xref: /php-src/ext/intl/converter/converter.c (revision 927adfb1)
1 /*
2    +----------------------------------------------------------------------+
3    | This source file is subject to version 3.01 of the PHP license,      |
4    | that is bundled with this package in the file LICENSE, and is        |
5    | available through the world-wide-web at the following url:           |
6    | https://www.php.net/license/3_01.txt                                 |
7    | If you did not receive a copy of the PHP license and are unable to   |
8    | obtain it through the world-wide-web, please send a note to          |
9    | license@php.net so we can mail you a copy immediately.               |
10    +----------------------------------------------------------------------+
11    | Authors: Sara Golemon <pollita@php.net>                              |
12    +----------------------------------------------------------------------+
13  */
14 
15 #include "converter.h"
16 #include "zend_exceptions.h"
17 
18 #include <unicode/utypes.h>
19 #include <unicode/utf8.h>
20 #include <unicode/utf16.h>
21 #include <unicode/ucnv.h>
22 #include <unicode/ustring.h>
23 
24 #include "../intl_error.h"
25 #include "../intl_common.h"
26 #include "converter_arginfo.h"
27 
28 typedef struct _php_converter_object {
29 	UConverter *src, *dest;
30 	zend_fcall_info to_cb, from_cb;
31 	zend_fcall_info_cache to_cache, from_cache;
32 	intl_error error;
33 	zend_object obj;
34 } php_converter_object;
35 
36 
php_converter_fetch_object(zend_object * obj)37 static inline php_converter_object *php_converter_fetch_object(zend_object *obj) {
38 	return (php_converter_object *)((char*)(obj) - XtOffsetOf(php_converter_object, obj));
39 }
40 #define Z_INTL_CONVERTER_P(zv) php_converter_fetch_object(Z_OBJ_P(zv))
41 
42 static zend_class_entry     *php_converter_ce;
43 static zend_object_handlers  php_converter_object_handlers;
44 
45 #define CONV_GET(pzv)  (Z_INTL_CONVERTER_P((pzv)))
46 #define THROW_UFAILURE(obj, fname, error) php_converter_throw_failure(obj, error, \
47                                           fname "() returned error " ZEND_LONG_FMT ": %s", (zend_long)error, u_errorName(error))
48 
49 /* {{{ php_converter_throw_failure */
php_converter_throw_failure(php_converter_object * objval,UErrorCode error,const char * format,...)50 static inline void php_converter_throw_failure(php_converter_object *objval, UErrorCode error, const char *format, ...) {
51 	intl_error *err = objval ? &(objval->error) : NULL;
52 	char message[1024];
53 	va_list vargs;
54 
55 	va_start(vargs, format);
56 	vsnprintf(message, sizeof(message), format, vargs);
57 	va_end(vargs);
58 
59 	intl_errors_set(err, error, message, 1);
60 }
61 /* }}} */
62 
63 /* {{{ php_converter_default_callback */
php_converter_default_callback(zval * return_value,zval * zobj,zend_long reason,zval * error)64 static void php_converter_default_callback(zval *return_value, zval *zobj, zend_long reason, zval *error) {
65 	/* Basic functionality so children can call parent::toUCallback() */
66 	switch (reason) {
67 		case UCNV_UNASSIGNED:
68 		case UCNV_ILLEGAL:
69 		case UCNV_IRREGULAR:
70 		{
71 			php_converter_object *objval = (php_converter_object*)CONV_GET(zobj);
72 			char chars[127];
73 			int8_t chars_len = sizeof(chars);
74 			UErrorCode uerror = U_ZERO_ERROR;
75 			if(!objval->src) {
76 				php_converter_throw_failure(objval, U_INVALID_STATE_ERROR, "Source Converter has not been initialized yet");
77 				chars[0] = 0x1A;
78 				chars[1] = 0;
79 				chars_len = 1;
80 				ZEND_TRY_ASSIGN_REF_LONG(error, U_INVALID_STATE_ERROR);
81 				RETVAL_STRINGL(chars, chars_len);
82 				return;
83 			}
84 
85 			/* Yes, this is fairly wasteful at first glance,
86 			 * but considering that the alternative is to store
87 			 * what's sent into setSubstChars() and the fact
88 			 * that this is an extremely unlikely codepath
89 			 * I'd rather take the CPU hit here, than waste time
90 			 * storing a value I'm unlikely to use.
