| 1 | /* Based on src/http/ngx_http_parse.c from NGINX copyright Igor Sysoev |
| 2 | * |
| 3 | * Additional changes are licensed under the same terms as NGINX and |
| 4 | * copyright Joyent, Inc. and other Node contributors. All rights reserved. |
| 5 | * |
| 6 | * Permission is hereby granted, free of charge, to any person obtaining a copy |
| 7 | * of this software and associated documentation files (the "Software"), to |
| 8 | * deal in the Software without restriction, including without limitation the |
| 9 | * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or |
| 10 | * sell copies of the Software, and to permit persons to whom the Software is |
| 11 | * furnished to do so, subject to the following conditions: |
| 12 | * |
| 13 | * The above copyright notice and this permission notice shall be included in |
| 14 | * all copies or substantial portions of the Software. |
| 15 | * |
| 16 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR |
| 17 | * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, |
| 18 | * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE |
| 19 | * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER |
| 20 | * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING |
| 21 | * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS |
| 22 | * IN THE SOFTWARE. |
| 23 | */ |
| 24 | #include "http_parser.h" |
| 25 | #include <assert.h> |
| 26 | #include <stddef.h> |
| 27 | #include <ctype.h> |
| 28 | #include <stdlib.h> |
| 29 | #include <string.h> |
| 30 | #include <limits.h> |
| 31 | |
| 32 | #ifndef ULLONG_MAX |
| 33 | # define ULLONG_MAX ((uint64_t) -1) /* 2^64-1 */ |
| 34 | #endif |
| 35 | |
| 36 | #ifndef MIN |
| 37 | # define MIN(a,b) ((a) < (b) ? (a) : (b)) |
| 38 | #endif |
| 39 | |
| 40 | #ifndef ARRAY_SIZE |
| 41 | # define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0])) |
| 42 | #endif |
| 43 | |
| 44 | #ifndef BIT_AT |
| 45 | # define BIT_AT(a, i) \ |
| 46 | (!!((unsigned int) (a)[(unsigned int) (i) >> 3] & \ |
| 47 | (1 << ((unsigned int) (i) & 7)))) |
| 48 | #endif |
| 49 | |
| 50 | #ifndef ELEM_AT |
| 51 | # define ELEM_AT(a, i, v) ((unsigned int) (i) < ARRAY_SIZE(a) ? (a)[(i)] : (v)) |
| 52 | #endif |
| 53 | |
| 54 | #if HTTP_PARSER_DEBUG |
| 55 | #define SET_ERRNO(e) \ |
| 56 | do { \ |
| 57 | parser->http_errno = (e); \ |
| 58 | parser->error_lineno = __LINE__; \ |
| 59 | } while (0) |
| 60 | #else |
| 61 | #define SET_ERRNO(e) \ |
| 62 | do { \ |
| 63 | parser->http_errno = (e); \ |
| 64 | } while(0) |
| 65 | #endif |
| 66 | |
| 67 | |
| 68 | /* Run the notify callback FOR, returning ER if it fails */ |
| 69 | #define CALLBACK_NOTIFY_(FOR, ER) \ |
| 70 | do { \ |
| 71 | assert(HTTP_PARSER_ERRNO(parser) == HPE_OK); \ |
| 72 | \ |
| 73 | if (settings->on_##FOR) { \ |
| 74 | if (0 != settings->on_##FOR(parser)) { \ |
| 75 | SET_ERRNO(HPE_CB_##FOR); \ |
| 76 | } \ |
| 77 | \ |
| 78 | /* We either errored above or got paused; get out */ \ |
| 79 | if (HTTP_PARSER_ERRNO(parser) != HPE_OK) { \ |
| 80 | return (ER); \ |
| 81 | } \ |
| 82 | } \ |
| 83 | } while (0) |
| 84 | |
| 85 | /* Run the notify callback FOR and consume the current byte */ |
| 86 | #define CALLBACK_NOTIFY(FOR) CALLBACK_NOTIFY_(FOR, p - data + 1) |
| 87 | |
| 88 | /* Run the notify callback FOR and don't consume the current byte */ |
| 89 | #define CALLBACK_NOTIFY_NOADVANCE(FOR) CALLBACK_NOTIFY_(FOR, p - data) |
| 90 | |
| 91 | /* Run data callback FOR with LEN bytes, returning ER if it fails */ |
| 92 | #define CALLBACK_DATA_(FOR, LEN, ER) \ |
| 93 | do { \ |
| 94 | assert(HTTP_PARSER_ERRNO(parser) == HPE_OK); \ |
| 95 | \ |
| 96 | if (FOR##_mark) { \ |
| 97 | if (settings->on_##FOR) { \ |
| 98 | if (0 != settings->on_##FOR(parser, FOR##_mark, (LEN))) { \ |
| 99 | SET_ERRNO(HPE_CB_##FOR); \ |
| 100 | } \ |
| 101 | \ |
| 102 | /* We either errored above or got paused; get out */ \ |
| 103 | if (HTTP_PARSER_ERRNO(parser) != HPE_OK) { \ |
| 104 | return (ER); \ |
| 105 | } \ |
| 106 | } \ |
| 107 | FOR##_mark = NULL; \ |
| 108 | } \ |
| 109 | } while (0) |
| 110 | |
| 111 | /* Run the data callback FOR and consume the current byte */ |
| 112 | #define CALLBACK_DATA(FOR) \ |
| 113 | CALLBACK_DATA_(FOR, p - FOR##_mark, p - data + 1) |
| 114 | |
| 115 | /* Run the data callback FOR and don't consume the current byte */ |
| 116 | #define CALLBACK_DATA_NOADVANCE(FOR) \ |
| 117 | CALLBACK_DATA_(FOR, p - FOR##_mark, p - data) |
| 118 | |
| 119 | /* Set the mark FOR; non-destructive if mark is already set */ |
| 120 | #define MARK(FOR) \ |
| 121 | do { \ |
| 122 | if (!FOR##_mark) { \ |
| 123 | FOR##_mark = p; \ |
| 124 | } \ |
| 125 | } while (0) |
| 126 | |
| 127 | |
| 128 | #define PROXY_CONNECTION "proxy-connection" |
| 129 | #define CONNECTION "connection" |
| 130 | #define CONTENT_LENGTH "content-length" |
| 131 | #define TRANSFER_ENCODING "transfer-encoding" |
| 132 | #define UPGRADE "upgrade" |
| 133 | #define CHUNKED "chunked" |
| 134 | #define KEEP_ALIVE "keep-alive" |
| 135 | #define CLOSE "close" |
| 136 | |
| 137 | |
| 138 | static const char *method_strings[] = |
| 139 | { |
| 140 | #define XX(num, name, string) #string, |
| 141 | HTTP_METHOD_MAP(XX) |
| 142 | #undef XX |
| 143 | }; |
| 144 | |
| 145 | |
| 146 | /* Tokens as defined by rfc 2616. Also lowercases them. |
| 147 | * token = 1*<any CHAR except CTLs or separators> |
| 148 | * separators = "(" | ")" | "<" | ">" | "@" |
| 149 | * | "," | ";" | ":" | "\" | <"> |
| 150 | * | "/" | "[" | "]" | "?" | "=" |
| 151 | * | "{" | "}" | SP | HT |
| 152 | */ |
| 153 | static const char tokens[256] = { |
| 154 | /* 0 nul 1 soh 2 stx 3 etx 4 eot 5 enq 6 ack 7 bel */ |
| 155 | 0, 0, 0, 0, 0, 0, 0, 0, |
| 156 | /* 8 bs 9 ht 10 nl 11 vt 12 np 13 cr 14 so 15 si */ |
| 157 | 0, 0, 0, 0, 0, 0, 0, 0, |
| 158 | /* 16 dle 17 dc1 18 dc2 19 dc3 20 dc4 21 nak 22 syn 23 etb */ |
| 159 | 0, 0, 0, 0, 0, 0, 0, 0, |
| 160 | /* 24 can 25 em 26 sub 27 esc 28 fs 29 gs 30 rs 31 us */ |
| 161 | 0, 0, 0, 0, 0, 0, 0, 0, |
| 162 | /* 32 sp 33 ! 34 " 35 # 36 $ 37 % 38 & 39 ' */ |
| 163 | 0, '!', 0, '#', '$', '%', '&', '\'', |
| 164 | /* 40 ( 41 ) 42 * 43 + 44 , 45 - 46 . 47 / */ |
| 165 | 0, 0, '*', '+', 0, '-', '.', 0, |
| 166 | /* 48 0 49 1 50 2 51 3 52 4 53 5 54 6 55 7 */ |
| 167 | '0', '1', '2', '3', '4', '5', '6', '7', |
| 168 | /* 56 8 57 9 58 : 59 ; 60 < 61 = 62 > 63 ? */ |
| 169 | '8', '9', 0, 0, 0, 0, 0, 0, |
| 170 | /* 64 @ 65 A 66 B 67 C 68 D 69 E 70 F 71 G */ |
| 171 | 0, 'a', 'b', 'c', 'd', 'e', 'f', 'g', |
| 172 | /* 72 H 73 I 74 J 75 K 76 L 77 M 78 N 79 O */ |
| 173 | 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', |
| 174 | /* 80 P 81 Q 82 R 83 S 84 T 85 U 86 V 87 W */ |
| 175 | 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', |
| 176 | /* 88 X 89 Y 90 Z 91 [ 92 \ 93 ] 94 ^ 95 _ */ |
| 177 | 'x', 'y', 'z', 0, 0, 0, '^', '_', |
| 178 | /* 96 ` 97 a 98 b 99 c 100 d 101 e 102 f 103 g */ |
| 179 | '`', 'a', 'b', 'c', 'd', 'e', 'f', 'g', |
| 180 | /* 104 h 105 i 106 j 107 k 108 l 109 m 110 n 111 o */ |
| 181 | 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', |
| 182 | /* 112 p 113 q 114 r 115 s 116 t 117 u 118 v 119 w */ |
| 183 | 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', |
| 184 | /* 120 x 121 y 122 z 123 { 124 | 125 } 126 ~ 127 del */ |
| 185 | 'x', 'y', 'z', 0, '|', 0, '~', 0 }; |
| 186 | |
| 187 | |
| 188 | static const int8_t unhex[256] = |
| 189 | {-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 |
| 190 | ,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 |
| 191 | ,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 |
| 192 | , 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,-1,-1,-1,-1,-1,-1 |
| 193 | ,-1,10,11,12,13,14,15,-1,-1,-1,-1,-1,-1,-1,-1,-1 |
| 194 | ,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 |
| 195 | ,-1,10,11,12,13,14,15,-1,-1,-1,-1,-1,-1,-1,-1,-1 |
| 196 | ,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 |
| 197 | }; |
| 198 | |
| 199 | |
| 200 | #if HTTP_PARSER_STRICT |
| 201 | # define T(v) 0 |
| 202 | #else |
| 203 | # define T(v) v |
| 204 | #endif |
| 205 | |
| 206 | |
| 207 | static const uint8_t normal_url_char[32] = { |
| 208 | /* 0 nul 1 soh 2 stx 3 etx 4 eot 5 enq 6 ack 7 bel */ |
| 209 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0, |
| 210 | /* 8 bs 9 ht 10 nl 11 vt 12 np 13 cr 14 so 15 si */ |
| 211 | 0 | T(2) | 0 | 0 | T(16) | 0 | 0 | 0, |
| 212 | /* 16 dle 17 dc1 18 dc2 19 dc3 20 dc4 21 nak 22 syn 23 etb */ |
| 213 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0, |
| 214 | /* 24 can 25 em 26 sub 27 esc 28 fs 29 gs 30 rs 31 us */ |
