/src/php-src/ext/uri/php_lexbor.c
Line | Count | Source (jump to first uncovered line) |
1 | | /* |
2 | | +----------------------------------------------------------------------+ |
3 | | | Copyright (c) The PHP Group | |
4 | | +----------------------------------------------------------------------+ |
5 | | | This source file is subject to version 3.01 of the PHP license, | |
6 | | | that is bundled with this package in the file LICENSE, and is | |
7 | | | available through the world-wide-web at the following url: | |
8 | | | https://www.php.net/license/3_01.txt | |
9 | | | If you did not receive a copy of the PHP license and are unable to | |
10 | | | obtain it through the world-wide-web, please send a note to | |
11 | | | license@php.net so we can mail you a copy immediately. | |
12 | | +----------------------------------------------------------------------+ |
13 | | | Authors: Máté Kocsis <kocsismate@php.net> | |
14 | | +----------------------------------------------------------------------+ |
15 | | */ |
16 | | |
17 | | #include "php.h" |
18 | | #include "php_lexbor.h" |
19 | | #include "php_uri_common.h" |
20 | | #include "Zend/zend_enum.h" |
21 | | #include "Zend/zend_smart_str.h" |
22 | | #include "Zend/zend_exceptions.h" |
23 | | #ifdef HAVE_ARPA_INET_H |
24 | | #include <arpa/inet.h> |
25 | | #endif |
26 | | |
27 | | ZEND_TLS lxb_url_parser_t lexbor_parser; |
28 | | ZEND_TLS unsigned short int lexbor_urls; |
29 | | |
30 | 0 | #define LEXBOR_MAX_URL_COUNT 500 |
31 | 268k | #define LEXBOR_MRAW_BYTE_SIZE 8192 |
32 | | |
33 | | static zend_always_inline void zval_string_or_null_to_lexbor_str(zval *value, lexbor_str_t *lexbor_str) |
34 | 0 | { |
35 | 0 | if (Z_TYPE_P(value) == IS_STRING && Z_STRLEN_P(value) > 0) { |
36 | 0 | lexbor_str->data = (lxb_char_t *) Z_STRVAL_P(value); |
37 | 0 | lexbor_str->length = Z_STRLEN_P(value); |
38 | 0 | } else { |
39 | 0 | ZEND_ASSERT(Z_ISNULL_P(value) || (Z_TYPE_P(value) == IS_STRING && Z_STRLEN_P(value) == 0)); |
40 | 0 | lexbor_str->data = (lxb_char_t *) ""; |
41 | 0 | lexbor_str->length = 0; |
42 | 0 | } |
43 | 0 | } |
44 | | |
45 | | static zend_always_inline void zval_long_or_null_to_lexbor_str(zval *value, lexbor_str_t *lexbor_str) |
46 | 0 | { |
47 | 0 | if (Z_TYPE_P(value) == IS_LONG) { |
48 | 0 | ZVAL_STR(value, zend_long_to_str(Z_LVAL_P(value))); |
49 | 0 | lexbor_str_init_append(lexbor_str, lexbor_parser.mraw, (const lxb_char_t *) Z_STRVAL_P(value), Z_STRLEN_P(value)); |
50 | 0 | zval_ptr_dtor_str(value); |
51 | 0 | } else { |
52 | 0 | ZEND_ASSERT(Z_ISNULL_P(value)); |
53 | 0 | lexbor_str->data = (lxb_char_t *) ""; |
54 | 0 | lexbor_str->length = 0; |
55 | 0 | } |
56 | 0 | } |
57 | | |
58 | | static void lexbor_cleanup_parser(void) |
59 | 0 | { |
60 | 0 | if (++lexbor_urls % LEXBOR_MAX_URL_COUNT == 0) { |
61 | 0 | lexbor_mraw_clean(lexbor_parser.mraw); |
62 | 0 | lexbor_urls = 0; |
63 | 0 | } |
64 | |
|
65 | 0 | lxb_url_parser_clean(&lexbor_parser); |
66 | 0 | } |
67 | | |
68 | | /** |
69 | | * Creates a Uri\WhatWg\UrlValidationError class by mapping error codes listed in |
70 | | * https://url.spec.whatwg.org/#writing to a Uri\WhatWg\UrlValidationErrorType enum. |