91 			 */
92 			ucnv_getSubstChars(objval->src, chars, &chars_len, &uerror);
93 			if (U_FAILURE(uerror)) {
94 				THROW_UFAILURE(objval, "ucnv_getSubstChars", uerror);
95 				chars[0] = 0x1A;
96 				chars[1] = 0;
97 				chars_len = 1;
98 			}
99 			ZEND_TRY_ASSIGN_REF_LONG(error, uerror);
100 			RETVAL_STRINGL(chars, chars_len);
101 		}
102 	}
103 }
104 /* }}} */
105 
106 /* {{{ */
PHP_METHOD(UConverter,toUCallback)107 PHP_METHOD(UConverter, toUCallback) {
108 	zend_long reason;
109 	zend_string *source, *codeUnits;
110 	zval *error;
111 
112 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "lSSz",
113 		&reason, &source, &codeUnits, &error) == FAILURE) {
114 		RETURN_THROWS();
115 	}
116 
117 	php_converter_default_callback(return_value, ZEND_THIS, reason, error);
118 }
119 /* }}} */
120 
121 /* {{{ */
PHP_METHOD(UConverter,fromUCallback)122 PHP_METHOD(UConverter, fromUCallback) {
123 	zend_long reason;
124 	zval *source, *error;
125 	zend_long codePoint;
126 
127 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "lalz",
128 		&reason, &source, &codePoint, &error) == FAILURE) {
129 		RETURN_THROWS();
130 	}
131 
132 	php_converter_default_callback(return_value, ZEND_THIS, reason, error);
133 }
134 /* }}} */
135 
136 /* {{{ php_converter_check_limits */
php_converter_check_limits(php_converter_object * objval,zend_long available,zend_long needed)137 static inline bool php_converter_check_limits(php_converter_object *objval, zend_long available, zend_long needed) {
138 	if (available < needed) {
139 		php_converter_throw_failure(objval, U_BUFFER_OVERFLOW_ERROR, "Buffer overrun " ZEND_LONG_FMT " bytes needed, " ZEND_LONG_FMT " available", needed, available);
140 		return 0;
141 	}
142 	return 1;
143 }
144 /* }}} */
145 
146 #define TARGET_CHECK(cnvargs, needed) php_converter_check_limits(objval, cnvargs->targetLimit - cnvargs->target, needed)
147 
148 /* {{{ php_converter_append_toUnicode_target */
php_converter_append_toUnicode_target(zval * val,UConverterToUnicodeArgs * args,php_converter_object * objval)149 static void php_converter_append_toUnicode_target(zval *val, UConverterToUnicodeArgs *args, php_converter_object *objval) {
150 	switch (Z_TYPE_P(val)) {
151 		case IS_NULL:
152 			/* Code unit is being skipped */
153 			return;
154 		case IS_LONG:
155 		{
156 			zend_long lval = Z_LVAL_P(val);
157 			if ((lval < 0) || (lval > 0x10FFFF)) {
158 				php_converter_throw_failure(objval, U_ILLEGAL_ARGUMENT_ERROR, "Invalid codepoint U+%04lx", lval);
159 				return;
160 			}
161 			if (lval > 0xFFFF) {
162 				/* Supplemental planes U+010000 - U+10FFFF */
163 				if (TARGET_CHECK(args, 2)) {
164 					/* TODO: Find the ICU call which does this properly */
165 					*(args->target++) = (UChar)(((lval - 0x10000) >> 10)   | 0xD800);
166 					*(args->target++) = (UChar)(((lval - 0x10000) & 0x3FF) | 0xDC00);
167 				}
168 				return;
169 			}
170 			/* Non-suggogate BMP codepoint */
171 			if (TARGET_CHECK(args, 1)) {
172 				*(args->target++) = (UChar)lval;
173 			}
174 			return;
175 		}
176 		case IS_STRING:
177 		{
178 			const char *strval = Z_STRVAL_P(val);
179 			int i = 0, strlen = Z_STRLEN_P(val);
180 
181 			while((i != strlen) && TARGET_CHECK(args, 1)) {
182 				UChar c;
183 				U8_NEXT(strval, i, strlen, c);
184 				*(args->target++) = c;
185 			}
186 			return;
187 		}
188 		case IS_ARRAY:
189 		{
190 			HashTable *ht = Z_ARRVAL_P(val);
191 			zval *tmpzval;
192 
193 			ZEND_HASH_FOREACH_VAL(ht, tmpzval) {
194 				php_converter_append_toUnicode_target(tmpzval, args, objval);
195 			} ZEND_HASH_FOREACH_END();
196 			return;
197 		}
198 		default:
199 			php_converter_throw_failure(objval, U_ILLEGAL_ARGUMENT_ERROR,
200 			                            "toUCallback() specified illegal type for substitution character");
201 	}
202 }
203 /* }}} */
204 
205 /* {{{ php_converter_to_u_callback */
php_converter_to_u_callback(const void * context,UConverterToUnicodeArgs * args,const char * codeUnits,int32_t length,UConverterCallbackReason reason,UErrorCode * pErrorCode)206 static void php_converter_to_u_callback(const void *context,
207                                         UConverterToUnicodeArgs *args,
208                                         const char *codeUnits, int32_t length,
209                                         UConverterCallbackReason reason,
210                                         UErrorCode *pErrorCode) {