| 215 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0, |
| 216 | /* 32 sp 33 ! 34 " 35 # 36 $ 37 % 38 & 39 ' */ |
| 217 | 0 | 2 | 4 | 0 | 16 | 32 | 64 | 128, |
| 218 | /* 40 ( 41 ) 42 * 43 + 44 , 45 - 46 . 47 / */ |
| 219 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 220 | /* 48 0 49 1 50 2 51 3 52 4 53 5 54 6 55 7 */ |
| 221 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 222 | /* 56 8 57 9 58 : 59 ; 60 < 61 = 62 > 63 ? */ |
| 223 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 0, |
| 224 | /* 64 @ 65 A 66 B 67 C 68 D 69 E 70 F 71 G */ |
| 225 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 226 | /* 72 H 73 I 74 J 75 K 76 L 77 M 78 N 79 O */ |
| 227 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 228 | /* 80 P 81 Q 82 R 83 S 84 T 85 U 86 V 87 W */ |
| 229 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 230 | /* 88 X 89 Y 90 Z 91 [ 92 \ 93 ] 94 ^ 95 _ */ |
| 231 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 232 | /* 96 ` 97 a 98 b 99 c 100 d 101 e 102 f 103 g */ |
| 233 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 234 | /* 104 h 105 i 106 j 107 k 108 l 109 m 110 n 111 o */ |
| 235 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 236 | /* 112 p 113 q 114 r 115 s 116 t 117 u 118 v 119 w */ |
| 237 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, |
| 238 | /* 120 x 121 y 122 z 123 { 124 | 125 } 126 ~ 127 del */ |
| 239 | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 0, }; |
| 240 | |
| 241 | #undef T |
| 242 | |
| 243 | enum state |
| 244 | { s_dead = 1 /* important that this is > 0 */ |
| 245 | |
| 246 | , s_start_req_or_res |
| 247 | , s_res_or_resp_H |
| 248 | , s_start_res |
| 249 | , s_res_H |
| 250 | , s_res_HT |
| 251 | , s_res_HTT |
| 252 | , s_res_HTTP |
| 253 | , s_res_first_http_major |
| 254 | , s_res_http_major |
| 255 | , s_res_first_http_minor |
| 256 | , s_res_http_minor |
| 257 | , s_res_first_status_code |
| 258 | , s_res_status_code |
| 259 | , s_res_status |
| 260 | , s_res_line_almost_done |
| 261 | |
| 262 | , s_start_req |
| 263 | |
| 264 | , s_req_method |
| 265 | , s_req_spaces_before_url |
| 266 | , s_req_schema |
| 267 | , s_req_schema_slash |
| 268 | , s_req_schema_slash_slash |
| 269 | , s_req_server_start |
| 270 | , s_req_server |
| 271 | , s_req_server_with_at |
| 272 | , s_req_path |
| 273 | , s_req_query_string_start |
| 274 | , s_req_query_string |
| 275 | , s_req_fragment_start |
| 276 | , s_req_fragment |
| 277 | , s_req_http_start |
| 278 | , s_req_http_H |
| 279 | , s_req_http_HT |
| 280 | , s_req_http_HTT |
| 281 | , s_req_http_HTTP |
| 282 | , s_req_first_http_major |
| 283 | , s_req_http_major |
| 284 | , s_req_first_http_minor |
| 285 | , s_req_http_minor |
| 286 | , s_req_line_almost_done |
| 287 | |
| 288 | , s_header_field_start |
| 289 | , s_header_field |
| 290 | , s_header_value_start |
| 291 | , s_header_value |
| 292 | , s_header_value_lws |
| 293 | |
| 294 | , s_header_almost_done |
| 295 | |
| 296 | , s_chunk_size_start |
| 297 | , s_chunk_size |
| 298 | , s_chunk_parameters |
| 299 | , s_chunk_size_almost_done |
| 300 | |
| 301 | , s_headers_almost_done |
| 302 | , s_headers_done |
| 303 | |
| 304 | /* Important: 's_headers_done' must be the last 'header' state. All |
| 305 | * states beyond this must be 'body' states. It is used for overflow |
| 306 | * checking. See the PARSING_HEADER() macro. |
| 307 | */ |
| 308 | |
| 309 | , s_chunk_data |
| 310 | , s_chunk_data_almost_done |
| 311 | , s_chunk_data_done |
| 312 | |
| 313 | , s_body_identity |
| 314 | , s_body_identity_eof |
| 315 | |
| 316 | , s_message_done |
| 317 | }; |
| 318 | |
| 319 | |
| 320 | #define PARSING_HEADER(state) (state <= s_headers_done) |
| 321 | |
| 322 | |
| 323 | enum header_states |
| 324 | { h_general = 0 |
| 325 | , h_C |
| 326 | , h_CO |
| 327 | , h_CON |
| 328 | |
| 329 | , h_matching_connection |
| 330 | , h_matching_proxy_connection |
| 331 | , h_matching_content_length |
| 332 | , h_matching_transfer_encoding |
| 333 | , h_matching_upgrade |
| 334 | |
| 335 | , h_connection |
| 336 | , h_content_length |
| 337 | , h_transfer_encoding |
| 338 | , h_upgrade |
| 339 | |
| 340 | , h_matching_transfer_encoding_chunked |
| 341 | , h_matching_connection_keep_alive |
| 342 | , h_matching_connection_close |
| 343 | |
| 344 | , h_transfer_encoding_chunked |
| 345 | , h_connection_keep_alive |
| 346 | , h_connection_close |
| 347 | }; |
| 348 | |
| 349 | enum http_host_state |
| 350 | { |
| 351 | s_http_host_dead = 1 |
| 352 | , s_http_userinfo_start |
| 353 | , s_http_userinfo |
| 354 | , s_http_host_start |
| 355 | , s_http_host_v6_start |
| 356 | , s_http_host |
| 357 | , s_http_host_v6 |
| 358 | , s_http_host_v6_end |
| 359 | , s_http_host_port_start |
| 360 | , s_http_host_port |
| 361 | }; |
| 362 | |
| 363 | /* Macros for character classes; depends on strict-mode */ |
| 364 | #define CR '\r' |
| 365 | #define LF '\n' |
| 366 | #define LOWER(c) (unsigned char)(c | 0x20) |
| 367 | #define IS_ALPHA(c) (LOWER(c) >= 'a' && LOWER(c) <= 'z') |
| 368 | #define IS_NUM(c) ((c) >= '0' && (c) <= '9') |
| 369 | #define IS_ALPHANUM(c) (IS_ALPHA(c) || IS_NUM(c)) |
| 370 | #define IS_HEX(c) (IS_NUM(c) || (LOWER(c) >= 'a' && LOWER(c) <= 'f')) |
| 371 | #define IS_MARK(c) ((c) == '-' || (c) == '_' || (c) == '.' || \ |
| 372 | (c) == '!' || (c) == '~' || (c) == '*' || (c) == '\'' || (c) == '(' || \ |
| 373 | (c) == ')') |
| 374 | #define IS_USERINFO_CHAR(c) (IS_ALPHANUM(c) || IS_MARK(c) || (c) == '%' || \ |
| 375 | (c) == ';' || (c) == ':' || (c) == '&' || (c) == '=' || (c) == '+' || \ |
| 376 | (c) == '$' || (c) == ',') |
| 377 | |
| 378 | #if HTTP_PARSER_STRICT |
| 379 | #define TOKEN(c) (tokens[(unsigned char)c]) |
| 380 | #define IS_URL_CHAR(c) (BIT_AT(normal_url_char, (unsigned char)c)) |
| 381 | #define IS_HOST_CHAR(c) (IS_ALPHANUM(c) || (c) == '.' || (c) == '-') |
| 382 | #else |
| 383 | #define TOKEN(c) ((c == ' ') ? ' ' : tokens[(unsigned char)c]) |
| 384 | #define IS_URL_CHAR(c) \ |
| 385 | (BIT_AT(normal_url_char, (unsigned char)c) || ((c) & 0x80)) |
| 386 | #define IS_HOST_CHAR(c) \ |
| 387 | (IS_ALPHANUM(c) || (c) == '.' || (c) == '-' || (c) == '_') |
| 388 | #endif |
| 389 | |
| 390 | |
| 391 | #define start_state (parser->type == HTTP_REQUEST ? s_start_req : s_start_res) |
| 392 | |
| 393 | |
| 394 | #if HTTP_PARSER_STRICT |
| 395 | # define STRICT_CHECK(cond) \ |
| 396 | do { \ |
| 397 | if (cond) { \ |
| 398 | SET_ERRNO(HPE_STRICT); \ |
| 399 | goto error; \ |
| 400 | } \ |
| 401 | } while (0) |
| 402 | # define NEW_MESSAGE() (http_should_keep_alive(parser) ? start_state : s_dead) |
| 403 | #else |
| 404 | # define STRICT_CHECK(cond) |
| 405 | # define NEW_MESSAGE() start_state |
| 406 | #endif |
| 407 | |
| 408 | |
| 409 | /* Map errno values to strings for human-readable output */ |
| 410 | #define HTTP_STRERROR_GEN(n, s) { "HPE_" #n, s }, |
| 411 | static struct { |
| 412 | const char *name; |
| 413 | const char *description; |
| 414 | } http_strerror_tab[] = { |
| 415 | HTTP_ERRNO_MAP(HTTP_STRERROR_GEN) |
| 416 | }; |
| 417 | #undef HTTP_STRERROR_GEN |
| 418 | |
| 419 | int http_message_needs_eof(const http_parser *parser); |
| 420 | |
| 421 | /* Our URL parser. |
| 422 | * |
| 423 | * This is designed to be shared by http_parser_execute() for URL validation, |
| 424 | * hence it has a state transition + byte-for-byte interface. In addition, it |
| 425 | * is meant to be embedded in http_parser_parse_url(), which does the dirty |
| 426 | * work of turning state transitions URL components for its API. |
| 427 | * |
| 428 | * This function should only be invoked with non-space characters. It is |
| 429 | * assumed that the caller cares about (and can detect) the transition between |
| 430 | * URL and non-URL states by looking for these. |
| 431 | */ |
| 432 | static enum state |
| 433 | parse_url_char(enum state s, const char ch) |
| 434 | { |
| 435 | if (ch == ' ' || ch == '\r' || ch == '\n') { |
| 436 | return s_dead; |
| 437 | } |
| 438 | |
| 439 | #if HTTP_PARSER_STRICT |
| 440 | if (ch == '\t' || ch == '\f') { |
| 441 | return s_dead; |
| 442 | } |
| 443 | #endif |
| 444 | |
| 445 | switch (s) { |
| 446 | case s_req_spaces_before_url: |
| 447 | /* Proxied requests are followed by scheme of an absolute URI (alpha). |
| 448 | * All methods except CONNECT are followed by '/' or '*'. |
| 449 | */ |
| 450 | |
| 451 | if (ch == '/' || ch == '*') { |
| 452 | return s_req_path; |
| 453 | } |
| 454 | |
| 455 | if (IS_ALPHA(ch)) { |
| 456 | return s_req_schema; |
| 457 | } |
| 458 | |
| 459 | break; |
| 460 | |
| 461 | case s_req_schema: |
| 462 | if (IS_ALPHA(ch)) { |
