src/HttpRequest.cc

   1 /*
   2  * Copyright (C) 1996-2014 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 73    HTTP Request */
  10
  11 #include "squid.h"
  12 #include "AccessLogEntry.h"
  13 #include "acl/AclSizeLimit.h"
  14 #include "acl/FilledChecklist.h"
  15 #include "client_side.h"
  16 #include "DnsLookupDetails.h"
  17 #include "err_detail_type.h"
  18 #include "globals.h"
  19 #include "gopher.h"
  20 #include "http.h"
  21 #include "HttpHdrCc.h"
  22 #include "HttpHeaderRange.h"
  23 #include "HttpRequest.h"
  24 #include "log/Config.h"
  25 #include "MemBuf.h"
  26 #include "SquidConfig.h"
  27 #include "Store.h"
  28 #include "URL.h"
  29
  30 #if USE_AUTH
  31 #include "auth/UserRequest.h"
  32 #endif
  33 #if ICAP_CLIENT
  34 #include "adaptation/icap/icap_log.h"
  35 #endif
  36
  37 HttpRequest::HttpRequest() :
  38         HttpMsg(hoRequest)
  39 {
  40     init();
  41 }
  42
  43 HttpRequest::HttpRequest(const HttpRequestMethod& aMethod, AnyP::ProtocolType aProtocol, const char *aUrlpath) :
  44         HttpMsg(hoRequest)
  45 {
  46     static unsigned int id = 1;
  47     debugs(93,7, HERE << "constructed, this=" << this << " id=" << ++id);
  48     init();
  49     initHTTP(aMethod, aProtocol, aUrlpath);
  50 }
  51
  52 HttpRequest::~HttpRequest()
  53 {
  54     clean();
  55     debugs(93,7, HERE << "destructed, this=" << this);
  56 }
  57
  58 void
  59 HttpRequest::initHTTP(const HttpRequestMethod& aMethod, AnyP::ProtocolType aProtocol, const char *aUrlpath)
  60 {
  61     method = aMethod;
  62     url.setScheme(aProtocol);
  63     urlpath = aUrlpath;
  64 }
  65
  66 void
  67 HttpRequest::init()
  68 {
  69     method = Http::METHOD_NONE;
  70     url.clear();
  71     urlpath = NULL;
  72     login[0] = '\0';
  73     host[0] = '\0';
  74     host_is_numeric = -1;
  75 #if USE_AUTH
  76     auth_user_request = NULL;
  77 #endif
  78     port = 0;
  79     canonical = NULL;
  80     memset(&flags, '\0', sizeof(flags));
  81     range = NULL;
  82     ims = -1;
  83     imslen = 0;
  84     lastmod = -1;
  85     client_addr.setEmpty();
  86     my_addr.setEmpty();
  87     body_pipe = NULL;
  88     // hier
  89     dnsWait = -1;
  90     errType = ERR_NONE;
  91     errDetail = ERR_DETAIL_NONE;
  92     peer_login = NULL;          // not allocated/deallocated by this class
  93     peer_domain = NULL;         // not allocated/deallocated by this class
  94     peer_host = NULL;
  95     vary_headers = NULL;
  96     myportname = null_string;
  97     tag = null_string;
  98 #if USE_AUTH
  99     extacl_user = null_string;
 100     extacl_passwd = null_string;
 101 #endif
 102     extacl_log = null_string;
 103     extacl_message = null_string;
 104     pstate = psReadyToParseStartLine;
 105 #if FOLLOW_X_FORWARDED_FOR
 106     indirect_client_addr.setEmpty();
 107 #endif /* FOLLOW_X_FORWARDED_FOR */
 108 #if USE_ADAPTATION
 109     adaptHistory_ = NULL;
 110 #endif
 111 #if ICAP_CLIENT
 112     icapHistory_ = NULL;
 113 #endif
 114     rangeOffsetLimit = -2; //a value of -2 means not checked yet
 115 }
 116
 117 void
 118 HttpRequest::clean()
 119 {
 120     // we used to assert that the pipe is NULL, but now the request only
 121     // points to a pipe that is owned and initiated by another object.
 122     body_pipe = NULL;
 123 #if USE_AUTH
 124     auth_user_request = NULL;
 125 #endif
 126     safe_free(canonical);
 127
 128     safe_free(vary_headers);
 129
 130     url.clear();
 131     urlpath.clean();
 132
 133     header.clean();
 134
 135     if (cache_control) {
 136         delete cache_control;
 137         cache_control = NULL;
 138     }
 139
 140     if (range) {
 141         delete range;
 142         range = NULL;
 143     }
 144
 145     myportname.clean();
 146
 147     notes = NULL;
 148
 149     tag.clean();
 150 #if USE_AUTH
 151     extacl_user.clean();
 152     extacl_passwd.clean();
 153 #endif
 154     extacl_log.clean();
 155
 156     extacl_message.clean();
 157
 158     etag.clean();
 159
 160 #if USE_ADAPTATION
 161     adaptHistory_ = NULL;
 162 #endif
 163 #if ICAP_CLIENT
 164     icapHistory_ = NULL;
 165 #endif
 166 }
 167
 168 void
 169 HttpRequest::reset()
 170 {
 171     clean();
 172     init();
 173 }
 174
 175 HttpRequest *
 176 HttpRequest::clone() const
 177 {
 178     HttpRequest *copy = new HttpRequest(method, url.getScheme(), urlpath.termedBuf());
 179     // TODO: move common cloning clone to Msg::copyTo() or copy ctor
 180     copy->header.append(&header);
 181     copy->hdrCacheInit();
 182     copy->hdr_sz = hdr_sz;
 183     copy->http_ver = http_ver;
 184     copy->pstate = pstate; // TODO: should we assert a specific state here?
