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     mb->init();
 131     packInto(mb);
 132     return mb;
 133 }
 134
 135 HttpReply *
 136 HttpReply::make304() const
 137 {
 138     static const http_hdr_type ImsEntries[] = {HDR_DATE, HDR_CONTENT_TYPE, HDR_EXPIRES, HDR_LAST_MODIFIED, /* eof */ HDR_OTHER};
 139
 140     HttpReply *rv = new HttpReply;
 141     int t;
 142     HttpHeaderEntry *e;
 143
 144     /* rv->content_length; */
 145     rv->date = date;
 146     rv->last_modified = last_modified;
 147     rv->expires = expires;
 148     rv->content_type = content_type;
 149     /* rv->cache_control */
 150     /* rv->content_range */
 151     /* rv->keep_alive */
 152     rv->sline.set(Http::ProtocolVersion(), Http::scNotModified, NULL);
 153
 154     for (t = 0; ImsEntries[t] != HDR_OTHER; ++t)
 155         if ((e = header.findEntry(ImsEntries[t])))
 156             rv->header.addEntry(e->clone());
 157
 158     /* rv->body */
 159     return rv;
 160 }
 161
 162 MemBuf *
 163 HttpReply::packed304Reply()
 164 {
 165     /* Not as efficient as skipping the header duplication,
 166      * but easier to maintain
 167      */
 168     HttpReply *temp = make304();
 169     MemBuf *rv = temp->pack();
 170     delete temp;
 171     return rv;
 172 }
 173
 174 void
 175 HttpReply::setHeaders(Http::StatusCode status, const char *reason,
 176                       const char *ctype, int64_t clen, time_t lmt, time_t expiresTime)
 177 {
 178     HttpHeader *hdr;
 179     sline.set(Http::ProtocolVersion(), status, reason);
 180     hdr = &header;
 181     hdr->putStr(HDR_SERVER, visible_appname_string);
 182     hdr->putStr(HDR_MIME_VERSION, "1.0");
 183     hdr->putTime(HDR_DATE, squid_curtime);
 184
 185     if (ctype) {
 186         hdr->putStr(HDR_CONTENT_TYPE, ctype);
 187         content_type = ctype;
 188     } else
 189         content_type = String();
 190
 191     if (clen >= 0)
 192         hdr->putInt64(HDR_CONTENT_LENGTH, clen);
 193
 194     if (expiresTime >= 0)
 195         hdr->putTime(HDR_EXPIRES, expiresTime);
 196
 197     if (lmt > 0)        /* this used to be lmt != 0 @?@ */
 198         hdr->putTime(HDR_LAST_MODIFIED, lmt);
 199
 200     date = squid_curtime;
 201
 202     content_length = clen;
 203
 204     expires = expiresTime;
 205
 206     last_modified = lmt;
 207 }
 208
 209 void
 210 HttpReply::redirect(Http::StatusCode status, const char *loc)
 211 {
 212     HttpHeader *hdr;
 213     sline.set(Http::ProtocolVersion(), status, NULL);
 214     hdr = &header;
 215     hdr->putStr(HDR_SERVER, APP_FULLNAME);
 216     hdr->putTime(HDR_DATE, squid_curtime);
 217     hdr->putInt64(HDR_CONTENT_LENGTH, 0);
 218     hdr->putStr(HDR_LOCATION, loc);
 219     date = squid_curtime;
 220     content_length = 0;
 221 }
 222
 223 /* compare the validators of two replies.