| 463 | return s; |
| 464 | } |
| 465 | |
| 466 | if (ch == ':') { |
| 467 | return s_req_schema_slash; |
| 468 | } |
| 469 | |
| 470 | break; |
| 471 | |
| 472 | case s_req_schema_slash: |
| 473 | if (ch == '/') { |
| 474 | return s_req_schema_slash_slash; |
| 475 | } |
| 476 | |
| 477 | break; |
| 478 | |
| 479 | case s_req_schema_slash_slash: |
| 480 | if (ch == '/') { |
| 481 | return s_req_server_start; |
| 482 | } |
| 483 | |
| 484 | break; |
| 485 | |
| 486 | case s_req_server_with_at: |
| 487 | if (ch == '@') { |
| 488 | return s_dead; |
| 489 | } |
| 490 | |
| 491 | /* FALLTHROUGH */ |
| 492 | case s_req_server_start: |
| 493 | case s_req_server: |
| 494 | if (ch == '/') { |
| 495 | return s_req_path; |
| 496 | } |
| 497 | |
| 498 | if (ch == '?') { |
| 499 | return s_req_query_string_start; |
| 500 | } |
| 501 | |
| 502 | if (ch == '@') { |
| 503 | return s_req_server_with_at; |
| 504 | } |
| 505 | |
| 506 | if (IS_USERINFO_CHAR(ch) || ch == '[' || ch == ']') { |
| 507 | return s_req_server; |
| 508 | } |
| 509 | |
| 510 | break; |
| 511 | |
| 512 | case s_req_path: |
| 513 | if (IS_URL_CHAR(ch)) { |
| 514 | return s; |
| 515 | } |
| 516 | |
| 517 | switch (ch) { |
| 518 | case '?': |
| 519 | return s_req_query_string_start; |
| 520 | |
| 521 | case '#': |
| 522 | return s_req_fragment_start; |
| 523 | } |
| 524 | |
| 525 | break; |
| 526 | |
| 527 | case s_req_query_string_start: |
| 528 | case s_req_query_string: |
| 529 | if (IS_URL_CHAR(ch)) { |
| 530 | return s_req_query_string; |
| 531 | } |
| 532 | |
| 533 | switch (ch) { |
| 534 | case '?': |
| 535 | /* allow extra '?' in query string */ |
| 536 | return s_req_query_string; |
| 537 | |
| 538 | case '#': |
| 539 | return s_req_fragment_start; |
| 540 | } |
| 541 | |
| 542 | break; |
| 543 | |
| 544 | case s_req_fragment_start: |
| 545 | if (IS_URL_CHAR(ch)) { |
| 546 | return s_req_fragment; |
| 547 | } |
| 548 | |
| 549 | switch (ch) { |
| 550 | case '?': |
| 551 | return s_req_fragment; |
| 552 | |
| 553 | case '#': |
| 554 | return s; |
| 555 | } |
| 556 | |
| 557 | break; |
| 558 | |
| 559 | case s_req_fragment: |
| 560 | if (IS_URL_CHAR(ch)) { |
| 561 | return s; |
| 562 | } |
| 563 | |
| 564 | switch (ch) { |
| 565 | case '?': |
| 566 | case '#': |
| 567 | return s; |
| 568 | } |
| 569 | |
| 570 | break; |
| 571 | |
| 572 | default: |
| 573 | break; |
| 574 | } |
| 575 | |
| 576 | /* We should never fall out of the switch above unless there's an error */ |
| 577 | return s_dead; |
| 578 | } |
| 579 | |
| 580 | size_t http_parser_execute (http_parser *parser, |
| 581 | const http_parser_settings *settings, |
| 582 | const char *data, |
| 583 | size_t len) |
| 584 | { |
| 585 | char c, ch; |
| 586 | int8_t unhex_val; |
| 587 | const char *p = data; |
| 588 | const char *header_field_mark = 0; |
| 589 | const char *header_value_mark = 0; |
| 590 | const char *url_mark = 0; |
| 591 | const char *body_mark = 0; |
| 592 | |
| 593 | /* We're in an error state. Don't bother doing anything. */ |
| 594 | if (HTTP_PARSER_ERRNO(parser) != HPE_OK) { |
| 595 | return 0; |
| 596 | } |
| 597 | |
| 598 | if (len == 0) { |
| 599 | switch (parser->state) { |
| 600 | case s_body_identity_eof: |
| 601 | /* Use of CALLBACK_NOTIFY() here would erroneously return 1 byte read if |
| 602 | * we got paused. |
| 603 | */ |
| 604 | CALLBACK_NOTIFY_NOADVANCE(message_complete); |
| 605 | return 0; |
| 606 | |
| 607 | case s_dead: |
| 608 | case s_start_req_or_res: |
| 609 | case s_start_res: |
| 610 | case s_start_req: |
| 611 | return 0; |
| 612 | |
| 613 | default: |
| 614 | SET_ERRNO(HPE_INVALID_EOF_STATE); |
| 615 | return 1; |
| 616 | } |
| 617 | } |
| 618 | |
| 619 | |
| 620 | if (parser->state == s_header_field) |
| 621 | header_field_mark = data; |
| 622 | if (parser->state == s_header_value) |
| 623 | header_value_mark = data; |
| 624 | switch (parser->state) { |
| 625 | case s_req_path: |
| 626 | case s_req_schema: |
| 627 | case s_req_schema_slash: |
| 628 | case s_req_schema_slash_slash: |
| 629 | case s_req_server_start: |
| 630 | case s_req_server: |
| 631 | case s_req_server_with_at: |
| 632 | case s_req_query_string_start: |
| 633 | case s_req_query_string: |
| 634 | case s_req_fragment_start: |
| 635 | case s_req_fragment: |
| 636 | url_mark = data; |
| 637 | break; |
| 638 | } |
| 639 | |
| 640 | for (p=data; p != data + len; p++) { |
| 641 | ch = *p; |
| 642 | |
| 643 | if (PARSING_HEADER(parser->state)) { |
| 644 | ++parser->nread; |
| 645 | /* Buffer overflow attack */ |
| 646 | if (parser->nread > HTTP_MAX_HEADER_SIZE) { |
| 647 | SET_ERRNO(HPE_HEADER_OVERFLOW); |
| 648 | goto error; |
| 649 | } |
| 650 | } |
| 651 | |
| 652 | reexecute_byte: |
| 653 | switch (parser->state) { |
| 654 | |
| 655 | case s_dead: |
| 656 | /* this state is used after a 'Connection: close' message |
| 657 | * the parser will error out if it reads another message |
| 658 | */ |
| 659 | if (ch == CR || ch == LF) |
| 660 | break; |
| 661 | |
| 662 | SET_ERRNO(HPE_CLOSED_CONNECTION); |
| 663 | goto error; |
| 664 | |
| 665 | case s_start_req_or_res: |
| 666 | { |
| 667 | if (ch == CR || ch == LF) |
| 668 | break; |
| 669 | parser->flags = 0; |
| 670 | parser->content_length = ULLONG_MAX; |
| 671 | |
| 672 | if (ch == 'H') { |
| 673 | parser->state = s_res_or_resp_H; |
| 674 | |
| 675 | CALLBACK_NOTIFY(message_begin); |
| 676 | } else { |
| 677 | parser->type = HTTP_REQUEST; |
| 678 | parser->state = s_start_req; |
| 679 | goto reexecute_byte; |
| 680 | } |
| 681 | |
| 682 | break; |
| 683 | } |
| 684 | |
| 685 | case s_res_or_resp_H: |
| 686 | if (ch == 'T') { |
| 687 | parser->type = HTTP_RESPONSE; |
| 688 | parser->state = s_res_HT; |
| 689 | } else { |
| 690 | if (ch != 'E') { |
| 691 | SET_ERRNO(HPE_INVALID_CONSTANT); |
| 692 | goto error; |
| 693 | } |
| 694 | |
| 695 | parser->type = HTTP_REQUEST; |
| 696 | parser->method = HTTP_HEAD; |
| 697 | parser->index = 2; |
| 698 | parser->state = s_req_method; |
| 699 | } |
| 700 | break; |
| 701 | |
| 702 | case s_start_res: |
| 703 | { |
| 704 | parser->flags = 0; |
| 705 | parser->content_length = ULLONG_MAX; |
| 706 | |
| 707 | switch (ch) { |
| 708 | case 'H': |
| 709 | parser->state = s_res_H; |
| 710 | break; |
| 711 | |
| 712 | case CR: |
| 713 | case LF: |
| 714 | break; |
| 715 | |
| 716 | default: |
| 717 | SET_ERRNO(HPE_INVALID_CONSTANT); |
| 718 | goto error; |
| 719 | } |
| 720 | |
| 721 | CALLBACK_NOTIFY(message_begin); |
| 722 | break; |
| 723 | } |
| 724 | |
| 725 | case s_res_H: |
| 726 | STRICT_CHECK(ch != 'T'); |
| 727 | parser->state = s_res_HT; |
| 728 | break; |
| 729 | |
| 730 | case s_res_HT: |
| 731 | STRICT_CHECK(ch != 'T'); |
| 732 | parser->state = s_res_HTT; |
| 733 | break; |
| 734 | |
| 735 | case s_res_HTT: |
| 736 | STRICT_CHECK(ch != 'P'); |
| 737 | parser->state = s_res_HTTP; |
| 738 | break; |
| 739 | |
| 740 | case s_res_HTTP: |
| 741 | STRICT_CHECK(ch != '/'); |
| 742 | parser->state = s_res_first_http_major; |
| 743 | break; |
| 744 | |
| 745 | case s_res_first_http_major: |
| 746 | if (ch < '0' || ch > '9') { |
| 747 | SET_ERRNO(HPE_INVALID_VERSION); |
| 748 | goto error; |
| 749 | } |
| 750 | |
| 751 | parser->http_major = ch - '0'; |
| 752 | parser->state = s_res_http_major; |
| 753 | break; |
| 754 | |
| 755 | /* major HTTP version or dot */ |
| 756 | case s_res_http_major: |
| 757 | { |
| 758 | if (ch == '.') { |
| 759 | parser->state = s_res_first_http_minor; |
| 760 | break; |
| 761 | } |
| 762 | |
| 763 | if (!IS_NUM(ch)) { |
| 764 | SET_ERRNO(HPE_INVALID_VERSION); |
| 765 | goto error; |
| 766 | } |
| 767 | |
| 768 | parser->http_major *= 10; |
| 769 | parser->http_major += ch - '0'; |
| 770 | |
| 771 | if (parser->http_major > 999) { |
| 772 | SET_ERRNO(HPE_INVALID_VERSION); |
| 773 | goto error; |
| 774 | } |
| 775 | |
| 776 | break; |
| 777 | } |
| 778 | |
| 779 | /* first digit of minor HTTP version */ |
| 780 | case s_res_first_http_minor: |
| 781 | if (!IS_NUM(ch)) { |
| 782 | SET_ERRNO(HPE_INVALID_VERSION); |
| 783 | goto error; |
| 784 | } |
| 785 | |
| 786 | parser->http_minor = ch - '0'; |
| 787 | parser->state = s_res_http_minor; |
| 788 | break; |
| 789 | |
| 790 | /* minor HTTP version or end of request line */ |
| 791 | case s_res_http_minor: |
| 792 | { |
| 793 | if (ch == ' ') { |
| 794 | parser->state = s_res_first_status_code; |
| 795 | break; |
| 796 | } |
| 797 | |
| 798 | if (!IS_NUM(ch)) { |
| 799 | SET_ERRNO(HPE_INVALID_VERSION); |
| 800 | goto error; |
| 801 | } |
| 802 | |
| 803 | parser->http_minor *= 10; |
| 804 | parser->http_minor += ch - '0'; |
| 805 | |