71 | | * The result is passed by reference to the errors parameter. |
72 | | * |
73 | | * When errors is NULL, the caller is not interested in the additional error information, |
74 | | * so the function does nothing. |
75 | | */ |
76 | | static const char *fill_errors(zval *errors) |
77 | 0 | { |
78 | 0 | if (errors == NULL) { |
79 | 0 | return NULL; |
80 | 0 | } |
81 | | |
82 | 0 | ZEND_ASSERT(Z_ISUNDEF_P(errors)); |
83 | | |
84 | 0 | array_init(errors); |
85 | |
|
86 | 0 | if (lexbor_parser.log == NULL) { |
87 | 0 | return NULL; |
88 | 0 | } |
89 | | |
90 | 0 | const char *result = NULL; |
91 | 0 | lexbor_plog_entry_t *lxb_error; |
92 | 0 | while ((lxb_error = lexbor_array_obj_pop(&lexbor_parser.log->list)) != NULL) { |
93 | 0 | zval error; |
94 | 0 | object_init_ex(&error, uri_whatwg_url_validation_error_ce); |
95 | 0 | zend_update_property_string(uri_whatwg_url_validation_error_ce, Z_OBJ(error), ZEND_STRL("context"), (const char *) lxb_error->data); |
96 | |
|
97 | 0 | const char *error_str; |
98 | 0 | zval failure; |
99 | 0 | switch (lxb_error->id) { |
100 | 0 | case LXB_URL_ERROR_TYPE_DOMAIN_TO_ASCII: |
101 | 0 | error_str = "DomainToAscii"; |
102 | 0 | ZVAL_TRUE(&failure); |
103 | 0 | break; |
104 | 0 | case LXB_URL_ERROR_TYPE_DOMAIN_TO_UNICODE: |
105 | 0 | error_str = "DomainToUnicode"; |
106 | 0 | ZVAL_FALSE(&failure); |
107 | 0 | break; |
108 | 0 | case LXB_URL_ERROR_TYPE_DOMAIN_INVALID_CODE_POINT: |
109 | 0 | error_str = "DomainInvalidCodePoint"; |
110 | 0 | ZVAL_TRUE(&failure); |
111 | 0 | break; |
112 | 0 | case LXB_URL_ERROR_TYPE_HOST_INVALID_CODE_POINT: |
113 | 0 | error_str = "HostInvalidCodePoint"; |
114 | 0 | ZVAL_TRUE(&failure); |
115 | 0 | break; |
116 | 0 | case LXB_URL_ERROR_TYPE_IPV4_EMPTY_PART: |
117 | 0 | error_str = "Ipv4EmptyPart"; |
118 | 0 | ZVAL_FALSE(&failure); |
119 | 0 | break; |
120 | 0 | case LXB_URL_ERROR_TYPE_IPV4_TOO_MANY_PARTS: |
121 | 0 | error_str = "Ipv4TooManyParts"; |
122 | 0 | ZVAL_TRUE(&failure); |
123 | 0 | break; |
124 | 0 | case LXB_URL_ERROR_TYPE_IPV4_NON_NUMERIC_PART: |
125 | 0 | error_str = "Ipv4NonNumericPart"; |
126 | 0 | ZVAL_TRUE(&failure); |
127 | 0 | break; |
128 | 0 | case LXB_URL_ERROR_TYPE_IPV4_NON_DECIMAL_PART: |
129 | 0 | error_str = "Ipv4NonDecimalPart"; |
130 | 0 | ZVAL_FALSE(&failure); |
131 | 0 | break; |
132 | 0 | case LXB_URL_ERROR_TYPE_IPV4_OUT_OF_RANGE_PART: |
133 | 0 | error_str = "Ipv4OutOfRangePart"; |
134 | 0 | ZVAL_TRUE(&failure); |
135 | 0 | break; |
136 | 0 | case LXB_URL_ERROR_TYPE_IPV6_UNCLOSED: |
137 | 0 | error_str = "Ipv6Unclosed"; |
138 | 0 | ZVAL_TRUE(&failure); |
139 | 0 | break; |
140 | 0 | case LXB_URL_ERROR_TYPE_IPV6_INVALID_COMPRESSION: |
141 | 0 | error_str = "Ipv6InvalidCompression"; |
142 | 0 | ZVAL_TRUE(&failure); |
143 | 0 | break; |
144 | 0 | case LXB_URL_ERROR_TYPE_IPV6_TOO_MANY_PIECES: |
145 | 0 | error_str = "Ipv6TooManyPieces"; |
146 | 0 | ZVAL_TRUE(&failure); |
147 | 0 | break; |
148 | 0 | case LXB_URL_ERROR_TYPE_IPV6_MULTIPLE_COMPRESSION: |
149 | 0 | error_str = "Ipv6MultipleCompression"; |
150 | 0 | ZVAL_TRUE(&failure); |
151 | 0 | break; |
152 | 0 | case LXB_URL_ERROR_TYPE_IPV6_INVALID_CODE_POINT: |