 224  * 1 = they match
 225  * 0 = they do not match
 226  */
 227 int
 228 HttpReply::validatorsMatch(HttpReply const * otherRep) const
 229 {
 230     String one,two;
 231     assert (otherRep);
 232     /* Numbers first - easiest to check */
 233     /* Content-Length */
 234     /* TODO: remove -1 bypass */
 235
 236     if (content_length != otherRep->content_length
 237             && content_length > -1 &&
 238             otherRep->content_length > -1)
 239         return 0;
 240
 241     /* ETag */
 242     one = header.getStrOrList(HDR_ETAG);
 243
 244     two = otherRep->header.getStrOrList(HDR_ETAG);
 245
 246     if (one.size()==0 || two.size()==0 || one.caseCmp(two)!=0 ) {
 247         one.clean();
 248         two.clean();
 249         return 0;
 250     }
 251
 252     if (last_modified != otherRep->last_modified)
 253         return 0;
 254
 255     /* MD5 */
 256     one = header.getStrOrList(HDR_CONTENT_MD5);
 257
 258     two = otherRep->header.getStrOrList(HDR_CONTENT_MD5);
 259
 260     if (one.size()==0 || two.size()==0 || one.caseCmp(two)!=0 ) {
 261         one.clean();
 262         two.clean();
 263         return 0;
 264     }
 265
 266     return 1;
 267 }
 268
 269 void
 270 HttpReply::updateOnNotModified(HttpReply const * freshRep)
 271 {
 272     assert(freshRep);
 273
 274     /* clean cache */
 275     hdrCacheClean();
 276     /* update raw headers */
 277     header.update(&freshRep->header,
 278                   (const HttpHeaderMask *) &Denied304HeadersMask);
 279
 280     header.compact();
 281     /* init cache */
 282     hdrCacheInit();
 283 }
 284
 285 /* internal routines */
 286
 287 time_t
 288 HttpReply::hdrExpirationTime()
 289 {
 290     /* The s-maxage and max-age directive takes priority over Expires */
 291
 292     if (cache_control) {
 293         if (date >= 0) {
 294             if (cache_control->hasSMaxAge())
 295                 return date + cache_control->sMaxAge();
 296
 297             if (cache_control->hasMaxAge())
 298                 return date + cache_control->maxAge();
 299         } else {
 300             /*
 301              * Conservatively handle the case when we have a max-age
 302              * header, but no Date for reference?
 303              */
 304
 305             if (cache_control->hasSMaxAge())
 306                 return squid_curtime;
 307
 308             if (cache_control->hasMaxAge())
 309                 return squid_curtime;
 310         }
 311     }
 312
 313     if (Config.onoff.vary_ignore_expire &&
 314             header.has(HDR_VARY)) {
 315         const time_t d = header.getTime(HDR_DATE);
 316         const time_t e = header.getTime(HDR_EXPIRES);
 317
 318         if (d == e)
 319             return -1;
 320     }
 321
 322     if (header.has(HDR_EXPIRES)) {
 323         const time_t e = header.getTime(HDR_EXPIRES);
 324         /*
 325          * HTTP/1.0 says that robust implementations should consider
 326          * bad or malformed Expires header as equivalent to "expires
 327          * immediately."
 328          */
 329         return e < 0 ? squid_curtime : e;
 330     }
 331
 332     return -1;
 333 }
 334
 335 /* sync this routine when you update HttpReply struct */
 336 void
 337 HttpReply::hdrCacheInit()
 338 {
 339     HttpMsg::hdrCacheInit();
 340
 341     http_ver = sline.version;
 342     content_length = header.getInt64(HDR_CONTENT_LENGTH);
 343     date = header.getTime(HDR_DATE);
 344     last_modified = header.getTime(HDR_LAST_MODIFIED);
 345     surrogate_control = header.getSc();
 346     content_range = header.getContRange();
 347     keep_alive = persistent() ? 1 : 0;
 348     const char *str = header.getStr(HDR_CONTENT_TYPE);
 349
 350     if (str)
 351         content_type.limitInit(str, strcspn(str, ";\t "));
 352     else
 353         content_type = String();
 354
 355     /* be sure to set expires after date and cache-control */
 356     expires = hdrExpirationTime();
 357 }
 358
 359 /* sync this routine when you update HttpReply struct */
 360 void
 361 HttpReply::hdrCacheClean()
 362 {
 363     content_type.clean();
 364
 365     if (cache_control) {
 366         delete cache_control;
 367         cache_control = NULL;
 368     }
 369
 370     if (surrogate_control) {
 371         delete surrogate_control;
 372         surrogate_control = NULL;
 373     }
 374
 375     if (content_range) {
 376         httpHdrContRangeDestroy(content_range);
 377         content_range = NULL;
 378     }
 379 }
 380
 381 /*
 382  * Returns the body size of a HTTP response
 383  */
 384 int64_t
 385 HttpReply::bodySize(const HttpRequestMethod& method) const
 386 {
 387     if (sline.version.major < 1)
 388         return -1;
 389     else if (method.id() == Http::METHOD_HEAD)
 390         return 0;
 391     else if (sline.status() == Http::scOkay)
 392         (void) 0;       /* common case, continue */
 393     else if (sline.status() == Http::scNoContent)
 394         return 0;
 395     else if (sline.status() == Http::scNotModified)
 396         return 0;
 397     else if (sline.status() < Http::scOkay)
 398         return 0;
 399
 400     return content_length;
 401 }
 402
 403 /**
 404  * Checks the first line of an HTTP Reply is valid.
 405  * currently only checks "HTTP/" exists.
 406  *
 407  * NP: not all error cases are detected yet. Some are left for detection later in parse.
 408  */
 409 bool
 410 HttpReply::sanityCheckStartLine(const char *buf, const size_t hdr_len, Http::StatusCode *error)
 411 {
 412     // hack warning: using psize instead of size here due to type mismatches with MemBuf.
