blob: ae315a82879675f55ae62b347439c910de6b6a1e [file] [log] [blame]
Willy Tarreau35b51c62018-09-10 15:38:55 +02001/*
2 * HTTP semantics
3 *
4 * Copyright 2000-2018 Willy Tarreau <w@1wt.eu>
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version
9 * 2 of the License, or (at your option) any later version.
10 *
11 */
12
13#include <ctype.h>
Willy Tarreau4c7e4b72020-05-27 12:58:42 +020014#include <haproxy/api.h>
Willy Tarreaucd72d8c2020-06-02 19:11:26 +020015#include <haproxy/http.h>
Willy Tarreau04f1e2d2018-09-10 18:04:24 +020016#include <common/standard.h>
Willy Tarreau35b51c62018-09-10 15:38:55 +020017
18/* It is about twice as fast on recent architectures to lookup a byte in a
19 * table than to perform a boolean AND or OR between two tests. Refer to
20 * RFC2616/RFC5234/RFC7230 for those chars. A token is any ASCII char that is
21 * neither a separator nor a CTL char. An http ver_token is any ASCII which can
22 * be found in an HTTP version, which includes 'H', 'T', 'P', '/', '.' and any
23 * digit. Note: please do not overwrite values in assignment since gcc-2.95
24 * will not handle them correctly. It's worth noting that chars 128..255 are
25 * nothing, not even control chars.
26 */
27const unsigned char http_char_classes[256] = {
28 [ 0] = HTTP_FLG_CTL,
29 [ 1] = HTTP_FLG_CTL,
30 [ 2] = HTTP_FLG_CTL,
31 [ 3] = HTTP_FLG_CTL,
32 [ 4] = HTTP_FLG_CTL,
33 [ 5] = HTTP_FLG_CTL,
34 [ 6] = HTTP_FLG_CTL,
35 [ 7] = HTTP_FLG_CTL,
36 [ 8] = HTTP_FLG_CTL,
37 [ 9] = HTTP_FLG_SPHT | HTTP_FLG_LWS | HTTP_FLG_SEP | HTTP_FLG_CTL,
38 [ 10] = HTTP_FLG_CRLF | HTTP_FLG_LWS | HTTP_FLG_CTL,
39 [ 11] = HTTP_FLG_CTL,
40 [ 12] = HTTP_FLG_CTL,
41 [ 13] = HTTP_FLG_CRLF | HTTP_FLG_LWS | HTTP_FLG_CTL,
42 [ 14] = HTTP_FLG_CTL,
43 [ 15] = HTTP_FLG_CTL,
44 [ 16] = HTTP_FLG_CTL,
45 [ 17] = HTTP_FLG_CTL,
46 [ 18] = HTTP_FLG_CTL,
47 [ 19] = HTTP_FLG_CTL,
48 [ 20] = HTTP_FLG_CTL,
49 [ 21] = HTTP_FLG_CTL,
50 [ 22] = HTTP_FLG_CTL,
51 [ 23] = HTTP_FLG_CTL,
52 [ 24] = HTTP_FLG_CTL,
53 [ 25] = HTTP_FLG_CTL,
54 [ 26] = HTTP_FLG_CTL,
55 [ 27] = HTTP_FLG_CTL,
56 [ 28] = HTTP_FLG_CTL,
57 [ 29] = HTTP_FLG_CTL,
58 [ 30] = HTTP_FLG_CTL,
59 [ 31] = HTTP_FLG_CTL,
60 [' '] = HTTP_FLG_SPHT | HTTP_FLG_LWS | HTTP_FLG_SEP,
61 ['!'] = HTTP_FLG_TOK,
62 ['"'] = HTTP_FLG_SEP,
63 ['#'] = HTTP_FLG_TOK,
64 ['$'] = HTTP_FLG_TOK,
65 ['%'] = HTTP_FLG_TOK,
66 ['&'] = HTTP_FLG_TOK,
67 [ 39] = HTTP_FLG_TOK,
68 ['('] = HTTP_FLG_SEP,
69 [')'] = HTTP_FLG_SEP,
70 ['*'] = HTTP_FLG_TOK,
71 ['+'] = HTTP_FLG_TOK,
72 [','] = HTTP_FLG_SEP,
73 ['-'] = HTTP_FLG_TOK,
74 ['.'] = HTTP_FLG_TOK | HTTP_FLG_VER,
75 ['/'] = HTTP_FLG_SEP | HTTP_FLG_VER,
76 ['0'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
77 ['1'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
78 ['2'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
79 ['3'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
80 ['4'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
81 ['5'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
82 ['6'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
83 ['7'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
84 ['8'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
85 ['9'] = HTTP_FLG_TOK | HTTP_FLG_VER | HTTP_FLG_DIG,
86 [':'] = HTTP_FLG_SEP,
87 [';'] = HTTP_FLG_SEP,
88 ['<'] = HTTP_FLG_SEP,
89 ['='] = HTTP_FLG_SEP,
90 ['>'] = HTTP_FLG_SEP,
91 ['?'] = HTTP_FLG_SEP,
92 ['@'] = HTTP_FLG_SEP,
93 ['A'] = HTTP_FLG_TOK,
94 ['B'] = HTTP_FLG_TOK,
95 ['C'] = HTTP_FLG_TOK,
96 ['D'] = HTTP_FLG_TOK,
97 ['E'] = HTTP_FLG_TOK,
98 ['F'] = HTTP_FLG_TOK,
99 ['G'] = HTTP_FLG_TOK,
100 ['H'] = HTTP_FLG_TOK | HTTP_FLG_VER,
101 ['I'] = HTTP_FLG_TOK,
102 ['J'] = HTTP_FLG_TOK,
103 ['K'] = HTTP_FLG_TOK,
104 ['L'] = HTTP_FLG_TOK,
105 ['M'] = HTTP_FLG_TOK,
106 ['N'] = HTTP_FLG_TOK,
107 ['O'] = HTTP_FLG_TOK,
108 ['P'] = HTTP_FLG_TOK | HTTP_FLG_VER,
109 ['Q'] = HTTP_FLG_TOK,
110 ['R'] = HTTP_FLG_TOK | HTTP_FLG_VER,
111 ['S'] = HTTP_FLG_TOK | HTTP_FLG_VER,
112 ['T'] = HTTP_FLG_TOK | HTTP_FLG_VER,
113 ['U'] = HTTP_FLG_TOK,
114 ['V'] = HTTP_FLG_TOK,
115 ['W'] = HTTP_FLG_TOK,
116 ['X'] = HTTP_FLG_TOK,
117 ['Y'] = HTTP_FLG_TOK,
118 ['Z'] = HTTP_FLG_TOK,
119 ['['] = HTTP_FLG_SEP,
120 [ 92] = HTTP_FLG_SEP,
121 [']'] = HTTP_FLG_SEP,
122 ['^'] = HTTP_FLG_TOK,
123 ['_'] = HTTP_FLG_TOK,
124 ['`'] = HTTP_FLG_TOK,
125 ['a'] = HTTP_FLG_TOK,
126 ['b'] = HTTP_FLG_TOK,
127 ['c'] = HTTP_FLG_TOK,
128 ['d'] = HTTP_FLG_TOK,
129 ['e'] = HTTP_FLG_TOK,
130 ['f'] = HTTP_FLG_TOK,