153 | 0 | error_str = "Ipv6InvalidCodePoint"; |
154 | 0 | ZVAL_TRUE(&failure); |
155 | 0 | break; |
156 | 0 | case LXB_URL_ERROR_TYPE_IPV6_TOO_FEW_PIECES: |
157 | 0 | error_str = "Ipv6TooFewPieces"; |
158 | 0 | ZVAL_TRUE(&failure); |
159 | 0 | break; |
160 | 0 | case LXB_URL_ERROR_TYPE_IPV4_IN_IPV6_TOO_MANY_PIECES: |
161 | 0 | error_str = "Ipv4InIpv6TooManyPieces"; |
162 | 0 | ZVAL_TRUE(&failure); |
163 | 0 | break; |
164 | 0 | case LXB_URL_ERROR_TYPE_IPV4_IN_IPV6_INVALID_CODE_POINT: |
165 | 0 | error_str = "Ipv4InIpv6InvalidCodePoint"; |
166 | 0 | ZVAL_TRUE(&failure); |
167 | 0 | break; |
168 | 0 | case LXB_URL_ERROR_TYPE_IPV4_IN_IPV6_OUT_OF_RANGE_PART: |
169 | 0 | error_str = "Ipv4InIpv6OutOfRangePart"; |
170 | 0 | ZVAL_TRUE(&failure); |
171 | 0 | break; |
172 | 0 | case LXB_URL_ERROR_TYPE_IPV4_IN_IPV6_TOO_FEW_PARTS: |
173 | 0 | error_str = "Ipv4InIpv6TooFewParts"; |
174 | 0 | ZVAL_TRUE(&failure); |
175 | 0 | break; |
176 | 0 | case LXB_URL_ERROR_TYPE_INVALID_URL_UNIT: |
177 | 0 | error_str = "InvalidUrlUnit"; |
178 | 0 | ZVAL_FALSE(&failure); |
179 | 0 | break; |
180 | 0 | case LXB_URL_ERROR_TYPE_SPECIAL_SCHEME_MISSING_FOLLOWING_SOLIDUS: |
181 | 0 | error_str = "SpecialSchemeMissingFollowingSolidus"; |
182 | 0 | ZVAL_FALSE(&failure); |
183 | 0 | break; |
184 | 0 | case LXB_URL_ERROR_TYPE_MISSING_SCHEME_NON_RELATIVE_URL: |
185 | 0 | error_str = "MissingSchemeNonRelativeUrl"; |
186 | 0 | ZVAL_TRUE(&failure); |
187 | 0 | break; |
188 | 0 | case LXB_URL_ERROR_TYPE_INVALID_REVERSE_SOLIDUS: |
189 | 0 | error_str = "InvalidReverseSoldius"; |
190 | 0 | ZVAL_FALSE(&failure); |
191 | 0 | break; |
192 | 0 | case LXB_URL_ERROR_TYPE_INVALID_CREDENTIALS: |
193 | 0 | error_str = "InvalidCredentials"; |
194 | 0 | ZVAL_FALSE(&failure); |
195 | 0 | break; |
196 | 0 | case LXB_URL_ERROR_TYPE_HOST_MISSING: |
197 | 0 | error_str = "HostMissing"; |
198 | 0 | ZVAL_TRUE(&failure); |
199 | 0 | break; |
200 | 0 | case LXB_URL_ERROR_TYPE_PORT_OUT_OF_RANGE: |
201 | 0 | error_str = "PortOutOfRange"; |
202 | 0 | ZVAL_TRUE(&failure); |
203 | 0 | break; |
204 | 0 | case LXB_URL_ERROR_TYPE_PORT_INVALID: |
205 | 0 | error_str = "PortInvalid"; |
206 | 0 | ZVAL_TRUE(&failure); |
207 | 0 | break; |
208 | 0 | case LXB_URL_ERROR_TYPE_FILE_INVALID_WINDOWS_DRIVE_LETTER: |
209 | 0 | error_str = "FileInvalidWindowsDriveLetter"; |
210 | 0 | ZVAL_FALSE(&failure); |
211 | 0 | break; |
212 | 0 | case LXB_URL_ERROR_TYPE_FILE_INVALID_WINDOWS_DRIVE_LETTER_HOST: |
213 | 0 | error_str = "FileInvalidWindowsDriveLetterHost"; |
214 | 0 | ZVAL_FALSE(&failure); |
215 | 0 | break; |
216 | 0 | EMPTY_SWITCH_DEFAULT_CASE() |
217 | 0 | } |
218 | | |
219 | 0 | zval error_type; |
220 | 0 | ZVAL_OBJ(&error_type, zend_enum_get_case_cstr(uri_whatwg_url_validation_error_type_ce, error_str)); |
221 | 0 | zend_update_property_ex(uri_whatwg_url_validation_error_ce, Z_OBJ(error), ZSTR_KNOWN(ZEND_STR_TYPE), &error_type); |
222 | |
|
223 | 0 | zend_update_property(uri_whatwg_url_validation_error_ce, Z_OBJ(error), ZEND_STRL("failure"), &failure); |
224 | |
|
225 | 0 | if (Z_TYPE(failure) == IS_TRUE) { |
226 | 0 | result = error_str; |
227 | 0 | } |
228 | |
|
