epan/req_resp_hdrs.c

   1 /* req_resp_hdrs.c
   2  * Routines handling protocols with a request/response line, headers,
   3  * a blank line, and an optional body.
   4  *
   5  * $Id$
   6  *
   7  * Wireshark - Network traffic analyzer
   8  * By Gerald Combs <gerald@wireshark.org>
   9  * Copyright 1998 Gerald Combs
  10  *
  11  * This program is free software; you can redistribute it and/or
  12  * modify it under the terms of the GNU General Public License
  13  * as published by the Free Software Foundation; either version 2
  14  * of the License, or (at your option) any later version.
  15  *
  16  * This program is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19  * GNU General Public License for more details.
  20  *
  21  * You should have received a copy of the GNU General Public License
  22  * along with this program; if not, write to the Free Software
  23  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  24  */
  25
  26 #include "config.h"
  27
  28 #include <glib.h>
  29 #include <stdio.h>
  30 #include <string.h>
  31
  32 #include <epan/packet.h>
  33 #include <epan/strutil.h>
  34
  35 #include <epan/req_resp_hdrs.h>
  36
  37 /*
  38  * Optionally do reassembly of the request/response line, headers, and body.
  39  */
  40 gboolean
  41 req_resp_hdrs_do_reassembly(tvbuff_t *tvb, const int offset, packet_info *pinfo,
  42     const gboolean desegment_headers, const gboolean desegment_body)
  43 {
  44         gint            next_offset;
  45         gint            next_offset_sav;
  46         gint            length_remaining, reported_length_remaining;
  47         int             linelen;
  48         gchar           *header_val;
  49         int             content_length;
  50         gboolean        content_length_found = FALSE;
  51         gboolean        content_type_found = FALSE;
  52         gboolean        chunked_encoding = FALSE;
  53         gboolean        keepalive_found = FALSE;
  54         gchar           *line;
  55         gchar           *content_type = NULL;
  56
  57         /*
  58          * Do header desegmentation if we've been told to.
  59          *
  60          * RFC 2616 defines HTTP messages as being either of the
  61          * Request or the Response type
  62          * (HTTP-message = Request | Response).
  63          * Request and Response are defined as:
  64          *     Request = Request-Line
  65          *         *(( general-header
  66          *         | request-header
  67          *         | entity-header ) CRLF)
  68          *         CRLF
  69          *         [ message-body ]
  70          *     Response = Status-Line
  71          *         *(( general-header
  72          *         | response-header
  73          *         | entity-header ) CRLF)
  74          *         CRLF
  75          *         [ message-body ]
  76          * that's why we can always assume two consecutive line
  77          * endings (we allow CR, LF, or CRLF, as some clients
  78          * or servers might not use a full CRLF) to mark the end
  79          * of the headers.  The worst thing that would happen
  80          * otherwise would be the packet not being desegmented
  81          * or being interpreted as only headers.
  82          *
  83          * RFC 2326 says RTSP works the same way; RFC 3261 says SIP
  84          * works the same way.
  85          */
  86
  87         /*
  88          * If header desegmentation is activated, check that all
  89          * headers are in this tvbuff (search for an empty line
  90          * marking end of headers) or request one more byte (we
  91          * don't know how many bytes we'll need, so we just ask
  92          * for one).
  93          */
  94         if (desegment_headers && pinfo->can_desegment) {
  95                 next_offset = offset;
  96                 for (;;) {
  97                         next_offset_sav = next_offset;
  98
  99                         reported_length_remaining =
 100                             tvb_reported_length_remaining(tvb, next_offset);
 101
 102                         /*
 103                          * Request one more byte if there're no
 104                          * bytes left in the reported data (if there're
 105                          * bytes left in the reported data, but not in
 106                          * the available data, requesting more bytes
 107                          * won't help, as those bytes weren't captured).
 108                          */
 109                         if (reported_length_remaining < 1) {
 110                                 pinfo->desegment_offset = offset;
 111                                 pinfo->desegment_len = DESEGMENT_ONE_MORE_SEGMENT;
 112                                 return FALSE;
 113                         }
 114
 115                         length_remaining = tvb_length_remaining(tvb,
 116                             next_offset);
 117
 118                         /*
 119                          * Request one more byte if we cannot find a
 120                          * header (i.e. a line end).
