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 #include <unicode/rbbi.h>
16 #include <memory>
17
18 extern "C" {
19 #define USE_BREAKITERATOR_POINTER 1
20 #include "breakiterator_class.h"
21 #include <zend_exceptions.h>
22 #include <limits.h>
23 }
24
25 #include "../intl_convertcpp.h"
26 #include "../intl_common.h"
27
28 using icu::RuleBasedBreakIterator;
29 using icu::Locale;
30
fetch_rbbi(BreakIterator_object * bio)31 static inline RuleBasedBreakIterator *fetch_rbbi(BreakIterator_object *bio) {
32 return (RuleBasedBreakIterator*)bio->biter;
33 }
34
_php_intlrbbi_constructor_body(INTERNAL_FUNCTION_PARAMETERS,zend_error_handling * error_handling,bool * error_handling_replaced)35 static void _php_intlrbbi_constructor_body(INTERNAL_FUNCTION_PARAMETERS, zend_error_handling *error_handling, bool *error_handling_replaced)
36 {
37 char *rules;
38 size_t rules_len;
39 bool compiled = false;
40 UErrorCode status = U_ZERO_ERROR;
41 BREAKITER_METHOD_INIT_VARS;
42 object = ZEND_THIS;
43
44 ZEND_PARSE_PARAMETERS_START(1, 2)
45 Z_PARAM_STRING(rules, rules_len)
46 Z_PARAM_OPTIONAL
47 Z_PARAM_BOOL(compiled)
48 ZEND_PARSE_PARAMETERS_END();
49
50 BREAKITER_METHOD_FETCH_OBJECT_NO_CHECK;
51 if (bio->biter) {
52 zend_throw_error(NULL, "IntlRuleBasedBreakIterator object is already constructed");
53 RETURN_THROWS();
54 }
55
56 zend_replace_error_handling(EH_THROW, IntlException_ce_ptr, error_handling);
57 *error_handling_replaced = 1;
58
59 // instantiation of ICU object
60 RuleBasedBreakIterator *rbbi;
61
62 if (!compiled) {
63 UnicodeString rulesStr;
64 UParseError parseError = UParseError();
65 if (intl_stringFromChar(rulesStr, rules, rules_len, &status)
66 == FAILURE) {
67 zend_throw_exception(IntlException_ce_ptr,
68 "IntlRuleBasedBreakIterator::__construct(): "
69 "rules were not a valid UTF-8 string", 0);
70 RETURN_THROWS();
71 }
72
73 rbbi = new RuleBasedBreakIterator(rulesStr, parseError, status);
74 intl_error_set_code(NULL, status);
75 if (U_FAILURE(status)) {
76 smart_str parse_error_str;
77 parse_error_str = intl_parse_error_to_string(&parseError);
78 zend_throw_exception_ex(IntlException_ce_ptr, 0,
79 "IntlRuleBasedBreakIterator::__construct(): "
80 "unable to create RuleBasedBreakIterator from rules (%s)",
81 parse_error_str.s ? ZSTR_VAL(parse_error_str.s) : "");
82 smart_str_free(&parse_error_str);
83 delete rbbi;
84 RETURN_THROWS();
85 }
86 } else { // compiled
87 rbbi = new RuleBasedBreakIterator((uint8_t*)rules, rules_len, status);
88 if (U_FAILURE(status)) {
89 zend_throw_exception(IntlException_ce_ptr,
90 "IntlRuleBasedBreakIterator::__construct(): "
91 "unable to create instance from compiled rules", 0);
92 delete rbbi;
93 RETURN_THROWS();
94 }
95 }
96
97 breakiterator_object_create(return_value, rbbi, 0);
98 }
99
PHP_METHOD(IntlRuleBasedBreakIterator,__construct)100 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, __construct)