229 | 0 | add_next_index_zval(errors, &error); |
230 | 0 | } |
231 | | |
232 | 0 | return result; |
233 | 0 | } |
234 | | |
235 | | static void throw_invalid_url_exception_during_write(zval *errors, const char *component) |
236 | 0 | { |
237 | 0 | const char *reason = fill_errors(errors); |
238 | 0 | zend_object *exception = zend_throw_exception_ex( |
239 | 0 | uri_whatwg_invalid_url_exception_ce, |
240 | 0 | 0, |
241 | 0 | "The specified %s is malformed%s%s%s", |
242 | 0 | component, |
243 | 0 | reason ? " (" : "", |
244 | 0 | reason ? reason : "", |
245 | 0 | reason ? ")" : "" |
246 | 0 | ); |
247 | 0 | zend_update_property(exception->ce, exception, ZEND_STRL("errors"), errors); |
248 | 0 | } |
249 | | |
250 | | static lxb_status_t lexbor_serialize_callback(const lxb_char_t *data, size_t length, void *ctx) |
251 | 0 | { |
252 | 0 | smart_str *uri_str = ctx; |
253 | |
|
254 | 0 | if (data != NULL && length > 0) { |
255 | 0 | smart_str_appendl(uri_str, (const char *) data, length); |
256 | 0 | } |
257 | |
|
258 | 0 | return LXB_STATUS_OK; |
259 | 0 | } |
260 | | |
261 | | static zend_result lexbor_read_scheme(const struct uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *retval) |
262 | 0 | { |
263 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
264 | |
|
265 | 0 | ZEND_ASSERT(lexbor_uri->scheme.type != LXB_URL_SCHEMEL_TYPE__UNDEF); |
266 | | |
267 | 0 | ZVAL_STRINGL(retval, (const char *) lexbor_uri->scheme.name.data, lexbor_uri->scheme.name.length); |
268 | |
|
269 | 0 | return SUCCESS; |
270 | 0 | } |
271 | | |
272 | | static zend_result lexbor_write_scheme(struct uri_internal_t *internal_uri, zval *value, zval *errors) |
273 | 0 | { |
274 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
275 | 0 | lexbor_str_t str = {0}; |
276 | |
|
277 | 0 | zval_string_or_null_to_lexbor_str(value, &str); |
278 | |
|
279 | 0 | if (lxb_url_api_protocol_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) { |
280 | 0 | throw_invalid_url_exception_during_write(errors, "scheme"); |
281 | |
|
282 | 0 | return FAILURE; |
283 | 0 | } |
284 | | |
285 | 0 | return SUCCESS; |
286 | 0 | } |
287 | | |
288 | | static zend_result lexbor_read_username(const struct uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *retval) |
289 | 0 | { |
290 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
291 | |
|
292 | 0 | if (lexbor_uri->username.length) { |
293 | 0 | ZVAL_STRINGL(retval, (const char *) lexbor_uri->username.data, lexbor_uri->username.length); |
294 | 0 | } else { |
295 | 0 | ZVAL_NULL(retval); |
296 | 0 | } |
297 | |
|
298 | 0 | return SUCCESS; |
299 | 0 | } |
300 | | |
301 | | static zend_result lexbor_write_username(uri_internal_t *internal_uri, zval *value, zval *errors) |
302 | 0 | { |
303 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
304 | 0 | lexbor_str_t str = {0}; |
305 | |
|
306 | 0 | zval_string_or_null_to_lexbor_str(value, &str); |
307 | |
|
308 | 0 | if (lxb_url_api_username_set(lexbor_uri, str.data, str.length) != LXB_STATUS_OK) { |
309 | 0 | throw_invalid_url_exception_during_write(errors, "username"); |
310 | |
|
311 | 0 | return FAILURE; |
312 | 0 | } |
313 | | |
314 | 0 | return SUCCESS; |
315 | 0 | } |