211 	php_converter_object *objval = (php_converter_object*)context;
212 	zval retval;
213 	zval zargs[4];
214 
215 	ZVAL_LONG(&zargs[0], reason);
216 	if (args->source) {
217 		ZVAL_STRINGL(&zargs[1], args->source, args->sourceLimit - args->source);
218 	} else {
219 		ZVAL_EMPTY_STRING(&zargs[1]);
220 	}
221 	if (codeUnits) {
222 		ZVAL_STRINGL(&zargs[2], codeUnits, length);
223 	} else {
224 		ZVAL_EMPTY_STRING(&zargs[2]);
225 	}
226 	ZVAL_LONG(&zargs[3], *pErrorCode);
227 	ZVAL_MAKE_REF(&zargs[3]);
228 
229 	objval->to_cb.param_count    = 4;
230 	objval->to_cb.params = zargs;
231 	objval->to_cb.retval = &retval;
232 	if (zend_call_function(&(objval->to_cb), &(objval->to_cache)) == FAILURE) {
233 		/* Unlikely */
234 		php_converter_throw_failure(objval, U_INTERNAL_PROGRAM_ERROR, "Unexpected failure calling toUCallback()");
235 	} else if (!Z_ISUNDEF(retval)) {
236 		php_converter_append_toUnicode_target(&retval, args, objval);
237 		zval_ptr_dtor(&retval);
238 	}
239 
240 	if (Z_TYPE(zargs[3]) == IS_LONG) {
241 		*pErrorCode = Z_LVAL(zargs[3]);
242 	} else if (Z_ISREF(zargs[3]) && Z_TYPE_P(Z_REFVAL(zargs[3])) == IS_LONG) {
243 		*pErrorCode = Z_LVAL_P(Z_REFVAL(zargs[3]));
244 	}
245 
246 	zval_ptr_dtor(&zargs[0]);
247 	zval_ptr_dtor(&zargs[1]);
248 	zval_ptr_dtor(&zargs[2]);
249 	zval_ptr_dtor(&zargs[3]);
250 }
251 /* }}} */
252 
253 /* {{{ php_converter_append_fromUnicode_target */
php_converter_append_fromUnicode_target(zval * val,UConverterFromUnicodeArgs * args,php_converter_object * objval)254 static void php_converter_append_fromUnicode_target(zval *val, UConverterFromUnicodeArgs *args, php_converter_object *objval) {
255 	switch (Z_TYPE_P(val)) {
256 		case IS_NULL:
257 			/* Ignore */
258 			return;
259 		case IS_LONG:
260 			if (TARGET_CHECK(args, 1)) {
261 				*(args->target++) = Z_LVAL_P(val);
262 			}
263 			return;
264 		case IS_STRING:
265 		{
266 			size_t vallen = Z_STRLEN_P(val);
267 			if (TARGET_CHECK(args, vallen)) {
268 				args->target = zend_mempcpy(args->target, Z_STRVAL_P(val), vallen);
269 			}
270 			return;
271 		}
272 		case IS_ARRAY:
273 		{
274 			HashTable *ht = Z_ARRVAL_P(val);
275 			zval *tmpzval;
276 			ZEND_HASH_FOREACH_VAL(ht, tmpzval) {
277 				php_converter_append_fromUnicode_target(tmpzval, args, objval);
278 			} ZEND_HASH_FOREACH_END();
279 			return;
280 		}
281 		default:
282 			php_converter_throw_failure(objval, U_ILLEGAL_ARGUMENT_ERROR, "fromUCallback() specified illegal type for substitution character");
283 	}
284 }
285 /* }}} */
286 
287 /* {{{ php_converter_from_u_callback */
php_converter_from_u_callback(const void * context,UConverterFromUnicodeArgs * args,const UChar * codeUnits,int32_t length,UChar32 codePoint,UConverterCallbackReason reason,UErrorCode * pErrorCode)288 static void php_converter_from_u_callback(const void *context,
289                                           UConverterFromUnicodeArgs *args,
290                                           const UChar *codeUnits, int32_t length, UChar32 codePoint,
291                                           UConverterCallbackReason reason,
292                                           UErrorCode *pErrorCode) {
293 	php_converter_object *objval = (php_converter_object*)context;
294 	zval retval;
295 	zval zargs[4];
296 	int i;
297 
298 	ZVAL_LONG(&zargs[0], reason);
299 	array_init(&zargs[1]);
300 	i = 0;
301 	while (i < length) {
302 		UChar32 c;
303 		U16_NEXT(codeUnits, i, length, c);
304 		add_next_index_long(&zargs[1], c);
305 	}
306 	ZVAL_LONG(&zargs[2], codePoint);
307 	ZVAL_LONG(&zargs[3], *pErrorCode);
308 	ZVAL_MAKE_REF(&zargs[3]);
309 
310 	objval->from_cb.param_count = 4;
311 	objval->from_cb.params = zargs;
312 	objval->from_cb.retval = &retval;
313 	if (zend_call_function(&(objval->from_cb), &(objval->from_cache)) == FAILURE) {
314 		/* Unlikely */
315 		php_converter_throw_failure(objval, U_INTERNAL_PROGRAM_ERROR, "Unexpected failure calling fromUCallback()");
316 	} else if (!Z_ISUNDEF(retval)) {
317 		php_converter_append_fromUnicode_target(&retval, args, objval);
318 		zval_ptr_dtor(&retval);
319 	}
320 
321 	if (Z_TYPE(zargs[3]) == IS_LONG) {
322 		*pErrorCode = Z_LVAL(zargs[3]);
323 	} else if (Z_ISREF(zargs[3]) && Z_TYPE_P(Z_REFVAL(zargs[3])) == IS_LONG) {
324 		*pErrorCode = Z_LVAL_P(Z_REFVAL(zargs[3]));
325 	}
326 
327 	zval_ptr_dtor(&zargs[0]);