131 ['g'] = HTTP_FLG_TOK,
132 ['h'] = HTTP_FLG_TOK,
133 ['i'] = HTTP_FLG_TOK,
134 ['j'] = HTTP_FLG_TOK,
135 ['k'] = HTTP_FLG_TOK,
136 ['l'] = HTTP_FLG_TOK,
137 ['m'] = HTTP_FLG_TOK,
138 ['n'] = HTTP_FLG_TOK,
139 ['o'] = HTTP_FLG_TOK,
140 ['p'] = HTTP_FLG_TOK,
141 ['q'] = HTTP_FLG_TOK,
142 ['r'] = HTTP_FLG_TOK,
143 ['s'] = HTTP_FLG_TOK,
144 ['t'] = HTTP_FLG_TOK,
145 ['u'] = HTTP_FLG_TOK,
146 ['v'] = HTTP_FLG_TOK,
147 ['w'] = HTTP_FLG_TOK,
148 ['x'] = HTTP_FLG_TOK,
149 ['y'] = HTTP_FLG_TOK,
150 ['z'] = HTTP_FLG_TOK,
151 ['{'] = HTTP_FLG_SEP,
152 ['|'] = HTTP_FLG_TOK,
153 ['}'] = HTTP_FLG_SEP,
154 ['~'] = HTTP_FLG_TOK,
155 [127] = HTTP_FLG_CTL,
156};
157
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200158const int http_err_codes[HTTP_ERR_SIZE] = {
159 [HTTP_ERR_200] = 200, /* used by "monitor-uri" */
160 [HTTP_ERR_400] = 400,
Christopher Faulet612f2ea2020-05-27 09:57:28 +0200161 [HTTP_ERR_401] = 401,
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200162 [HTTP_ERR_403] = 403,
Florian Tham9205fea2020-01-08 13:35:30 +0100163 [HTTP_ERR_404] = 404,
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200164 [HTTP_ERR_405] = 405,
Christopher Faulet612f2ea2020-05-27 09:57:28 +0200165 [HTTP_ERR_407] = 407,
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200166 [HTTP_ERR_408] = 408,
Florian Tham272e29b2020-01-08 10:19:05 +0100167 [HTTP_ERR_410] = 410,
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200168 [HTTP_ERR_421] = 421,
169 [HTTP_ERR_425] = 425,
170 [HTTP_ERR_429] = 429,
171 [HTTP_ERR_500] = 500,
172 [HTTP_ERR_502] = 502,
173 [HTTP_ERR_503] = 503,
174 [HTTP_ERR_504] = 504,
175};
176
Christopher Fauleta7b677c2018-11-29 16:48:49 +0100177const char *http_err_msgs[HTTP_ERR_SIZE] = {
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200178 [HTTP_ERR_200] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200179 "HTTP/1.1 200 OK\r\n"
180 "Content-length: 58\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200181 "Cache-Control: no-cache\r\n"
182 "Connection: close\r\n"
183 "Content-Type: text/html\r\n"
184 "\r\n"
185 "<html><body><h1>200 OK</h1>\nService ready.\n</body></html>\n",
186
187 [HTTP_ERR_400] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200188 "HTTP/1.1 400 Bad request\r\n"
189 "Content-length: 90\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200190 "Cache-Control: no-cache\r\n"
191 "Connection: close\r\n"
192 "Content-Type: text/html\r\n"
193 "\r\n"
194 "<html><body><h1>400 Bad request</h1>\nYour browser sent an invalid request.\n</body></html>\n",
195
Christopher Faulet612f2ea2020-05-27 09:57:28 +0200196 [HTTP_ERR_401] =
197 "HTTP/1.1 401 Unauthorized\r\n"
198 "Content-length: 112\r\n"
199 "Cache-Control: no-cache\r\n"
200 "Connection: close\r\n"
201 "Content-Type: text/html\r\n"
202 "\r\n"
203 "<html><body><h1>401 Unauthorized</h1>\nYou need a valid user and password to access this content.\n</body></html>\n",
204
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200205 [HTTP_ERR_403] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200206 "HTTP/1.1 403 Forbidden\r\n"
207 "Content-length: 93\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200208 "Cache-Control: no-cache\r\n"
209 "Connection: close\r\n"
210 "Content-Type: text/html\r\n"
211 "\r\n"
212 "<html><body><h1>403 Forbidden</h1>\nRequest forbidden by administrative rules.\n</body></html>\n",
213
Florian Tham9205fea2020-01-08 13:35:30 +0100214 [HTTP_ERR_404] =
215 "HTTP/1.1 404 Not Found\r\n"
216 "Content-length: 83\r\n"
217 "Cache-Control: no-cache\r\n"
218 "Connection: close\r\n"
219 "Content-Type: text/html\r\n"
220 "\r\n"
221 "<html><body><h1>404 Not Found</h1>\nThe resource could not be found.\n</body></html>\n",
222
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200223 [HTTP_ERR_405] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200224 "HTTP/1.1 405 Method Not Allowed\r\n"
225 "Content-length: 146\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200226 "Cache-Control: no-cache\r\n"
227 "Connection: close\r\n"
228 "Content-Type: text/html\r\n"
229 "\r\n"
230 "<html><body><h1>405 Method Not Allowed</h1>\nA request was made of a resource using a request method not supported by that resource\n</body></html>\n",
231
Christopher Faulet612f2ea2020-05-27 09:57:28 +0200232 [HTTP_ERR_407] =
233 "HTTP/1.1 407 Unauthorized\r\n"
234 "Content-length: 112\r\n"
235 "Cache-Control: no-cache\r\n"
236 "Connection: close\r\n"
237 "Content-Type: text/html\r\n"
238 "\r\n"
