src/HttpReply.cc

   1 /*
   2  * Copyright (C) 1996-2015 The Squid Software Foundation and contributors
   3  *
   4  * Squid software is distributed under GPLv2+ license and includes
   5  * contributions from numerous individuals and organizations.
   6  * Please see the COPYING and CONTRIBUTORS files for details.
   7  */
   8
   9 /* DEBUG: section 58    HTTP Reply (Response) */
  10
  11 #include "squid.h"
  12 #include "acl/AclSizeLimit.h"
  13 #include "acl/FilledChecklist.h"
  14 #include "globals.h"
  15 #include "HttpBody.h"
  16 #include "HttpHdrCc.h"
  17 #include "HttpHdrContRange.h"
  18 #include "HttpHdrSc.h"
  19 #include "HttpReply.h"
  20 #include "HttpRequest.h"
  21 #include "MemBuf.h"
  22 #include "SquidConfig.h"
  23 #include "SquidTime.h"
  24 #include "Store.h"
  25 #include "StrList.h"
  26
  27 /* local constants */
  28
  29 /* If we receive a 304 from the origin during a cache revalidation, we must
  30  * update the headers of the existing entry. Specifically, we need to update all
  31  * end-to-end headers and not any hop-by-hop headers (rfc2616 13.5.3).
  32  *
  33  * This is not the whole story though: since it is possible for a faulty/malicious
  34  * origin server to set headers it should not in a 304, we must explicitly ignore
  35  * these too. Specifically all entity-headers except those permitted in a 304
  36  * (rfc2616 10.3.5) must be ignored.
  37  *
  38  * The list of headers we don't update is made up of:
  39  *     all hop-by-hop headers
  40  *     all entity-headers except Expires and Content-Location
  41  */
  42 static HttpHeaderMask Denied304HeadersMask;
  43 static http_hdr_type Denied304HeadersArr[] = {
  44     // hop-by-hop headers
  45     HDR_CONNECTION, HDR_KEEP_ALIVE, HDR_PROXY_AUTHENTICATE, HDR_PROXY_AUTHORIZATION,
  46     HDR_TE, HDR_TRAILER, HDR_TRANSFER_ENCODING, HDR_UPGRADE,
  47     // entity headers
  48     HDR_ALLOW, HDR_CONTENT_ENCODING, HDR_CONTENT_LANGUAGE, HDR_CONTENT_LENGTH,
  49     HDR_CONTENT_MD5, HDR_CONTENT_RANGE, HDR_CONTENT_TYPE, HDR_LAST_MODIFIED
  50 };
  51
  52 /* module initialization */
  53 void
  54 httpReplyInitModule(void)
  55 {
  56     assert(Http::scNone == 0); // HttpReply::parse() interface assumes that
  57     httpHeaderMaskInit(&Denied304HeadersMask, 0);
  58     httpHeaderCalcMask(&Denied304HeadersMask, Denied304HeadersArr, countof(Denied304HeadersArr));
  59 }
  60
  61 HttpReply::HttpReply() : HttpMsg(hoReply), date (0), last_modified (0),
  62     expires (0), surrogate_control (NULL), content_range (NULL), keep_alive (0),
  63     protoPrefix("HTTP/"), bodySizeMax(-2)
  64 {
  65     init();
  66 }
  67
  68 HttpReply::~HttpReply()
  69 {
  70     if (do_clean)
  71         clean();
  72 }
  73
  74 void
  75 HttpReply::init()
  76 {
  77     hdrCacheInit();
  78     sline.init();
  79     pstate = psReadyToParseStartLine;
  80     do_clean = true;
  81 }
  82
  83 void HttpReply::reset()
  84 {
  85
  86     // reset should not reset the protocol; could have made protoPrefix a
  87     // virtual function instead, but it is not clear whether virtual methods
  88     // are allowed with MEMPROXY_CLASS() and whether some cbdata void*
  89     // conversions are not going to kill virtual tables
  90     const String pfx = protoPrefix;
  91     clean();
  92     init();
  93     protoPrefix = pfx;
  94 }
  95
  96 void
  97 HttpReply::clean()
  98 {
  99     // we used to assert that the pipe is NULL, but now the message only
 100     // points to a pipe that is owned and initiated by another object.