| 806 | if (parser->http_minor > 999) { |
| 807 | SET_ERRNO(HPE_INVALID_VERSION); |
| 808 | goto error; |
| 809 | } |
| 810 | |
| 811 | break; |
| 812 | } |
| 813 | |
| 814 | case s_res_first_status_code: |
| 815 | { |
| 816 | if (!IS_NUM(ch)) { |
| 817 | if (ch == ' ') { |
| 818 | break; |
| 819 | } |
| 820 | |
| 821 | SET_ERRNO(HPE_INVALID_STATUS); |
| 822 | goto error; |
| 823 | } |
| 824 | parser->status_code = ch - '0'; |
| 825 | parser->state = s_res_status_code; |
| 826 | break; |
| 827 | } |
| 828 | |
| 829 | case s_res_status_code: |
| 830 | { |
| 831 | if (!IS_NUM(ch)) { |
| 832 | switch (ch) { |
| 833 | case ' ': |
| 834 | parser->state = s_res_status; |
| 835 | break; |
| 836 | case CR: |
| 837 | parser->state = s_res_line_almost_done; |
| 838 | break; |
| 839 | case LF: |
| 840 | parser->state = s_header_field_start; |
| 841 | break; |
| 842 | default: |
| 843 | SET_ERRNO(HPE_INVALID_STATUS); |
| 844 | goto error; |
| 845 | } |
| 846 | break; |
| 847 | } |
| 848 | |
| 849 | parser->status_code *= 10; |
| 850 | parser->status_code += ch - '0'; |
| 851 | |
| 852 | if (parser->status_code > 999) { |
| 853 | SET_ERRNO(HPE_INVALID_STATUS); |
| 854 | goto error; |
| 855 | } |
| 856 | |
| 857 | break; |
| 858 | } |
| 859 | |
| 860 | case s_res_status: |
| 861 | /* the human readable status. e.g. "NOT FOUND" |
| 862 | * we are not humans so just ignore this */ |
| 863 | if (ch == CR) { |
| 864 | parser->state = s_res_line_almost_done; |
| 865 | break; |
| 866 | } |
| 867 | |
| 868 | if (ch == LF) { |
| 869 | parser->state = s_header_field_start; |
| 870 | break; |
| 871 | } |
| 872 | break; |
| 873 | |
| 874 | case s_res_line_almost_done: |
| 875 | STRICT_CHECK(ch != LF); |
| 876 | parser->state = s_header_field_start; |
| 877 | break; |
| 878 | |
| 879 | case s_start_req: |
| 880 | { |
| 881 | if (ch == CR || ch == LF) |
| 882 | break; |
| 883 | parser->flags = 0; |
| 884 | parser->content_length = ULLONG_MAX; |
| 885 | |
| 886 | if (!IS_ALPHA(ch)) { |
| 887 | SET_ERRNO(HPE_INVALID_METHOD); |
| 888 | goto error; |
| 889 | } |
| 890 | |
| 891 | parser->method = (enum http_method) 0; |
| 892 | parser->index = 1; |
| 893 | switch (ch) { |
| 894 | case 'A': parser->method = HTTP_ANNOUNCE; break; |
| 895 | case 'C': parser->method = HTTP_CONNECT; /* or COPY, CHECKOUT */ break; |
| 896 | case 'D': parser->method = HTTP_DELETE; /* or DESCRIBE */ break; |
| 897 | case 'F': parser->method = HTTP_FLUSH; break; |
| 898 | case 'G': parser->method = HTTP_GET; /* or GET_PARAMETER */ break; |
| 899 | case 'H': parser->method = HTTP_HEAD; break; |
| 900 | case 'L': parser->method = HTTP_LOCK; break; |
| 901 | case 'M': parser->method = HTTP_MKCOL; /* or MOVE, MKACTIVITY, MERGE, M-SEARCH */ break; |
| 902 | case 'N': parser->method = HTTP_NOTIFY; break; |
| 903 | case 'O': parser->method = HTTP_OPTIONS; break; |
| 904 | case 'P': parser->method = HTTP_POST; |
| 905 | /* or PROPFIND|PROPPATCH|PUT|PATCH|PURGE|PLAY|PAUSE */ |
| 906 | break; |
| 907 | case 'R': parser->method = HTTP_REPORT; /* or REDIRECT, RECORD */ break; |
| 908 | case 'S': parser->method = HTTP_SUBSCRIBE; /* or SEARCH, SETUP, SET_PARAMETER */ break; |
| 909 | case 'T': parser->method = HTTP_TRACE; /* or TEARDOWN */ break; |
| 910 | case 'U': parser->method = HTTP_UNLOCK; /* or UNSUBSCRIBE */ break; |
| 911 | default: |
| 912 | SET_ERRNO(HPE_INVALID_METHOD); |
| 913 | goto error; |
| 914 | } |
| 915 | parser->state = s_req_method; |
| 916 | |
| 917 | CALLBACK_NOTIFY(message_begin); |
| 918 | |
| 919 | break; |
| 920 | } |
| 921 | |
| 922 | case s_req_method: |
| 923 | { |
| 924 | const char *matcher; |
| 925 | if (ch == '\0') { |
| 926 | SET_ERRNO(HPE_INVALID_METHOD); |
| 927 | goto error; |
| 928 | } |
| 929 | |
| 930 | matcher = method_strings[parser->method]; |
| 931 | if (ch == ' ' && matcher[parser->index] == '\0') { |
| 932 | parser->state = s_req_spaces_before_url; |
| 933 | } else if (ch == matcher[parser->index]) { |
| 934 | ; /* nada */ |
| 935 | } else if (parser->method == HTTP_CONNECT) { |
| 936 | if (parser->index == 1 && ch == 'H') { |
| 937 | parser->method = HTTP_CHECKOUT; |
| 938 | } else if (parser->index == 2 && ch == 'P') { |
| 939 | parser->method = HTTP_COPY; |
| 940 | } else { |
| 941 | goto error; |
| 942 | } |
| 943 | } else if (parser->index == 2 && parser->method == HTTP_DELETE && ch == 'S') { |
| 944 | parser->method = HTTP_DESCRIBE; |
| 945 | } else if (parser->index == 3 && parser->method == HTTP_GET && ch == '_') { |
| 946 | parser->method = HTTP_GET_PARAMETER; |
| 947 | } else if (parser->method == HTTP_MKCOL) { |
| 948 | if (parser->index == 1 && ch == 'O') { |
| 949 | parser->method = HTTP_MOVE; |
| 950 | } else if (parser->index == 1 && ch == 'E') { |
| 951 | parser->method = HTTP_MERGE; |
| 952 | } else if (parser->index == 1 && ch == '-') { |
| 953 | parser->method = HTTP_MSEARCH; |
| 954 | } else if (parser->index == 2 && ch == 'A') { |
| 955 | parser->method = HTTP_MKACTIVITY; |
| 956 | } else { |
| 957 | goto error; |
| 958 | } |
| 959 | } else if (parser->method == HTTP_SUBSCRIBE) { |
| 960 | if (parser->index == 1 && ch == 'E') { |
| 961 | parser->method = HTTP_SEARCH; /* or HTTP_SETUP or HTTP_SET_PARAMETER */ |
| 962 | } else { |
| 963 | goto error; |
| 964 | } |
| 965 | } else if (parser->method == HTTP_TRACE) { |
| 966 | if (parser->index == 1 && ch == 'E') { |
| 967 | parser->method = HTTP_TEARDOWN; |
| 968 | } else { |
| 969 | goto error; |
| 970 | } |
| 971 | } else if (parser->index == 1 && parser->method == HTTP_POST) { |
| 972 | if (ch == 'R') { |
| 973 | parser->method = HTTP_PROPFIND; /* or HTTP_PROPPATCH */ |
| 974 | } else if (ch == 'U') { |
| 975 | parser->method = HTTP_PUT; /* or HTTP_PURGE */ |
| 976 | } else if (ch == 'A') { |
| 977 | parser->method = HTTP_PATCH; /* or HTTP_PAUSE */ |
| 978 | } else if (ch == 'L') { |
| 979 | parser->method = HTTP_PLAY; |
| 980 | } else { |
| 981 | goto error; |
| 982 | } |
| 983 | } else if (parser->index == 2) { |
| 984 | if (parser->method == HTTP_PUT) { |
| 985 | if (ch == 'R') parser->method = HTTP_PURGE; |
| 986 | } else if (parser->method == HTTP_PATCH) { |
| 987 | if (ch == 'U') parser->method = HTTP_PAUSE; |
| 988 | } else if (parser->method == HTTP_REPORT && ch == 'D') { |
| 989 | parser->method = HTTP_REDIRECT; |
| 990 | } else if (parser->method == HTTP_REPORT && ch == 'C') { |
| 991 | parser->method = HTTP_RECORD; |
| 992 | } else if (parser->method == HTTP_SEARCH) { |
| 993 | if (ch == 'T') parser->method = HTTP_SETUP; /* or HTTP_SET_PARAMETER */ |
| 994 | } else if (parser->method == HTTP_UNLOCK) { |
| 995 | if (ch == 'S') parser->method = HTTP_UNSUBSCRIBE; |
| 996 | } else { |
| 997 | goto error; |
| 998 | } |
| 999 | } else if (parser->index == 3 && parser->method == HTTP_SETUP && ch == '_') { |
| 1000 | parser->method = HTTP_SET_PARAMETER; |
| 1001 | } else if (parser->index == 4 && parser->method == HTTP_PROPFIND && ch == 'P') { |
| 1002 | parser->method = HTTP_PROPPATCH; |
| 1003 | } else { |
| 1004 | SET_ERRNO(HPE_INVALID_METHOD); |
| 1005 | goto error; |
| 1006 | } |
| 1007 | |
| 1008 | ++parser->index; |
| 1009 | break; |
| 1010 | } |
| 1011 | |
| 1012 | case s_req_spaces_before_url: |
| 1013 | { |
| 1014 | if (ch == ' ') break; |
| 1015 | |
| 1016 | MARK(url); |
| 1017 | if (parser->method == HTTP_CONNECT) { |
| 1018 | parser->state = s_req_server_start; |
| 1019 | } |
| 1020 | |
| 1021 | parser->state = parse_url_char((enum state)parser->state, ch); |
| 1022 | if (parser->state == s_dead) { |
| 1023 | SET_ERRNO(HPE_INVALID_URL); |
| 1024 | goto error; |
| 1025 | } |
| 1026 | |
| 1027 | break; |
| 1028 | } |
| 1029 | |
| 1030 | case s_req_schema: |
| 1031 | case s_req_schema_slash: |
| 1032 | case s_req_schema_slash_slash: |
| 1033 | case s_req_server_start: |
| 1034 | { |
| 1035 | switch (ch) { |
| 1036 | /* No whitespace allowed here */ |
| 1037 | case ' ': |
| 1038 | case CR: |
| 1039 | case LF: |
| 1040 | SET_ERRNO(HPE_INVALID_URL); |
| 1041 | goto error; |
| 1042 | default: |
| 1043 | parser->state = parse_url_char((enum state)parser->state, ch); |
| 1044 | if (parser->state == s_dead) { |
| 1045 | SET_ERRNO(HPE_INVALID_URL); |
| 1046 | goto error; |
| 1047 | } |
| 1048 | } |
| 1049 | |
| 1050 | break; |
| 1051 | } |
| 1052 | |
| 1053 | case s_req_server: |
| 1054 | case s_req_server_with_at: |
| 1055 | case s_req_path: |
| 1056 | case s_req_query_string_start: |
| 1057 | case s_req_query_string: |
| 1058 | case s_req_fragment_start: |
| 1059 | case s_req_fragment: |
| 1060 | { |
| 1061 | switch (ch) { |
| 1062 | case ' ': |
| 1063 | parser->state = s_req_http_start; |