239 "<html><body><h1>407 Unauthorized</h1>\nYou need a valid user and password to access this content.\n</body></html>\n",
240
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200241 [HTTP_ERR_408] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200242 "HTTP/1.1 408 Request Time-out\r\n"
243 "Content-length: 110\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200244 "Cache-Control: no-cache\r\n"
245 "Connection: close\r\n"
246 "Content-Type: text/html\r\n"
247 "\r\n"
248 "<html><body><h1>408 Request Time-out</h1>\nYour browser didn't send a complete request in time.\n</body></html>\n",
249
Florian Tham272e29b2020-01-08 10:19:05 +0100250 [HTTP_ERR_410] =
251 "HTTP/1.1 410 Gone\r\n"
252 "Content-length: 114\r\n"
253 "Cache-Control: no-cache\r\n"
254 "Connection: close\r\n"
255 "Content-Type: text/html\r\n"
256 "\r\n"
257 "<html><body><h1>410 Gone</h1>\nThe resource is no longer available and will not be available again.\n</body></html>\n",
258
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200259 [HTTP_ERR_421] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200260 "HTTP/1.1 421 Misdirected Request\r\n"
261 "Content-length: 104\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200262 "Cache-Control: no-cache\r\n"
263 "Connection: close\r\n"
264 "Content-Type: text/html\r\n"
265 "\r\n"
266 "<html><body><h1>421 Misdirected Request</h1>\nRequest sent to a non-authoritative server.\n</body></html>\n",
267
268 [HTTP_ERR_425] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200269 "HTTP/1.1 425 Too Early\r\n"
270 "Content-length: 80\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200271 "Cache-Control: no-cache\r\n"
272 "Connection: close\r\n"
273 "Content-Type: text/html\r\n"
274 "\r\n"
275 "<html><body><h1>425 Too Early</h1>\nYour browser sent early data.\n</body></html>\n",
276
277 [HTTP_ERR_429] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200278 "HTTP/1.1 429 Too Many Requests\r\n"
279 "Content-length: 117\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200280 "Cache-Control: no-cache\r\n"
281 "Connection: close\r\n"
282 "Content-Type: text/html\r\n"
283 "\r\n"
284 "<html><body><h1>429 Too Many Requests</h1>\nYou have sent too many requests in a given amount of time.\n</body></html>\n",
285
286 [HTTP_ERR_500] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200287 "HTTP/1.1 500 Internal Server Error\r\n"
288 "Content-length: 96\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200289 "Cache-Control: no-cache\r\n"
290 "Connection: close\r\n"
291 "Content-Type: text/html\r\n"
292 "\r\n"
293 "<html><body><h1>500 Internal Server Error</h1>\nAn internal server error occured.\n</body></html>\n",
294
295 [HTTP_ERR_502] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200296 "HTTP/1.1 502 Bad Gateway\r\n"
297 "Content-length: 107\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200298 "Cache-Control: no-cache\r\n"
299 "Connection: close\r\n"
300 "Content-Type: text/html\r\n"
301 "\r\n"
302 "<html><body><h1>502 Bad Gateway</h1>\nThe server returned an invalid or incomplete response.\n</body></html>\n",
303
304 [HTTP_ERR_503] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200305 "HTTP/1.1 503 Service Unavailable\r\n"
306 "Content-length: 107\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200307 "Cache-Control: no-cache\r\n"
308 "Connection: close\r\n"
309 "Content-Type: text/html\r\n"
310 "\r\n"
311 "<html><body><h1>503 Service Unavailable</h1>\nNo server is available to handle this request.\n</body></html>\n",
312
313 [HTTP_ERR_504] =
Willy Tarreaub5ba2b02019-06-11 16:08:25 +0200314 "HTTP/1.1 504 Gateway Time-out\r\n"
315 "Content-length: 92\r\n"
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200316 "Cache-Control: no-cache\r\n"
317 "Connection: close\r\n"
318 "Content-Type: text/html\r\n"
319 "\r\n"
320 "<html><body><h1>504 Gateway Time-out</h1>\nThe server didn't respond in time.\n</body></html>\n",
321
322};
323
Willy Tarreau35b51c62018-09-10 15:38:55 +0200324const struct ist http_known_methods[HTTP_METH_OTHER] = {
325 [HTTP_METH_OPTIONS] = IST("OPTIONS"),
326 [HTTP_METH_GET] = IST("GET"),
327 [HTTP_METH_HEAD] = IST("HEAD"),
328 [HTTP_METH_POST] = IST("POST"),
329 [HTTP_METH_PUT] = IST("PUT"),
330 [HTTP_METH_DELETE] = IST("DELETE"),
331 [HTTP_METH_TRACE] = IST("TRACE"),
332 [HTTP_METH_CONNECT] = IST("CONNECT"),
333};
334
335/*
336 * returns a known method among HTTP_METH_* or HTTP_METH_OTHER for all unknown
337 * ones.