 101     body_pipe = NULL;
 102
 103     body.clear();
 104     hdrCacheClean();
 105     header.clean();
 106     sline.clean();
 107     bodySizeMax = -2; // hack: make calculatedBodySizeMax() false
 108 }
 109
 110 void
 111 HttpReply::packHeadersInto(Packable * p) const
 112 {
 113     sline.packInto(p);
 114     header.packInto(p);
 115     p->append("\r\n", 2);
 116 }
 117
 118 void
 119 HttpReply::packInto(Packable * p)
 120 {
 121     packHeadersInto(p);
 122     body.packInto(p);
 123 }
 124
 125 /* create memBuf, create mem-based packer, pack, destroy packer, return MemBuf */
 126 MemBuf *
 127 HttpReply::pack()
 128 {
 129     MemBuf *mb = new MemBuf;
 130     Packer p;
 131
 132     mb->init();
 133     packerToMemInit(&p, mb);
 134     packInto(&p);
 135     return mb;
 136 }
 137
 138 HttpReply *
 139 HttpReply::make304() const
 140 {
 141     static const http_hdr_type ImsEntries[] = {HDR_DATE, HDR_CONTENT_TYPE, HDR_EXPIRES, HDR_LAST_MODIFIED, /* eof */ HDR_OTHER};
 142
 143     HttpReply *rv = new HttpReply;
 144     int t;
 145     HttpHeaderEntry *e;
 146
 147     /* rv->content_length; */
 148     rv->date = date;
 149     rv->last_modified = last_modified;
 150     rv->expires = expires;
 151     rv->content_type = content_type;
 152     /* rv->cache_control */
 153     /* rv->content_range */
 154     /* rv->keep_alive */
 155     rv->sline.set(Http::ProtocolVersion(), Http::scNotModified, NULL);
 156
 157     for (t = 0; ImsEntries[t] != HDR_OTHER; ++t)
 158         if ((e = header.findEntry(ImsEntries[t])))
 159             rv->header.addEntry(e->clone());
 160
 161     /* rv->body */
 162     return rv;
 163 }
 164
 165 MemBuf *
 166 HttpReply::packed304Reply()
 167 {
 168     /* Not as efficient as skipping the header duplication,
 169      * but easier to maintain
 170      */
 171     HttpReply *temp = make304();
 172     MemBuf *rv = temp->pack();
 173     delete temp;
 174     return rv;
 175 }
 176
 177 void
 178 HttpReply::setHeaders(Http::StatusCode status, const char *reason,
 179                       const char *ctype, int64_t clen, time_t lmt, time_t expiresTime)
 180 {
 181     HttpHeader *hdr;
 182     sline.set(Http::ProtocolVersion(), status, reason);
 183     hdr = &header;
 184     hdr->putStr(HDR_SERVER, visible_appname_string);
 185     hdr->putStr(HDR_MIME_VERSION, "1.0");
 186     hdr->putTime(HDR_DATE, squid_curtime);
 187
 188     if (ctype) {
 189         hdr->putStr(HDR_CONTENT_TYPE, ctype);
 190         content_type = ctype;
 191     } else
 192         content_type = String();
 193
 194     if (clen >= 0)
 195         hdr->putInt64(HDR_CONTENT_LENGTH, clen);
 196
 197     if (expiresTime >= 0)
 198         hdr->putTime(HDR_EXPIRES, expiresTime);
 199
 200     if (lmt > 0)        /* this used to be lmt != 0 @?@ */
 201         hdr->putTime(HDR_LAST_MODIFIED, lmt);
 202
 203     date = squid_curtime;
 204
 205     content_length = clen;
 206
 207     expires = expiresTime;
 208
 209     last_modified = lmt;
 210 }
 211
 212 void
 213 HttpReply::redirect(Http::StatusCode status, const char *loc)
 214 {
 215     HttpHeader *hdr;
 216     sline.set(Http::ProtocolVersion(), status, NULL);
 217     hdr = &header;
 218     hdr->putStr(HDR_SERVER, APP_FULLNAME);
 219     hdr->putTime(HDR_DATE, squid_curtime);
 220     hdr->putInt64(HDR_CONTENT_LENGTH, 0);
 221     hdr->putStr(HDR_LOCATION, loc);
 222     date = squid_curtime;
 223     content_length = 0;
 224 }
 225
 226 /* compare the validators of two replies.