| 1064 | CALLBACK_DATA(url); |
| 1065 | break; |
| 1066 | case CR: |
| 1067 | case LF: |
| 1068 | parser->http_major = 0; |
| 1069 | parser->http_minor = 9; |
| 1070 | parser->state = (ch == CR) ? |
| 1071 | s_req_line_almost_done : |
| 1072 | s_header_field_start; |
| 1073 | CALLBACK_DATA(url); |
| 1074 | break; |
| 1075 | default: |
| 1076 | parser->state = parse_url_char((enum state)parser->state, ch); |
| 1077 | if (parser->state == s_dead) { |
| 1078 | SET_ERRNO(HPE_INVALID_URL); |
| 1079 | goto error; |
| 1080 | } |
| 1081 | } |
| 1082 | break; |
| 1083 | } |
| 1084 | |
| 1085 | case s_req_http_start: |
| 1086 | switch (ch) { |
| 1087 | case 'H': |
| 1088 | case 'R': |
| 1089 | parser->state = s_req_http_H; |
| 1090 | break; |
| 1091 | case ' ': |
| 1092 | break; |
| 1093 | default: |
| 1094 | SET_ERRNO(HPE_INVALID_CONSTANT); |
| 1095 | goto error; |
| 1096 | } |
| 1097 | break; |
| 1098 | |
| 1099 | case s_req_http_H: |
| 1100 | STRICT_CHECK(ch != 'T'); |
| 1101 | parser->state = s_req_http_HT; |
| 1102 | break; |
| 1103 | |
| 1104 | case s_req_http_HT: |
| 1105 | STRICT_CHECK(ch != 'T'); |
| 1106 | parser->state = s_req_http_HTT; |
| 1107 | break; |
| 1108 | |
| 1109 | case s_req_http_HTT: |
| 1110 | STRICT_CHECK(ch != 'P'); |
| 1111 | parser->state = s_req_http_HTTP; |
| 1112 | break; |
| 1113 | |
| 1114 | case s_req_http_HTTP: |
| 1115 | STRICT_CHECK(ch != '/'); |
| 1116 | parser->state = s_req_first_http_major; |
| 1117 | break; |
| 1118 | |
| 1119 | /* first digit of major HTTP version */ |
| 1120 | case s_req_first_http_major: |
| 1121 | if (ch < '1' || ch > '9') { |
| 1122 | SET_ERRNO(HPE_INVALID_VERSION); |
| 1123 | goto error; |
| 1124 | } |
| 1125 | |
| 1126 | parser->http_major = ch - '0'; |
| 1127 | parser->state = s_req_http_major; |
| 1128 | break; |
| 1129 | |
| 1130 | /* major HTTP version or dot */ |
| 1131 | case s_req_http_major: |
| 1132 | { |
| 1133 | if (ch == '.') { |
| 1134 | parser->state = s_req_first_http_minor; |
| 1135 | break; |
| 1136 | } |
| 1137 | |
| 1138 | if (!IS_NUM(ch)) { |
| 1139 | SET_ERRNO(HPE_INVALID_VERSION); |
| 1140 | goto error; |
| 1141 | } |
| 1142 | |
| 1143 | parser->http_major *= 10; |
| 1144 | parser->http_major += ch - '0'; |
| 1145 | |
| 1146 | if (parser->http_major > 999) { |
| 1147 | SET_ERRNO(HPE_INVALID_VERSION); |
| 1148 | goto error; |
| 1149 | } |
| 1150 | |
| 1151 | break; |
| 1152 | } |
| 1153 | |
| 1154 | /* first digit of minor HTTP version */ |
| 1155 | case s_req_first_http_minor: |
| 1156 | if (!IS_NUM(ch)) { |
| 1157 | SET_ERRNO(HPE_INVALID_VERSION); |
| 1158 | goto error; |
| 1159 | } |
| 1160 | |
| 1161 | parser->http_minor = ch - '0'; |
| 1162 | parser->state = s_req_http_minor; |
| 1163 | break; |
| 1164 | |
| 1165 | /* minor HTTP version or end of request line */ |
| 1166 | case s_req_http_minor: |
| 1167 | { |
| 1168 | if (ch == CR) { |
| 1169 | parser->state = s_req_line_almost_done; |
| 1170 | break; |
| 1171 | } |
| 1172 | |
| 1173 | if (ch == LF) { |
| 1174 | parser->state = s_header_field_start; |
| 1175 | break; |
| 1176 | } |
| 1177 | |
| 1178 | /* XXX allow spaces after digit? */ |
| 1179 | |
| 1180 | if (!IS_NUM(ch)) { |
| 1181 | SET_ERRNO(HPE_INVALID_VERSION); |
| 1182 | goto error; |
| 1183 | } |
| 1184 | |
| 1185 | parser->http_minor *= 10; |
| 1186 | parser->http_minor += ch - '0'; |
| 1187 | |
| 1188 | if (parser->http_minor > 999) { |
| 1189 | SET_ERRNO(HPE_INVALID_VERSION); |
| 1190 | goto error; |
| 1191 | } |
| 1192 | |
| 1193 | break; |
| 1194 | } |
| 1195 | |
| 1196 | /* end of request line */ |
| 1197 | case s_req_line_almost_done: |
| 1198 | { |
| 1199 | if (ch != LF) { |
| 1200 | SET_ERRNO(HPE_LF_EXPECTED); |
| 1201 | goto error; |
| 1202 | } |
| 1203 | |
| 1204 | parser->state = s_header_field_start; |
| 1205 | break; |
| 1206 | } |
| 1207 | |
| 1208 | case s_header_field_start: |
| 1209 | { |
| 1210 | if (ch == CR) { |
| 1211 | parser->state = s_headers_almost_done; |
| 1212 | break; |
| 1213 | } |
| 1214 | |
| 1215 | if (ch == LF) { |
| 1216 | /* they might be just sending \n instead of \r\n so this would be |
| 1217 | * the second \n to denote the end of headers*/ |
| 1218 | parser->state = s_headers_almost_done; |
| 1219 | goto reexecute_byte; |
| 1220 | } |
| 1221 | |
| 1222 | c = TOKEN(ch); |
| 1223 | |
| 1224 | if (!c) { |
| 1225 | SET_ERRNO(HPE_INVALID_HEADER_TOKEN); |
| 1226 | goto error; |
| 1227 | } |
| 1228 | |
| 1229 | MARK(header_field); |
| 1230 | |
| 1231 | parser->index = 0; |
| 1232 | parser->state = s_header_field; |
| 1233 | |
| 1234 | switch (c) { |
| 1235 | case 'c': |
| 1236 | parser->header_state = h_C; |
| 1237 | break; |
| 1238 | |
| 1239 | case 'p': |
| 1240 | parser->header_state = h_matching_proxy_connection; |
| 1241 | break; |
| 1242 | |
| 1243 | case 't': |
| 1244 | parser->header_state = h_matching_transfer_encoding; |
| 1245 | break; |
| 1246 | |
| 1247 | case 'u': |
| 1248 | parser->header_state = h_matching_upgrade; |
| 1249 | break; |
| 1250 | |
| 1251 | default: |
| 1252 | parser->header_state = h_general; |
| 1253 | break; |
| 1254 | } |
| 1255 | break; |
| 1256 | } |
| 1257 | |
| 1258 | case s_header_field: |
| 1259 | { |
| 1260 | c = TOKEN(ch); |
| 1261 | |
| 1262 | if (c) { |
| 1263 | switch (parser->header_state) { |
| 1264 | case h_general: |
| 1265 | break; |
| 1266 | |
| 1267 | case h_C: |
| 1268 | parser->index++; |
| 1269 | parser->header_state = (c == 'o' ? h_CO : h_general); |
| 1270 | break; |
| 1271 | |
| 1272 | case h_CO: |
| 1273 | parser->index++; |
| 1274 | parser->header_state = (c == 'n' ? h_CON : h_general); |
| 1275 | break; |
| 1276 | |
| 1277 | case h_CON: |
| 1278 | parser->index++; |
| 1279 | switch (c) { |
| 1280 | case 'n': |
| 1281 | parser->header_state = h_matching_connection; |
| 1282 | break; |
| 1283 | case 't': |
| 1284 | parser->header_state = h_matching_content_length; |
| 1285 | break; |
| 1286 | default: |
| 1287 | parser->header_state = h_general; |
| 1288 | break; |
| 1289 | } |
| 1290 | break; |
| 1291 | |
| 1292 | /* connection */ |
| 1293 | |
| 1294 | case h_matching_connection: |
| 1295 | parser->index++; |
| 1296 | if (parser->index > sizeof(CONNECTION)-1 |
| 1297 | || c != CONNECTION[parser->index]) { |
| 1298 | parser->header_state = h_general; |
| 1299 | } else if (parser->index == sizeof(CONNECTION)-2) { |
| 1300 | parser->header_state = h_connection; |
| 1301 | } |
| 1302 | break; |
| 1303 | |
| 1304 | /* proxy-connection */ |
| 1305 | |
| 1306 | case h_matching_proxy_connection: |
| 1307 | parser->index++; |
| 1308 | if (parser->index > sizeof(PROXY_CONNECTION)-1 |
| 1309 | || c != PROXY_CONNECTION[parser->index]) { |
| 1310 | parser->header_state = h_general; |
| 1311 | } else if (parser->index == sizeof(PROXY_CONNECTION)-2) { |
| 1312 | parser->header_state = h_connection; |
| 1313 | } |
| 1314 | break; |
| 1315 | |
| 1316 | /* content-length */ |
| 1317 | |
| 1318 | case h_matching_content_length: |
| 1319 | parser->index++; |
| 1320 | if (parser->index > sizeof(CONTENT_LENGTH)-1 |
| 1321 | || c != CONTENT_LENGTH[parser->index]) { |
| 1322 | parser->header_state = h_general; |
| 1323 | } else if (parser->index == sizeof(CONTENT_LENGTH)-2) { |
| 1324 | parser->header_state = h_content_length; |
| 1325 | } |
| 1326 | break; |
| 1327 | |
| 1328 | /* transfer-encoding */ |
| 1329 | |
| 1330 | case h_matching_transfer_encoding: |
| 1331 | parser->index++; |
| 1332 | if (parser->index > sizeof(TRANSFER_ENCODING)-1 |
| 1333 | || c != TRANSFER_ENCODING[parser->index]) { |
| 1334 | parser->header_state = h_general; |
| 1335 | } else if (parser->index == sizeof(TRANSFER_ENCODING)-2) { |
| 1336 | parser->header_state = h_transfer_encoding; |
| 1337 | } |
| 1338 | break; |
| 1339 | |
| 1340 | /* upgrade */ |
| 1341 | |
| 1342 | case h_matching_upgrade: |
| 1343 | parser->index++; |
| 1344 | if (parser->index > sizeof(UPGRADE)-1 |
| 1345 | || c != UPGRADE[parser->index]) { |
| 1346 | parser->header_state = h_general; |
| 1347 | } else if (parser->index == sizeof(UPGRADE)-2) { |
| 1348 | parser->header_state = h_upgrade; |
| 1349 | } |
| 1350 | break; |
| 1351 | |
| 1352 | case h_connection: |
| 1353 | case h_content_length: |
| 1354 | case h_transfer_encoding: |
| 1355 | case h_upgrade: |
| 1356 | if (ch != ' ') parser->header_state = h_general; |
| 1357 | break; |
| 1358 | |
| 1359 | default: |
| 1360 | assert(0 && "Unknown header_state"); |
| 1361 | break; |
| 1362 | } |
| 1363 | break; |
| 1364 | } |
| 1365 | |
| 1366 | if (ch == ':') { |
| 1367 | parser->state = s_header_value_start; |
| 1368 | CALLBACK_DATA(header_field); |
| 1369 | break; |