 413
 414     // content is long enough to possibly hold a reply
 415     // 4 being magic size of a 3-digit number plus space delimiter
 416     if (hdr_len < (size_t)(protoPrefix.psize() + 4)) {
 417         if (hdr_len > 0) {
 418             debugs(58, 3, "Too small reply header (" << hdr_len << " bytes)");
 419             *error = Http::scInvalidHeader;
 420         }
 421         return false;
 422     }
 423
 424     int pos;
 425     // catch missing or mismatched protocol identifier
 426     // allow special-case for ICY protocol (non-HTTP identifier) in response to faked HTTP request.
 427     if (strncmp(buf, "ICY", 3) == 0) {
 428         protoPrefix = "ICY";
 429         pos = protoPrefix.psize();
 430     } else {
 431
 432         if (protoPrefix.cmp(buf, protoPrefix.size()) != 0) {
 433             debugs(58, 3, "missing protocol prefix (" << protoPrefix << ") in '" << buf << "'");
 434             *error = Http::scInvalidHeader;
 435             return false;
 436         }
 437
 438         // catch missing or negative status value (negative '-' is not a digit)
 439         pos = protoPrefix.psize();
 440
 441         // skip arbitrary number of digits and a dot in the verion portion
 442         while ((size_t)pos <= hdr_len && (*(buf+pos) == '.' || xisdigit(*(buf+pos)) ) ) ++pos;
 443
 444         // catch missing version info
 445         if (pos == protoPrefix.psize()) {
 446             debugs(58, 3, "missing protocol version numbers (ie. " << protoPrefix << "/1.0) in '" << buf << "'");
 447             *error = Http::scInvalidHeader;
 448             return false;
 449         }
 450     }
 451
 452     // skip arbitrary number of spaces...
 453     while ((size_t)pos <= hdr_len && (char)*(buf+pos) == ' ') ++pos;
 454
 455     if ((size_t)pos < hdr_len && !xisdigit(*(buf+pos))) {
 456         debugs(58, 3, "missing or invalid status number in '" << buf << "'");
 457         *error = Http::scInvalidHeader;
 458         return false;
 459     }
 460
 461     return true;
 462 }
 463
 464 bool
 465 HttpReply::parseFirstLine(const char *blk_start, const char *blk_end)
 466 {
 467     return sline.parse(protoPrefix, blk_start, blk_end);
 468 }
 469
 470 /* handy: resets and returns -1 */
 471 int
 472 HttpReply::httpMsgParseError()
 473 {
 474     int result(HttpMsg::httpMsgParseError());
 475     /* indicate an error in the status line */
 476     sline.set(Http::ProtocolVersion(), Http::scInvalidHeader);
 477     return result;
 478 }
 479
 480 /*
 481  * Indicate whether or not we would usually expect an entity-body
 482  * along with this response
 483  */
 484 bool
 485 HttpReply::expectingBody(const HttpRequestMethod& req_method, int64_t& theSize) const
 486 {
 487     bool expectBody = true;
 488
 489     if (req_method == Http::METHOD_HEAD)
 490         expectBody = false;
 491     else if (sline.status() == Http::scNoContent)
 492         expectBody = false;
 493     else if (sline.status() == Http::scNotModified)
 494         expectBody = false;
 495     else if (sline.status() < Http::scOkay)
 496         expectBody = false;
 497     else
 498         expectBody = true;
 499
 500     if (expectBody) {
 501         if (header.chunked())
 502             theSize = -1;
 503         else if (content_length >= 0)
 504             theSize = content_length;
 505         else
 506             theSize = -1;
 507     }
 508
 509     return expectBody;
 510 }
 511
 512 bool
 513 HttpReply::receivedBodyTooLarge(HttpRequest& request, int64_t receivedSize)
 514 {
 515     calcMaxBodySize(request);
 516     debugs(58, 3, HERE << receivedSize << " >? " << bodySizeMax);
 517     return bodySizeMax >= 0 && receivedSize > bodySizeMax;
 518 }
 519
 520 bool
 521 HttpReply::expectedBodyTooLarge(HttpRequest& request)
 522 {
 523     calcMaxBodySize(request);
 524     debugs(58, 7, HERE << "bodySizeMax=" << bodySizeMax);
 525
 526     if (bodySizeMax < 0) // no body size limit
 527         return false;
 528
 529     int64_t expectedSize = -1;
 530     if (!expectingBody(request.method, expectedSize))
 531         return false;
 532
 533     debugs(58, 6, HERE << expectedSize << " >? " << bodySizeMax);
 534
 535     if (expectedSize < 0) // expecting body of an unknown length
 536         return false;
 537
 538     return expectedSize > bodySizeMax;