338 */
339enum http_meth_t find_http_meth(const char *str, const int len)
340{
341 const struct ist m = ist2(str, len);
342
343 if (isteq(m, ist("GET"))) return HTTP_METH_GET;
344 else if (isteq(m, ist("HEAD"))) return HTTP_METH_HEAD;
345 else if (isteq(m, ist("POST"))) return HTTP_METH_POST;
346 else if (isteq(m, ist("CONNECT"))) return HTTP_METH_CONNECT;
347 else if (isteq(m, ist("PUT"))) return HTTP_METH_PUT;
348 else if (isteq(m, ist("OPTIONS"))) return HTTP_METH_OPTIONS;
349 else if (isteq(m, ist("DELETE"))) return HTTP_METH_DELETE;
350 else if (isteq(m, ist("TRACE"))) return HTTP_METH_TRACE;
351 else return HTTP_METH_OTHER;
352}
Willy Tarreau6b952c82018-09-10 17:45:34 +0200353
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200354/* This function returns HTTP_ERR_<num> (enum) matching http status code.
355 * Returned value should match codes from http_err_codes.
356 */
Willy Tarreau8de1df92019-04-15 21:27:18 +0200357int http_get_status_idx(unsigned int status)
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200358{
359 switch (status) {
360 case 200: return HTTP_ERR_200;
361 case 400: return HTTP_ERR_400;
Christopher Faulet612f2ea2020-05-27 09:57:28 +0200362 case 401: return HTTP_ERR_401;
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200363 case 403: return HTTP_ERR_403;
Florian Tham9205fea2020-01-08 13:35:30 +0100364 case 404: return HTTP_ERR_404;
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200365 case 405: return HTTP_ERR_405;
Christopher Faulet612f2ea2020-05-27 09:57:28 +0200366 case 407: return HTTP_ERR_407;
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200367 case 408: return HTTP_ERR_408;
Florian Tham272e29b2020-01-08 10:19:05 +0100368 case 410: return HTTP_ERR_410;
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200369 case 421: return HTTP_ERR_421;
370 case 425: return HTTP_ERR_425;
371 case 429: return HTTP_ERR_429;
372 case 500: return HTTP_ERR_500;
373 case 502: return HTTP_ERR_502;
374 case 503: return HTTP_ERR_503;
375 case 504: return HTTP_ERR_504;
376 default: return HTTP_ERR_500;
377 }
378}
379
380/* This function returns a reason associated with the HTTP status.
381 * This function never fails, a message is always returned.
382 */
383const char *http_get_reason(unsigned int status)
384{
385 switch (status) {
386 case 100: return "Continue";
387 case 101: return "Switching Protocols";
388 case 102: return "Processing";
389 case 200: return "OK";
390 case 201: return "Created";
391 case 202: return "Accepted";
392 case 203: return "Non-Authoritative Information";
393 case 204: return "No Content";
394 case 205: return "Reset Content";
395 case 206: return "Partial Content";
396 case 207: return "Multi-Status";
397 case 210: return "Content Different";
398 case 226: return "IM Used";
399 case 300: return "Multiple Choices";
400 case 301: return "Moved Permanently";
401 case 302: return "Moved Temporarily";
402 case 303: return "See Other";
403 case 304: return "Not Modified";
404 case 305: return "Use Proxy";
405 case 307: return "Temporary Redirect";
406 case 308: return "Permanent Redirect";
407 case 310: return "Too many Redirects";
408 case 400: return "Bad Request";
409 case 401: return "Unauthorized";
410 case 402: return "Payment Required";
411 case 403: return "Forbidden";
412 case 404: return "Not Found";
413 case 405: return "Method Not Allowed";
414 case 406: return "Not Acceptable";
415 case 407: return "Proxy Authentication Required";
416 case 408: return "Request Time-out";
417 case 409: return "Conflict";
418 case 410: return "Gone";
419 case 411: return "Length Required";
420 case 412: return "Precondition Failed";
421 case 413: return "Request Entity Too Large";
422 case 414: return "Request-URI Too Long";
423 case 415: return "Unsupported Media Type";
424 case 416: return "Requested range unsatisfiable";
425 case 417: return "Expectation failed";
426 case 418: return "I'm a teapot";
427 case 421: return "Misdirected Request";
428 case 422: return "Unprocessable entity";
429 case 423: return "Locked";
430 case 424: return "Method failure";
431 case 425: return "Too Early";
432 case 426: return "Upgrade Required";
433 case 428: return "Precondition Required";
434 case 429: return "Too Many Requests";
435 case 431: return "Request Header Fields Too Large";