328 	zval_ptr_dtor(&zargs[1]);
329 	zval_ptr_dtor(&zargs[2]);
330 	zval_ptr_dtor(&zargs[3]);
331 }
332 /* }}} */
333 
334 /* {{{ php_converter_set_callbacks */
php_converter_set_callbacks(php_converter_object * objval,UConverter * cnv)335 static inline bool php_converter_set_callbacks(php_converter_object *objval, UConverter *cnv) {
336 	bool ret = 1;
337 	UErrorCode error = U_ZERO_ERROR;
338 
339 	if (objval->obj.ce == php_converter_ce) {
340 		/* Short-circuit having to go through method calls and data marshalling
341 		 * when we're using default behavior
342 		 */
343 		return 1;
344 	}
345 
346 	ucnv_setToUCallBack(cnv, (UConverterToUCallback)php_converter_to_u_callback, (const void*)objval,
347 	                    NULL, NULL, &error);
348 	if (U_FAILURE(error)) {
349 		THROW_UFAILURE(objval, "ucnv_setToUCallBack", error);
350 		ret = 0;
351 	}
352 
353 	error = U_ZERO_ERROR;
354 	ucnv_setFromUCallBack(cnv, (UConverterFromUCallback)php_converter_from_u_callback, (const void*)objval,
355 	                      NULL, NULL, &error);
356 	if (U_FAILURE(error)) {
357 		THROW_UFAILURE(objval, "ucnv_setFromUCallBack", error);
358 		ret = 0;
359 	}
360 	return ret;
361 }
362 /* }}} */
363 
364 /* {{{ php_converter_set_encoding */
php_converter_set_encoding(php_converter_object * objval,UConverter ** pcnv,const char * enc,size_t enc_len)365 static bool php_converter_set_encoding(php_converter_object *objval,
366                                        UConverter **pcnv,
367                                        const char *enc, size_t enc_len) {
368 	UErrorCode error = U_ZERO_ERROR;
369 	UConverter *cnv = ucnv_open(enc, &error);
370 
371 	if (error == U_AMBIGUOUS_ALIAS_WARNING) {
372 		UErrorCode getname_error = U_ZERO_ERROR;
373 		const char *actual_encoding = ucnv_getName(cnv, &getname_error);
374 		if (U_FAILURE(getname_error)) {
375 			/* Should never happen */
376 			actual_encoding = "(unknown)";
377 		}
378 		php_error_docref(NULL, E_WARNING, "Ambiguous encoding specified, using %s", actual_encoding);
379 	} else if (U_FAILURE(error)) {
380 		if (objval) {
381 			THROW_UFAILURE(objval, "ucnv_open", error);
382 		} else {
383 			php_error_docref(NULL, E_WARNING, "Error setting encoding: %d - %s", (int)error, u_errorName(error));
384 		}
385 		return 0;
386 	}
387 
388 	if (objval && !php_converter_set_callbacks(objval, cnv)) {
389 		return 0;
390 	}
391 
392 	if (*pcnv) {
393 		ucnv_close(*pcnv);
394 	}
395 	*pcnv = cnv;
396 	return 1;
397 }
398 /* }}} */
399 
400 /* {{{ php_converter_do_set_encoding */
php_converter_do_set_encoding(UConverter ** pcnv,INTERNAL_FUNCTION_PARAMETERS)401 static void php_converter_do_set_encoding(UConverter **pcnv, INTERNAL_FUNCTION_PARAMETERS) {
402 	php_converter_object *objval = CONV_GET(ZEND_THIS);
403 	char *enc;
404 	size_t enc_len;
405 
406 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "s", &enc, &enc_len) == FAILURE) {
407 		RETURN_THROWS();
408 	}
409 	intl_errors_reset(&objval->error);
410 
411 	RETURN_BOOL(php_converter_set_encoding(objval, pcnv, enc, enc_len));
412 }
413 /* }}} */
414 
415 /* {{{ */
PHP_METHOD(UConverter,setSourceEncoding)416 PHP_METHOD(UConverter, setSourceEncoding) {
417 	php_converter_object *objval = CONV_GET(ZEND_THIS);
418 	php_converter_do_set_encoding(&(objval->src), INTERNAL_FUNCTION_PARAM_PASSTHRU);
419 }
420 /* }}} */
421 
422 /* {{{ */
PHP_METHOD(UConverter,setDestinationEncoding)423 PHP_METHOD(UConverter, setDestinationEncoding) {
424 	php_converter_object *objval = CONV_GET(ZEND_THIS);
425 	php_converter_do_set_encoding(&(objval->dest), INTERNAL_FUNCTION_PARAM_PASSTHRU);
426 }
427 /* }}} */
428 
429 /* {{{ php_converter_do_get_encoding */
php_converter_do_get_encoding(php_converter_object * objval,UConverter * cnv,INTERNAL_FUNCTION_PARAMETERS)430 static void php_converter_do_get_encoding(php_converter_object *objval, UConverter *cnv, INTERNAL_FUNCTION_PARAMETERS) {
431 	const char *name;
432 
433 	if (zend_parse_parameters_none() == FAILURE) {
434 		RETURN_THROWS();
435 	}
436 
437 	intl_errors_reset(&objval->error);
438 
439 	if (!cnv) {
440 		RETURN_NULL();
441 	}
442 
443 	name = ucnv_getName(cnv, &objval->error.code);
444 	if (U_FAILURE(objval->error.code)) {
445 		THROW_UFAILURE(objval, "ucnv_getName()", objval->error.code);
446 		RETURN_FALSE;
447 	}
448 
449 	RETURN_STRING(name);
450 }
451 /* }}} */
452 
453 /* {{{ */
