Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 1 | /* |
Willy Tarreau | 3667d5d | 2009-10-18 19:50:43 +0200 | [diff] [blame] | 2 | * include/types/proto_http.h |
| 3 | * This file contains HTTP protocol definitions. |
| 4 | * |
| 5 | * Copyright (C) 2000-2009 Willy Tarreau - w@1wt.eu |
| 6 | * |
| 7 | * This library is free software; you can redistribute it and/or |
| 8 | * modify it under the terms of the GNU Lesser General Public |
| 9 | * License as published by the Free Software Foundation, version 2.1 |
| 10 | * exclusively. |
| 11 | * |
| 12 | * This library is distributed in the hope that it will be useful, |
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 15 | * Lesser General Public License for more details. |
| 16 | * |
| 17 | * You should have received a copy of the GNU Lesser General Public |
| 18 | * License along with this library; if not, write to the Free Software |
| 19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
| 20 | */ |
Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 21 | |
| 22 | #ifndef _TYPES_PROTO_HTTP_H |
| 23 | #define _TYPES_PROTO_HTTP_H |
| 24 | |
Willy Tarreau | e3ba5f0 | 2006-06-29 18:54:54 +0200 | [diff] [blame] | 25 | #include <common/config.h> |
Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 26 | |
Willy Tarreau | 3bac9ff | 2007-03-18 17:31:28 +0100 | [diff] [blame] | 27 | #include <types/buffers.h> |
| 28 | #include <types/hdr_idx.h> |
| 29 | |
Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 30 | /* |
| 31 | * FIXME: break this into HTTP state and TCP socket state. |
Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 32 | */ |
| 33 | |
Willy Tarreau | b686644 | 2008-07-14 23:54:42 +0200 | [diff] [blame] | 34 | /* different possible states for the client side */ |
Willy Tarreau | 67f0eea | 2008-08-10 22:55:22 +0200 | [diff] [blame] | 35 | #define CL_STDATA 0 |
| 36 | #define CL_STSHUTR 1 |
| 37 | #define CL_STSHUTW 2 |
| 38 | #define CL_STCLOSE 3 |
Willy Tarreau | b686644 | 2008-07-14 23:54:42 +0200 | [diff] [blame] | 39 | |
Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 40 | /* different possible states for the server side */ |
| 41 | #define SV_STIDLE 0 |
Willy Tarreau | adfb856 | 2008-08-11 15:24:42 +0200 | [diff] [blame] | 42 | #define SV_STCONN 1 |
Willy Tarreau | f5483bf | 2008-08-14 18:35:40 +0200 | [diff] [blame] | 43 | #define SV_STDATA 2 |
| 44 | #define SV_STSHUTR 3 |
| 45 | #define SV_STSHUTW 4 |
| 46 | #define SV_STCLOSE 5 |
Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 47 | |
Willy Tarreau | 3d30059 | 2007-03-18 18:34:41 +0100 | [diff] [blame] | 48 | /* |
| 49 | * Transaction flags moved from session |
| 50 | */ |
| 51 | |
| 52 | |
| 53 | /* action flags */ |
| 54 | #define TX_CLDENY 0x00000001 /* a client header matches a deny regex */ |
| 55 | #define TX_CLALLOW 0x00000002 /* a client header matches an allow regex */ |
| 56 | #define TX_SVDENY 0x00000004 /* a server header matches a deny regex */ |
| 57 | #define TX_SVALLOW 0x00000008 /* a server header matches an allow regex */ |
| 58 | #define TX_CLTARPIT 0x00000010 /* the session is tarpitted (anti-dos) */ |
| 59 | /* unused: 0x00000020 */ |
| 60 | |
| 61 | /* transaction flags dedicated to cookies : bits values 0x40, 0x80 (0-3 shift 6) */ |
| 62 | #define TX_CK_NONE 0x00000000 /* this session had no cookie */ |
| 63 | #define TX_CK_INVALID 0x00000040 /* this session had a cookie which matches no server */ |
| 64 | #define TX_CK_DOWN 0x00000080 /* this session had cookie matching a down server */ |
