4365 0 : RETURN_FALSE;
4366 : }
4367 : }
4368 :
4369 513 : convd = mbfl_buffer_converter_new(no_encoding, no_encoding, 0);
4370 513 : if (convd == NULL) {
4371 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unable to create converter");
4372 0 : RETURN_FALSE;
4373 : }
4374 513 : mbfl_buffer_converter_illegal_mode(convd, MBFL_OUTPUTFILTER_ILLEGAL_MODE_NONE);
4375 513 : mbfl_buffer_converter_illegal_substchar(convd, 0);
4376 :
4377 : /* initialize string */
4378 513 : mbfl_string_init_set(&string, mbfl_no_language_neutral, no_encoding);
4379 513 : mbfl_string_init(&result);
4380 :
4381 513 : string.val = (unsigned char *)var;
4382 513 : string.len = var_len;
4383 513 : ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
4384 513 : illegalchars = mbfl_buffer_illegalchars(convd);
4385 513 : mbfl_buffer_converter_delete(convd);
4386 :
4387 513 : RETVAL_FALSE;
4388 513 : if (ret != NULL) {
4389 513 : if (illegalchars == 0 && string.len == result.len && memcmp(string.val, result.val, string.len) == 0) {
4390 376 : RETVAL_TRUE;
4391 : }
4392 513 : mbfl_string_clear(&result);
4393 : }
4394 : }
4395 : /* }}} */
4396 :
4397 : /* {{{ MBSTRING_API int php_mb_encoding_translation() */
4398 : MBSTRING_API int php_mb_encoding_translation(TSRMLS_D)
4399 0 : {
4400 0 : return MBSTRG(encoding_translation);
4401 : }
4402 : /* }}} */
4403 :
4404 : /* {{{ MBSTRING_API size_t php_mb_mbchar_bytes_ex() */
4405 : MBSTRING_API size_t php_mb_mbchar_bytes_ex(const char *s, const mbfl_encoding *enc)
4406 0 : {
4407 0 : if (enc != NULL) {
4408 0 : if (enc->flag & MBFL_ENCTYPE_MBCS) {
4409 0 : if (enc->mblen_table != NULL) {
4410 0 : if (s != NULL) return enc->mblen_table[*(unsigned char *)s];
4411 : }
4412 0 : } else if (enc->flag & (MBFL_ENCTYPE_WCS2BE | MBFL_ENCTYPE_WCS2LE)) {
4413 0 : return 2;
4414 0 : } else if (enc->flag & (MBFL_ENCTYPE_WCS4BE | MBFL_ENCTYPE_WCS4LE)) {
4415 0 : return 4;
4416 : }
4417 : }
4418 0 : return 1;
4419 : }
4420 : /* }}} */
4421 :
4422 : /* {{{ MBSTRING_API size_t php_mb_mbchar_bytes() */
4423 : MBSTRING_API size_t php_mb_mbchar_bytes(const char *s TSRMLS_DC)
4424 0 : {
4425 0 : return php_mb_mbchar_bytes_ex(s,
4426 : mbfl_no2encoding(MBSTRG(internal_encoding)));
4427 : }
4428 : /* }}} */
4429 :
4430 : /* {{{ MBSTRING_API char *php_mb_safe_strrchr_ex() */
4431 : MBSTRING_API char *php_mb_safe_strrchr_ex(const char *s, unsigned int c, size_t nbytes, const mbfl_encoding *enc)
4432 0 : {
4433 0 : register const char *p = s;
4434 0 : char *last=NULL;
4435 :
4436 0 : if (nbytes == (size_t)-1) {
4437 0 : size_t nb = 0;
4438 :
4439 0 : while (*p != '\0') {
4440 0 : if (nb == 0) {
4441 0 : if ((unsigned char)*p == (unsigned char)c) {
4442 0 : last = (char *)p;
4443 : }
4444 0 : nb = php_mb_mbchar_bytes_ex(p, enc);
4445 0 : if (nb == 0) {
4446 0 : return NULL; /* something is going wrong! */
4447 : }
4448 : }
4449 0 : --nb;
4450 0 : ++p;
4451 : }
4452 : } else {
4453 0 : register size_t bcnt = nbytes;
4454 : register size_t nbytes_char;
4455 0 : while (bcnt > 0) {
4456 0 : if ((unsigned char)*p == (unsigned char)c) {
4457 0 : last = (char *)p;
4458 : }
4459 0 : nbytes_char = php_mb_mbchar_bytes_ex(p, enc);
4460 0 : if (bcnt < nbytes_char) {
4461 0 : return NULL;
4462 : }
4463 0 : p += nbytes_char;
4464 0 : bcnt -= nbytes_char;
4465 : }
4466 : }
4467 0 : return last;
4468 : }
4469 : /* }}} */
4470 :
4471 : /* {{{ MBSTRING_API char *php_mb_safe_strrchr() */
4472 : MBSTRING_API char *php_mb_safe_strrchr(const char *s, unsigned int c, size_t nbytes TSRMLS_DC)
4473 0 : {
4474 0 : return php_mb_safe_strrchr_ex(s, c, nbytes,