PHP_METHOD(UConverter,getSourceEncoding)454 PHP_METHOD(UConverter, getSourceEncoding) {
455 	php_converter_object *objval = CONV_GET(ZEND_THIS);
456 	php_converter_do_get_encoding(objval, objval->src, INTERNAL_FUNCTION_PARAM_PASSTHRU);
457 }
458 /* }}} */
459 
460 /* {{{ */
PHP_METHOD(UConverter,getDestinationEncoding)461 PHP_METHOD(UConverter, getDestinationEncoding) {
462 	php_converter_object *objval = CONV_GET(ZEND_THIS);
463 	php_converter_do_get_encoding(objval, objval->dest, INTERNAL_FUNCTION_PARAM_PASSTHRU);
464 }
465 /* }}} */
466 
467 /* {{{ php_converter_do_get_type */
php_converter_do_get_type(php_converter_object * objval,UConverter * cnv,INTERNAL_FUNCTION_PARAMETERS)468 static void php_converter_do_get_type(php_converter_object *objval, UConverter *cnv, INTERNAL_FUNCTION_PARAMETERS) {
469 	UConverterType t;
470 
471 	if (zend_parse_parameters_none() == FAILURE) {
472 		RETURN_THROWS();
473 	}
474 	intl_errors_reset(&objval->error);
475 
476 	if (!cnv) {
477 		RETURN_NULL();
478 	}
479 
480 	t = ucnv_getType(cnv);
481 	if (U_FAILURE(objval->error.code)) {
482 		THROW_UFAILURE(objval, "ucnv_getType", objval->error.code);
483 		RETURN_FALSE;
484 	}
485 
486 	RETURN_LONG(t);
487 }
488 /* }}} */
489 
490 /* {{{ */
PHP_METHOD(UConverter,getSourceType)491 PHP_METHOD(UConverter, getSourceType) {
492 	php_converter_object *objval = CONV_GET(ZEND_THIS);
493 	php_converter_do_get_type(objval, objval->src, INTERNAL_FUNCTION_PARAM_PASSTHRU);
494 }
495 /* }}} */
496 
497 /* {{{ */
PHP_METHOD(UConverter,getDestinationType)498 PHP_METHOD(UConverter, getDestinationType) {
499 	php_converter_object *objval = CONV_GET(ZEND_THIS);
500 	php_converter_do_get_type(objval, objval->dest, INTERNAL_FUNCTION_PARAM_PASSTHRU);
501 }
502 /* }}} */
503 
504 /* {{{ php_converter_resolve_callback */
php_converter_resolve_callback(zval * zobj,php_converter_object * objval,const char * callback_name,zend_fcall_info * finfo,zend_fcall_info_cache * fcache)505 static void php_converter_resolve_callback(zval *zobj,
506                                            php_converter_object *objval,
507                                            const char *callback_name,
508                                            zend_fcall_info *finfo,
509                                            zend_fcall_info_cache *fcache) {
510 	char *errstr = NULL;
511 	zval caller;
512 
513 	array_init(&caller);
514 	Z_ADDREF_P(zobj);
515 	add_index_zval(&caller, 0, zobj);
516 	add_index_string(&caller, 1, callback_name);
517 	if (zend_fcall_info_init(&caller, 0, finfo, fcache, NULL, &errstr) == FAILURE) {
518 		php_converter_throw_failure(objval, U_INTERNAL_PROGRAM_ERROR, "Error setting converter callback: %s", errstr);
519 	}
520 	zend_array_destroy(Z_ARR(caller));
521 	ZVAL_UNDEF(&finfo->function_name);
522 	if (errstr) {
523 		efree(errstr);
524 	}
525 }
526 /* }}} */
527 
528 /* {{{ */
PHP_METHOD(UConverter,__construct)529 PHP_METHOD(UConverter, __construct) {
530 	php_converter_object *objval = CONV_GET(ZEND_THIS);
531 	char *src = "utf-8";
532 	size_t src_len = sizeof("utf-8") - 1;
533 	char *dest = src;
534 	size_t dest_len = src_len;
535 
536 	intl_error_reset(NULL);
537 
538 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "|s!s!", &dest, &dest_len, &src, &src_len) == FAILURE) {
539 		RETURN_THROWS();
540 	}
541 
542 	php_converter_set_encoding(objval, &(objval->src),  src,  src_len );
543 	php_converter_set_encoding(objval, &(objval->dest), dest, dest_len);
544 	php_converter_resolve_callback(ZEND_THIS, objval, "toUCallback",   &(objval->to_cb),   &(objval->to_cache));
545 	php_converter_resolve_callback(ZEND_THIS, objval, "fromUCallback", &(objval->from_cb), &(objval->from_cache));
546 }
547 /* }}} */
548 
549 /* {{{ */
PHP_METHOD(UConverter,setSubstChars)550 PHP_METHOD(UConverter, setSubstChars) {
551 	php_converter_object *objval = CONV_GET(ZEND_THIS);
552 	char *chars;
553 	size_t chars_len;
554 	int ret = 1;
555 
556 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "s", &chars, &chars_len) == FAILURE) {
557 		RETURN_THROWS();
558 	}
559 	intl_errors_reset(&objval->error);
560 
561 	if (objval->src) {
562 		UErrorCode error = U_ZERO_ERROR;
563 		ucnv_setSubstChars(objval->src, chars, chars_len, &error);
564 		if (U_FAILURE(error)) {
565 			THROW_UFAILURE(objval, "ucnv_setSubstChars", error);
566 			ret = 0;
567 		}
568 	} else {