| 65 | #define TX_CK_VALID 0x000000C0 /* this session had cookie matching a valid server */ |
| 66 | #define TX_CK_MASK 0x000000C0 /* mask to get this session's cookie flags */ |
| 67 | #define TX_CK_SHIFT 6 /* bit shift */ |
| 68 | |
| 69 | /* cookie information, bits values 0x100 to 0x800 (0-8 shift 8) */ |
| 70 | #define TX_SCK_NONE 0x00000000 /* no set-cookie seen for the server cookie */ |
| 71 | #define TX_SCK_DELETED 0x00000100 /* existing set-cookie deleted or changed */ |
| 72 | #define TX_SCK_INSERTED 0x00000200 /* new set-cookie inserted or changed existing one */ |
| 73 | #define TX_SCK_SEEN 0x00000400 /* set-cookie seen for the server cookie */ |
| 74 | #define TX_SCK_MASK 0x00000700 /* mask to get the set-cookie field */ |
| 75 | #define TX_SCK_ANY 0x00000800 /* at least one set-cookie seen (not to be counted) */ |
| 76 | #define TX_SCK_SHIFT 8 /* bit shift */ |
| 77 | |
| 78 | /* cacheability management, bits values 0x1000 to 0x3000 (0-3 shift 12) */ |
| 79 | #define TX_CACHEABLE 0x00001000 /* at least part of the response is cacheable */ |
| 80 | #define TX_CACHE_COOK 0x00002000 /* a cookie in the response is cacheable */ |
| 81 | #define TX_CACHE_SHIFT 12 /* bit shift */ |
| 82 | |
Willy Tarreau | 5b15447 | 2009-12-21 20:11:07 +0100 | [diff] [blame] | 83 | /* request and response HTTP version */ |
| 84 | #define TX_REQ_VER_11 0x00004000 /* the request is HTTP/1.1 or above */ |
| 85 | #define TX_RES_VER_11 0x00008000 /* the response is HTTP/1.1 or above */ |
Willy Tarreau | 3667d5d | 2009-10-18 19:50:43 +0200 | [diff] [blame] | 86 | |
| 87 | /* report presence of transfer-encoding:chunked and content-length headers */ |
| 88 | #define TX_REQ_CNT_LEN 0x00010000 /* content-length present in the request */ |
| 89 | #define TX_REQ_TE_CHNK 0x00020000 /* transfer-encoding: chunked present in the request */ |
| 90 | #define TX_RES_CNT_LEN 0x00040000 /* content-length present in the response */ |
| 91 | #define TX_RES_TE_CHNK 0x00080000 /* transfer-encoding: chunked present in the response */ |
Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 92 | |
Willy Tarreau | 5b15447 | 2009-12-21 20:11:07 +0100 | [diff] [blame] | 93 | /* indicate how we *want* the connection to behave, regardless of what is in |
| 94 | * the headers. We have 4 possible values right now : |
| 95 | * - WANT_TUN : will be a tunnel (default when nothing configured or with CONNECT). |
| 96 | * - WANT_KAL : try to maintain keep-alive |
| 97 | * - WANT_SCL : enforce close on the server side |
| 98 | * - WANT_CLO : enforce close on both sides |
| 99 | */ |
| 100 | #define TX_CON_WANT_TUN 0x00000000 /* note: it's important that it is 0 (init) */ |
| 101 | #define TX_CON_WANT_KAL 0x00100000 |
| 102 | #define TX_CON_WANT_SCL 0x00200000 |
| 103 | #define TX_CON_WANT_CLO 0x00300000 |
| 104 | #define TX_CON_WANT_MSK 0x00300000 /* this is the mask to get the bits */ |
| 105 | |
| 106 | #define TX_CON_HDR_PARS 0x00400000 /* "connection" header already parsed (req or res) */ |
| 107 | #define TX_REQ_CONN_CLO 0x00800000 /* request asks for "Connection: close" mode */ |
| 108 | |
| 109 | |
Willy Tarreau | 8d5d7f2 | 2007-01-21 19:16:41 +0100 | [diff] [blame] | 110 | /* The HTTP parser is more complex than it looks like, because we have to |
| 111 | * support multi-line headers and any number of spaces between the colon and |
| 112 | * the value. |
| 113 | * |
| 114 | * All those examples must work : |
| 115 | |
| 116 | Hdr1:val1\r\n |
| 117 | Hdr1: val1\r\n |
| 118 | Hdr1:\t val1\r\n |
| 119 | Hdr1: \r\n |
| 120 | val1\r\n |
| 121 | Hdr1:\r\n |
| 122 | val1\n |