436 case 449: return "Retry With";
437 case 450: return "Blocked by Windows Parental Controls";
438 case 451: return "Unavailable For Legal Reasons";
439 case 456: return "Unrecoverable Error";
440 case 499: return "client has closed connection";
441 case 500: return "Internal Server Error";
442 case 501: return "Not Implemented";
443 case 502: return "Bad Gateway or Proxy Error";
444 case 503: return "Service Unavailable";
445 case 504: return "Gateway Time-out";
446 case 505: return "HTTP Version not supported";
447 case 506: return "Variant also negociate";
448 case 507: return "Insufficient storage";
449 case 508: return "Loop detected";
450 case 509: return "Bandwidth Limit Exceeded";
451 case 510: return "Not extended";
452 case 511: return "Network authentication required";
453 case 520: return "Web server is returning an unknown error";
454 default:
455 switch (status) {
456 case 100 ... 199: return "Informational";
457 case 200 ... 299: return "Success";
458 case 300 ... 399: return "Redirection";
459 case 400 ... 499: return "Client Error";
460 case 500 ... 599: return "Server Error";
461 default: return "Other";
462 }
463 }
464}
465
Christopher Faulet16fdc552019-10-08 14:56:58 +0200466/* Parse the uri and looks for the authority, between the scheme and the
467 * path. if no_userinfo is not zero, the part before the '@' (including it) is
468 * skipped. If not found, an empty ist is returned. Otherwise, the ist pointing
469 * on the authority is returned.
470 */
471struct ist http_get_authority(const struct ist uri, int no_userinfo)
472{
473 const char *ptr, *start, *end;
474
475 if (!uri.len)
476 goto not_found;
477
478 ptr = uri.ptr;
479 start = ptr;
480 end = ptr + uri.len;
481
482 /* RFC7230, par. 2.7 :
483 * Request-URI = "*" | absuri | abspath | authority
484 */
485
486 if (*ptr == '*' || *ptr == '/')
487 goto not_found;
488
489 if (isalpha((unsigned char)*ptr)) {
490 /* this is a scheme as described by RFC3986, par. 3.1, or only
491 * an authority (in case of a CONNECT method).
492 */
493 ptr++;
494 while (ptr < end &&
495 (isalnum((unsigned char)*ptr) || *ptr == '+' || *ptr == '-' || *ptr == '.'))
496 ptr++;
497 /* skip '://' or take the whole as authority if not found */
498 if (ptr == end || *ptr++ != ':')
499 goto authority;
500 if (ptr == end || *ptr++ != '/')
501 goto authority;
502 if (ptr == end || *ptr++ != '/')
503 goto authority;
504 }
505
506 start = ptr;
507 while (ptr < end && *ptr != '/') {
508 if (*ptr++ == '@' && no_userinfo)
509 start = ptr;
510 }
511
512 /* OK, ptr point on the '/' or the end */
513 end = ptr;
514
515 authority:
516 return ist2(start, end - start);
517
518 not_found:
Tim Duesterhus241e29e2020-03-05 17:56:30 +0100519 return IST_NULL;
Christopher Faulet16fdc552019-10-08 14:56:58 +0200520}
521
Willy Tarreau6b952c82018-09-10 17:45:34 +0200522/* Parse the URI from the given transaction (which is assumed to be in request
523 * phase) and look for the "/" beginning the PATH. If not found, ist2(0,0) is
524 * returned. Otherwise the pointer and length are returned.
525 */
526struct ist http_get_path(const struct ist uri)
527{
528 const char *ptr, *end;
529
530 if (!uri.len)
531 goto not_found;
532
533 ptr = uri.ptr;
534 end = ptr + uri.len;
535
536 /* RFC7230, par. 2.7 :
537 * Request-URI = "*" | absuri | abspath | authority
538 */
539
540 if (*ptr == '*')
541 goto not_found;
542
543 if (isalpha((unsigned char)*ptr)) {
544 /* this is a scheme as described by RFC3986, par. 3.1 */
545 ptr++;
546 while (ptr < end &&
547 (isalnum((unsigned char)*ptr) || *ptr == '+' || *ptr == '-' || *ptr == '.'))
548 ptr++;
549 /* skip '://' */
550 if (ptr == end || *ptr++ != ':')
551 goto not_found;
552 if (ptr == end || *ptr++ != '/')
553 goto not_found;
554 if (ptr == end || *ptr++ != '/')
555 goto not_found;
556 }
557 /* skip [user[:passwd]@]host[:[port]] */
558
559 while (ptr < end && *ptr != '/')
560 ptr++;
561
562 if (ptr == end)
563 goto not_found;
564
565 /* OK, we got the '/' ! */
566 return ist2(ptr, end - ptr);
567
568 not_found:
Tim Duesterhus241e29e2020-03-05 17:56:30 +0100569 return IST_NULL;
Willy Tarreau6b952c82018-09-10 17:45:34 +0200570}
Willy Tarreau04f1e2d2018-09-10 18:04:24 +0200571
Willy Tarreauab813a42018-09-10 18:41:28 +0200572/*
573 * Checks if <hdr> is exactly <name> for <len> chars, and ends with a colon.