 227  * 1 = they match
 228  * 0 = they do not match
 229  */
 230 int
 231 HttpReply::validatorsMatch(HttpReply const * otherRep) const
 232 {
 233     String one,two;
 234     assert (otherRep);
 235     /* Numbers first - easiest to check */
 236     /* Content-Length */
 237     /* TODO: remove -1 bypass */
 238
 239     if (content_length != otherRep->content_length
 240             && content_length > -1 &&
 241             otherRep->content_length > -1)
 242         return 0;
 243
 244     /* ETag */
 245     one = header.getStrOrList(HDR_ETAG);
 246
 247     two = otherRep->header.getStrOrList(HDR_ETAG);
 248
 249     if (one.size()==0 || two.size()==0 || one.caseCmp(two)!=0 ) {
 250         one.clean();
 251         two.clean();
 252         return 0;
 253     }
 254
 255     if (last_modified != otherRep->last_modified)
 256         return 0;
 257
 258     /* MD5 */
 259     one = header.getStrOrList(HDR_CONTENT_MD5);
 260
 261     two = otherRep->header.getStrOrList(HDR_CONTENT_MD5);
 262
 263     if (one.size()==0 || two.size()==0 || one.caseCmp(two)!=0 ) {
 264         one.clean();
 265         two.clean();
 266         return 0;
 267     }
 268
 269     return 1;
 270 }
 271
 272 void
 273 HttpReply::updateOnNotModified(HttpReply const * freshRep)
 274 {
 275     assert(freshRep);
 276
 277     /* clean cache */
 278     hdrCacheClean();
 279     /* update raw headers */
 280     header.update(&freshRep->header,
 281                   (const HttpHeaderMask *) &Denied304HeadersMask);
 282
 283     header.compact();
 284     /* init cache */
 285     hdrCacheInit();
 286 }
 287
 288 /* internal routines */
 289
 290 time_t
 291 HttpReply::hdrExpirationTime()
 292 {
 293     /* The s-maxage and max-age directive takes priority over Expires */
 294
 295     if (cache_control) {
 296         if (date >= 0) {
 297             if (cache_control->hasSMaxAge())
 298                 return date + cache_control->sMaxAge();
 299
 300             if (cache_control->hasMaxAge())
 301                 return date + cache_control->maxAge();
 302         } else {
 303             /*
 304              * Conservatively handle the case when we have a max-age
 305              * header, but no Date for reference?
 306              */
 307
 308             if (cache_control->hasSMaxAge())
 309                 return squid_curtime;
 310
 311             if (cache_control->hasMaxAge())
 312                 return squid_curtime;
 313         }
 314     }
 315
 316     if (Config.onoff.vary_ignore_expire &&
 317             header.has(HDR_VARY)) {
 318         const time_t d = header.getTime(HDR_DATE);
 319         const time_t e = header.getTime(HDR_EXPIRES);
 320
 321         if (d == e)
 322             return -1;
 323     }
 324
 325     if (header.has(HDR_EXPIRES)) {
 326         const time_t e = header.getTime(HDR_EXPIRES);
 327         /*
 328          * HTTP/1.0 says that robust implementations should consider
 329          * bad or malformed Expires header as equivalent to "expires
 330          * immediately."
