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 | http://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 #include <unicode/rbbi.h>
16
17 extern "C" {
18 #define USE_BREAKITERATOR_POINTER 1
19 #include "breakiterator_class.h"
20 #include <zend_exceptions.h>
21 #include <limits.h>
22 }
23
24 #include "../intl_convertcpp.h"
25 #include "../intl_common.h"
26
27 using icu::RuleBasedBreakIterator;
28 using icu::Locale;
29
fetch_rbbi(BreakIterator_object * bio)30 static inline RuleBasedBreakIterator *fetch_rbbi(BreakIterator_object *bio) {
31 return (RuleBasedBreakIterator*)bio->biter;
32 }
33
_php_intlrbbi_constructor_body(INTERNAL_FUNCTION_PARAMETERS)34 static void _php_intlrbbi_constructor_body(INTERNAL_FUNCTION_PARAMETERS)
35 {
36 char *rules;
37 size_t rules_len;
38 zend_bool compiled = 0;
39 UErrorCode status = U_ZERO_ERROR;
40 BREAKITER_METHOD_INIT_VARS;
41 object = ZEND_THIS;
42
43 if (zend_parse_parameters(ZEND_NUM_ARGS(), "s|b",
44 &rules, &rules_len, &compiled) == FAILURE) {
45 RETURN_THROWS();
46 }
47
48 BREAKITER_METHOD_FETCH_OBJECT_NO_CHECK;
49 if (bio->biter) {
50 zend_throw_error(NULL, "IntlRuleBasedBreakIterator object is already constructed");
51 RETURN_THROWS();
52 }
53
54 // instantiation of ICU object
55 RuleBasedBreakIterator *rbbi;
56
57 if (!compiled) {
58 UnicodeString rulesStr;
59 UParseError parseError = UParseError();
60 if (intl_stringFromChar(rulesStr, rules, rules_len, &status)
61 == FAILURE) {
62 zend_throw_exception(IntlException_ce_ptr,
63 "IntlRuleBasedBreakIterator::__construct(): "
64 "rules were not a valid UTF-8 string", 0);
65 RETURN_THROWS();
66 }
67
68 rbbi = new RuleBasedBreakIterator(rulesStr, parseError, status);
69 intl_error_set_code(NULL, status);
70 if (U_FAILURE(status)) {
71 smart_str parse_error_str;
72 parse_error_str = intl_parse_error_to_string(&parseError);
73 zend_throw_exception_ex(IntlException_ce_ptr, 0,
74 "IntlRuleBasedBreakIterator::__construct(): "
75 "unable to create RuleBasedBreakIterator from rules (%s)",
76 parse_error_str.s ? ZSTR_VAL(parse_error_str.s) : "");
77 smart_str_free(&parse_error_str);
78 delete rbbi;
79 RETURN_THROWS();
80 }
81 } else { // compiled
82 rbbi = new RuleBasedBreakIterator((uint8_t*)rules, rules_len, status);
83 if (U_FAILURE(status)) {
84 zend_throw_exception(IntlException_ce_ptr,
85 "IntlRuleBasedBreakIterator::__construct(): "
86 "unable to create instance from compiled rules", 0);
87 delete rbbi;
88 RETURN_THROWS();
89 }
90 }
91
92 breakiterator_object_create(return_value, rbbi, 0);
93 }
94
PHP_METHOD(IntlRuleBasedBreakIterator,__construct)95 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, __construct)
96 {
97 zend_error_handling error_handling;
98
99 zend_replace_error_handling(EH_THROW, IntlException_ce_ptr, &error_handling);
100 return_value = ZEND_THIS;
101 _php_intlrbbi_constructor_body(INTERNAL_FUNCTION_PARAM_PASSTHRU);
102 zend_restore_error_handling(&error_handling);
103 }
104
