1 /*
2 +--------------------------------------------------------------------+
3 | PECL :: http |
4 +--------------------------------------------------------------------+
5 | Redistribution and use in source and binary forms, with or without |
6 | modification, are permitted provided that the conditions mentioned |
7 | in the accompanying LICENSE file are met. |
8 +--------------------------------------------------------------------+
9 | Copyright (c) 2004-2014, Michael Wallner <mike@php.net> |
10 +--------------------------------------------------------------------+
11 */
12
13 #include "php_http_api.h"
14
15 #ifndef DBG_PARSER
16 # define DBG_PARSER 0
17 #endif
18
19 typedef struct php_http_header_parser_state_spec {
20 php_http_header_parser_state_t state;
21 unsigned need_data:1;
22 } php_http_header_parser_state_spec_t;
23
24 static const php_http_header_parser_state_spec_t php_http_header_parser_states[] = {
25 {PHP_HTTP_HEADER_PARSER_STATE_START, 1},
26 {PHP_HTTP_HEADER_PARSER_STATE_KEY, 1},
27 {PHP_HTTP_HEADER_PARSER_STATE_VALUE, 1},
28 {PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX, 0},
29 {PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE, 0},
30 {PHP_HTTP_HEADER_PARSER_STATE_DONE, 0}
31 };
32
php_http_header_parser_init(php_http_header_parser_t * parser)33 php_http_header_parser_t *php_http_header_parser_init(php_http_header_parser_t *parser)
34 {
35 if (!parser) {
36 parser = emalloc(sizeof(*parser));
37 }
38 memset(parser, 0, sizeof(*parser));
39
40 return parser;
41 }
42
43 #define php_http_header_parser_state_push(parser, state) zend_ptr_stack_push(&(parser)->stack, (void *) (state)), (state)
44 #define php_http_header_parser_state_ex(parser) ((parser)->stack.top \
45 ? (php_http_header_parser_state_t) (parser)->stack.elements[(parser)->stack.top - 1] \
46 : PHP_HTTP_HEADER_PARSER_STATE_START)
47
php_http_header_parser_state_is(php_http_header_parser_t * parser)48 php_http_header_parser_state_t php_http_header_parser_state_is(php_http_header_parser_t *parser)
49 {
50 return php_http_header_parser_state_ex(parser);
51 }
52
53 #define php_http_header_parser_state_pop(parser) ((parser)->stack.top \
54 ? (php_http_header_parser_state_t) zend_ptr_stack_pop(&(parser)->stack) \
55 : PHP_HTTP_HEADER_PARSER_STATE_START)
56
php_http_header_parser_dtor(php_http_header_parser_t * parser)57 void php_http_header_parser_dtor(php_http_header_parser_t *parser)
58 {
59 zend_ptr_stack_destroy(&parser->stack);
60 php_http_info_dtor(&parser->info);
61 PTR_FREE(parser->_key.str);
62 PTR_FREE(parser->_val.str);
63 }
64
php_http_header_parser_free(php_http_header_parser_t ** parser)65 void php_http_header_parser_free(php_http_header_parser_t **parser)
66 {
67 if (*parser) {
68 php_http_header_parser_dtor(*parser);
69 efree(*parser);
70 *parser = NULL;
71 }
72 }
73
74 /* NOTE: 'str' has to be null terminated */
php_http_header_parser_error(size_t valid_len,char * str,size_t len,const char * eol_str)75 static void php_http_header_parser_error(size_t valid_len, char *str, size_t len, const char *eol_str )
76 {
77 zend_string *escaped_str, *zstr_str = zend_string_init(str, len, 0);
78
79 #if PHP_VERSION_ID < 70300
80 escaped_str = php_addcslashes(zstr_str, 1, ZEND_STRL("\x0..\x1F\x7F..\xFF"));
81 #else
82 escaped_str = php_addcslashes(zstr_str, ZEND_STRL("\x0..\x1F\x7F..\xFF"));
83 zend_string_release_ex(zstr_str, 0);
84 #endif
85
86 if (valid_len != len && (!eol_str || (str+valid_len) != eol_str)) {
87 php_error_docref(NULL, E_WARNING, "Failed to parse headers: unexpected character '\\%03o' at pos %zu of '%s'", str[valid_len], valid_len, escaped_str->val);
88 } else if (eol_str) {
89 php_error_docref(NULL, E_WARNING, "Failed to parse headers: unexpected end of line at pos %zu of '%s'", eol_str - str, escaped_str->val);