 331          */
 332         return e < 0 ? squid_curtime : e;
 333     }
 334
 335     return -1;
 336 }
 337
 338 /* sync this routine when you update HttpReply struct */
 339 void
 340 HttpReply::hdrCacheInit()
 341 {
 342     HttpMsg::hdrCacheInit();
 343
 344     http_ver = sline.version;
 345     content_length = header.getInt64(HDR_CONTENT_LENGTH);
 346     date = header.getTime(HDR_DATE);
 347     last_modified = header.getTime(HDR_LAST_MODIFIED);
 348     surrogate_control = header.getSc();
 349     content_range = header.getContRange();
 350     keep_alive = persistent() ? 1 : 0;
 351     const char *str = header.getStr(HDR_CONTENT_TYPE);
 352
 353     if (str)
 354         content_type.limitInit(str, strcspn(str, ";\t "));
 355     else
 356         content_type = String();
 357
 358     /* be sure to set expires after date and cache-control */
 359     expires = hdrExpirationTime();
 360 }
 361
 362 /* sync this routine when you update HttpReply struct */
 363 void
 364 HttpReply::hdrCacheClean()
 365 {
 366     content_type.clean();
 367
 368     if (cache_control) {
 369         delete cache_control;
 370         cache_control = NULL;
 371     }
 372
 373     if (surrogate_control) {
 374         delete surrogate_control;
 375         surrogate_control = NULL;
 376     }
 377
 378     if (content_range) {
 379         httpHdrContRangeDestroy(content_range);
 380         content_range = NULL;
 381     }
 382 }
 383
 384 /*
 385  * Returns the body size of a HTTP response
 386  */
 387 int64_t
 388 HttpReply::bodySize(const HttpRequestMethod& method) const
 389 {
 390     if (sline.version.major < 1)
 391         return -1;
 392     else if (method.id() == Http::METHOD_HEAD)
 393         return 0;
 394     else if (sline.status() == Http::scOkay)
 395         (void) 0;       /* common case, continue */
 396     else if (sline.status() == Http::scNoContent)
 397         return 0;
 398     else if (sline.status() == Http::scNotModified)
 399         return 0;
 400     else if (sline.status() < Http::scOkay)
 401         return 0;
 402
 403     return content_length;
 404 }
 405
 406 /**
 407  * Checks the first line of an HTTP Reply is valid.
 408  * currently only checks "HTTP/" exists.
 409  *
 410  * NP: not all error cases are detected yet. Some are left for detection later in parse.
 411  */
 412 bool
 413 HttpReply::sanityCheckStartLine(const char *buf, const size_t hdr_len, Http::StatusCode *error)
 414 {
 415     // hack warning: using psize instead of size here due to type mismatches with MemBuf.
 416
 417     // content is long enough to possibly hold a reply
 418     // 4 being magic size of a 3-digit number plus space delimiter
 419     if (hdr_len < (size_t)(protoPrefix.psize() + 4)) {
 420         if (hdr_len > 0) {
 421             debugs(58, 3, "Too small reply header (" << hdr_len << " bytes)");
 422             *error = Http::scInvalidHeader;
 423         }
 424         return false;
 425     }
 426
 427     int pos;
 428     // catch missing or mismatched protocol identifier
 429     // allow special-case for ICY protocol (non-HTTP identifier) in response to faked HTTP request.
 430     if (strncmp(buf, "ICY", 3) == 0) {
 431         protoPrefix = "ICY";
 432         pos = protoPrefix.psize();
 433     } else {
 434
 435         if (protoPrefix.cmp(buf, protoPrefix.size()) != 0) {
 436             debugs(58, 3, "missing protocol prefix (" << protoPrefix << ") in '" << buf << "'");
 437             *error = Http::scInvalidHeader;
 438             return false;
 439         }
 440
 441         // catch missing or negative status value (negative '-' is not a digit)
 442         pos = protoPrefix.psize();
 443
 444         // skip arbitrary number of digits and a dot in the verion portion
 445         while ((size_t)pos <= hdr_len && (*(buf+pos) == '.' || xisdigit(*(buf+pos)) ) ) ++pos;
 446
 447         // catch missing version info
 448         if (pos == protoPrefix.psize()) {
 449             debugs(58, 3, "missing protocol version numbers (ie. " << protoPrefix << "/1.0) in '" << buf << "'");
 450             *error = Http::scInvalidHeader;
 451             return false;
 452         }
 453     }
 454
 455     // skip arbitrary number of spaces...