PHP_METHOD(IntlRuleBasedBreakIterator,getRules)105 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, getRules)
106 {
107 BREAKITER_METHOD_INIT_VARS;
108 object = ZEND_THIS;
109
110 if (zend_parse_parameters_none() == FAILURE) {
111 RETURN_THROWS();
112 }
113
114 BREAKITER_METHOD_FETCH_OBJECT;
115
116 zend_string *u8str;
117 const UnicodeString rules = fetch_rbbi(bio)->getRules();
118
119 u8str = intl_charFromString(rules, BREAKITER_ERROR_CODE_P(bio));
120 if (!u8str)
121 {
122 intl_errors_set(BREAKITER_ERROR_P(bio), BREAKITER_ERROR_CODE(bio),
123 "rbbi_hash_code: Error converting result to UTF-8 string",
124 0);
125 RETURN_FALSE;
126 }
127 RETVAL_STR(u8str);
128 }
129
PHP_METHOD(IntlRuleBasedBreakIterator,getRuleStatus)130 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, getRuleStatus)
131 {
132 BREAKITER_METHOD_INIT_VARS;
133 object = ZEND_THIS;
134
135 if (zend_parse_parameters_none() == FAILURE) {
136 RETURN_THROWS();
137 }
138
139 BREAKITER_METHOD_FETCH_OBJECT;
140
141 RETURN_LONG(fetch_rbbi(bio)->getRuleStatus());
142 }
143
PHP_METHOD(IntlRuleBasedBreakIterator,getRuleStatusVec)144 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, getRuleStatusVec)
145 {
146 BREAKITER_METHOD_INIT_VARS;
147 object = ZEND_THIS;
148
149 if (zend_parse_parameters_none() == FAILURE) {
150 RETURN_THROWS();
151 }
152
153 BREAKITER_METHOD_FETCH_OBJECT;
154
155 int32_t num_rules = fetch_rbbi(bio)->getRuleStatusVec(NULL, 0,
156 BREAKITER_ERROR_CODE(bio));
157
158 ZEND_ASSERT(BREAKITER_ERROR_CODE(bio) == U_BUFFER_OVERFLOW_ERROR);
159 BREAKITER_ERROR_CODE(bio) = U_ZERO_ERROR;
160
161 int32_t *rules = new int32_t[num_rules];
162 num_rules = fetch_rbbi(bio)->getRuleStatusVec(rules, num_rules,
163 BREAKITER_ERROR_CODE(bio));
164 if (U_FAILURE(BREAKITER_ERROR_CODE(bio))) {
165 delete[] rules;
166 intl_errors_set(BREAKITER_ERROR_P(bio), BREAKITER_ERROR_CODE(bio),
167 "rbbi_get_rule_status_vec: failed obtaining the status values",
168 0);
169 RETURN_FALSE;
170 }
171
172 array_init_size(return_value, num_rules);
173 for (int32_t i = 0; i < num_rules; i++) {
174 add_next_index_long(return_value, rules[i]);
175 }
176 delete[] rules;
177 }
178
PHP_METHOD(IntlRuleBasedBreakIterator,getBinaryRules)179 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, getBinaryRules)
180 {
181 BREAKITER_METHOD_INIT_VARS;
182 object = ZEND_THIS;
183
184 if (zend_parse_parameters_none() == FAILURE) {
185 RETURN_THROWS();
186 }
187
188 BREAKITER_METHOD_FETCH_OBJECT;
189
190 uint32_t rules_len;
191 const uint8_t *rules = fetch_rbbi(bio)->getBinaryRules(rules_len);
192
193 if (rules_len > INT_MAX - 1) {
194 intl_errors_set(BREAKITER_ERROR_P(bio), BREAKITER_ERROR_CODE(bio),
195 "rbbi_get_binary_rules: the rules are too large",
196 0);
197 RETURN_FALSE;
198 }
199
200 zend_string *ret_rules = zend_string_alloc(rules_len, 0);
201 memcpy(ZSTR_VAL(ret_rules), rules, rules_len);
202 ZSTR_VAL(ret_rules)[rules_len] = '\0';
203
204 RETURN_STR(ret_rules);
205 }
206