101 {
102 zend_error_handling error_handling;
103 bool error_handling_replaced = 0;
104
105 return_value = ZEND_THIS;
106 _php_intlrbbi_constructor_body(INTERNAL_FUNCTION_PARAM_PASSTHRU, &error_handling, &error_handling_replaced);
107 if (error_handling_replaced) {
108 zend_restore_error_handling(&error_handling);
109 }
110 }
111
PHP_METHOD(IntlRuleBasedBreakIterator,getRules)112 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, getRules)
113 {
114 BREAKITER_METHOD_INIT_VARS;
115 object = ZEND_THIS;
116
117 ZEND_PARSE_PARAMETERS_NONE();
118
119 BREAKITER_METHOD_FETCH_OBJECT;
120
121 zend_string *u8str;
122 const UnicodeString rules = fetch_rbbi(bio)->getRules();
123
124 u8str = intl_charFromString(rules, BREAKITER_ERROR_CODE_P(bio));
125 if (!u8str)
126 {
127 intl_errors_set(BREAKITER_ERROR_P(bio), BREAKITER_ERROR_CODE(bio),
128 "rbbi_hash_code: Error converting result to UTF-8 string",
129 0);
130 RETURN_FALSE;
131 }
132 RETVAL_STR(u8str);
133 }
134
PHP_METHOD(IntlRuleBasedBreakIterator,getRuleStatus)135 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, getRuleStatus)
136 {
137 BREAKITER_METHOD_INIT_VARS;
138 object = ZEND_THIS;
139
140 ZEND_PARSE_PARAMETERS_NONE();
141
142 BREAKITER_METHOD_FETCH_OBJECT;
143
144 RETURN_LONG(fetch_rbbi(bio)->getRuleStatus());
145 }
146
PHP_METHOD(IntlRuleBasedBreakIterator,getRuleStatusVec)147 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, getRuleStatusVec)
148 {
149 BREAKITER_METHOD_INIT_VARS;
150 object = ZEND_THIS;
151
152 ZEND_PARSE_PARAMETERS_NONE();
153
154 BREAKITER_METHOD_FETCH_OBJECT;
155
156 int32_t num_rules = fetch_rbbi(bio)->getRuleStatusVec(NULL, 0,
157 BREAKITER_ERROR_CODE(bio));
158
159 ZEND_ASSERT(BREAKITER_ERROR_CODE(bio) == U_BUFFER_OVERFLOW_ERROR);
160 BREAKITER_ERROR_CODE(bio) = U_ZERO_ERROR;
161
162 std::unique_ptr<int32_t[]> rules = std::unique_ptr<int32_t[]>(new int32_t[num_rules]);
163 num_rules = fetch_rbbi(bio)->getRuleStatusVec(rules.get(), num_rules,
164 BREAKITER_ERROR_CODE(bio));
165 if (U_FAILURE(BREAKITER_ERROR_CODE(bio))) {
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 }
177
PHP_METHOD(IntlRuleBasedBreakIterator,getBinaryRules)178 U_CFUNC PHP_METHOD(IntlRuleBasedBreakIterator, getBinaryRules)
179 {
180 BREAKITER_METHOD_INIT_VARS;
181 object = ZEND_THIS;
182
183 ZEND_PARSE_PARAMETERS_NONE();
184
185 BREAKITER_METHOD_FETCH_OBJECT;
186
187 uint32_t rules_len;
188 const uint8_t *rules = fetch_rbbi(bio)->getBinaryRules(rules_len);
189
190 if (rules_len > INT_MAX - 1) {
191 intl_errors_set(BREAKITER_ERROR_P(bio), BREAKITER_ERROR_CODE(bio),
192 "rbbi_get_binary_rules: the rules are too large",
193 0);
194 RETURN_FALSE;
195 }
196
197 zend_string *ret_rules = zend_string_alloc(rules_len, 0);
198 memcpy(ZSTR_VAL(ret_rules), rules, rules_len);
199 ZSTR_VAL(ret_rules)[rules_len] = '\0';
200
201 RETURN_STR(ret_rules);
202 }
203