| 1370 | } |
| 1371 | |
| 1372 | if (ch == CR) { |
| 1373 | parser->state = s_header_almost_done; |
| 1374 | CALLBACK_DATA(header_field); |
| 1375 | break; |
| 1376 | } |
| 1377 | |
| 1378 | if (ch == LF) { |
| 1379 | parser->state = s_header_field_start; |
| 1380 | CALLBACK_DATA(header_field); |
| 1381 | break; |
| 1382 | } |
| 1383 | |
| 1384 | SET_ERRNO(HPE_INVALID_HEADER_TOKEN); |
| 1385 | goto error; |
| 1386 | } |
| 1387 | |
| 1388 | case s_header_value_start: |
| 1389 | { |
| 1390 | if (ch == ' ' || ch == '\t') break; |
| 1391 | |
| 1392 | MARK(header_value); |
| 1393 | |
| 1394 | parser->state = s_header_value; |
| 1395 | parser->index = 0; |
| 1396 | |
| 1397 | if (ch == CR) { |
| 1398 | parser->header_state = h_general; |
| 1399 | parser->state = s_header_almost_done; |
| 1400 | CALLBACK_DATA(header_value); |
| 1401 | break; |
| 1402 | } |
| 1403 | |
| 1404 | if (ch == LF) { |
| 1405 | parser->state = s_header_field_start; |
| 1406 | CALLBACK_DATA(header_value); |
| 1407 | break; |
| 1408 | } |
| 1409 | |
| 1410 | c = LOWER(ch); |
| 1411 | |
| 1412 | switch (parser->header_state) { |
| 1413 | case h_upgrade: |
| 1414 | parser->flags |= F_UPGRADE; |
| 1415 | parser->header_state = h_general; |
| 1416 | break; |
| 1417 | |
| 1418 | case h_transfer_encoding: |
| 1419 | /* looking for 'Transfer-Encoding: chunked' */ |
| 1420 | if ('c' == c) { |
| 1421 | parser->header_state = h_matching_transfer_encoding_chunked; |
| 1422 | } else { |
| 1423 | parser->header_state = h_general; |
| 1424 | } |
| 1425 | break; |
| 1426 | |
| 1427 | case h_content_length: |
| 1428 | if (!IS_NUM(ch)) { |
| 1429 | SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); |
| 1430 | goto error; |
| 1431 | } |
| 1432 | |
| 1433 | parser->content_length = ch - '0'; |
| 1434 | break; |
| 1435 | |
| 1436 | case h_connection: |
| 1437 | /* looking for 'Connection: keep-alive' */ |
| 1438 | if (c == 'k') { |
| 1439 | parser->header_state = h_matching_connection_keep_alive; |
| 1440 | /* looking for 'Connection: close' */ |
| 1441 | } else if (c == 'c') { |
| 1442 | parser->header_state = h_matching_connection_close; |
| 1443 | } else { |
| 1444 | parser->header_state = h_general; |
| 1445 | } |
| 1446 | break; |
| 1447 | |
| 1448 | default: |
| 1449 | parser->header_state = h_general; |
| 1450 | break; |
| 1451 | } |
| 1452 | break; |
| 1453 | } |
| 1454 | |
| 1455 | case s_header_value: |
| 1456 | { |
| 1457 | |
| 1458 | if (ch == CR) { |
| 1459 | parser->state = s_header_almost_done; |
| 1460 | CALLBACK_DATA(header_value); |
| 1461 | break; |
| 1462 | } |
| 1463 | |
| 1464 | if (ch == LF) { |
| 1465 | parser->state = s_header_almost_done; |
| 1466 | CALLBACK_DATA_NOADVANCE(header_value); |
| 1467 | goto reexecute_byte; |
| 1468 | } |
| 1469 | |
| 1470 | c = LOWER(ch); |
| 1471 | |
| 1472 | switch (parser->header_state) { |
| 1473 | case h_general: |
| 1474 | break; |
| 1475 | |
| 1476 | case h_connection: |
| 1477 | case h_transfer_encoding: |
| 1478 | assert(0 && "Shouldn't get here."); |
| 1479 | break; |
| 1480 | |
| 1481 | case h_content_length: |
| 1482 | { |
| 1483 | uint64_t t; |
| 1484 | |
| 1485 | if (ch == ' ') break; |
| 1486 | |
| 1487 | if (!IS_NUM(ch)) { |
| 1488 | SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); |
| 1489 | goto error; |
| 1490 | } |
| 1491 | |
| 1492 | t = parser->content_length; |
| 1493 | t *= 10; |
| 1494 | t += ch - '0'; |
| 1495 | |
| 1496 | /* Overflow? */ |
| 1497 | if (t < parser->content_length || t == ULLONG_MAX) { |
| 1498 | SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); |
| 1499 | goto error; |
| 1500 | } |
| 1501 | |
| 1502 | parser->content_length = t; |
| 1503 | break; |
| 1504 | } |
| 1505 | |
| 1506 | /* Transfer-Encoding: chunked */ |
| 1507 | case h_matching_transfer_encoding_chunked: |
| 1508 | parser->index++; |
| 1509 | if (parser->index > sizeof(CHUNKED)-1 |
| 1510 | || c != CHUNKED[parser->index]) { |
| 1511 | parser->header_state = h_general; |
| 1512 | } else if (parser->index == sizeof(CHUNKED)-2) { |
| 1513 | parser->header_state = h_transfer_encoding_chunked; |
| 1514 | } |
| 1515 | break; |
| 1516 | |
| 1517 | /* looking for 'Connection: keep-alive' */ |
| 1518 | case h_matching_connection_keep_alive: |
| 1519 | parser->index++; |
| 1520 | if (parser->index > sizeof(KEEP_ALIVE)-1 |
| 1521 | || c != KEEP_ALIVE[parser->index]) { |
| 1522 | parser->header_state = h_general; |
| 1523 | } else if (parser->index == sizeof(KEEP_ALIVE)-2) { |
| 1524 | parser->header_state = h_connection_keep_alive; |
| 1525 | } |
| 1526 | break; |
| 1527 | |
| 1528 | /* looking for 'Connection: close' */ |
| 1529 | case h_matching_connection_close: |
| 1530 | parser->index++; |
| 1531 | if (parser->index > sizeof(CLOSE)-1 || c != CLOSE[parser->index]) { |
| 1532 | parser->header_state = h_general; |
| 1533 | } else if (parser->index == sizeof(CLOSE)-2) { |
| 1534 | parser->header_state = h_connection_close; |
| 1535 | } |
| 1536 | break; |
| 1537 | |
| 1538 | case h_transfer_encoding_chunked: |
| 1539 | case h_connection_keep_alive: |
| 1540 | case h_connection_close: |
| 1541 | if (ch != ' ') parser->header_state = h_general; |
| 1542 | break; |
| 1543 | |
| 1544 | default: |
| 1545 | parser->state = s_header_value; |
| 1546 | parser->header_state = h_general; |
| 1547 | break; |
| 1548 | } |
| 1549 | break; |
| 1550 | } |
| 1551 | |
| 1552 | case s_header_almost_done: |
| 1553 | { |
| 1554 | STRICT_CHECK(ch != LF); |
| 1555 | |
| 1556 | parser->state = s_header_value_lws; |
| 1557 | |
| 1558 | switch (parser->header_state) { |
| 1559 | case h_connection_keep_alive: |
| 1560 | parser->flags |= F_CONNECTION_KEEP_ALIVE; |
| 1561 | break; |
| 1562 | case h_connection_close: |
| 1563 | parser->flags |= F_CONNECTION_CLOSE; |
| 1564 | break; |
| 1565 | case h_transfer_encoding_chunked: |
| 1566 | parser->flags |= F_CHUNKED; |
| 1567 | break; |
| 1568 | default: |
| 1569 | break; |
| 1570 | } |
| 1571 | |
| 1572 | break; |
| 1573 | } |
| 1574 | |
| 1575 | case s_header_value_lws: |
| 1576 | { |
| 1577 | if (ch == ' ' || ch == '\t') |
| 1578 | parser->state = s_header_value_start; |
| 1579 | else |
| 1580 | { |
| 1581 | parser->state = s_header_field_start; |
| 1582 | goto reexecute_byte; |
| 1583 | } |
| 1584 | break; |
| 1585 | } |
| 1586 | |
| 1587 | case s_headers_almost_done: |
| 1588 | { |
| 1589 | STRICT_CHECK(ch != LF); |
| 1590 | |
| 1591 | if (parser->flags & F_TRAILING) { |
| 1592 | /* End of a chunked request */ |
| 1593 | parser->state = NEW_MESSAGE(); |
| 1594 | CALLBACK_NOTIFY(message_complete); |
| 1595 | break; |
| 1596 | } |
| 1597 | |
| 1598 | parser->state = s_headers_done; |
| 1599 | |
| 1600 | /* Set this here so that on_headers_complete() callbacks can see it */ |
| 1601 | parser->upgrade = |
| 1602 | (parser->flags & F_UPGRADE || parser->method == HTTP_CONNECT); |
| 1603 | |
| 1604 | /* Here we call the headers_complete callback. This is somewhat |
| 1605 | * different than other callbacks because if the user returns 1, we |
| 1606 | * will interpret that as saying that this message has no body. This |
| 1607 | * is needed for the annoying case of recieving a response to a HEAD |
| 1608 | * request. |
| 1609 | * |
| 1610 | * We'd like to use CALLBACK_NOTIFY_NOADVANCE() here but we cannot, so |
| 1611 | * we have to simulate it by handling a change in errno below. |
| 1612 | */ |
| 1613 | if (settings->on_headers_complete) { |
| 1614 | switch (settings->on_headers_complete(parser)) { |
| 1615 | case 0: |
| 1616 | break; |
| 1617 | |
| 1618 | case 1: |
| 1619 | parser->flags |= F_SKIPBODY; |
| 1620 | break; |
| 1621 | |
| 1622 | default: |
| 1623 | SET_ERRNO(HPE_CB_headers_complete); |
| 1624 | return p - data; /* Error */ |
| 1625 | } |
| 1626 | } |
| 1627 | |
| 1628 | if (HTTP_PARSER_ERRNO(parser) != HPE_OK) { |
| 1629 | return p - data; |
| 1630 | } |
| 1631 | |
| 1632 | goto reexecute_byte; |
| 1633 | } |
| 1634 | |
| 1635 | case s_headers_done: |
| 1636 | { |
| 1637 | STRICT_CHECK(ch != LF); |
| 1638 | |
| 1639 | parser->nread = 0; |
| 1640 | |
| 1641 | /* Exit, the rest of the connect is in a different protocol. */ |
| 1642 | if (parser->upgrade) { |
| 1643 | parser->state = NEW_MESSAGE(); |
| 1644 | CALLBACK_NOTIFY(message_complete); |
| 1645 | return (p - data) + 1; |
| 1646 | } |
| 1647 | |
| 1648 | if (parser->flags & F_SKIPBODY) { |
| 1649 | parser->state = NEW_MESSAGE(); |
| 1650 | CALLBACK_NOTIFY(message_complete); |
| 1651 | } else if (parser->flags & F_CHUNKED) { |
| 1652 | /* chunked encoding - ignore Content-Length header */ |
| 1653 | parser->state = s_chunk_size_start; |