 121                          */
 122                         linelen = tvb_find_line_end(tvb, next_offset,
 123                             -1, &next_offset, TRUE);
 124                         if (linelen == -1 &&
 125                             length_remaining >= reported_length_remaining) {
 126                                 /*
 127                                  * Not enough data; ask for one more
 128                                  * byte.
 129                                  */
 130                                 pinfo->desegment_offset = offset;
 131                                 pinfo->desegment_len = DESEGMENT_ONE_MORE_SEGMENT;
 132                                 return FALSE;
 133                         }
 134
 135                         if (linelen == 0) {
 136                                 /*
 137                                  * We found the end of the headers.
 138                                  */
 139                                 break;
 140                         }
 141
 142                         /*
 143                          * Is this a Content-Length or Transfer-Encoding
 144                          * header?  If not, it either means that we are in
 145                          * a different header line, or that we are
 146                          * at the end of the headers, or that there
 147                          * isn't enough data; the two latter cases
 148                          * have already been handled above.
 149                          */
 150                         if (desegment_body) {
 151                                 /* Optimization to avoid fetching the whole (potentially very long)
 152                                  * line and doing expensive string comparisons if the first
 153                                  * character doesn't match. Shaves about 20% off the load time of
 154                                  * one of my sample files that's HTTP-alike. */
 155                                 guchar first_byte = tvb_get_guint8(tvb, next_offset_sav);
 156                                 if (! (first_byte == 'c' || first_byte == 'C' ||
 157                                        first_byte == 't' || first_byte == 'T')) {
 158                                         continue;
 159                                 }
 160
 161                                 /*
 162                                  * Check if we've found Content-Length.
 163                                  */
 164                                 line = tvb_get_string(wmem_packet_scope(), tvb, next_offset_sav, linelen);
 165                                 if (g_ascii_strncasecmp(line, "Content-Length:", 15) == 0) {
 166                                         /* XXX - what if it doesn't fit in an int?
 167                                            (Do not "fix" that by making this
 168                                            a "long"; make it a gint64 or a
 169                                            guint64.) */
 170                                         if (sscanf(line+15,"%i", &content_length) == 1)
 171                                                 content_length_found = TRUE;
 172                                 } else if (g_ascii_strncasecmp(line, "Content-Type:", 13) == 0) {
 173                                         content_type_found = TRUE;
 174                                         content_type = line+13;
 175                                         while (*content_type == ' ') {
 176                                                 content_type++;
 177                                         }
 178                                 } else if (g_ascii_strncasecmp(line, "Connection:", 11) == 0) {
 179                                         /* Check for keep-alive */
 180                                         header_val = line+11;
 181                                         if(header_val){
 182                                                 while(*header_val==' '){
 183                                                         header_val++;
 184                                                 }
 185                                                 if(!g_ascii_strncasecmp(header_val, "Keep-Alive", 10)){
 186                                                         keepalive_found = TRUE;
 187                                                 }
 188                                         }
 189                                 } else if (g_ascii_strncasecmp( line, "Transfer-Encoding:", 18) == 0) {
 190                                         /*
 191                                          * Find out if this Transfer-Encoding is
 192                                          * chunked.  It should be, since there
 193                                          * really aren't any other types, but
 194                                          * RFC 2616 allows for them.
 195                                          */
 196                                         gchar *p;
 197                                         guint len;
 198
 199                                         header_val = line+18;
 200                                         p = header_val;
 201                                         len = (guint) strlen(header_val);
 202                                         /* Skip white space */
 203                                         while (p < header_val + len &&
 204                                             (*p == ' ' || *p == '\t'))
 205                                                 p++;
 206                                         if (p <= header_val + len) {
 207                                                 if (g_ascii_strncasecmp(p, "chunked", 7)
 208                                                     == 0) {
 209                                                         /*
 210                                                          * Don't bother looking
 211                                                          * for extensions;
 212                                                          * since we don't
 213                                                          * understand them,
 214                                                          * they should be
 215                                                          * ignored.
