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: Gustavo Lopes <cataphract@php.net> |
12 +----------------------------------------------------------------------+
13 */
14
15 #ifdef HAVE_CONFIG_H
16 #include "config.h"
17 #endif
18
19 #include <unicode/brkiter.h>
20 #include "codepointiterator_internal.h"
21
22 #include "breakiterator_iterators.h"
23
24 extern "C" {
25 #include "../php_intl.h"
26 #define USE_BREAKITERATOR_POINTER 1
27 #include "breakiterator_class.h"
28 #include "../locale/locale.h"
29 #include <zend_exceptions.h>
30 #include <zend_interfaces.h>
31 }
32
33 using PHP::CodePointBreakIterator;
34 using icu::BreakIterator;
35 using icu::Locale;
36
PHP_METHOD(IntlBreakIterator,__construct)37 U_CFUNC PHP_METHOD(IntlBreakIterator, __construct)
38 {
39 zend_throw_exception( NULL,
40 "An object of this type cannot be created with the new operator",
41 0 );
42 }
43
_breakiter_factory(const char * func_name,BreakIterator * (* func)(const Locale &,UErrorCode &),INTERNAL_FUNCTION_PARAMETERS)44 static void _breakiter_factory(const char *func_name,
45 BreakIterator *(*func)(const Locale&, UErrorCode&),
46 INTERNAL_FUNCTION_PARAMETERS)
47 {
48 BreakIterator *biter;
49 char *locale_str = NULL;
50 size_t dummy;
51 char *msg;
52 UErrorCode status = UErrorCode();
53 intl_error_reset(NULL);
54
55 ZEND_PARSE_PARAMETERS_START(0, 1)
56 Z_PARAM_OPTIONAL
57 Z_PARAM_STRING_OR_NULL(locale_str, dummy)
58 ZEND_PARSE_PARAMETERS_END();
59
60 if (locale_str == NULL) {
61 locale_str = (char *)intl_locale_get_default();
62 }
63
64 biter = func(Locale::createFromName(locale_str), status);
65 intl_error_set_code(NULL, status);
66 if (U_FAILURE(status)) {
67 spprintf(&msg, 0, "%s: error creating BreakIterator",
68 func_name);
69 intl_error_set_custom_msg(NULL, msg, 1);
70 efree(msg);
71 RETURN_NULL();
72 }
73
74 breakiterator_object_create(return_value, biter, 1);
75 }
76
PHP_METHOD(IntlBreakIterator,createWordInstance)77 U_CFUNC PHP_METHOD(IntlBreakIterator, createWordInstance)
78 {
79 _breakiter_factory("breakiter_create_word_instance",
80 &BreakIterator::createWordInstance,
81 INTERNAL_FUNCTION_PARAM_PASSTHRU);
82 }
83
PHP_METHOD(IntlBreakIterator,createLineInstance)84 U_CFUNC PHP_METHOD(IntlBreakIterator, createLineInstance)
85 {
86 _breakiter_factory("breakiter_create_line_instance",
87 &BreakIterator::createLineInstance,
88 INTERNAL_FUNCTION_PARAM_PASSTHRU);
89 }
90
PHP_METHOD(IntlBreakIterator,createCharacterInstance)91 U_CFUNC PHP_METHOD(IntlBreakIterator, createCharacterInstance)
92 {
93 _breakiter_factory("breakiter_create_character_instance",
94 &BreakIterator::createCharacterInstance,
95 INTERNAL_FUNCTION_PARAM_PASSTHRU);
96 }
97
PHP_METHOD(IntlBreakIterator,createSentenceInstance)98 U_CFUNC PHP_METHOD(IntlBreakIterator, createSentenceInstance)
99 {