574 * If so, returns the position of the first non-space character relative to
575 * <hdr>, or <end>-<hdr> if not found before. If no value is found, it tries
576 * to return a pointer to the place after the first space. Returns 0 if the
577 * header name does not match. Checks are case-insensitive.
578 */
579int http_header_match2(const char *hdr, const char *end,
580 const char *name, int len)
581{
582 const char *val;
583
584 if (hdr + len >= end)
585 return 0;
586 if (hdr[len] != ':')
587 return 0;
588 if (strncasecmp(hdr, name, len) != 0)
589 return 0;
590 val = hdr + len + 1;
591 while (val < end && HTTP_IS_SPHT(*val))
592 val++;
593 if ((val >= end) && (len + 2 <= end - hdr))
594 return len + 2; /* we may replace starting from second space */
595 return val - hdr;
596}
597
598/* Find the end of the header value contained between <s> and <e>. See RFC7230,
599 * par 3.2 for more information. Note that it requires a valid header to return
600 * a valid result. This works for headers defined as comma-separated lists.
601 */
602char *http_find_hdr_value_end(char *s, const char *e)
603{
604 int quoted, qdpair;
605
606 quoted = qdpair = 0;
607
Willy Tarreau02ac9502020-02-21 16:31:22 +0100608#ifdef HA_UNALIGNED_LE
Willy Tarreauab813a42018-09-10 18:41:28 +0200609 /* speedup: skip everything not a comma nor a double quote */
610 for (; s <= e - sizeof(int); s += sizeof(int)) {
611 unsigned int c = *(int *)s; // comma
612 unsigned int q = c; // quote
613
614 c ^= 0x2c2c2c2c; // contains one zero on a comma
615 q ^= 0x22222222; // contains one zero on a quote
616
617 c = (c - 0x01010101) & ~c; // contains 0x80 below a comma
618 q = (q - 0x01010101) & ~q; // contains 0x80 below a quote
619
620 if ((c | q) & 0x80808080)
621 break; // found a comma or a quote
622 }
623#endif
624 for (; s < e; s++) {
625 if (qdpair) qdpair = 0;
626 else if (quoted) {
627 if (*s == '\\') qdpair = 1;
628 else if (*s == '"') quoted = 0;
629 }
630 else if (*s == '"') quoted = 1;
631 else if (*s == ',') return s;
632 }
633 return s;
634}
635
636/* Find the end of a cookie value contained between <s> and <e>. It works the
637 * same way as with headers above except that the semi-colon also ends a token.
638 * See RFC2965 for more information. Note that it requires a valid header to
639 * return a valid result.
640 */
641char *http_find_cookie_value_end(char *s, const char *e)
642{
643 int quoted, qdpair;
644
645 quoted = qdpair = 0;
646 for (; s < e; s++) {
647 if (qdpair) qdpair = 0;
648 else if (quoted) {
649 if (*s == '\\') qdpair = 1;
650 else if (*s == '"') quoted = 0;
651 }
652 else if (*s == '"') quoted = 1;
653 else if (*s == ',' || *s == ';') return s;
654 }
655 return s;
656}
657
658/* Try to find the next occurrence of a cookie name in a cookie header value.
659 * The lookup begins at <hdr>. The pointer and size of the next occurrence of
660 * the cookie value is returned into *value and *value_l, and the function
661 * returns a pointer to the next pointer to search from if the value was found.
662 * Otherwise if the cookie was not found, NULL is returned and neither value
663 * nor value_l are touched. The input <hdr> string should first point to the
664 * header's value, and the <hdr_end> pointer must point to the first character
665 * not part of the value. <list> must be non-zero if value may represent a list
666 * of values (cookie headers). This makes it faster to abort parsing when no
667 * list is expected.
668 */
669char *http_extract_cookie_value(char *hdr, const char *hdr_end,
670 char *cookie_name, size_t cookie_name_l,
671 int list, char **value, size_t *value_l)
672{
673 char *equal, *att_end, *att_beg, *val_beg, *val_end;
674 char *next;
675
676 /* we search at least a cookie name followed by an equal, and more
677 * generally something like this :
678 * Cookie: NAME1 = VALUE 1 ; NAME2 = VALUE2 ; NAME3 = VALUE3\r\n
679 */
680 for (att_beg = hdr; att_beg + cookie_name_l + 1 < hdr_end; att_beg = next + 1) {
681 /* Iterate through all cookies on this line */
682
683 while (att_beg < hdr_end && HTTP_IS_SPHT(*att_beg))
684 att_beg++;
685
686 /* find att_end : this is the first character after the last non
687 * space before the equal. It may be equal to hdr_end.
688 */
689 equal = att_end = att_beg;
690
691 while (equal < hdr_end) {
692 if (*equal == '=' || *equal == ';' || (list && *equal == ','))
693 break;
694 if (HTTP_IS_SPHT(*equal++))
695 continue;
696 att_end = equal;
697 }
698
699 /* here, <equal> points to '=', a delimitor or the end. <att_end>
700 * is between <att_beg> and <equal>, both may be identical.
701 */
702
703 /* look for end of cookie if there is an equal sign */
704 if (equal < hdr_end && *equal == '=') {
705 /* look for the beginning of the value */
706 val_beg = equal + 1;
707 while (val_beg < hdr_end && HTTP_IS_SPHT(*val_beg))
708 val_beg++;
709
710 /* find the end of the value, respecting quotes */
711 next = http_find_cookie_value_end(val_beg, hdr_end);
712
713 /* make val_end point to the first white space or delimitor after the value */
714 val_end = next;
715 while (val_end > val_beg && HTTP_IS_SPHT(*(val_end - 1)))
716 val_end--;
717 } else {
718 val_beg = val_end = next = equal;
719 }
720
721 /* We have nothing to do with attributes beginning with '$'. However,
722 * they will automatically be removed if a header before them is removed,
723 * since they're supposed to be linked together.