 216                                                          */
 217                                                         chunked_encoding = TRUE;
 218                                                 }
 219                                         }
 220                                 }
 221                         }
 222                 }
 223         }
 224
 225         /*
 226          * The above loop ends when we reached the end of the headers, so
 227          * there should be content_length bytes after the 4 terminating bytes
 228          * and next_offset points to after the end of the headers.
 229          */
 230         if (desegment_body) {
 231                 if (chunked_encoding) {
 232                         /*
 233                          * This data is chunked, so we need to keep pulling
 234                          * data until we reach the end of the stream, or a
 235                          * zero sized chunk.
 236                          *
 237                          * XXX
 238                          * This doesn't bother with trailing headers; I don't
 239                          * think they are really used, and we'd have to use
 240                          * is_http_request_or_reply() to determine if it was
 241                          * a trailing header, or the start of a new response.
 242                          */
 243                         gboolean done_chunking = FALSE;
 244
 245                         while (!done_chunking) {
 246                                 guint chunk_size = 0;
 247                                 gint  chunk_offset = 0;
 248                                 gchar *chunk_string = NULL;
 249                                 gchar *c = NULL;
 250
 251                                 reported_length_remaining =
 252                                     tvb_reported_length_remaining(tvb,
 253                                     next_offset);
 254
 255                                 if (reported_length_remaining < 1) {
 256                                         pinfo->desegment_offset = offset;
 257                                         pinfo->desegment_len = DESEGMENT_ONE_MORE_SEGMENT;
 258                                         return FALSE;
 259                                 }
 260
 261                                 length_remaining = tvb_length_remaining(tvb,
 262                                     next_offset);
 263
 264                                 linelen = tvb_find_line_end(tvb, next_offset,
 265                                                 -1, &chunk_offset, TRUE);
 266
 267                                 if (linelen == -1 &&
 268                                     length_remaining >=
 269                                     reported_length_remaining) {
 270                                          pinfo->desegment_offset = offset;
 271                                          pinfo->desegment_len = DESEGMENT_ONE_MORE_SEGMENT;
 272                                          return FALSE;
 273                                 }
 274
 275                                 /* We have a line with the chunk size in it.*/
 276                                 chunk_string = tvb_get_string(wmem_packet_scope(), tvb, next_offset,
 277                                     linelen);
 278                                 c = chunk_string;
 279
 280                                 /*
 281                                  * We don't care about the extensions.
 282                                  */
 283                                 if ((c = strchr(c, ';'))) {
 284                                         *c = '\0';
 285                                 }
 286
 287                                 if (sscanf(chunk_string, "%x", &chunk_size) < 1) {
 288                                         /* We couldn't get the chunk size,
 289                                          * so stop trying.
 290                                          */
 291                                         return TRUE;
 292                                 }
 293                                 if (chunk_size > (guint)1<<31) {
 294                                         /* Chunk size is unreasonable. */
 295                                         /* XXX What /is/ reasonable? */
 296                                         return TRUE;
 297                                 }
 298
 299                                 if (chunk_size == 0) {
 300                                         /*
 301                                          * This is the last chunk.  Let's pull in the
 302                                          * trailing CRLF.
 303                                          */
 304                                         linelen = tvb_find_line_end(tvb,
 305                                             chunk_offset, -1, &chunk_offset, TRUE);
 306
 307                                         if (linelen == -1 &&
 308                                             length_remaining >=
 309                                             reported_length_remaining) {
 310                                                 pinfo->desegment_offset = offset;
 311                                                 pinfo->desegment_len = DESEGMENT_ONE_MORE_SEGMENT;
 312                                                 return FALSE;
 313                                         }
 314
 315                                         pinfo->desegment_offset = chunk_offset;
 316                                         pinfo->desegment_len = 0;
 317                                         done_chunking = TRUE;
 318                                 } else {
 319                                         /*
 320                                          * Skip to the next chunk if we
 321                                          * already have it
 322                                          */
 323                                         if (reported_length_remaining >
 324                                                 (gint) chunk_size) {
 325
 326                                                 next_offset = chunk_offset
 327                                                     + chunk_size + 2;
 328                                         } else {
 329                                                 /*
 330                                                  * Fetch this chunk, plus the
 331                                                  * trailing CRLF.