| 123 | \tval2\r\n |
| 124 | val3\n |
| 125 | |
| 126 | * |
| 127 | */ |
| 128 | |
Willy Tarreau | 58f10d7 | 2006-12-04 02:26:12 +0100 | [diff] [blame] | 129 | /* Possible states while parsing HTTP messages (request|response) */ |
Willy Tarreau | 8d5d7f2 | 2007-01-21 19:16:41 +0100 | [diff] [blame] | 130 | #define HTTP_MSG_RQBEFORE 0 // request: leading LF, before start line |
| 131 | #define HTTP_MSG_RQBEFORE_CR 1 // request: leading CRLF, before start line |
| 132 | |
| 133 | /* these ones define a request start line */ |
| 134 | #define HTTP_MSG_RQMETH 2 // parsing the Method |
| 135 | #define HTTP_MSG_RQMETH_SP 3 // space(s) after the ethod |
| 136 | #define HTTP_MSG_RQURI 4 // parsing the Request URI |
| 137 | #define HTTP_MSG_RQURI_SP 5 // space(s) after the Request URI |
| 138 | #define HTTP_MSG_RQVER 6 // parsing the Request Version |
| 139 | #define HTTP_MSG_RQLINE_END 7 // end of request line (CR or LF) |
| 140 | |
| 141 | #define HTTP_MSG_RPBEFORE 8 // response: leading LF, before start line |
| 142 | #define HTTP_MSG_RPBEFORE_CR 9 // response: leading CRLF, before start line |
| 143 | |
| 144 | /* these ones define a response start line */ |
| 145 | #define HTTP_MSG_RPVER 10 // parsing the Response Version |
| 146 | #define HTTP_MSG_RPVER_SP 11 // space(s) after the Response Version |
| 147 | #define HTTP_MSG_RPCODE 12 // response code |
| 148 | #define HTTP_MSG_RPCODE_SP 13 // space(s) after the response code |
| 149 | #define HTTP_MSG_RPREASON 14 // response reason |
| 150 | #define HTTP_MSG_RPLINE_END 15 // end of response line (CR or LF) |
| 151 | |
| 152 | /* common header processing */ |
| 153 | |
| 154 | #define HTTP_MSG_HDR_FIRST 16 // waiting for first header or last CRLF (no LWS possible) |
| 155 | #define HTTP_MSG_HDR_NAME 17 // parsing header name |
| 156 | #define HTTP_MSG_HDR_COL 18 // parsing header colon |
| 157 | #define HTTP_MSG_HDR_L1_SP 19 // parsing header LWS (SP|HT) before value |
| 158 | #define HTTP_MSG_HDR_L1_LF 20 // parsing header LWS (LF) before value |
| 159 | #define HTTP_MSG_HDR_L1_LWS 21 // checking whether it's a new header or an LWS |
| 160 | #define HTTP_MSG_HDR_VAL 22 // parsing header value |
| 161 | #define HTTP_MSG_HDR_L2_LF 23 // parsing header LWS (LF) inside/after value |
| 162 | #define HTTP_MSG_HDR_L2_LWS 24 // checking whether it's a new header or an LWS |
| 163 | |
| 164 | #define HTTP_MSG_LAST_LF 25 // parsing last LF |
Willy Tarreau | 655dce9 | 2009-11-08 13:10:58 +0100 | [diff] [blame] | 165 | |
| 166 | /* error state : must be before HTTP_MSG_BODY so that (>=BODY) always indicates |
| 167 | * that data are being processed. |
| 168 | */ |
| 169 | |
| 170 | #define HTTP_MSG_ERROR 26 // an error occurred |
| 171 | |
| 172 | /* Body processing. |
| 173 | * The state HTTP_MSG_BODY is a delimiter to know if we're waiting for headers |
| 174 | * or body. All the sub-states below also indicate we're processing the body, |
| 175 | * with some additional information. |
| 176 | */ |
| 177 | #define HTTP_MSG_BODY 27 // parsing body at end of headers |
| 178 | #define HTTP_MSG_100_SENT 28 // parsing body after a 100-Continue was sent |
| 179 | #define HTTP_MSG_CHUNK_SIZE 29 // parsing the chunk size (RFC2616 #3.6.1) |
| 180 | #define HTTP_MSG_DATA 30 // skipping data chunk / content-length data |
| 181 | #define HTTP_MSG_DATA_CRLF 31 // skipping CRLF after data chunk |
| 182 | #define HTTP_MSG_TRAILERS 32 // trailers (post-data entity headers) |
Willy Tarreau | 8d5d7f2 | 2007-01-21 19:16:41 +0100 | [diff] [blame] | 183 | |