724 */
725 if (*att_beg == '$')
726 continue;
727
728 /* Ignore cookies with no equal sign */
729 if (equal == next)
730 continue;
731
732 /* Now we have the cookie name between att_beg and att_end, and
733 * its value between val_beg and val_end.
734 */
735
736 if (att_end - att_beg == cookie_name_l &&
737 memcmp(att_beg, cookie_name, cookie_name_l) == 0) {
738 /* let's return this value and indicate where to go on from */
739 *value = val_beg;
740 *value_l = val_end - val_beg;
741 return next + 1;
742 }
743
744 /* Set-Cookie headers only have the name in the first attr=value part */
745 if (!list)
746 break;
747 }
748
749 return NULL;
750}
751
Joseph Herlant942eea32018-11-15 13:57:22 -0800752/* Parses a qvalue and returns it multiplied by 1000, from 0 to 1000. If the
Willy Tarreauab813a42018-09-10 18:41:28 +0200753 * value is larger than 1000, it is bound to 1000. The parser consumes up to
754 * 1 digit, one dot and 3 digits and stops on the first invalid character.
755 * Unparsable qvalues return 1000 as "q=1.000".
756 */
757int http_parse_qvalue(const char *qvalue, const char **end)
758{
759 int q = 1000;
760
761 if (!isdigit((unsigned char)*qvalue))
762 goto out;
763 q = (*qvalue++ - '0') * 1000;
764
765 if (*qvalue++ != '.')
766 goto out;
767
768 if (!isdigit((unsigned char)*qvalue))
769 goto out;
770 q += (*qvalue++ - '0') * 100;
771
772 if (!isdigit((unsigned char)*qvalue))
773 goto out;
774 q += (*qvalue++ - '0') * 10;
775
776 if (!isdigit((unsigned char)*qvalue))
777 goto out;
778 q += (*qvalue++ - '0') * 1;
779 out:
780 if (q > 1000)
781 q = 1000;
782 if (end)
783 *end = qvalue;
784 return q;
785}
786
787/*
Joseph Herlant942eea32018-11-15 13:57:22 -0800788 * Given a url parameter, find the starting position of the first occurrence,
Willy Tarreauab813a42018-09-10 18:41:28 +0200789 * or NULL if the parameter is not found.
790 *
791 * Example: if query_string is "yo=mama;ye=daddy" and url_param_name is "ye",
792 * the function will return query_string+8.
793 *
794 * Warning: this function returns a pointer that can point to the first chunk
795 * or the second chunk. The caller must be check the position before using the
796 * result.
797 */
798const char *http_find_url_param_pos(const char **chunks,
799 const char* url_param_name, size_t url_param_name_l,
800 char delim)
801{
802 const char *pos, *last, *equal;
803 const char **bufs = chunks;
804 int l1, l2;
805
806
807 pos = bufs[0];
808 last = bufs[1];
809 while (pos < last) {
810 /* Check the equal. */
811 equal = pos + url_param_name_l;
812 if (fix_pointer_if_wrap(chunks, &equal)) {
813 if (equal >= chunks[3])
814 return NULL;
815 } else {
816 if (equal >= chunks[1])
817 return NULL;
818 }
819 if (*equal == '=') {
820 if (pos + url_param_name_l > last) {
821 /* process wrap case, we detect a wrap. In this case, the
822 * comparison is performed in two parts.
823 */
824
825 /* This is the end, we dont have any other chunk. */
826 if (bufs != chunks || !bufs[2])
827 return NULL;
828
829 /* Compute the length of each part of the comparison. */
830 l1 = last - pos;
831 l2 = url_param_name_l - l1;
832
833 /* The second buffer is too short to contain the compared string. */
834 if (bufs[2] + l2 > bufs[3])
835 return NULL;
836
837 if (memcmp(pos, url_param_name, l1) == 0 &&
838 memcmp(bufs[2], url_param_name+l1, l2) == 0)
839 return pos;
840
841 /* Perform wrapping and jump the string who fail the comparison. */
842 bufs += 2;
843 pos = bufs[0] + l2;
844 last = bufs[1];
845
846 } else {
847 /* process a simple comparison. */
848 if (memcmp(pos, url_param_name, url_param_name_l) == 0)
849 return pos;
850 pos += url_param_name_l + 1;
851 if (fix_pointer_if_wrap(chunks, &pos))
852 last = bufs[2];
853 }
854 }
855
856 while (1) {
857 /* Look for the next delimiter. */
858 while (pos < last && !http_is_param_delimiter(*pos, delim))
859 pos++;
860 if (pos < last)
861 break;
862 /* process buffer wrapping. */
863 if (bufs != chunks || !bufs[2])
864 return NULL;
865 bufs += 2;
866 pos = bufs[0];
867 last = bufs[1];
868 }
869 pos++;
870 }
871 return NULL;
872}
873
874/*
875 * Given a url parameter name and a query string, find the next value.
876 * An empty url_param_name matches the first available parameter.
877 * If the parameter is found, 1 is returned and *vstart / *vend are updated to
878 * respectively provide a pointer to the value and its end.
879 * Otherwise, 0 is returned and vstart/vend are not modified.