4475 : mbfl_no2encoding(MBSTRG(internal_encoding)));
4476 : }
4477 : /* }}} */
4478 :
4479 : /* {{{ MBSTRING_API char *php_mb_strrchr() */
4480 : MBSTRING_API char *php_mb_strrchr(const char *s, char c TSRMLS_DC)
4481 0 : {
4482 0 : return php_mb_safe_strrchr(s, c, -1 TSRMLS_CC);
4483 : }
4484 : /* }}} */
4485 :
4486 : /* {{{ MBSTRING_API size_t php_mb_gpc_mbchar_bytes() */
4487 : MBSTRING_API size_t php_mb_gpc_mbchar_bytes(const char *s TSRMLS_DC)
4488 0 : {
4489 :
4490 0 : if (MBSTRG(http_input_identify) != mbfl_no_encoding_invalid){
4491 0 : return php_mb_mbchar_bytes_ex(s,
4492 : mbfl_no2encoding(MBSTRG(http_input_identify)));
4493 : } else {
4494 0 : return php_mb_mbchar_bytes_ex(s,
4495 : mbfl_no2encoding(MBSTRG(internal_encoding)));
4496 : }
4497 : }
4498 : /* }}} */
4499 :
4500 : /* {{{ MBSTRING_API int php_mb_gpc_encoding_converter() */
4501 : MBSTRING_API int php_mb_gpc_encoding_converter(char **str, int *len, int num, const char *encoding_to, const char *encoding_from TSRMLS_DC)
4502 0 : {
4503 : int i;
4504 0 : mbfl_string string, result, *ret = NULL;
4505 : enum mbfl_no_encoding from_encoding, to_encoding;
4506 : mbfl_buffer_converter *convd;
4507 :
4508 0 : if (encoding_to) {
4509 : /* new encoding */
4510 0 : to_encoding = mbfl_name2no_encoding(encoding_to);
4511 0 : if (to_encoding == mbfl_no_encoding_invalid) {
4512 0 : return -1;
4513 : }
4514 : } else {
4515 0 : to_encoding = MBSTRG(current_internal_encoding);
4516 : }
4517 0 : if (encoding_from) {
4518 : /* old encoding */
4519 0 : from_encoding = mbfl_name2no_encoding(encoding_from);
4520 0 : if (from_encoding == mbfl_no_encoding_invalid) {
4521 0 : return -1;
4522 : }
4523 : } else {
4524 0 : from_encoding = MBSTRG(http_input_identify);
4525 : }
4526 :
4527 0 : if (from_encoding == mbfl_no_encoding_pass) {
4528 0 : return 0;
4529 : }
4530 :
4531 : /* initialize string */
4532 0 : mbfl_string_init(&string);
4533 0 : mbfl_string_init(&result);
4534 0 : string.no_encoding = from_encoding;
4535 0 : string.no_language = MBSTRG(language);
4536 :
4537 0 : for (i=0; i<num; i++){
4538 0 : string.val = (unsigned char *)str[i];
4539 0 : string.len = len[i];
4540 :
4541 : /* initialize converter */
4542 0 : convd = mbfl_buffer_converter_new(from_encoding, to_encoding, string.len);
4543 0 : if (convd == NULL) {
4544 0 : return -1;
4545 : }
4546 0 : mbfl_buffer_converter_illegal_mode(convd, MBSTRG(current_filter_illegal_mode));
4547 0 : mbfl_buffer_converter_illegal_substchar(convd, MBSTRG(current_filter_illegal_substchar));
4548 :
4549 : /* do it */
4550 0 : ret = mbfl_buffer_converter_feed_result(convd, &string, &result);
4551 0 : if (ret != NULL) {
4552 0 : efree(str[i]);
4553 0 : str[i] = (char *)ret->val;
4554 0 : len[i] = (int)ret->len;
4555 : }
4556 :
4557 0 : MBSTRG(illegalchars) += mbfl_buffer_illegalchars(convd);
4558 0 : mbfl_buffer_converter_delete(convd);
4559 : }
4560 :
4561 0 : return ret ? 0 : -1;
4562 : }
4563 : /* }}} */
4564 :
4565 : /* {{{ MBSTRING_API int php_mb_gpc_encoding_detector()
4566 : */
4567 : MBSTRING_API int php_mb_gpc_encoding_detector(char **arg_string, int *arg_length, int num, char *arg_list TSRMLS_DC)
4568 0 : {
4569 : mbfl_string string;
4570 : enum mbfl_no_encoding *elist;
4571 0 : enum mbfl_no_encoding encoding = mbfl_no_encoding_invalid;
4572 0 : mbfl_encoding_detector *identd = NULL;
4573 :
4574 : int size;
4575 : enum mbfl_no_encoding *list;
4576 :
4577 0 : if (MBSTRG(http_input_list_size) == 1 &&
4578 : MBSTRG(http_input_list)[0] == mbfl_no_encoding_pass) {
4579 0 : MBSTRG(http_input_identify) = mbfl_no_encoding_pass;