90 } else {
91 php_error_docref(NULL, E_WARNING, "Failed to parse headers: unexpected end of input at pos %zu of '%s'", len, escaped_str->val);
92 }
93
94 efree(escaped_str);
95 }
96
php_http_header_parser_parse(php_http_header_parser_t * parser,php_http_buffer_t * buffer,unsigned flags,HashTable * headers,php_http_info_callback_t callback_func,void * callback_arg)97 php_http_header_parser_state_t php_http_header_parser_parse(php_http_header_parser_t *parser, php_http_buffer_t *buffer, unsigned flags, HashTable *headers, php_http_info_callback_t callback_func, void *callback_arg)
98 {
99 while (buffer->used || !php_http_header_parser_states[php_http_header_parser_state_ex(parser)].need_data) {
100 #if DBG_PARSER
101 const char *state[] = {"START", "KEY", "VALUE", "VALUE_EX", "HEADER_DONE", "DONE"};
102 fprintf(stderr, "#HP: %s (avail:%zu, num:%d cleanup:%u)\n", php_http_header_parser_state_is(parser) < 0 ? "FAILURE" : state[php_http_header_parser_state_is(parser)], buffer->used, headers?zend_hash_num_elements(headers):0, flags);
103 _dpf(0, buffer->data, buffer->used);
104 #endif
105 switch (php_http_header_parser_state_pop(parser)) {
106 case PHP_HTTP_HEADER_PARSER_STATE_FAILURE:
107 php_error_docref(NULL, E_WARNING, "Failed to parse headers");
108 return php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_FAILURE);
109
110 case PHP_HTTP_HEADER_PARSER_STATE_START: {
111 char *ptr = buffer->data;
112
113 while (ptr - buffer->data < buffer->used && PHP_HTTP_IS_CTYPE(space, *ptr)) {
114 ++ptr;
115 }
116
117 php_http_buffer_cut(buffer, 0, ptr - buffer->data);
118 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_KEY);
119 break;
120 }
121
122 case PHP_HTTP_HEADER_PARSER_STATE_KEY: {
123 const char *colon, *eol_str = NULL;
124 int eol_len = 0;
125
126 /* fix buffer here, so eol_str pointer doesn't become obsolete afterwards */
127 php_http_buffer_fix(buffer);
128
129 if (buffer->data == (eol_str = php_http_locate_bin_eol(buffer->data, buffer->used, &eol_len))) {
130 /* end of headers */
131 php_http_buffer_cut(buffer, 0, eol_len);
132 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_DONE);
133 } else if (php_http_info_parse(&parser->info, buffer->data)) {
134 /* new message starting with request/response line */
135 if (callback_func) {
136 callback_func(callback_arg, &headers, &parser->info);
137 }
138 php_http_info_dtor(&parser->info);
139 php_http_buffer_cut(buffer, 0, eol_str + eol_len - buffer->data);
140 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE);
141 } else if ((colon = memchr(buffer->data, ':', buffer->used)) && (!eol_str || eol_str > colon)) {
142 /* header: string */
143 size_t valid_len;
144
145 parser->_key.len = colon - buffer->data;
146 parser->_key.str = estrndup(buffer->data, parser->_key.len);
147
148 valid_len = strspn(parser->_key.str, PHP_HTTP_HEADER_NAME_CHARS);
149 if (valid_len != parser->_key.len) {
150 php_http_header_parser_error(valid_len, parser->_key.str, parser->_key.len, eol_str);
151 PTR_SET(parser->_key.str, NULL);
152 return php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_FAILURE);
153 }
154 while (PHP_HTTP_IS_CTYPE(space, *++colon) && *colon != '\n' && *colon != '\r');
155 php_http_buffer_cut(buffer, 0, colon - buffer->data);
156 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_VALUE);
157 } else if (eol_str || (flags & PHP_HTTP_HEADER_PARSER_CLEANUP)) {
158 /* neither reqeust/response line nor 'header:' string, or injected new line or NUL etc. */
159 php_http_header_parser_error(strspn(buffer->data, PHP_HTTP_HEADER_NAME_CHARS), buffer->data, buffer->used, eol_str);
160 return php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_FAILURE);
161 } else {
162 /* keep feeding */
163 return php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_KEY);