569 		php_converter_throw_failure(objval, U_INVALID_STATE_ERROR, "Source Converter has not been initialized yet");
570 		ret = 0;
571 	}
572 
573 	if (objval->dest) {
574 		UErrorCode error = U_ZERO_ERROR;
575 		ucnv_setSubstChars(objval->dest, chars, chars_len, &error);
576 		if (U_FAILURE(error)) {
577 			THROW_UFAILURE(objval, "ucnv_setSubstChars", error);
578 			ret = 0;
579 		}
580 	} else {
581 		php_converter_throw_failure(objval, U_INVALID_STATE_ERROR, "Destination Converter has not been initialized yet");
582 		ret = 0;
583 	}
584 
585 	RETURN_BOOL(ret);
586 }
587 /* }}} */
588 
589 /* {{{ */
PHP_METHOD(UConverter,getSubstChars)590 PHP_METHOD(UConverter, getSubstChars) {
591 	php_converter_object *objval = CONV_GET(ZEND_THIS);
592 	char chars[127];
593 	int8_t chars_len = sizeof(chars);
594 	UErrorCode error = U_ZERO_ERROR;
595 
596 	if (zend_parse_parameters_none() == FAILURE) {
597 		RETURN_THROWS();
598 	}
599 	intl_errors_reset(&objval->error);
600 
601 	if (!objval->src) {
602 		RETURN_NULL();
603 	}
604 
605 	/* src and dest get the same subst chars set,
606 	 * so it doesn't really matter which one we read from
607 	 */
608 	ucnv_getSubstChars(objval->src, chars, &chars_len, &error);
609 	if (U_FAILURE(error)) {
610 		THROW_UFAILURE(objval, "ucnv_getSubstChars", error);
611 		RETURN_FALSE;
612 	}
613 
614 	RETURN_STRINGL(chars, chars_len);
615 }
616 /* }}} */
617 
618 /* {{{ php_converter_do_convert */
php_converter_do_convert(UConverter * dest_cnv,UConverter * src_cnv,const char * src,int32_t src_len,php_converter_object * objval)619 static zend_string* php_converter_do_convert(UConverter *dest_cnv,
620                                              UConverter *src_cnv,  const char *src, int32_t src_len,
621                                              php_converter_object *objval
622                                             ) {
623 	UErrorCode	error = U_ZERO_ERROR;
624 	int32_t		temp_len, ret_len;
625 	zend_string	*ret;
626 	UChar		*temp;
627 
628 	if (!src_cnv || !dest_cnv) {
629 		php_converter_throw_failure(objval, U_INVALID_STATE_ERROR,
630 		                            "Internal converters not initialized");
631 		return NULL;
632 	}
633 
634 	/* Get necessary buffer size first */
635 	temp_len = 1 + ucnv_toUChars(src_cnv, NULL, 0, src, src_len, &error);
636 	if (U_FAILURE(error) && error != U_BUFFER_OVERFLOW_ERROR) {
637 		THROW_UFAILURE(objval, "ucnv_toUChars", error);
638 		return NULL;
639 	}
640 	temp = safe_emalloc(sizeof(UChar), temp_len, sizeof(UChar));
641 
642 	/* Convert to intermediate UChar* array */
643 	error = U_ZERO_ERROR;
644 	temp_len = ucnv_toUChars(src_cnv, temp, temp_len, src, src_len, &error);
645 	if (U_FAILURE(error)) {
646 		THROW_UFAILURE(objval, "ucnv_toUChars", error);
647 		efree(temp);
648 		return NULL;
649 	}
650 	temp[temp_len] = 0;
651 
652 	/* Get necessary output buffer size */
653 	ret_len = ucnv_fromUChars(dest_cnv, NULL, 0, temp, temp_len, &error);
654 	if (U_FAILURE(error) && error != U_BUFFER_OVERFLOW_ERROR) {
655 		THROW_UFAILURE(objval, "ucnv_fromUChars", error);
656 		efree(temp);
657 		return NULL;
658 	}
659 
660 	ret = zend_string_alloc(ret_len, 0);
661 
662 	/* Convert to final encoding */
663 	error = U_ZERO_ERROR;
664 	ZSTR_LEN(ret) = ucnv_fromUChars(dest_cnv, ZSTR_VAL(ret), ret_len+1, temp, temp_len, &error);
665 	efree(temp);
666 	if (U_FAILURE(error)) {
667 		THROW_UFAILURE(objval, "ucnv_fromUChars", error);
668 		zend_string_efree(ret);
669 		return NULL;
670 	}
671 
672 	return ret;
673 }
674 /* }}} */
675 
676 /* {{{ */
677 #define UCNV_REASON_CASE(v) case (UCNV_ ## v) : RETURN_STRINGL( "REASON_" #v , sizeof( "REASON_" #v ) - 1);
PHP_METHOD(UConverter,reasonText)678 PHP_METHOD(UConverter, reasonText) {
679 	zend_long reason;
680 
681 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "l", &reason) == FAILURE) {
682 		RETURN_THROWS();
683 	}
684 	intl_error_reset(NULL);
685 
686 	switch (reason) {
687 		UCNV_REASON_CASE(UNASSIGNED)
688 		UCNV_REASON_CASE(ILLEGAL)
689 		UCNV_REASON_CASE(IRREGULAR)
690 		UCNV_REASON_CASE(RESET)
691 		UCNV_REASON_CASE(CLOSE)
692 		UCNV_REASON_CASE(CLONE)
693 		default:
694 			zend_argument_value_error(1, "must be a UConverter::REASON_* constant");
695 			RETURN_THROWS();
696 	}
697 }
698 /* }}} */
699 
700 /* {{{ */
PHP_METHOD(UConverter,convert)701 PHP_METHOD(UConverter, convert) {
702 	php_converter_object *objval = CONV_GET(ZEND_THIS);