316 | | |
317 | | static zend_result lexbor_read_password(const struct uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *retval) |
318 | 0 | { |
319 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
320 | |
|
321 | 0 | if (lexbor_uri->password.length > 0) { |
322 | 0 | ZVAL_STRINGL(retval, (const char *) lexbor_uri->password.data, lexbor_uri->password.length); |
323 | 0 | } else { |
324 | 0 | ZVAL_NULL(retval); |
325 | 0 | } |
326 | |
|
327 | 0 | return SUCCESS; |
328 | 0 | } |
329 | | |
330 | | static zend_result lexbor_write_password(struct uri_internal_t *internal_uri, zval *value, zval *errors) |
331 | 0 | { |
332 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
333 | 0 | lexbor_str_t str = {0}; |
334 | |
|
335 | 0 | zval_string_or_null_to_lexbor_str(value, &str); |
336 | |
|
337 | 0 | if (lxb_url_api_password_set(lexbor_uri, str.data, str.length) != LXB_STATUS_OK) { |
338 | 0 | throw_invalid_url_exception_during_write(errors, "password"); |
339 | |
|
340 | 0 | return FAILURE; |
341 | 0 | } |
342 | | |
343 | 0 | return SUCCESS; |
344 | 0 | } |
345 | | |
346 | | static zend_result init_idna(void) |
347 | 0 | { |
348 | 0 | if (lexbor_parser.idna != NULL) { |
349 | 0 | return SUCCESS; |
350 | 0 | } |
351 | | |
352 | 0 | lexbor_parser.idna = lxb_unicode_idna_create(); |
353 | |
|
354 | 0 | return lxb_unicode_idna_init(lexbor_parser.idna) == LXB_STATUS_OK ? SUCCESS : FAILURE; |
355 | 0 | } |
356 | | |
357 | | static zend_result lexbor_read_host(const struct uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *retval) |
358 | 0 | { |
359 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
360 | |
|
361 | 0 | if (lexbor_uri->host.type == LXB_URL_HOST_TYPE_IPV4) { |
362 | 0 | smart_str host_str = {0}; |
363 | |
|
364 | 0 | lxb_url_serialize_host_ipv4(lexbor_uri->host.u.ipv4, lexbor_serialize_callback, &host_str); |
365 | |
|
366 | 0 | ZVAL_NEW_STR(retval, smart_str_extract(&host_str)); |
367 | 0 | } else if (lexbor_uri->host.type == LXB_URL_HOST_TYPE_IPV6) { |
368 | 0 | smart_str host_str = {0}; |
369 | |
|
370 | 0 | smart_str_appendc(&host_str, '['); |
371 | 0 | lxb_url_serialize_host_ipv6(lexbor_uri->host.u.ipv6, lexbor_serialize_callback, &host_str); |
372 | 0 | smart_str_appendc(&host_str, ']'); |
373 | |
|
374 | 0 | ZVAL_NEW_STR(retval, smart_str_extract(&host_str)); |
375 | 0 | } else if (lexbor_uri->host.type == LXB_URL_HOST_TYPE_EMPTY) { |
376 | 0 | ZVAL_EMPTY_STRING(retval); |
377 | 0 | } else if (lexbor_uri->host.type != LXB_URL_HOST_TYPE__UNDEF) { |
378 | 0 | switch (read_mode) { |
379 | 0 | case URI_COMPONENT_READ_NORMALIZED_UNICODE: { |
380 | 0 | smart_str host_str = {0}; |
381 | 0 | if (init_idna() == FAILURE) { |
382 | 0 | return FAILURE; |
383 | 0 | } |
384 | 0 | lxb_url_serialize_host_unicode(lexbor_parser.idna, &lexbor_uri->host, lexbor_serialize_callback, &host_str); |
385 | 0 | lxb_unicode_idna_clean(lexbor_parser.idna); |
386 | |
|
387 | 0 | ZVAL_NEW_STR(retval, smart_str_extract(&host_str)); |
388 | 0 | break; |
389 | 0 | } |
390 | 0 | case URI_COMPONENT_READ_NORMALIZED_ASCII: |
391 | 0 | ZEND_FALLTHROUGH; |
392 | 0 | case URI_COMPONENT_READ_RAW: |