880 */
881int http_find_next_url_param(const char **chunks,
882 const char* url_param_name, size_t url_param_name_l,
883 const char **vstart, const char **vend, char delim)
884{
885 const char *arg_start, *qs_end;
886 const char *value_start, *value_end;
887
888 arg_start = chunks[0];
889 qs_end = chunks[1];
890 if (url_param_name_l) {
891 /* Looks for an argument name. */
892 arg_start = http_find_url_param_pos(chunks,
893 url_param_name, url_param_name_l,
894 delim);
895 /* Check for wrapping. */
896 if (arg_start >= qs_end)
897 qs_end = chunks[3];
898 }
899 if (!arg_start)
900 return 0;
901
902 if (!url_param_name_l) {
903 while (1) {
904 /* looks for the first argument. */
905 value_start = memchr(arg_start, '=', qs_end - arg_start);
906 if (!value_start) {
907 /* Check for wrapping. */
908 if (arg_start >= chunks[0] &&
909 arg_start < chunks[1] &&
910 chunks[2]) {
911 arg_start = chunks[2];
912 qs_end = chunks[3];
913 continue;
914 }
915 return 0;
916 }
917 break;
918 }
919 value_start++;
920 }
921 else {
922 /* Jump the argument length. */
923 value_start = arg_start + url_param_name_l + 1;
924
925 /* Check for pointer wrapping. */
926 if (fix_pointer_if_wrap(chunks, &value_start)) {
927 /* Update the end pointer. */
928 qs_end = chunks[3];
929
930 /* Check for overflow. */
931 if (value_start >= qs_end)
932 return 0;
933 }
934 }
935
936 value_end = value_start;
937
938 while (1) {
939 while ((value_end < qs_end) && !http_is_param_delimiter(*value_end, delim))
940 value_end++;
941 if (value_end < qs_end)
942 break;
943 /* process buffer wrapping. */
944 if (value_end >= chunks[0] &&
945 value_end < chunks[1] &&
946 chunks[2]) {
947 value_end = chunks[2];
948 qs_end = chunks[3];
949 continue;
950 }
951 break;
952 }
953
954 *vstart = value_start;
955 *vend = value_end;
956 return 1;
957}
958
Christopher Faulet8277ca72018-10-22 15:12:04 +0200959/* Parses a single header line (without the CRLF) and splits it into its name
960 * and its value. The parsing is pretty naive and just skip spaces.
961 */
962int http_parse_header(const struct ist hdr, struct ist *name, struct ist *value)
963{
964 char *p = hdr.ptr;
965 char *end = p + hdr.len;
966
967 name->len = value->len = 0;
968
969 /* Skip leading spaces */
970 for (; p < end && HTTP_IS_SPHT(*p); p++);
971
972 /* Set the header name */
973 name->ptr = p;
974 for (; p < end && HTTP_IS_TOKEN(*p); p++);
975 name->len = p - name->ptr;
976
977 /* Skip the ':' and spaces before and after it */
978 for (; p < end && HTTP_IS_SPHT(*p); p++);
979 if (p < end && *p == ':') p++;
980 for (; p < end && HTTP_IS_SPHT(*p); p++);
981
982 /* Set the header value */
983 value->ptr = p;
984 value->len = end - p;
985
986 return 1;
987}
988
989/* Parses a single start line (without the CRLF) and splits it into 3 parts. The
990 * parsing is pretty naive and just skip spaces.
991 */
992int http_parse_stline(const struct ist line, struct ist *p1, struct ist *p2, struct ist *p3)
993{
994 char *p = line.ptr;
995 char *end = p + line.len;
996
997 p1->len = p2->len = p3->len = 0;
998
999 /* Skip leading spaces */
1000 for (; p < end && HTTP_IS_SPHT(*p); p++);
1001
1002 /* Set the first part */
1003 p1->ptr = p;
1004 for (; p < end && HTTP_IS_TOKEN(*p); p++);
1005 p1->len = p - p1->ptr;
1006
1007 /* Skip spaces between p1 and p2 */
1008 for (; p < end && HTTP_IS_SPHT(*p); p++);
1009
1010 /* Set the second part */
1011 p2->ptr = p;
1012 for (; p < end && !HTTP_IS_SPHT(*p); p++);
1013 p2->len = p - p2->ptr;
1014
1015 /* Skip spaces between p2 and p3 */
1016 for (; p < end && HTTP_IS_SPHT(*p); p++);
1017
1018 /* The remaing is the third value */
1019 p3->ptr = p;
1020 p3->len = end - p;
1021
1022 return 1;
1023}
Christopher Faulet341fac12019-09-16 11:37:05 +02001024
1025/* Parses value of a Status header with the following format: "Status: Code[
1026 * Reason]". The parsing is pretty naive and just skip spaces. It return the
1027 * numeric value of the status code.
1028 */
1029int http_parse_status_val(const struct ist value, struct ist *status, struct ist *reason)
1030{
1031 char *p = value.ptr;
1032 char *end = p + value.len;
1033 uint16_t code;
1034
1035 status->len = reason->len = 0;
1036
1037 /* Skip leading spaces */
1038 for (; p < end && HTTP_IS_SPHT(*p); p++);
1039
1040 /* Set the status part */
1041 status->ptr = p;
1042 for (; p < end && HTTP_IS_TOKEN(*p); p++);
1043 status->len = p - status->ptr;
1044
1045 /* Skip spaces between status and reason */
1046 for (; p < end && HTTP_IS_SPHT(*p); p++);
1047
1048 /* the remaining is the reason */
1049 reason->ptr = p;
1050 reason->len = end - p;
1051
1052 code = strl2ui(status->ptr, status->len);
1053 return code;
1054}