703 	char *str;
704 	size_t str_len;
705 	zend_string *ret;
706 	bool reverse = 0;
707 
708 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|b",
709 	                          &str, &str_len, &reverse) == FAILURE) {
710 		RETURN_THROWS();
711 	}
712 	intl_errors_reset(&objval->error);
713 
714 	ret = php_converter_do_convert(reverse ? objval->src : objval->dest,
715 	                               reverse ? objval->dest : objval->src,
716 	                               str,   str_len,
717 	                               objval);
718 	if (ret) {
719 		RETURN_NEW_STR(ret);
720 	} else {
721 		RETURN_FALSE;
722 	}
723 }
724 /* }}} */
725 
726 /* {{{ */
PHP_METHOD(UConverter,transcode)727 PHP_METHOD(UConverter, transcode) {
728 	char *str, *src, *dest;
729 	size_t str_len, src_len, dest_len;
730 	zval *options = NULL;
731 	UConverter *src_cnv = NULL, *dest_cnv = NULL;
732 
733 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "sss|a!",
734 			&str, &str_len, &dest, &dest_len, &src, &src_len, &options) == FAILURE) {
735 		RETURN_THROWS();
736 	}
737 	intl_error_reset(NULL);
738 
739 	if (php_converter_set_encoding(NULL, &src_cnv,  src,  src_len) &&
740 	    php_converter_set_encoding(NULL, &dest_cnv, dest, dest_len)) {
741 	    zend_string *ret;
742 		UErrorCode error = U_ZERO_ERROR;
743 
744 		if (options && zend_hash_num_elements(Z_ARRVAL_P(options))) {
745 			zval *tmpzval;
746 
747 			if (U_SUCCESS(error) &&
748 				(tmpzval = zend_hash_str_find(Z_ARRVAL_P(options), "from_subst", sizeof("from_subst") - 1)) != NULL &&
749 				Z_TYPE_P(tmpzval) == IS_STRING) {
750 				error = U_ZERO_ERROR;
751 				ucnv_setSubstChars(src_cnv, Z_STRVAL_P(tmpzval), Z_STRLEN_P(tmpzval) & 0x7F, &error);
752 			}
753 			if (U_SUCCESS(error) &&
754 				(tmpzval = zend_hash_str_find(Z_ARRVAL_P(options), "to_subst", sizeof("to_subst") - 1)) != NULL &&
755 				Z_TYPE_P(tmpzval) == IS_STRING) {
756 				error = U_ZERO_ERROR;
757 				ucnv_setSubstChars(dest_cnv, Z_STRVAL_P(tmpzval), Z_STRLEN_P(tmpzval) & 0x7F, &error);
758 			}
759 		}
760 
761 		if (U_SUCCESS(error) &&
762 			(ret = php_converter_do_convert(dest_cnv, src_cnv, str, str_len, NULL)) != NULL) {
763 			RETVAL_NEW_STR(ret);
764 		}
765 
766 		if (U_FAILURE(error)) {
767 			THROW_UFAILURE(NULL, "transcode", error);
768 			RETVAL_FALSE;
769 		}
770 	} else {
771 		RETVAL_FALSE;
772 	}
773 
774 	if (src_cnv) {
775 		ucnv_close(src_cnv);
776 	}
777 	if (dest_cnv) {
778 		ucnv_close(dest_cnv);
779 	}
780 }
781 /* }}} */
782 
783 /* {{{ */
PHP_METHOD(UConverter,getErrorCode)784 PHP_METHOD(UConverter, getErrorCode) {
785 	php_converter_object *objval = CONV_GET(ZEND_THIS);
786 
787 	if (zend_parse_parameters_none() == FAILURE) {
788 		RETURN_THROWS();
789 	}
790 
791 	RETURN_LONG(intl_error_get_code(&(objval->error)));
792 }
793 /* }}} */
794 
795 /* {{{ */
PHP_METHOD(UConverter,getErrorMessage)796 PHP_METHOD(UConverter, getErrorMessage) {
797 	php_converter_object *objval = CONV_GET(ZEND_THIS);
798 
799 	if (zend_parse_parameters_none() == FAILURE) {
800 		RETURN_THROWS();
801 	}
802 
803 	zend_string *message = intl_error_get_message(&(objval->error));
804 	if (message) {
805 		RETURN_STR(message);
806 	} else {
807 		RETURN_NULL();
808 	}
809 }
810 /* }}} */
811 
812 /* {{{ */
PHP_METHOD(UConverter,getAvailable)813 PHP_METHOD(UConverter, getAvailable) {
814 	int32_t i,
815 			count = ucnv_countAvailable();
816 
817 	if (zend_parse_parameters_none() == FAILURE) {
818 		RETURN_THROWS();
819 	}
820 	intl_error_reset(NULL);
821 
822 	array_init(return_value);
823 	for(i = 0; i < count; i++) {
824 		const char *name = ucnv_getAvailableName(i);
825 		add_next_index_string(return_value, name);
826 	}
827 }
828 /* }}} */
829 
830 /* {{{ */
PHP_METHOD(UConverter,getAliases)831 PHP_METHOD(UConverter, getAliases) {
832 	char *name;
833 	size_t name_len;
834 	UErrorCode error = U_ZERO_ERROR;
835 	uint16_t i, count;
836 
837 	if (zend_parse_parameters(ZEND_NUM_ARGS(), "s", &name, &name_len) == FAILURE) {
838 		RETURN_THROWS();
839 	}
840 	intl_error_reset(NULL);
841 
842 	count = ucnv_countAliases(name, &error);
843 	if (U_FAILURE(error)) {
844 		THROW_UFAILURE(NULL, "ucnv_countAliases", error);
845 		RETURN_FALSE;
846 	}
847 
848 	array_init(return_value);
849 	for(i = 0; i < count; i++) {
850 		const char *alias;
851 
852 		error = U_ZERO_ERROR;
853 		alias = ucnv_getAlias(name, i, &error);