164 }
165 break;
166 }
167
168 case PHP_HTTP_HEADER_PARSER_STATE_VALUE: {
169 const char *eol_str;
170 int eol_len;
171
172 #define SET_ADD_VAL(slen, eol_len) \
173 do { \
174 const char *ptr = buffer->data; \
175 size_t len = slen; \
176 \
177 while (len > 0 && PHP_HTTP_IS_CTYPE(space, *ptr)) { \
178 ++ptr; \
179 --len; \
180 } \
181 while (len > 0 && PHP_HTTP_IS_CTYPE(space, ptr[len - 1])) { \
182 --len; \
183 } \
184 \
185 if (len > 0) { \
186 if (parser->_val.str) { \
187 parser->_val.str = erealloc(parser->_val.str, parser->_val.len + len + 2); \
188 parser->_val.str[parser->_val.len++] = ' '; \
189 memcpy(&parser->_val.str[parser->_val.len], ptr, len); \
190 parser->_val.len += len; \
191 parser->_val.str[parser->_val.len] = '\0'; \
192 } else { \
193 parser->_val.len = len; \
194 parser->_val.str = estrndup(ptr, len); \
195 } \
196 } \
197 php_http_buffer_cut(buffer, 0, slen + eol_len); \
198 } while (0)
199
200 if ((eol_str = php_http_locate_bin_eol(buffer->data, buffer->used, &eol_len))) {
201 SET_ADD_VAL(eol_str - buffer->data, eol_len);
202 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX);
203 } else if (flags & PHP_HTTP_HEADER_PARSER_CLEANUP) {
204 if (buffer->used) {
205 SET_ADD_VAL(buffer->used, 0);
206 }
207 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE);
208 } else {
209 return php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_VALUE);
210 }
211 break;
212 }
213
214 case PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX:
215 if (buffer->used && (*buffer->data == ' ' || *buffer->data == '\t')) {
216 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_VALUE);
217 } else if (buffer->used || (flags & PHP_HTTP_HEADER_PARSER_CLEANUP)) {
218 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE);
219 } else {
220 /* keep feeding */
221 return php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX);
222 }
223 break;
224
225 case PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE:
226 if (parser->_key.str && parser->_val.str) {
227 zval tmp, *exist;
228 size_t valid_len = strlen(parser->_val.str);
229
230 /* check for truncation */
231 if (valid_len != parser->_val.len) {
232 php_http_header_parser_error(valid_len, parser->_val.str, parser->_val.len, NULL);
233
234 PTR_SET(parser->_key.str, NULL);
235 PTR_SET(parser->_val.str, NULL);
236
237 return php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_FAILURE);
238 }
239
240 if (!headers && callback_func) {
241 callback_func(callback_arg, &headers, NULL);
242 }
243
244 php_http_pretty_key(parser->_key.str, parser->_key.len, 1, 1);
245 if ((exist = zend_symtable_str_find(headers, parser->_key.str, parser->_key.len))) {
246 convert_to_array(exist);
247 add_next_index_str(exist, php_http_cs2zs(parser->_val.str, parser->_val.len));
248 } else {
249 ZVAL_STR(&tmp, php_http_cs2zs(parser->_val.str, parser->_val.len));
250 zend_symtable_str_update(headers, parser->_key.str, parser->_key.len, &tmp);
251 }
252 parser->_val.str = NULL;
253 }
254
255 PTR_SET(parser->_key.str, NULL);
256 PTR_SET(parser->_val.str, NULL);
257
258 php_http_header_parser_state_push(parser, PHP_HTTP_HEADER_PARSER_STATE_KEY);
259 break;
260
261 case PHP_HTTP_HEADER_PARSER_STATE_DONE:
262 return PHP_HTTP_HEADER_PARSER_STATE_DONE;
263 }
264 }
265
266 return php_http_header_parser_state_is(parser);
267 }
268
php_http_header_parser_parse_stream(php_http_header_parser_t * parser,php_http_buffer_t * buf,php_stream * s,unsigned flags,HashTable * headers,php_http_info_callback_t callback_func,void * callback_arg)269 php_http_header_parser_state_t php_http_header_parser_parse_stream(php_http_header_parser_t *parser, php_http_buffer_t *buf, php_stream *s, unsigned flags, HashTable *headers, php_http_info_callback_t callback_func, void *callback_arg)