Willy Tarreau | 58f10d7 | 2006-12-04 02:26:12 +0100 | [diff] [blame] | 184 | |
Willy Tarreau | c0dde7a | 2007-01-01 21:38:07 +0100 | [diff] [blame] | 185 | /* various data sources for the responses */ |
| 186 | #define DATA_SRC_NONE 0 |
| 187 | #define DATA_SRC_STATS 1 |
| 188 | |
| 189 | /* data transmission states for the stats responses */ |
| 190 | enum { |
| 191 | DATA_ST_INIT = 0, |
| 192 | DATA_ST_HEAD, |
| 193 | DATA_ST_INFO, |
| 194 | DATA_ST_LIST, |
| 195 | DATA_ST_END, |
| 196 | DATA_ST_FIN, |
| 197 | }; |
| 198 | |
| 199 | /* data transmission states for the stats responses inside a proxy */ |
| 200 | enum { |
| 201 | DATA_ST_PX_INIT = 0, |
| 202 | DATA_ST_PX_TH, |
| 203 | DATA_ST_PX_FE, |
Krzysztof Piotr Oledzki | aeebf9b | 2009-10-04 15:43:17 +0200 | [diff] [blame] | 204 | DATA_ST_PX_LI, |
Willy Tarreau | c0dde7a | 2007-01-01 21:38:07 +0100 | [diff] [blame] | 205 | DATA_ST_PX_SV, |
| 206 | DATA_ST_PX_BE, |
| 207 | DATA_ST_PX_END, |
| 208 | DATA_ST_PX_FIN, |
| 209 | }; |
| 210 | |
Willy Tarreau | 79da469 | 2008-11-19 20:03:04 +0100 | [diff] [blame] | 211 | /* Redirect flags */ |
| 212 | enum { |
| 213 | REDIRECT_FLAG_NONE = 0, |
| 214 | REDIRECT_FLAG_DROP_QS = 1, /* drop query string */ |
| 215 | }; |
Willy Tarreau | b463dfb | 2008-06-07 23:08:56 +0200 | [diff] [blame] | 216 | |
| 217 | /* Redirect types (location, prefix, extended ) */ |
| 218 | enum { |
| 219 | REDIRECT_TYPE_NONE = 0, /* no redirection */ |
| 220 | REDIRECT_TYPE_LOCATION, /* location redirect */ |
| 221 | REDIRECT_TYPE_PREFIX, /* prefix redirect */ |
| 222 | }; |
| 223 | |
Willy Tarreau | 3bac9ff | 2007-03-18 17:31:28 +0100 | [diff] [blame] | 224 | /* Known HTTP methods */ |
| 225 | typedef enum { |
| 226 | HTTP_METH_NONE = 0, |
| 227 | HTTP_METH_OPTIONS, |
| 228 | HTTP_METH_GET, |
| 229 | HTTP_METH_HEAD, |
| 230 | HTTP_METH_POST, |
| 231 | HTTP_METH_PUT, |
| 232 | HTTP_METH_DELETE, |
| 233 | HTTP_METH_TRACE, |
| 234 | HTTP_METH_CONNECT, |
| 235 | HTTP_METH_OTHER, |
| 236 | } http_meth_t; |
| 237 | |
| 238 | /* This is an HTTP message, as described in RFC2616. It can be either a request |
| 239 | * message or a response message. |
| 240 | * |
| 241 | * The values there are a little bit obscure, because their meaning can change |
| 242 | * during the parsing : |
| 243 | * |
| 244 | * - som (Start of Message) : relative offset in the buffer of first byte of |
| 245 | * the request being processed or parsed. Reset to |
| 246 | * zero during accept(). |
| 247 | * - eoh (End of Headers) : relative offset in the buffer of first byte that |
| 248 | * is not part of a completely processed header. |
| 249 | * During parsing, it points to last header seen |
Willy Tarreau | fa355d4 | 2009-11-29 18:12:29 +0100 | [diff] [blame] | 250 | * for states after START. When in HTTP_MSG_BODY, |
| 251 | * eoh points to the first byte of the last CRLF |
| 252 | * preceeding data. |
| 253 | * - col and sov : When in HTTP_MSG_BODY, will point to the first |
| 254 | * byte of data. |
Willy Tarreau | 3bac9ff | 2007-03-18 17:31:28 +0100 | [diff] [blame] | 255 | * - eol (End of Line) : relative offset in the buffer of the first byte |
| 256 | * which marks the end of the line (LF or CRLF). |
| 257 | */ |
| 258 | struct http_msg { |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 259 | unsigned int msg_state; /* where we are in the current message parsing */ |
Willy Tarreau | fa355d4 | 2009-11-29 18:12:29 +0100 | [diff] [blame] | 260 | unsigned int col, sov; /* current header: colon, start of value */ |