393 | 0 | ZVAL_STRINGL(retval, (const char *) lexbor_uri->host.u.domain.data, lexbor_uri->host.u.domain.length); |
394 | 0 | break; |
395 | 0 | EMPTY_SWITCH_DEFAULT_CASE() |
396 | 0 | } |
397 | 0 | } else { |
398 | 0 | ZVAL_NULL(retval); |
399 | 0 | } |
400 | | |
401 | 0 | return SUCCESS; |
402 | 0 | } |
403 | | |
404 | | static zend_result lexbor_write_host(struct uri_internal_t *internal_uri, zval *value, zval *errors) |
405 | 0 | { |
406 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
407 | 0 | lexbor_str_t str = {0}; |
408 | |
|
409 | 0 | zval_string_or_null_to_lexbor_str(value, &str); |
410 | |
|
411 | 0 | if (lxb_url_api_hostname_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) { |
412 | 0 | throw_invalid_url_exception_during_write(errors, "host"); |
413 | |
|
414 | 0 | return FAILURE; |
415 | 0 | } |
416 | | |
417 | 0 | return SUCCESS; |
418 | 0 | } |
419 | | |
420 | | static zend_result lexbor_read_port(const struct uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *retval) |
421 | 0 | { |
422 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
423 | |
|
424 | 0 | if (lexbor_uri->has_port) { |
425 | 0 | ZVAL_LONG(retval, lexbor_uri->port); |
426 | 0 | } else { |
427 | 0 | ZVAL_NULL(retval); |
428 | 0 | } |
429 | |
|
430 | 0 | return SUCCESS; |
431 | 0 | } |
432 | | |
433 | | static zend_result lexbor_write_port(struct uri_internal_t *internal_uri, zval *value, zval *errors) |
434 | 0 | { |
435 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
436 | 0 | lexbor_str_t str = {0}; |
437 | |
|
438 | 0 | zval_long_or_null_to_lexbor_str(value, &str); |
439 | |
|
440 | 0 | if (lxb_url_api_port_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) { |
441 | 0 | throw_invalid_url_exception_during_write(errors, "port"); |
442 | |
|
443 | 0 | return FAILURE; |
444 | 0 | } |
445 | | |
446 | 0 | return SUCCESS; |
447 | 0 | } |
448 | | |
449 | | static zend_result lexbor_read_path(const struct uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *retval) |
450 | 0 | { |
451 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
452 | |
|
453 | 0 | if (lexbor_uri->path.str.length) { |
454 | 0 | ZVAL_STRINGL(retval, (const char *) lexbor_uri->path.str.data, lexbor_uri->path.str.length); |
455 | 0 | } else { |
456 | 0 | ZVAL_EMPTY_STRING(retval); |
457 | 0 | } |
458 | |
|
459 | 0 | return SUCCESS; |
460 | 0 | } |
461 | | |
462 | | static zend_result lexbor_write_path(struct uri_internal_t *internal_uri, zval *value, zval *errors) |
463 | 0 | { |
464 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
465 | 0 | lexbor_str_t str = {0}; |
466 | |
|
467 | 0 | zval_string_or_null_to_lexbor_str(value, &str); |
468 | |
|
469 | 0 | if (lxb_url_api_pathname_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) { |
470 | 0 | throw_invalid_url_exception_during_write(errors, "path"); |
471 | |
|
472 | 0 | return FAILURE; |
473 | 0 | } |
474 | | |
475 | 0 | return SUCCESS; |
476 | 0 | } |
477 | | |
478 | | static zend_result lexbor_read_query(const struct uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *retval) |
479 | 0 | { |
480 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
481 | |
|
482 | 0 | if (lexbor_uri->query.length) { |