100 _breakiter_factory("breakiter_create_sentence_instance",
101 &BreakIterator::createSentenceInstance,
102 INTERNAL_FUNCTION_PARAM_PASSTHRU);
103 }
104
PHP_METHOD(IntlBreakIterator,createTitleInstance)105 U_CFUNC PHP_METHOD(IntlBreakIterator, createTitleInstance)
106 {
107 _breakiter_factory("breakiter_create_title_instance",
108 &BreakIterator::createTitleInstance,
109 INTERNAL_FUNCTION_PARAM_PASSTHRU);
110 }
111
PHP_METHOD(IntlBreakIterator,createCodePointInstance)112 U_CFUNC PHP_METHOD(IntlBreakIterator, createCodePointInstance)
113 {
114 intl_error_reset(NULL);
115
116 ZEND_PARSE_PARAMETERS_NONE();
117
118 CodePointBreakIterator *cpbi = new CodePointBreakIterator();
119 breakiterator_object_create(return_value, cpbi, 1);
120 }
121
PHP_METHOD(IntlBreakIterator,getText)122 U_CFUNC PHP_METHOD(IntlBreakIterator, getText)
123 {
124 BREAKITER_METHOD_INIT_VARS;
125 object = ZEND_THIS;
126
127 ZEND_PARSE_PARAMETERS_NONE();
128
129 BREAKITER_METHOD_FETCH_OBJECT;
130
131 if (Z_ISUNDEF(bio->text)) {
132 RETURN_NULL();
133 } else {
134 ZVAL_COPY(return_value, &bio->text);
135 }
136 }
137
PHP_METHOD(IntlBreakIterator,setText)138 U_CFUNC PHP_METHOD(IntlBreakIterator, setText)
139 {
140 UText *ut = NULL;
141 zend_string *text;
142 BREAKITER_METHOD_INIT_VARS;
143 object = ZEND_THIS;
144
145 ZEND_PARSE_PARAMETERS_START(1, 1)
146 Z_PARAM_STR(text)
147 ZEND_PARSE_PARAMETERS_END();
148
149 BREAKITER_METHOD_FETCH_OBJECT;
150
151 ut = utext_openUTF8(ut, ZSTR_VAL(text), ZSTR_LEN(text), BREAKITER_ERROR_CODE_P(bio));
152 INTL_METHOD_CHECK_STATUS(bio, "breakiter_set_text: error opening UText");
153
154 bio->biter->setText(ut, BREAKITER_ERROR_CODE(bio));
155 utext_close(ut); /* ICU shallow clones the UText */
156 INTL_METHOD_CHECK_STATUS(bio, "breakiter_set_text: error calling "
157 "BreakIterator::setText()");
158
159 /* When ICU clones the UText, it does not copy the buffer, so we have to
160 * keep the string buffer around by holding a reference to its zval. This
161 * also allows a faste implementation of getText() */
162 zval_ptr_dtor(&bio->text);
163 ZVAL_STR_COPY(&bio->text, text);
164
165 RETURN_TRUE;
166 }
167
_breakiter_no_args_ret_int32(int32_t (BreakIterator::* func)(),INTERNAL_FUNCTION_PARAMETERS)168 static void _breakiter_no_args_ret_int32(
169 int32_t (BreakIterator::*func)(),
170 INTERNAL_FUNCTION_PARAMETERS)
171 {
172 BREAKITER_METHOD_INIT_VARS;
173 object = ZEND_THIS;
174
175 ZEND_PARSE_PARAMETERS_NONE();
176
177 BREAKITER_METHOD_FETCH_OBJECT;
178
179 int32_t res = (bio->biter->*func)();
180
181 RETURN_LONG((zend_long)res);
182 }
183
_breakiter_int32_ret_int32(int32_t (BreakIterator::* func)(int32_t),INTERNAL_FUNCTION_PARAMETERS)184 static void _breakiter_int32_ret_int32(
185 int32_t (BreakIterator::*func)(int32_t),
186 INTERNAL_FUNCTION_PARAMETERS)
187 {
188 zend_long arg;