 539 }
 540
 541 void
 542 HttpReply::calcMaxBodySize(HttpRequest& request) const
 543 {
 544     // hack: -2 is used as "we have not calculated max body size yet" state
 545     if (bodySizeMax != -2) // already tried
 546         return;
 547     bodySizeMax = -1;
 548
 549     // short-circuit ACL testing if there are none configured
 550     if (!Config.ReplyBodySize)
 551         return;
 552
 553     ACLFilledChecklist ch(NULL, &request, NULL);
 554     // XXX: cont-cast becomes irrelevant when checklist is HttpReply::Pointer
 555     ch.reply = const_cast<HttpReply *>(this);
 556     HTTPMSGLOCK(ch.reply);
 557     for (AclSizeLimit *l = Config.ReplyBodySize; l; l = l -> next) {
 558         /* if there is no ACL list or if the ACLs listed match use this size value */
 559         if (!l->aclList || ch.fastCheck(l->aclList) == ACCESS_ALLOWED) {
 560             debugs(58, 4, HERE << "bodySizeMax=" << bodySizeMax);
 561             bodySizeMax = l->size; // may be -1
 562             break;
 563         }
 564     }
 565 }
 566
 567 // XXX: check that this is sufficient for eCAP cloning
 568 HttpReply *
 569 HttpReply::clone() const
 570 {
 571     HttpReply *rep = new HttpReply();
 572     rep->sline = sline; // used in hdrCacheInit() call below
 573     rep->header.append(&header);
 574     rep->hdrCacheInit();
 575     rep->hdr_sz = hdr_sz;
 576     rep->http_ver = http_ver;
 577     rep->pstate = pstate;
 578     rep->body_pipe = body_pipe;
 579
 580     // keep_alive is handled in hdrCacheInit()
 581     return rep;
 582 }
 583
 584 bool HttpReply::inheritProperties(const HttpMsg *aMsg)
 585 {
 586     const HttpReply *aRep = dynamic_cast<const HttpReply*>(aMsg);
 587     if (!aRep)
 588         return false;
 589     keep_alive = aRep->keep_alive;
 590     return true;
 591 }
 592
 593 void HttpReply::removeStaleWarnings()
 594 {
 595     String warning;
 596     if (header.getList(HDR_WARNING, &warning)) {
 597         const String newWarning = removeStaleWarningValues(warning);
 598         if (warning.size() && warning.size() == newWarning.size())
 599             return; // some warnings are there and none changed
 600         header.delById(HDR_WARNING);
 601         if (newWarning.size()) { // some warnings left
 602             HttpHeaderEntry *const e =
 603                 new HttpHeaderEntry(HDR_WARNING, NULL, newWarning.termedBuf());
 604             header.addEntry(e);
 605         }
 606     }
 607 }
 608
 609 /**
 610  * Remove warning-values with warn-date different from Date value from
 611  * a single header entry. Returns a string with all valid warning-values.
 612  */
 613 String HttpReply::removeStaleWarningValues(const String &value)
 614 {
 615     String newValue;
 616     const char *item = 0;
 617     int len = 0;
 618     const char *pos = 0;
 619     while (strListGetItem(&value, ',', &item, &len, &pos)) {
 620         bool keep = true;
 621         // Does warning-value have warn-date (which contains quoted date)?
 622         // We scan backwards, looking for two quoted strings.
 623         // warning-value = warn-code SP warn-agent SP warn-text [SP warn-date]
 624         const char *p = item + len - 1;
 625
 626         while (p >= item && xisspace(*p)) --p; // skip whitespace
 627
 628         // warning-value MUST end with quote
 629         if (p >= item && *p == '"') {
 630             const char *const warnDateEnd = p;
 631             --p;
 632             while (p >= item && *p != '"') --p; // find the next quote
 633
 634             const char *warnDateBeg = p + 1;
 635             --p;
 636             while (p >= item && xisspace(*p)) --p; // skip whitespace
 637
 638             if (p >= item && *p == '"' && warnDateBeg - p > 2) {
 639                 // found warn-text
 640                 String warnDate;
 641                 warnDate.append(warnDateBeg, warnDateEnd - warnDateBeg);
 642                 const time_t time = parse_rfc1123(warnDate.termedBuf());
 643                 keep = (time > 0 && time == date); // keep valid and matching date
 644             }
 645         }
 646
 647         if (keep) {
 648             if (newValue.size())
 649                 newValue.append(", ");
 650             newValue.append(item, len);
 651         }
 652     }
 653
 654     return newValue;
 655 }
 656