483 | 0 | ZVAL_STRINGL(retval, (const char *) lexbor_uri->query.data, lexbor_uri->query.length); |
484 | 0 | } else { |
485 | 0 | ZVAL_NULL(retval); |
486 | 0 | } |
487 | |
|
488 | 0 | return SUCCESS; |
489 | 0 | } |
490 | | |
491 | | static zend_result lexbor_write_query(struct uri_internal_t *internal_uri, zval *value, zval *errors) |
492 | 0 | { |
493 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
494 | 0 | lexbor_str_t str = {0}; |
495 | |
|
496 | 0 | zval_string_or_null_to_lexbor_str(value, &str); |
497 | |
|
498 | 0 | if (lxb_url_api_search_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) { |
499 | 0 | throw_invalid_url_exception_during_write(errors, "query string"); |
500 | |
|
501 | 0 | return FAILURE; |
502 | 0 | } |
503 | | |
504 | 0 | return SUCCESS; |
505 | 0 | } |
506 | | |
507 | | static zend_result lexbor_read_fragment(const struct uri_internal_t *internal_uri, uri_component_read_mode_t read_mode, zval *retval) |
508 | 0 | { |
509 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
510 | |
|
511 | 0 | if (lexbor_uri->fragment.length) { |
512 | 0 | ZVAL_STRINGL(retval, (const char *) lexbor_uri->fragment.data, lexbor_uri->fragment.length); |
513 | 0 | } else { |
514 | 0 | ZVAL_NULL(retval); |
515 | 0 | } |
516 | |
|
517 | 0 | return SUCCESS; |
518 | 0 | } |
519 | | |
520 | | static zend_result lexbor_write_fragment(struct uri_internal_t *internal_uri, zval *value, zval *errors) |
521 | 0 | { |
522 | 0 | lxb_url_t *lexbor_uri = internal_uri->uri; |
523 | 0 | lexbor_str_t str = {0}; |
524 | |
|
525 | 0 | zval_string_or_null_to_lexbor_str(value, &str); |
526 | |
|
527 | 0 | if (lxb_url_api_hash_set(lexbor_uri, &lexbor_parser, str.data, str.length) != LXB_STATUS_OK) { |
528 | 0 | throw_invalid_url_exception_during_write(errors, "fragment"); |
529 | |
|
530 | 0 | return FAILURE; |
531 | 0 | } |
532 | | |
533 | 0 | return SUCCESS; |
534 | 0 | } |
535 | | |
536 | | zend_result lexbor_request_init(void) |
537 | 268k | { |
538 | 268k | lexbor_mraw_t *mraw = lexbor_mraw_create(); |
539 | 268k | lxb_status_t status = lexbor_mraw_init(mraw, LEXBOR_MRAW_BYTE_SIZE); |
540 | 268k | if (status != LXB_STATUS_OK) { |
541 | 0 | lexbor_mraw_destroy(mraw, true); |
542 | 0 | return FAILURE; |
543 | 0 | } |
544 | | |
545 | 268k | status = lxb_url_parser_init(&lexbor_parser, mraw); |
546 | 268k | if (status != LXB_STATUS_OK) { |
547 | 0 | lxb_url_parser_destroy(&lexbor_parser, false); |
548 | 0 | lexbor_mraw_destroy(mraw, true); |
549 | 0 | return FAILURE; |
550 | 0 | } |
551 | | |
552 | 268k | lexbor_urls = 0; |
553 | | |
554 | 268k | return SUCCESS; |
555 | 268k | } |
556 | | |
557 | | void lexbor_request_shutdown(void) |
558 | 268k | { |
559 | 268k | lxb_url_parser_memory_destroy(&lexbor_parser); |
560 | 268k | lxb_url_parser_destroy(&lexbor_parser, false); |
561 | | |
562 | 268k | lexbor_urls = 0; |
563 | 268k | } |
564 | | |
565 | | lxb_url_t *lexbor_parse_uri_ex(const zend_string *uri_str, const lxb_url_t *lexbor_base_url, zval *errors, bool silent) |
566 | 0 | { |
567 | 0 | lexbor_cleanup_parser(); |
568 | |
|
569 | 0 | lxb_url_t *url = lxb_url_parse(&lexbor_parser, lexbor_base_url, (unsigned char *) ZSTR_VAL(uri_str), ZSTR_LEN(uri_str)); |