189 BREAKITER_METHOD_INIT_VARS;
190 object = ZEND_THIS;
191
192 ZEND_PARSE_PARAMETERS_START(1, 1)
193 Z_PARAM_LONG(arg)
194 ZEND_PARSE_PARAMETERS_END();
195
196 BREAKITER_METHOD_FETCH_OBJECT;
197
198 if (UNEXPECTED(arg < INT32_MIN || arg > INT32_MAX)) {
199 zend_argument_value_error(1, "must be between %d and %d", INT32_MIN, INT32_MAX);
200 RETURN_THROWS();
201 }
202
203 int32_t res = (bio->biter->*func)((int32_t)arg);
204
205 RETURN_LONG((zend_long)res);
206 }
207
PHP_METHOD(IntlBreakIterator,first)208 U_CFUNC PHP_METHOD(IntlBreakIterator, first)
209 {
210 _breakiter_no_args_ret_int32(&BreakIterator::first,
211 INTERNAL_FUNCTION_PARAM_PASSTHRU);
212 }
213
PHP_METHOD(IntlBreakIterator,last)214 U_CFUNC PHP_METHOD(IntlBreakIterator, last)
215 {
216 _breakiter_no_args_ret_int32(&BreakIterator::last,
217 INTERNAL_FUNCTION_PARAM_PASSTHRU);
218 }
219
PHP_METHOD(IntlBreakIterator,previous)220 U_CFUNC PHP_METHOD(IntlBreakIterator, previous)
221 {
222 _breakiter_no_args_ret_int32(&BreakIterator::previous,
223 INTERNAL_FUNCTION_PARAM_PASSTHRU);
224 }
225
PHP_METHOD(IntlBreakIterator,next)226 U_CFUNC PHP_METHOD(IntlBreakIterator, next)
227 {
228 zval *arg = NULL;
229
230 ZEND_PARSE_PARAMETERS_START(0, 1)
231 Z_PARAM_OPTIONAL
232 Z_PARAM_ZVAL_OR_NULL(arg)
233 ZEND_PARSE_PARAMETERS_END();
234
235 if (arg == NULL) {
236 ZEND_NUM_ARGS() = 0;
237 _breakiter_no_args_ret_int32(&BreakIterator::next,
238 INTERNAL_FUNCTION_PARAM_PASSTHRU);
239 } else {
240 _breakiter_int32_ret_int32(&BreakIterator::next,
241 INTERNAL_FUNCTION_PARAM_PASSTHRU);
242 }
243 }
244
PHP_METHOD(IntlBreakIterator,current)245 U_CFUNC PHP_METHOD(IntlBreakIterator, current)
246 {
247 BREAKITER_METHOD_INIT_VARS;
248 object = ZEND_THIS;
249
250 ZEND_PARSE_PARAMETERS_NONE();
251
252 BREAKITER_METHOD_FETCH_OBJECT;
253
254 int32_t res = bio->biter->current();
255
256 RETURN_LONG((zend_long)res);
257 }
258
PHP_METHOD(IntlBreakIterator,following)259 U_CFUNC PHP_METHOD(IntlBreakIterator, following)
260 {
261 _breakiter_int32_ret_int32(
262 &BreakIterator::following,
263 INTERNAL_FUNCTION_PARAM_PASSTHRU);
264 }
265
PHP_METHOD(IntlBreakIterator,preceding)266 U_CFUNC PHP_METHOD(IntlBreakIterator, preceding)
267 {
268 _breakiter_int32_ret_int32(
269 &BreakIterator::preceding,
270 INTERNAL_FUNCTION_PARAM_PASSTHRU);
271 }
272
PHP_METHOD(IntlBreakIterator,isBoundary)273 U_CFUNC PHP_METHOD(IntlBreakIterator, isBoundary)
274 {
275 zend_long offset;
276 BREAKITER_METHOD_INIT_VARS;
277 object = ZEND_THIS;
278
279 ZEND_PARSE_PARAMETERS_START(1, 1)
280 Z_PARAM_LONG(offset)
281 ZEND_PARSE_PARAMETERS_END();
282
283 if (UNEXPECTED(offset < INT32_MIN || offset > INT32_MAX)) {
284 zend_argument_value_error(1, "must be between %d and %d", INT32_MIN, INT32_MAX);
285 RETURN_THROWS();
286 }
287
288 BREAKITER_METHOD_FETCH_OBJECT;
289