 456     while ((size_t)pos <= hdr_len && (char)*(buf+pos) == ' ') ++pos;
 457
 458     if ((size_t)pos < hdr_len && !xisdigit(*(buf+pos))) {
 459         debugs(58, 3, "missing or invalid status number in '" << buf << "'");
 460         *error = Http::scInvalidHeader;
 461         return false;
 462     }
 463
 464     return true;
 465 }
 466
 467 bool
 468 HttpReply::parseFirstLine(const char *blk_start, const char *blk_end)
 469 {
 470     return sline.parse(protoPrefix, blk_start, blk_end);
 471 }
 472
 473 /* handy: resets and returns -1 */
 474 int
 475 HttpReply::httpMsgParseError()
 476 {
 477     int result(HttpMsg::httpMsgParseError());
 478     /* indicate an error in the status line */
 479     sline.set(Http::ProtocolVersion(), Http::scInvalidHeader);
 480     return result;
 481 }
 482
 483 /*
 484  * Indicate whether or not we would usually expect an entity-body
 485  * along with this response
 486  */
 487 bool
 488 HttpReply::expectingBody(const HttpRequestMethod& req_method, int64_t& theSize) const
 489 {
 490     bool expectBody = true;
 491
 492     if (req_method == Http::METHOD_HEAD)
 493         expectBody = false;
 494     else if (sline.status() == Http::scNoContent)
 495         expectBody = false;
 496     else if (sline.status() == Http::scNotModified)
 497         expectBody = false;
 498     else if (sline.status() < Http::scOkay)
 499         expectBody = false;
 500     else
 501         expectBody = true;
 502
 503     if (expectBody) {
 504         if (header.chunked())
 505             theSize = -1;
 506         else if (content_length >= 0)
 507             theSize = content_length;
 508         else
 509             theSize = -1;
 510     }
 511
 512     return expectBody;
 513 }
 514
 515 bool
 516 HttpReply::receivedBodyTooLarge(HttpRequest& request, int64_t receivedSize)
 517 {
 518     calcMaxBodySize(request);
 519     debugs(58, 3, HERE << receivedSize << " >? " << bodySizeMax);
 520     return bodySizeMax >= 0 && receivedSize > bodySizeMax;
 521 }
 522
 523 bool
 524 HttpReply::expectedBodyTooLarge(HttpRequest& request)
 525 {
 526     calcMaxBodySize(request);
 527     debugs(58, 7, HERE << "bodySizeMax=" << bodySizeMax);
 528
 529     if (bodySizeMax < 0) // no body size limit
 530         return false;
 531
 532     int64_t expectedSize = -1;
 533     if (!expectingBody(request.method, expectedSize))
 534         return false;
 535
 536     debugs(58, 6, HERE << expectedSize << " >? " << bodySizeMax);
 537
 538     if (expectedSize < 0) // expecting body of an unknown length
 539         return false;
 540
 541     return expectedSize > bodySizeMax;
 542 }
 543
 544 void
 545 HttpReply::calcMaxBodySize(HttpRequest& request) const
 546 {
 547     // hack: -2 is used as "we have not calculated max body size yet" state
 548     if (bodySizeMax != -2) // already tried
 549         return;
 550     bodySizeMax = -1;
 551
 552     // short-circuit ACL testing if there are none configured
 553     if (!Config.ReplyBodySize)
 554         return;
 555
 556     ACLFilledChecklist ch(NULL, &request, NULL);
 557     // XXX: cont-cast becomes irrelevant when checklist is HttpReply::Pointer
 558     ch.reply = const_cast<HttpReply *>(this);
 559     HTTPMSGLOCK(ch.reply);
 560     for (AclSizeLimit *l = Config.ReplyBodySize; l; l = l -> next) {
 561         /* if there is no ACL list or if the ACLs listed match use this size value */
 562         if (!l->aclList || ch.fastCheck(l->aclList) == ACCESS_ALLOWED) {