570 | 0 | const char *reason = fill_errors(errors); |
571 | |
|
572 | 0 | if (url == NULL && !silent) { |
573 | 0 | zend_object *exception = zend_throw_exception_ex(uri_whatwg_invalid_url_exception_ce, 0, "The specified URI is malformed%s%s%s", reason ? " (" : "", reason ? reason : "", reason ? ")" : ""); |
574 | 0 | zend_update_property(exception->ce, exception, ZEND_STRL("errors"), errors); |
575 | 0 | } |
576 | |
|
577 | 0 | return url; |
578 | 0 | } |
579 | | |
580 | | static void *lexbor_parse_uri(const zend_string *uri_str, const void *base_url, zval *errors, bool silent) |
581 | 0 | { |
582 | 0 | return lexbor_parse_uri_ex(uri_str, base_url, errors, silent); |
583 | 0 | } |
584 | | |
585 | | static void *lexbor_clone_uri(void *uri) |
586 | 0 | { |
587 | 0 | lxb_url_t *lexbor_uri = (lxb_url_t *) uri; |
588 | |
|
589 | 0 | return lxb_url_clone(lexbor_parser.mraw, lexbor_uri); |
590 | 0 | } |
591 | | |
592 | | static zend_string *lexbor_uri_to_string(void *uri, uri_recomposition_mode_t recomposition_mode, bool exclude_fragment) |
593 | 0 | { |
594 | 0 | lxb_url_t *lexbor_uri = (lxb_url_t *) uri; |
595 | 0 | smart_str uri_str = {0}; |
596 | |
|
597 | 0 | switch (recomposition_mode) { |
598 | 0 | case URI_RECOMPOSITION_RAW_UNICODE: |
599 | 0 | ZEND_FALLTHROUGH; |
600 | 0 | case URI_RECOMPOSITION_NORMALIZED_UNICODE: |
601 | 0 | if (init_idna() == FAILURE) { |
602 | 0 | return NULL; |
603 | 0 | } |
604 | 0 | lxb_url_serialize_idna(lexbor_parser.idna, lexbor_uri, lexbor_serialize_callback, &uri_str, exclude_fragment); |
605 | 0 | lxb_unicode_idna_clean(lexbor_parser.idna); |
606 | 0 | break; |
607 | 0 | case URI_RECOMPOSITION_RAW_ASCII: |
608 | 0 | ZEND_FALLTHROUGH; |
609 | 0 | case URI_RECOMPOSITION_NORMALIZED_ASCII: |
610 | 0 | lxb_url_serialize(lexbor_uri, lexbor_serialize_callback, &uri_str, exclude_fragment); |
611 | 0 | break; |
612 | 0 | EMPTY_SWITCH_DEFAULT_CASE() |
613 | 0 | } |
614 | | |
615 | 0 | return smart_str_extract(&uri_str); |
616 | 0 | } |
617 | | |
618 | | static void lexbor_free_uri(void *uri) |
619 | 0 | { |
620 | 0 | } |
621 | | |
622 | | const uri_handler_t lexbor_uri_handler = { |
623 | | .name = URI_PARSER_WHATWG, |
624 | | .parse_uri = lexbor_parse_uri, |
625 | | .clone_uri = lexbor_clone_uri, |
626 | | .uri_to_string = lexbor_uri_to_string, |
627 | | .free_uri = lexbor_free_uri, |
628 | | { |
629 | | .scheme = {.read_func = lexbor_read_scheme, .write_func = lexbor_write_scheme}, |
630 | | .username = {.read_func = lexbor_read_username, .write_func = lexbor_write_username}, |
631 | | .password = {.read_func = lexbor_read_password, .write_func = lexbor_write_password}, |
632 | | .host = {.read_func = lexbor_read_host, .write_func = lexbor_write_host}, |
633 | | .port = {.read_func = lexbor_read_port, .write_func = lexbor_write_port}, |
634 | | .path = {.read_func = lexbor_read_path, .write_func = lexbor_write_path}, |
635 | | .query = {.read_func = lexbor_read_query, .write_func = lexbor_write_query}, |
636 | | .fragment = {.read_func = lexbor_read_fragment, .write_func = lexbor_write_fragment}, |
637 | | } |
638 | | }; |