| 1654 | } else { |
| 1655 | if (parser->content_length == 0) { |
| 1656 | /* Content-Length header given but zero: Content-Length: 0\r\n */ |
| 1657 | parser->state = NEW_MESSAGE(); |
| 1658 | CALLBACK_NOTIFY(message_complete); |
| 1659 | } else if (parser->content_length != ULLONG_MAX) { |
| 1660 | /* Content-Length header given and non-zero */ |
| 1661 | parser->state = s_body_identity; |
| 1662 | } else { |
| 1663 | if (parser->type == HTTP_REQUEST || |
| 1664 | !http_message_needs_eof(parser)) { |
| 1665 | /* Assume content-length 0 - read the next */ |
| 1666 | parser->state = NEW_MESSAGE(); |
| 1667 | CALLBACK_NOTIFY(message_complete); |
| 1668 | } else { |
| 1669 | /* Read body until EOF */ |
| 1670 | parser->state = s_body_identity_eof; |
| 1671 | } |
| 1672 | } |
| 1673 | } |
| 1674 | |
| 1675 | break; |
| 1676 | } |
| 1677 | |
| 1678 | case s_body_identity: |
| 1679 | { |
| 1680 | uint64_t to_read = MIN(parser->content_length, |
| 1681 | (uint64_t) ((data + len) - p)); |
| 1682 | |
| 1683 | assert(parser->content_length != 0 |
| 1684 | && parser->content_length != ULLONG_MAX); |
| 1685 | |
| 1686 | /* The difference between advancing content_length and p is because |
| 1687 | * the latter will automaticaly advance on the next loop iteration. |
| 1688 | * Further, if content_length ends up at 0, we want to see the last |
| 1689 | * byte again for our message complete callback. |
| 1690 | */ |
| 1691 | MARK(body); |
| 1692 | parser->content_length -= to_read; |
| 1693 | p += to_read - 1; |
| 1694 | |
| 1695 | if (parser->content_length == 0) { |
| 1696 | parser->state = s_message_done; |
| 1697 | |
| 1698 | /* Mimic CALLBACK_DATA_NOADVANCE() but with one extra byte. |
| 1699 | * |
| 1700 | * The alternative to doing this is to wait for the next byte to |
| 1701 | * trigger the data callback, just as in every other case. The |
| 1702 | * problem with this is that this makes it difficult for the test |
| 1703 | * harness to distinguish between complete-on-EOF and |
| 1704 | * complete-on-length. It's not clear that this distinction is |
| 1705 | * important for applications, but let's keep it for now. |
| 1706 | */ |
| 1707 | CALLBACK_DATA_(body, p - body_mark + 1, p - data); |
| 1708 | goto reexecute_byte; |
| 1709 | } |
| 1710 | |
| 1711 | break; |
| 1712 | } |
| 1713 | |
| 1714 | /* read until EOF */ |
| 1715 | case s_body_identity_eof: |
| 1716 | MARK(body); |
| 1717 | p = data + len - 1; |
| 1718 | |
| 1719 | break; |
| 1720 | |
| 1721 | case s_message_done: |
| 1722 | parser->state = NEW_MESSAGE(); |
| 1723 | CALLBACK_NOTIFY(message_complete); |
| 1724 | break; |
| 1725 | |
| 1726 | case s_chunk_size_start: |
| 1727 | { |
| 1728 | assert(parser->nread == 1); |
| 1729 | assert(parser->flags & F_CHUNKED); |
| 1730 | |
| 1731 | unhex_val = unhex[(unsigned char)ch]; |
| 1732 | if (unhex_val == -1) { |
| 1733 | SET_ERRNO(HPE_INVALID_CHUNK_SIZE); |
| 1734 | goto error; |
| 1735 | } |
| 1736 | |
| 1737 | parser->content_length = unhex_val; |
| 1738 | parser->state = s_chunk_size; |
| 1739 | break; |
| 1740 | } |
| 1741 | |
| 1742 | case s_chunk_size: |
| 1743 | { |
| 1744 | uint64_t t; |
| 1745 | |
| 1746 | assert(parser->flags & F_CHUNKED); |
| 1747 | |
| 1748 | if (ch == CR) { |
| 1749 | parser->state = s_chunk_size_almost_done; |
| 1750 | break; |
| 1751 | } |
| 1752 | |
| 1753 | unhex_val = unhex[(unsigned char)ch]; |
| 1754 | |
| 1755 | if (unhex_val == -1) { |
| 1756 | if (ch == ';' || ch == ' ') { |
| 1757 | parser->state = s_chunk_parameters; |
| 1758 | break; |
| 1759 | } |
| 1760 | |
| 1761 | SET_ERRNO(HPE_INVALID_CHUNK_SIZE); |
| 1762 | goto error; |
| 1763 | } |
| 1764 | |
| 1765 | t = parser->content_length; |
| 1766 | t *= 16; |
| 1767 | t += unhex_val; |
| 1768 | |
| 1769 | /* Overflow? */ |
| 1770 | if (t < parser->content_length || t == ULLONG_MAX) { |
| 1771 | SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); |
| 1772 | goto error; |
| 1773 | } |
| 1774 | |
| 1775 | parser->content_length = t; |
| 1776 | break; |
| 1777 | } |
| 1778 | |
| 1779 | case s_chunk_parameters: |
| 1780 | { |
| 1781 | assert(parser->flags & F_CHUNKED); |
| 1782 | /* just ignore this shit. TODO check for overflow */ |
| 1783 | if (ch == CR) { |
| 1784 | parser->state = s_chunk_size_almost_done; |
| 1785 | break; |
| 1786 | } |
| 1787 | break; |
| 1788 | } |
| 1789 | |
| 1790 | case s_chunk_size_almost_done: |
| 1791 | { |
| 1792 | assert(parser->flags & F_CHUNKED); |
| 1793 | STRICT_CHECK(ch != LF); |
| 1794 | |
| 1795 | parser->nread = 0; |
| 1796 | |
| 1797 | if (parser->content_length == 0) { |
| 1798 | parser->flags |= F_TRAILING; |
| 1799 | parser->state = s_header_field_start; |
| 1800 | } else { |
| 1801 | parser->state = s_chunk_data; |
| 1802 | } |
| 1803 | break; |
| 1804 | } |
| 1805 | |
| 1806 | case s_chunk_data: |
| 1807 | { |
| 1808 | uint64_t to_read = MIN(parser->content_length, |
| 1809 | (uint64_t) ((data + len) - p)); |
| 1810 | |
| 1811 | assert(parser->flags & F_CHUNKED); |
| 1812 | assert(parser->content_length != 0 |
| 1813 | && parser->content_length != ULLONG_MAX); |
| 1814 | |
| 1815 | /* See the explanation in s_body_identity for why the content |
| 1816 | * length and data pointers are managed this way. |
| 1817 | */ |
| 1818 | MARK(body); |
| 1819 | parser->content_length -= to_read; |
| 1820 | p += to_read - 1; |
| 1821 | |
| 1822 | if (parser->content_length == 0) { |
| 1823 | parser->state = s_chunk_data_almost_done; |
| 1824 | } |
| 1825 | |
| 1826 | break; |
| 1827 | } |
| 1828 | |
| 1829 | case s_chunk_data_almost_done: |
| 1830 | assert(parser->flags & F_CHUNKED); |
| 1831 | assert(parser->content_length == 0); |
| 1832 | STRICT_CHECK(ch != CR); |
| 1833 | parser->state = s_chunk_data_done; |
| 1834 | CALLBACK_DATA(body); |
| 1835 | break; |
| 1836 | |
| 1837 | case s_chunk_data_done: |
| 1838 | assert(parser->flags & F_CHUNKED); |
| 1839 | STRICT_CHECK(ch != LF); |
| 1840 | parser->nread = 0; |
| 1841 | parser->state = s_chunk_size_start; |
| 1842 | break; |
| 1843 | |
| 1844 | default: |
| 1845 | assert(0 && "unhandled state"); |
| 1846 | SET_ERRNO(HPE_INVALID_INTERNAL_STATE); |
| 1847 | goto error; |
| 1848 | } |
| 1849 | } |
| 1850 | |
| 1851 | /* Run callbacks for any marks that we have leftover after we ran our of |
| 1852 | * bytes. There should be at most one of these set, so it's OK to invoke |
| 1853 | * them in series (unset marks will not result in callbacks). |
| 1854 | * |
| 1855 | * We use the NOADVANCE() variety of callbacks here because 'p' has already |
| 1856 | * overflowed 'data' and this allows us to correct for the off-by-one that |
| 1857 | * we'd otherwise have (since CALLBACK_DATA() is meant to be run with a 'p' |
| 1858 | * value that's in-bounds). |
| 1859 | */ |
| 1860 | |
| 1861 | assert(((header_field_mark ? 1 : 0) + |
| 1862 | (header_value_mark ? 1 : 0) + |
| 1863 | (url_mark ? 1 : 0) + |
| 1864 | (body_mark ? 1 : 0)) <= 1); |
| 1865 | |
| 1866 | CALLBACK_DATA_NOADVANCE(header_field); |
| 1867 | CALLBACK_DATA_NOADVANCE(header_value); |
| 1868 | CALLBACK_DATA_NOADVANCE(url); |
| 1869 | CALLBACK_DATA_NOADVANCE(body); |
| 1870 | |
| 1871 | return len; |
| 1872 | |
| 1873 | error: |
| 1874 | if (HTTP_PARSER_ERRNO(parser) == HPE_OK) { |
| 1875 | SET_ERRNO(HPE_UNKNOWN); |
| 1876 | } |
| 1877 | |
| 1878 | return (p - data); |
| 1879 | } |
| 1880 | |
| 1881 | |
| 1882 | /* Does the parser need to see an EOF to find the end of the message? */ |
| 1883 | int |
| 1884 | http_message_needs_eof (const http_parser *parser) |
| 1885 | { |
| 1886 | if (parser->type == HTTP_REQUEST) { |
| 1887 | return 0; |
| 1888 | } |
| 1889 | |
| 1890 | /* See RFC 2616 section 4.4 */ |
| 1891 | if (parser->status_code / 100 == 1 || /* 1xx e.g. Continue */ |
| 1892 | parser->status_code == 204 || /* No Content */ |
| 1893 | parser->status_code == 304 || /* Not Modified */ |
| 1894 | parser->flags & F_SKIPBODY) { /* response to a HEAD request */ |
| 1895 | return 0; |
| 1896 | } |
| 1897 | |
| 1898 | if ((parser->flags & F_CHUNKED) || parser->content_length != ULLONG_MAX) { |
| 1899 | return 0; |
| 1900 | } |
| 1901 | |
| 1902 | return 1; |
| 1903 | } |
| 1904 | |
| 1905 | |
| 1906 | int |
| 1907 | http_should_keep_alive (const http_parser *parser) |
| 1908 | { |
| 1909 | if (parser->http_major > 0 && parser->http_minor > 0) { |
| 1910 | /* HTTP/1.1 */ |
| 1911 | if (parser->flags & F_CONNECTION_CLOSE) { |
| 1912 | return 0; |
| 1913 | } |
| 1914 | } else { |
| 1915 | /* HTTP/1.0 or earlier */ |
| 1916 | if (!(parser->flags & F_CONNECTION_KEEP_ALIVE)) { |
| 1917 | return 0; |
| 1918 | } |
| 1919 | } |