270 {
271 php_http_header_parser_state_t state = PHP_HTTP_HEADER_PARSER_STATE_START;
272
273 if (!buf->data) {
274 php_http_buffer_resize_ex(buf, 0x1000, 1, 0);
275 }
276 while (1) {
277 size_t justread = 0;
278 #if DBG_PARSER
279 const char *states[] = {"START", "KEY", "VALUE", "VALUE_EX", "HEADER_DONE", "DONE"};
280 fprintf(stderr, "#SHP: %s (f:%u)\n", states[state], flags);
281 #endif
282 /* resize if needed */
283 if (buf->free < 0x1000) {
284 php_http_buffer_resize_ex(buf, 0x1000, 1, 0);
285 }
286 switch (state) {
287 case PHP_HTTP_HEADER_PARSER_STATE_FAILURE:
288 case PHP_HTTP_HEADER_PARSER_STATE_DONE:
289 return state;
290
291 default:
292 /* read line */
293 php_stream_get_line(s, buf->data + buf->used, buf->free, &justread);
294 /* if we fail reading a whole line, try a single char */
295 if (!justread) {
296 int c = php_stream_getc(s);
297
298 if (c != EOF) {
299 char s[1] = {c};
300 justread = php_http_buffer_append(buf, s, 1);
301 }
302 }
303 php_http_buffer_account(buf, justread);
304 }
305
306 if (justread) {
307 state = php_http_header_parser_parse(parser, buf, flags, headers, callback_func, callback_arg);
308 } else if (php_stream_eof(s)) {
309 return php_http_header_parser_parse(parser, buf, flags | PHP_HTTP_HEADER_PARSER_CLEANUP, headers, callback_func, callback_arg);
310 } else {
311 return state;
312 }
313 }
314
315 return PHP_HTTP_HEADER_PARSER_STATE_DONE;
316 }
317
318 static zend_class_entry *php_http_header_parser_class_entry;
php_http_get_header_parser_class_entry(void)319 zend_class_entry *php_http_get_header_parser_class_entry(void)
320 {
321 return php_http_header_parser_class_entry;
322 }
323 static zend_object_handlers php_http_header_parser_object_handlers;
324
php_http_header_parser_object_new(zend_class_entry * ce)325 zend_object *php_http_header_parser_object_new(zend_class_entry *ce)
326 {
327 return &php_http_header_parser_object_new_ex(ce, NULL)->zo;
328 }
329
php_http_header_parser_object_new_ex(zend_class_entry * ce,php_http_header_parser_t * parser)330 php_http_header_parser_object_t *php_http_header_parser_object_new_ex(zend_class_entry *ce, php_http_header_parser_t *parser)
331 {
332 php_http_header_parser_object_t *o;
333
334 o = ecalloc(1, sizeof(php_http_header_parser_object_t) + zend_object_properties_size(ce));
335 zend_object_std_init(&o->zo, ce);
336 object_properties_init(&o->zo, ce);
337
338 if (parser) {
339 o->parser = parser;
340 } else {
341 o->parser = php_http_header_parser_init(NULL);
342 }
343 o->buffer = php_http_buffer_new();
344
345 o->zo.handlers = &php_http_header_parser_object_handlers;
346
347 return o;
348 }
349
php_http_header_parser_object_free(zend_object * object)350 void php_http_header_parser_object_free(zend_object *object)
351 {
352 php_http_header_parser_object_t *o = PHP_HTTP_OBJ(object, NULL);
353
354 if (o->parser) {
355 php_http_header_parser_free(&o->parser);
356 }
357 if (o->buffer) {
358 php_http_buffer_free(&o->buffer);
359 }
360 zend_object_std_dtor(object);
361 }
362
363 ZEND_BEGIN_ARG_INFO_EX(ai_HttpHeaderParser_getState, 0, 0, 0)
364 ZEND_END_ARG_INFO();
PHP_METHOD(HttpHeaderParser,getState)365 static PHP_METHOD(HttpHeaderParser, getState)
366 {
367 php_http_header_parser_object_t *parser_obj = PHP_HTTP_OBJ(NULL, getThis());
368
369 zend_parse_parameters_none();
370 /* always return the real state */
371 RETVAL_LONG(php_http_header_parser_state_is(parser_obj->parser));
372 }
373
374 ZEND_BEGIN_ARG_INFO_EX(ai_HttpHeaderParser_parse, 0, 0, 3)
375 ZEND_ARG_INFO(0, data)
376 ZEND_ARG_INFO(0, flags)
377 ZEND_ARG_ARRAY_INFO(1, headers, 1)
378 ZEND_END_ARG_INFO();
PHP_METHOD(HttpHeaderParser,parse)379 static PHP_METHOD(HttpHeaderParser, parse)
380 {
381 php_http_header_parser_object_t *parser_obj;
382 zval *zmsg;