4580 0 : return SUCCESS;
4581 : }
4582 :
4583 0 : if (MBSTRG(http_input_list_size) == 1 &&
4584 : MBSTRG(http_input_list)[0] != mbfl_no_encoding_auto &&
4585 : mbfl_no_encoding2name(MBSTRG(http_input_list)[0]) != NULL) {
4586 0 : MBSTRG(http_input_identify) = MBSTRG(http_input_list)[0];
4587 0 : return SUCCESS;
4588 : }
4589 :
4590 0 : if (arg_list && strlen(arg_list)>0) {
4591 : /* make encoding list */
4592 0 : list = NULL;
4593 0 : size = 0;
4594 0 : php_mb_parse_encoding_list(arg_list, strlen(arg_list), &list, &size, 0 TSRMLS_CC);
4595 :
4596 0 : if (size > 0 && list != NULL) {
4597 0 : elist = list;
4598 : } else {
4599 0 : elist = MBSTRG(current_detect_order_list);
4600 0 : size = MBSTRG(current_detect_order_list_size);
4601 0 : if (size <= 0){
4602 0 : elist = MBSTRG(default_detect_order_list);
4603 0 : size = MBSTRG(default_detect_order_list_size);
4604 : }
4605 : }
4606 : } else {
4607 0 : elist = MBSTRG(current_detect_order_list);
4608 0 : size = MBSTRG(current_detect_order_list_size);
4609 0 : if (size <= 0){
4610 0 : elist = MBSTRG(default_detect_order_list);
4611 0 : size = MBSTRG(default_detect_order_list_size);
4612 : }
4613 : }
4614 :
4615 0 : mbfl_string_init(&string);
4616 0 : string.no_language = MBSTRG(language);
4617 :
4618 0 : identd = mbfl_encoding_detector_new(elist, size, MBSTRG(strict_detection));
4619 :
4620 0 : if (identd) {
4621 0 : int n = 0;
4622 0 : while(n < num){
4623 0 : string.val = (unsigned char *)arg_string[n];
4624 0 : string.len = arg_length[n];
4625 0 : if (mbfl_encoding_detector_feed(identd, &string)) {
4626 0 : break;
4627 : }
4628 0 : n++;
4629 : }
4630 0 : encoding = mbfl_encoding_detector_judge(identd);
4631 0 : mbfl_encoding_detector_delete(identd);
4632 : }
4633 :
4634 0 : if (encoding != mbfl_no_encoding_invalid) {
4635 0 : MBSTRG(http_input_identify) = encoding;
4636 0 : return SUCCESS;
4637 : } else {
4638 0 : return FAILURE;
4639 : }
4640 : }
4641 : /* }}} */
4642 :
4643 : /* {{{ MBSTRING_API int php_mb_stripos()
4644 : */
4645 : 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 TSRMLS_DC)
4646 566 : {
4647 : int n;
4648 : mbfl_string haystack, needle;
4649 566 : n = -1;
4650 :
4651 566 : mbfl_string_init(&haystack);
4652 566 : mbfl_string_init(&needle);
4653 566 : haystack.no_language = MBSTRG(language);
4654 566 : haystack.no_encoding = MBSTRG(current_internal_encoding);
4655 566 : needle.no_language = MBSTRG(language);
4656 566 : needle.no_encoding = MBSTRG(current_internal_encoding);
4657 :
4658 : do {
4659 566 : size_t len = 0;
4660 566 : haystack.val = (unsigned char *)php_unicode_convert_case(PHP_UNICODE_CASE_UPPER, (char *)old_haystack, old_haystack_len, &len, from_encoding TSRMLS_CC);
4661 566 : haystack.len = len;
4662 :
4663 566 : if (!haystack.val) {
4664 40 : break;
4665 : }
4666 :
4667 526 : if (haystack.len <= 0) {
4668 32 : break;
4669 : }
4670 :
4671 494 : needle.val = (unsigned char *)php_unicode_convert_case(PHP_UNICODE_CASE_UPPER, (char *)old_needle, old_needle_len, &len, from_encoding TSRMLS_CC);
4672 494 : needle.len = len;
4673 :
4674 494 : if (!needle.val) {
4675 0 : break;
4676 : }
4677 :
4678 494 : if (needle.len <= 0) {
4679 16 : break;
4680 : }
4681 :
4682 478 : haystack.no_encoding = needle.no_encoding = mbfl_name2no_encoding(from_encoding);
4683 478 : if (haystack.no_encoding == mbfl_no_encoding_invalid) {
4684 0 : php_error_docref(NULL TSRMLS_CC, E_WARNING, "Unknown encoding \"%s\"", from_encoding);
4685 0 : break;
4686 : }
4687 :
4688 : {
4689 478 : int haystack_char_len = mbfl_strlen(&haystack);
4690 :
4691 478 : if (mode) {
4692 239 : |