| 1920 | |
| 1921 | return !http_message_needs_eof(parser); |
| 1922 | } |
| 1923 | |
| 1924 | |
| 1925 | const char * |
| 1926 | http_method_str (enum http_method m) |
| 1927 | { |
| 1928 | return ELEM_AT(method_strings, m, "<unknown>"); |
| 1929 | } |
| 1930 | |
| 1931 | |
| 1932 | void |
| 1933 | http_parser_init (http_parser *parser, enum http_parser_type t) |
| 1934 | { |
| 1935 | void *data = parser->data; /* preserve application data */ |
| 1936 | memset(parser, 0, sizeof(*parser)); |
| 1937 | parser->data = data; |
| 1938 | parser->type = t; |
| 1939 | parser->state = (t == HTTP_REQUEST ? s_start_req : (t == HTTP_RESPONSE ? s_start_res : s_start_req_or_res)); |
| 1940 | parser->http_errno = HPE_OK; |
| 1941 | } |
| 1942 | |
| 1943 | const char * |
| 1944 | http_errno_name(enum http_errno err) { |
| 1945 | assert(err < (sizeof(http_strerror_tab)/sizeof(http_strerror_tab[0]))); |
| 1946 | return http_strerror_tab[err].name; |
| 1947 | } |
| 1948 | |
| 1949 | const char * |
| 1950 | http_errno_description(enum http_errno err) { |
| 1951 | assert(err < (sizeof(http_strerror_tab)/sizeof(http_strerror_tab[0]))); |
| 1952 | return http_strerror_tab[err].description; |
| 1953 | } |
| 1954 | |
| 1955 | static enum http_host_state |
| 1956 | http_parse_host_char(enum http_host_state s, const char ch) { |
| 1957 | switch(s) { |
| 1958 | case s_http_userinfo: |
| 1959 | case s_http_userinfo_start: |
| 1960 | if (ch == '@') { |
| 1961 | return s_http_host_start; |
| 1962 | } |
| 1963 | |
| 1964 | if (IS_USERINFO_CHAR(ch)) { |
| 1965 | return s_http_userinfo; |
| 1966 | } |
| 1967 | break; |
| 1968 | |
| 1969 | case s_http_host_start: |
| 1970 | if (ch == '[') { |
| 1971 | return s_http_host_v6_start; |
| 1972 | } |
| 1973 | |
| 1974 | if (IS_HOST_CHAR(ch)) { |
| 1975 | return s_http_host; |
| 1976 | } |
| 1977 | |
| 1978 | break; |
| 1979 | |
| 1980 | case s_http_host: |
| 1981 | if (IS_HOST_CHAR(ch)) { |
| 1982 | return s_http_host; |
| 1983 | } |
| 1984 | |
| 1985 | /* FALLTHROUGH */ |
| 1986 | case s_http_host_v6_end: |
| 1987 | if (ch == ':') { |
| 1988 | return s_http_host_port_start; |
| 1989 | } |
| 1990 | |
| 1991 | break; |
| 1992 | |
| 1993 | case s_http_host_v6: |
| 1994 | if (ch == ']') { |
| 1995 | return s_http_host_v6_end; |
| 1996 | } |
| 1997 | |
| 1998 | /* FALLTHROUGH */ |
| 1999 | case s_http_host_v6_start: |
| 2000 | if (IS_HEX(ch) || ch == ':') { |
| 2001 | return s_http_host_v6; |
| 2002 | } |
| 2003 | |
| 2004 | break; |
| 2005 | |
| 2006 | case s_http_host_port: |
| 2007 | case s_http_host_port_start: |
| 2008 | if (IS_NUM(ch)) { |
| 2009 | return s_http_host_port; |
| 2010 | } |
| 2011 | |
| 2012 | break; |
| 2013 | |
| 2014 | default: |
| 2015 | break; |
| 2016 | } |
| 2017 | return s_http_host_dead; |
| 2018 | } |
| 2019 | |
| 2020 | static int |
| 2021 | http_parse_host(const char * buf, struct http_parser_url *u, int found_at) { |
| 2022 | enum http_host_state s; |
| 2023 | |
| 2024 | const char *p; |
| 2025 | size_t buflen = u->field_data[UF_HOST].off + u->field_data[UF_HOST].len; |
| 2026 | |
| 2027 | u->field_data[UF_HOST].len = 0; |
| 2028 | |
| 2029 | s = found_at ? s_http_userinfo_start : s_http_host_start; |
| 2030 | |
| 2031 | for (p = buf + u->field_data[UF_HOST].off; p < buf + buflen; p++) { |
| 2032 | enum http_host_state new_s = http_parse_host_char(s, *p); |
| 2033 | |
| 2034 | if (new_s == s_http_host_dead) { |
| 2035 | return 1; |
| 2036 | } |
| 2037 | |
| 2038 | switch(new_s) { |
| 2039 | case s_http_host: |
| 2040 | if (s != s_http_host) { |
| 2041 | u->field_data[UF_HOST].off = p - buf; |
| 2042 | } |
| 2043 | u->field_data[UF_HOST].len++; |
| 2044 | break; |
| 2045 | |
| 2046 | case s_http_host_v6: |
| 2047 | if (s != s_http_host_v6) { |
| 2048 | u->field_data[UF_HOST].off = p - buf; |
| 2049 | } |
| 2050 | u->field_data[UF_HOST].len++; |
| 2051 | break; |
| 2052 | |
| 2053 | case s_http_host_port: |
| 2054 | if (s != s_http_host_port) { |
| 2055 | u->field_data[UF_PORT].off = p - buf; |
| 2056 | u->field_data[UF_PORT].len = 0; |
| 2057 | u->field_set |= (1 << UF_PORT); |
| 2058 | } |
| 2059 | u->field_data[UF_PORT].len++; |
| 2060 | break; |
| 2061 | |
| 2062 | case s_http_userinfo: |
| 2063 | if (s != s_http_userinfo) { |
| 2064 | u->field_data[UF_USERINFO].off = p - buf ; |
| 2065 | u->field_data[UF_USERINFO].len = 0; |
| 2066 | u->field_set |= (1 << UF_USERINFO); |
| 2067 | } |
| 2068 | u->field_data[UF_USERINFO].len++; |
| 2069 | break; |
| 2070 | |
| 2071 | default: |
| 2072 | break; |
| 2073 | } |
| 2074 | s = new_s; |
| 2075 | } |
| 2076 | |
| 2077 | /* Make sure we don't end somewhere unexpected */ |
| 2078 | switch (s) { |
| 2079 | case s_http_host_start: |
| 2080 | case s_http_host_v6_start: |
| 2081 | case s_http_host_v6: |
| 2082 | case s_http_host_port_start: |
| 2083 | case s_http_userinfo: |
| 2084 | case s_http_userinfo_start: |
| 2085 | return 1; |
| 2086 | default: |
| 2087 | break; |
| 2088 | } |
| 2089 | |
| 2090 | return 0; |
| 2091 | } |
| 2092 | |
| 2093 | int |
| 2094 | http_parser_parse_url(const char *buf, size_t buflen, int is_connect, |
| 2095 | struct http_parser_url *u) |
| 2096 | { |
| 2097 | enum state s; |
| 2098 | const char *p; |
| 2099 | enum http_parser_url_fields uf, old_uf; |
| 2100 | int found_at = 0; |
| 2101 | |
| 2102 | u->port = u->field_set = 0; |
| 2103 | s = is_connect ? s_req_server_start : s_req_spaces_before_url; |
| 2104 | uf = old_uf = UF_MAX; |
| 2105 | |
| 2106 | for (p = buf; p < buf + buflen; p++) { |
| 2107 | s = parse_url_char(s, *p); |
| 2108 | |
| 2109 | /* Figure out the next field that we're operating on */ |
| 2110 | switch (s) { |
| 2111 | case s_dead: |
| 2112 | return 1; |
| 2113 | |
| 2114 | /* Skip delimeters */ |
| 2115 | case s_req_schema_slash: |
| 2116 | case s_req_schema_slash_slash: |
| 2117 | case s_req_server_start: |
| 2118 | case s_req_query_string_start: |
| 2119 | case s_req_fragment_start: |
| 2120 | continue; |
| 2121 | |
| 2122 | case s_req_schema: |
| 2123 | uf = UF_SCHEMA; |
| 2124 | break; |
| 2125 | |
| 2126 | case s_req_server_with_at: |
| 2127 | found_at = 1; |
| 2128 | |
| 2129 | /* FALLTROUGH */ |
| 2130 | case s_req_server: |
| 2131 | uf = UF_HOST; |
| 2132 | break; |
| 2133 | |
| 2134 | case s_req_path: |
| 2135 | uf = UF_PATH; |
| 2136 | break; |
| 2137 | |
| 2138 | case s_req_query_string: |
| 2139 | uf = UF_QUERY; |
| 2140 | break; |
| 2141 | |
| 2142 | case s_req_fragment: |
| 2143 | uf = UF_FRAGMENT; |
| 2144 | break; |
| 2145 | |
| 2146 | default: |
| 2147 | assert(!"Unexpected state"); |
| 2148 | return 1; |
| 2149 | } |
| 2150 | |
| 2151 | /* Nothing's changed; soldier on */ |
| 2152 | if (uf == old_uf) { |
| 2153 | u->field_data[uf].len++; |
| 2154 | continue; |
| 2155 | } |
| 2156 | |
| 2157 | u->field_data[uf].off = p - buf; |
| 2158 | u->field_data[uf].len = 1; |
| 2159 | |
| 2160 | u->field_set |= (1 << uf); |
| 2161 | old_uf = uf; |
| 2162 | } |
| 2163 | |
| 2164 | /* host must be present if there is a schema */ |
| 2165 | /* parsing http:///toto will fail */ |
| 2166 | if ((u->field_set & ((1 << UF_SCHEMA) | (1 << UF_HOST))) != 0) { |
| 2167 | if (http_parse_host(buf, u, found_at) != 0) { |
| 2168 | return 1; |
| 2169 | } |
| 2170 | } |
| 2171 | |
| 2172 | /* CONNECT requests can only contain "hostname:port" */ |
| 2173 | if (is_connect && u->field_set != ((1 << UF_HOST)|(1 << UF_PORT))) { |
| 2174 | return 1; |
| 2175 | } |
| 2176 | |
| 2177 | if (u->field_set & (1 << UF_PORT)) { |
| 2178 | /* Don't bother with endp; we've already validated the string */ |
| 2179 | unsigned long v = strtoul(buf + u->field_data[UF_PORT].off, NULL, 10); |
| 2180 | |
| 2181 | /* Ports have a max value of 2^16 */ |
| 2182 | if (v > 0xffff) { |
| 2183 | return 1; |
| 2184 | } |
| 2185 | |
| 2186 | u->port = (uint16_t) v; |
| 2187 | } |
| 2188 | |
| 2189 | return 0; |
| 2190 | } |
| 2191 | |
| 2192 | void |
| 2193 | http_parser_pause(http_parser *parser, int paused) { |
| 2194 | /* Users should only be pausing/unpausing a parser that is not in an error |
| 2195 | * state. In non-debug builds, there's not much that we can do about this |
| 2196 | * other than ignore it. |
| 2197 | */ |
| 2198 | if (HTTP_PARSER_ERRNO(parser) == HPE_OK || |
| 2199 | HTTP_PARSER_ERRNO(parser) == HPE_PAUSED) { |
| 2200 | SET_ERRNO((paused) ? HPE_PAUSED : HPE_OK); |
| 2201 | } else { |
| 2202 | assert(0 && "Attempting to pause parser in error state"); |
| 2203 | } |
| 2204 | } |
| 2205 | |
| 2206 | int |
| 2207 | http_body_is_final(const struct http_parser *parser) { |
| 2208 | return parser->state == s_message_done; |
| 2209 | } |