854 		if (U_FAILURE(error)) {
855 			THROW_UFAILURE(NULL, "ucnv_getAlias", error);
856 			zend_array_destroy(Z_ARR_P(return_value));
857 			RETURN_NULL();
858 		}
859 		add_next_index_string(return_value, alias);
860 	}
861 }
862 /* }}} */
863 
864 /* {{{ */
PHP_METHOD(UConverter,getStandards)865 PHP_METHOD(UConverter, getStandards) {
866 	uint16_t i, count;
867 
868 	if (zend_parse_parameters_none() == FAILURE) {
869 		RETURN_THROWS();
870 	}
871 	intl_error_reset(NULL);
872 
873 	array_init(return_value);
874 	count = ucnv_countStandards();
875 	for(i = 0; i < count; i++) {
876 		UErrorCode error = U_ZERO_ERROR;
877 		const char *name = ucnv_getStandard(i, &error);
878 		if (U_FAILURE(error)) {
879 			THROW_UFAILURE(NULL, "ucnv_getStandard", error);
880 			zend_array_destroy(Z_ARR_P(return_value));
881 			RETURN_NULL();
882 		}
883 		add_next_index_string(return_value, name);
884 	}
885 }
886 /* }}} */
887 
888 /* {{{ Converter create/clone/destroy */
php_converter_free_object(zend_object * obj)889 static void php_converter_free_object(zend_object *obj) {
890 	php_converter_object *objval = php_converter_fetch_object(obj);
891 
892 	if (objval->src) {
893 		ucnv_close(objval->src);
894 	}
895 
896 	if (objval->dest) {
897 		ucnv_close(objval->dest);
898 	}
899 
900 	intl_error_reset(&objval->error);
901 	zend_object_std_dtor(obj);
902 }
903 
php_converter_object_ctor(zend_class_entry * ce,php_converter_object ** pobjval)904 static zend_object *php_converter_object_ctor(zend_class_entry *ce, php_converter_object **pobjval) {
905 	php_converter_object *objval;
906 
907 	objval = zend_object_alloc(sizeof(php_converter_object), ce);
908 
909 	zend_object_std_init(&objval->obj, ce);
910 	object_properties_init(&objval->obj, ce);
911 	intl_error_init(&(objval->error));
912 
913 	*pobjval = objval;
914 
915 	return &objval->obj;
916 }
917 
php_converter_create_object(zend_class_entry * ce)918 static zend_object *php_converter_create_object(zend_class_entry *ce) {
919 	php_converter_object *objval = NULL;
920 	zend_object *retval = php_converter_object_ctor(ce, &objval);
921 
922 	object_properties_init(&(objval->obj), ce);
923 
924 	return retval;
925 }
926 
php_converter_clone_object(zend_object * object)927 static zend_object *php_converter_clone_object(zend_object *object) {
928 	php_converter_object *objval, *oldobj = php_converter_fetch_object(object);
929 	zend_object *retval = php_converter_object_ctor(object->ce, &objval);
930 	UErrorCode error = U_ZERO_ERROR;
931 
932 	intl_errors_reset(&oldobj->error);
933 
934 #if U_ICU_VERSION_MAJOR_NUM > 70
935 	objval->src = ucnv_clone(oldobj->src, &error);
936 #else
937 	objval->src = ucnv_safeClone(oldobj->src, NULL, NULL, &error);
938 #endif
939 	if (U_SUCCESS(error)) {
940 		error = U_ZERO_ERROR;
941 #if U_ICU_VERSION_MAJOR_NUM > 70
942 		objval->dest = ucnv_clone(oldobj->dest, &error);
943 #else
944 		objval->dest = ucnv_safeClone(oldobj->dest, NULL, NULL, &error);
945 #endif
946 	}
947 	if (U_FAILURE(error)) {
948 		zend_string *err_msg;
949 		THROW_UFAILURE(oldobj, "ucnv_safeClone", error);
950 
951 		err_msg = intl_error_get_message(&oldobj->error);
952 		zend_throw_exception(NULL, ZSTR_VAL(err_msg), 0);
953 		zend_string_release_ex(err_msg, 0);
954 
955 		return retval;
956 	}
957 
958 	/* Update contexts for converter error handlers */
959 	php_converter_set_callbacks(objval, objval->src );
960 	php_converter_set_callbacks(objval, objval->dest);
961 
962 	zend_objects_clone_members(&(objval->obj), &(oldobj->obj));
963 
964 	/* Newly cloned object deliberately does not inherit error state from original object */
965 
966 	return retval;
967 }
968 /* }}} */
969 
970 /* {{{ php_converter_minit */
php_converter_minit(INIT_FUNC_ARGS)971 int php_converter_minit(INIT_FUNC_ARGS) {
972 	php_converter_ce = register_class_UConverter();
973 	php_converter_ce->create_object = php_converter_create_object;
974 	php_converter_ce->default_object_handlers = &php_converter_object_handlers;
975 	memcpy(&php_converter_object_handlers, &std_object_handlers, sizeof(zend_object_handlers));
976 	php_converter_object_handlers.offset = XtOffsetOf(php_converter_object, obj);
977 	php_converter_object_handlers.clone_obj = php_converter_clone_object;
978 	php_converter_object_handlers.free_obj = php_converter_free_object;
979 
980 	return SUCCESS;
981 }
982 /* }}} */
983