383 char *data_str;
384 size_t data_len;
385 zend_long flags;
386
387 php_http_expect(SUCCESS == zend_parse_parameters(ZEND_NUM_ARGS(), "slz", &data_str, &data_len, &flags, &zmsg), invalid_arg, return);
388
389 ZVAL_DEREF(zmsg);
390 if (Z_TYPE_P(zmsg) != IS_ARRAY) {
391 zval_dtor(zmsg);
392 array_init(zmsg);
393 }
394 parser_obj = PHP_HTTP_OBJ(NULL, getThis());
395 php_http_buffer_append(parser_obj->buffer, data_str, data_len);
396 RETVAL_LONG(php_http_header_parser_parse(parser_obj->parser, parser_obj->buffer, flags, Z_ARRVAL_P(zmsg), NULL, NULL));
397 }
398
399 ZEND_BEGIN_ARG_INFO_EX(ai_HttpHeaderParser_stream, 0, 0, 3)
400 ZEND_ARG_INFO(0, stream)
401 ZEND_ARG_INFO(0, flags)
402 ZEND_ARG_ARRAY_INFO(1, headers, 1)
403 ZEND_END_ARG_INFO();
PHP_METHOD(HttpHeaderParser,stream)404 static PHP_METHOD(HttpHeaderParser, stream)
405 {
406 php_http_header_parser_object_t *parser_obj;
407 zend_error_handling zeh;
408 zval *zmsg, *zstream;
409 php_stream *s;
410 zend_long flags;
411
412 php_http_expect(SUCCESS == zend_parse_parameters(ZEND_NUM_ARGS(), "rlz", &zstream, &flags, &zmsg), invalid_arg, return);
413
414 zend_replace_error_handling(EH_THROW, php_http_get_exception_unexpected_val_class_entry(), &zeh);
415 php_stream_from_zval(s, zstream);
416 zend_restore_error_handling(&zeh);
417
418 ZVAL_DEREF(zmsg);
419 if (Z_TYPE_P(zmsg) != IS_ARRAY) {
420 zval_dtor(zmsg);
421 array_init(zmsg);
422 }
423 parser_obj = PHP_HTTP_OBJ(NULL, getThis());
424 RETVAL_LONG(php_http_header_parser_parse_stream(parser_obj->parser, parser_obj->buffer, s, flags, Z_ARRVAL_P(zmsg), NULL, NULL));
425 }
426
427 static zend_function_entry php_http_header_parser_methods[] = {
428 PHP_ME(HttpHeaderParser, getState, ai_HttpHeaderParser_getState, ZEND_ACC_PUBLIC)
429 PHP_ME(HttpHeaderParser, parse, ai_HttpHeaderParser_parse, ZEND_ACC_PUBLIC)
430 PHP_ME(HttpHeaderParser, stream, ai_HttpHeaderParser_stream, ZEND_ACC_PUBLIC)
431 {NULL, NULL, NULL}
432 };
433
PHP_MINIT_FUNCTION(http_header_parser)434 PHP_MINIT_FUNCTION(http_header_parser)
435 {
436 zend_class_entry ce;
437
438 INIT_NS_CLASS_ENTRY(ce, "http\\Header", "Parser", php_http_header_parser_methods);
439 php_http_header_parser_class_entry = zend_register_internal_class(&ce);
440 memcpy(&php_http_header_parser_object_handlers, zend_get_std_object_handlers(), sizeof(zend_object_handlers));
441 php_http_header_parser_class_entry->create_object = php_http_header_parser_object_new;
442 php_http_header_parser_object_handlers.offset = XtOffsetOf(php_http_header_parser_object_t, zo);
443 php_http_header_parser_object_handlers.clone_obj = NULL;
444 php_http_header_parser_object_handlers.free_obj = php_http_header_parser_object_free;
445
446 zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("CLEANUP"), PHP_HTTP_HEADER_PARSER_CLEANUP);
447
448 zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_FAILURE"), PHP_HTTP_HEADER_PARSER_STATE_FAILURE);
449 zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_START"), PHP_HTTP_HEADER_PARSER_STATE_START);
450 zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_KEY"), PHP_HTTP_HEADER_PARSER_STATE_KEY);
451 zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_VALUE"), PHP_HTTP_HEADER_PARSER_STATE_VALUE);
452 zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_VALUE_EX"), PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX);
453 zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_HEADER_DONE"), PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE);
454 zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_DONE"), PHP_HTTP_HEADER_PARSER_STATE_DONE);
455
456 return SUCCESS;
457 }
458
459 /*
460 * Local variables:
461 * tab-width: 4
462 * c-basic-offset: 4
463 * End:
464 * vim600: noet sw=4 ts=4 fdm=marker
465 * vim<600: noet sw=4 ts=4
466 */
467
468