 563             debugs(58, 4, HERE << "bodySizeMax=" << bodySizeMax);
 564             bodySizeMax = l->size; // may be -1
 565             break;
 566         }
 567     }
 568 }
 569
 570 // XXX: check that this is sufficient for eCAP cloning
 571 HttpReply *
 572 HttpReply::clone() const
 573 {
 574     HttpReply *rep = new HttpReply();
 575     rep->sline = sline; // used in hdrCacheInit() call below
 576     rep->header.append(&header);
 577     rep->hdrCacheInit();
 578     rep->hdr_sz = hdr_sz;
 579     rep->http_ver = http_ver;
 580     rep->pstate = pstate;
 581     rep->body_pipe = body_pipe;
 582
 583     // keep_alive is handled in hdrCacheInit()
 584     return rep;
 585 }
 586
 587 bool HttpReply::inheritProperties(const HttpMsg *aMsg)
 588 {
 589     const HttpReply *aRep = dynamic_cast<const HttpReply*>(aMsg);
 590     if (!aRep)
 591         return false;
 592     keep_alive = aRep->keep_alive;
 593     return true;
 594 }
 595
 596 void HttpReply::removeStaleWarnings()
 597 {
 598     String warning;
 599     if (header.getList(HDR_WARNING, &warning)) {
 600         const String newWarning = removeStaleWarningValues(warning);
 601         if (warning.size() && warning.size() == newWarning.size())
 602             return; // some warnings are there and none changed
 603         header.delById(HDR_WARNING);
 604         if (newWarning.size()) { // some warnings left
 605             HttpHeaderEntry *const e =
 606                 new HttpHeaderEntry(HDR_WARNING, NULL, newWarning.termedBuf());
 607             header.addEntry(e);
 608         }
 609     }
 610 }
 611
 612 /**
 613  * Remove warning-values with warn-date different from Date value from
 614  * a single header entry. Returns a string with all valid warning-values.
 615  */
 616 String HttpReply::removeStaleWarningValues(const String &value)
 617 {
 618     String newValue;
 619     const char *item = 0;
 620     int len = 0;
 621     const char *pos = 0;
 622     while (strListGetItem(&value, ',', &item, &len, &pos)) {
 623         bool keep = true;
 624         // Does warning-value have warn-date (which contains quoted date)?
 625         // We scan backwards, looking for two quoted strings.
 626         // warning-value = warn-code SP warn-agent SP warn-text [SP warn-date]
 627         const char *p = item + len - 1;
 628
 629         while (p >= item && xisspace(*p)) --p; // skip whitespace
 630
 631         // warning-value MUST end with quote
 632         if (p >= item && *p == '"') {
 633             const char *const warnDateEnd = p;
 634             --p;
 635             while (p >= item && *p != '"') --p; // find the next quote
 636
 637             const char *warnDateBeg = p + 1;
 638             --p;
 639             while (p >= item && xisspace(*p)) --p; // skip whitespace
 640
 641             if (p >= item && *p == '"' && warnDateBeg - p > 2) {
 642                 // found warn-text
 643                 String warnDate;
 644                 warnDate.append(warnDateBeg, warnDateEnd - warnDateBeg);
 645                 const time_t time = parse_rfc1123(warnDate.termedBuf());
 646                 keep = (time > 0 && time == date); // keep valid and matching date
 647             }
 648         }
 649
 650         if (keep) {
 651             if (newValue.size())
 652                 newValue.append(", ");
 653             newValue.append(item, len);
 654         }
 655     }
 656
 657     return newValue;
 658 }
 659