290 UBool res = bio->biter->isBoundary((int32_t)offset);
291
292 RETURN_BOOL((zend_long)res);
293 }
294
PHP_METHOD(IntlBreakIterator,getLocale)295 U_CFUNC PHP_METHOD(IntlBreakIterator, getLocale)
296 {
297 zend_long locale_type;
298 BREAKITER_METHOD_INIT_VARS;
299 object = ZEND_THIS;
300
301 ZEND_PARSE_PARAMETERS_START(1, 1)
302 Z_PARAM_LONG(locale_type)
303 ZEND_PARSE_PARAMETERS_END();
304
305 /* Change to ValueError? */
306 if (locale_type != ULOC_ACTUAL_LOCALE && locale_type != ULOC_VALID_LOCALE) {
307 intl_error_set(NULL, U_ILLEGAL_ARGUMENT_ERROR,
308 "breakiter_get_locale: invalid locale type", 0);
309 RETURN_FALSE;
310 }
311
312 BREAKITER_METHOD_FETCH_OBJECT;
313
314 Locale locale = bio->biter->getLocale((ULocDataLocaleType)locale_type,
315 BREAKITER_ERROR_CODE(bio));
316 INTL_METHOD_CHECK_STATUS(bio,
317 "breakiter_get_locale: Call to ICU method has failed");
318
319 RETURN_STRING(locale.getName());
320 }
321
PHP_METHOD(IntlBreakIterator,getPartsIterator)322 U_CFUNC PHP_METHOD(IntlBreakIterator, getPartsIterator)
323 {
324 zend_long key_type = 0;
325 BREAKITER_METHOD_INIT_VARS;
326 object = ZEND_THIS;
327
328 ZEND_PARSE_PARAMETERS_START(0, 1)
329 Z_PARAM_OPTIONAL
330 Z_PARAM_LONG(key_type)
331 ZEND_PARSE_PARAMETERS_END();
332
333 if (key_type != PARTS_ITERATOR_KEY_SEQUENTIAL
334 && key_type != PARTS_ITERATOR_KEY_LEFT
335 && key_type != PARTS_ITERATOR_KEY_RIGHT) {
336 zend_argument_value_error(1, "must be one of IntlPartsIterator::KEY_SEQUENTIAL, "
337 "IntlPartsIterator::KEY_LEFT, or IntlPartsIterator::KEY_RIGHT");
338 RETURN_THROWS();
339 }
340
341 BREAKITER_METHOD_FETCH_OBJECT;
342
343 IntlIterator_from_BreakIterator_parts(
344 object, return_value, (parts_iter_key_type)key_type);
345 }
346
PHP_METHOD(IntlBreakIterator,getErrorCode)347 U_CFUNC PHP_METHOD(IntlBreakIterator, getErrorCode)
348 {
349 BREAKITER_METHOD_INIT_VARS;
350 object = ZEND_THIS;
351
352 ZEND_PARSE_PARAMETERS_NONE();
353
354 /* Fetch the object (without resetting its last error code ). */
355 bio = Z_INTL_BREAKITERATOR_P(object);
356 RETURN_LONG((zend_long)BREAKITER_ERROR_CODE(bio));
357 }
358
PHP_METHOD(IntlBreakIterator,getErrorMessage)359 U_CFUNC PHP_METHOD(IntlBreakIterator, getErrorMessage)
360 {
361 zend_string* message = NULL;
362 BREAKITER_METHOD_INIT_VARS;
363 object = ZEND_THIS;
364
365 ZEND_PARSE_PARAMETERS_NONE();
366
367 /* Fetch the object (without resetting its last error code ). */
368 bio = Z_INTL_BREAKITERATOR_P(object);
369
370 /* Return last error message. */
371 message = intl_error_get_message(BREAKITER_ERROR_P(bio));
372 RETURN_STR(message);
373 }
374
PHP_METHOD(IntlBreakIterator,getIterator)375 U_CFUNC PHP_METHOD(IntlBreakIterator, getIterator)
376 {
377 ZEND_PARSE_PARAMETERS_NONE();
378
379 zend_create_internal_iterator_zval(return_value, ZEND_THIS);
380 }
381