 332                                                  */
 333                                                 pinfo->desegment_offset = offset;
 334                                                 pinfo->desegment_len = DESEGMENT_ONE_MORE_SEGMENT;
 335                                                 return FALSE;
 336                                         }
 337                                 }
 338
 339                         }
 340                 } else if (content_length_found) {
 341                         if (content_length >= 128*1024) { /* MS-RPCH stipulate that the content-length must be between 128K and 2G */
 342                                 gchar *tmp;
 343                                 if (content_type_found &&
 344                                 strncmp(content_type, "application/rpc", 15) == 0) {
 345                                         /* It looks like a RPC_IN_DATA request or a RPC_OUT_DATA response
 346                                          * in which the content-length is meaningless
 347                                          */
 348                                         return TRUE;
 349                                 }
 350                                 /* Following sizeof will return the length of the string + \0 we need to not count it*/
 351                                 tmp = tvb_get_string(wmem_packet_scope(), tvb, 0, sizeof("RPC_OUT_DATA") - 1);
 352                                 if ((strncmp(tmp, "RPC_IN_DATA", sizeof("RPC_IN_DATA") - 1) == 0) ||
 353                                     (strncmp(tmp, "RPC_OUT_DATA", sizeof("RPC_OUT_DATA") - 1) == 0)) {
 354                                         return TRUE;
 355                                 }
 356                         }
 357                         /* next_offset has been set to the end of the headers */
 358                         if (!tvb_bytes_exist(tvb, next_offset, content_length)) {
 359                                 length_remaining = tvb_length_remaining(tvb,
 360                                     next_offset);
 361                                 reported_length_remaining =
 362                                     tvb_reported_length_remaining(tvb, next_offset);
 363                                 if (length_remaining < reported_length_remaining) {
 364                                         /*
 365                                          * It's a waste of time asking for more
 366                                          * data, because that data wasn't captured.
 367                                          */
 368                                         return TRUE;
 369                                 }
 370                                 if (length_remaining == -1)
 371                                         length_remaining = 0;
 372                                 pinfo->desegment_offset = offset;
 373                                 pinfo->desegment_len =
 374                                     content_length - length_remaining;
 375                                 return FALSE;
 376                         }
 377                 } else if (content_type_found && pinfo->can_desegment) {
 378                         /* We found a content-type but no content-length.
 379                          * This is probably a HTTP header for a session with
 380                          * only one HTTP PDU and where the content spans
 381                          * until the end of the tcp session, unless there
 382                          * is a keepalive header present in which case we
 383                          * assume there is no message body at all and thus
 384                          * we wont do any reassembly.
 385                          * Set up tcp reassembly until the end of this session.
 386                          */
 387                         length_remaining = tvb_length_remaining(tvb, next_offset);
 388                         reported_length_remaining = tvb_reported_length_remaining(tvb, next_offset);
 389                         if (length_remaining < reported_length_remaining) {
 390                                 /*
 391                                  * It's a waste of time asking for more
 392                                  * data, because that data wasn't captured.
 393                                  */
 394                                 return TRUE;
 395                         }
 396
 397                         if (keepalive_found) {
 398                                 /* We have a keep-alive but no content-length.
 399                                  * Assume there is no message body and dont
 400                                  * do any reassembly.
 401                                  */
 402                                 return TRUE;
 403                         }
 404
 405                         pinfo->desegment_offset = offset;
 406                         pinfo->desegment_len = DESEGMENT_UNTIL_FIN;
 407
 408                         return FALSE;
 409                 }
 410
 411         }
 412
 413         /*
 414          * No further desegmentation needed.
 415          */
 416         return TRUE;
 417 }
 418
 419 /*
 420  * Editor modelines  -  http://www.wireshark.org/tools/modelines.html
 421  *
 422  * Local variables:
 423  * c-basic-offset: 8
 424  * tab-width: 8
 425  * indent-tabs-mode: t
 426  * End:
 427  *
 428  * vi: set shiftwidth=8 tabstop=8 noexpandtab:
 429  * :indentSize=8:tabSize=8:noTabs=false:
 430  */