| 261 | unsigned int eoh; /* End Of Headers, relative to buffer */ |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 262 | char *sol; /* start of line, also start of message when fully parsed */ |
| 263 | char *eol; /* end of line */ |
| 264 | unsigned int som; /* Start Of Message, relative to buffer */ |
Willy Tarreau | fa355d4 | 2009-11-29 18:12:29 +0100 | [diff] [blame] | 265 | int err_pos; /* err handling: -2=block, -1=pass, 0+=detected */ |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 266 | union { /* useful start line pointers, relative to buffer */ |
Willy Tarreau | 3bac9ff | 2007-03-18 17:31:28 +0100 | [diff] [blame] | 267 | struct { |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 268 | int l; /* request line length (not including CR) */ |
| 269 | int m_l; /* METHOD length (method starts at ->som) */ |
| 270 | int u, u_l; /* URI, length */ |
| 271 | int v, v_l; /* VERSION, length */ |
| 272 | } rq; /* request line : field, length */ |
Willy Tarreau | 3bac9ff | 2007-03-18 17:31:28 +0100 | [diff] [blame] | 273 | struct { |
matt.farnsworth@nokia.com | 1c2ab96 | 2008-04-14 20:47:37 +0200 | [diff] [blame] | 274 | int l; /* status line length (not including CR) */ |
| 275 | int v_l; /* VERSION length (version starts at ->som) */ |
| 276 | int c, c_l; /* CODE, length */ |
| 277 | int r, r_l; /* REASON, length */ |
| 278 | } st; /* status line : field, length */ |
| 279 | } sl; /* start line */ |
Willy Tarreau | 3667d5d | 2009-10-18 19:50:43 +0200 | [diff] [blame] | 280 | unsigned long long hdr_content_len; /* cache for parsed header value or for chunk-size if present */ |
Willy Tarreau | fa355d4 | 2009-11-29 18:12:29 +0100 | [diff] [blame] | 281 | char **cap; /* array of captured headers (may be NULL) */ |
Willy Tarreau | 3bac9ff | 2007-03-18 17:31:28 +0100 | [diff] [blame] | 282 | }; |
| 283 | |
| 284 | /* This is an HTTP transaction. It contains both a request message and a |
| 285 | * response message (which can be empty). |
| 286 | */ |
| 287 | struct http_txn { |
| 288 | http_meth_t meth; /* HTTP method */ |
| 289 | struct hdr_idx hdr_idx; /* array of header indexes (max: MAX_HTTP_HDR) */ |
| 290 | struct chunk auth_hdr; /* points to 'Authorization:' header */ |
| 291 | struct http_msg req, rsp; /* HTTP request and response messages */ |
| 292 | |
| 293 | char *uri; /* first line if log needed, NULL otherwise */ |
| 294 | char *cli_cookie; /* cookie presented by the client, in capture mode */ |
| 295 | char *srv_cookie; /* cookie presented by the server, in capture mode */ |
| 296 | int status; /* HTTP status from the server, negative if from proxy */ |
Willy Tarreau | 3d30059 | 2007-03-18 18:34:41 +0100 | [diff] [blame] | 297 | unsigned int flags; /* transaction flags */ |
Willy Tarreau | 3bac9ff | 2007-03-18 17:31:28 +0100 | [diff] [blame] | 298 | }; |
| 299 | |
Willy Tarreau | 33a7e69 | 2007-06-10 19:45:56 +0200 | [diff] [blame] | 300 | /* This structure is used by http_find_header() to return values of headers. |
| 301 | * The header starts at <line>, the value at <line>+<val> for <vlen> bytes. |
| 302 | */ |
| 303 | struct hdr_ctx { |
| 304 | const char *line; |
| 305 | int idx; |
| 306 | int val; /* relative to line */ |
| 307 | int vlen; /* relative to line+val */ |
| 308 | }; |
Willy Tarreau | 58f10d7 | 2006-12-04 02:26:12 +0100 | [diff] [blame] | 309 | |
Willy Tarreau | baaee00 | 2006-06-26 02:48:02 +0200 | [diff] [blame] | 310 | #endif /* _TYPES_PROTO_HTTP_H */ |
| 311 | |
| 312 | /* |
| 313 | * Local variables: |
| 314 | * c-indent-level: 8 |
| 315 | * c-basic-offset: 8 |
| 316 | * End: |
| 317 | */ |