 185     copy->body_pipe = body_pipe;
 186
 187     strncpy(copy->login, login, sizeof(login)); // MAX_LOGIN_SZ
 188     strncpy(copy->host, host, sizeof(host)); // SQUIDHOSTNAMELEN
 189     copy->host_addr = host_addr;
 190
 191     copy->port = port;
 192     // urlPath handled in ctor
 193     copy->canonical = canonical ? xstrdup(canonical) : NULL;
 194
 195     // range handled in hdrCacheInit()
 196     copy->ims = ims;
 197     copy->imslen = imslen;
 198     copy->hier = hier; // Is it safe to copy? Should we?
 199
 200     copy->errType = errType;
 201
 202     // XXX: what to do with copy->peer_login?
 203
 204     copy->lastmod = lastmod;
 205     copy->etag = etag;
 206     copy->vary_headers = vary_headers ? xstrdup(vary_headers) : NULL;
 207     // XXX: what to do with copy->peer_domain?
 208
 209     copy->tag = tag;
 210     copy->extacl_log = extacl_log;
 211     copy->extacl_message = extacl_message;
 212
 213     const bool inheritWorked = copy->inheritProperties(this);
 214     assert(inheritWorked);
 215
 216     return copy;
 217 }
 218
 219 bool
 220 HttpRequest::inheritProperties(const HttpMsg *aMsg)
 221 {
 222     const HttpRequest* aReq = dynamic_cast<const HttpRequest*>(aMsg);
 223     if (!aReq)
 224         return false;
 225
 226     client_addr = aReq->client_addr;
 227 #if FOLLOW_X_FORWARDED_FOR
 228     indirect_client_addr = aReq->indirect_client_addr;
 229 #endif
 230     my_addr = aReq->my_addr;
 231
 232     dnsWait = aReq->dnsWait;
 233
 234 #if USE_ADAPTATION
 235     adaptHistory_ = aReq->adaptHistory();
 236 #endif
 237 #if ICAP_CLIENT
 238     icapHistory_ = aReq->icapHistory();
 239 #endif
 240
 241     // This may be too conservative for the 204 No Content case
 242     // may eventually need cloneNullAdaptationImmune() for that.
 243     flags = aReq->flags.cloneAdaptationImmune();
 244
 245     errType = aReq->errType;
 246     errDetail = aReq->errDetail;
 247 #if USE_AUTH
 248     auth_user_request = aReq->auth_user_request;
 249     extacl_user = aReq->extacl_user;
 250     extacl_passwd = aReq->extacl_passwd;
 251 #endif
 252
 253     myportname = aReq->myportname;
 254
 255     // main property is which connection the request was received on (if any)
 256     clientConnectionManager = aReq->clientConnectionManager;
 257
 258     notes = aReq->notes;
 259     return true;
 260 }
 261
 262 /**
 263  * Checks the first line of an HTTP request is valid
 264  * currently just checks the request method is present.
 265  *
 266  * NP: Other errors are left for detection later in the parse.
 267  */
 268 bool
 269 HttpRequest::sanityCheckStartLine(MemBuf *buf, const size_t hdr_len, Http::StatusCode *error)
 270 {
 271     // content is long enough to possibly hold a reply
 272     // 2 being magic size of a 1-byte request method plus space delimiter
 273     if ( buf->contentSize() < 2 ) {
 274         // this is ony a real error if the headers apparently complete.
 275         if (hdr_len > 0) {
 276             debugs(58, 3, HERE << "Too large request header (" << hdr_len << " bytes)");
 277             *error = Http::scInvalidHeader;
 278         }
 279         return false;
 280     }
 281
 282     /* See if the request buffer starts with a known HTTP request method. */
 283     if (HttpRequestMethod(buf->content(),NULL) == Http::METHOD_NONE) {
 284         debugs(73, 3, "HttpRequest::sanityCheckStartLine: did not find HTTP request method");
 285         *error = Http::scInvalidHeader;
 286         return false;
 287     }
 288
 289     return true;
 290 }
 291
 292 bool
 293 HttpRequest::parseFirstLine(const char *start, const char *end)
 294 {
 295     const char *t = start + strcspn(start, w_space);
 296     method = HttpRequestMethod(start, t);
 297
 298     if (method == Http::METHOD_NONE)
 299         return false;
 300
 301     start = t + strspn(t, w_space);
 302
 303     const char *ver = findTrailingHTTPVersion(start, end);
 304
 305     if (ver) {
 306         end = ver - 1;
 307
 308         while (xisspace(*end)) // find prev non-space
 309             --end;
 310
 311         ++end;                 // back to space
 312
 313         if (2 != sscanf(ver + 5, "%d.%d", &http_ver.major, &http_ver.minor)) {
 314             debugs(73, DBG_IMPORTANT, "parseRequestLine: Invalid HTTP identifier.");
 315             return false;
 316         }
 317     } else {
 318         http_ver.major = 0;
 319         http_ver.minor = 9;
 320     }
 321
 322     if (end < start)   // missing URI
 323         return false;
 324
 325     char save = *end;
 326
 327     * (char *) end = '\0';     // temp terminate URI, XXX dangerous?
 328
 329     HttpRequest *tmp = urlParse(method, (char *) start, this);
 330
 331     * (char *) end = save;
 332
 333     if (NULL == tmp)
 334         return false;
 335
 336     return true;
 337 }
 338
 339 int
 340 HttpRequest::parseHeader(const char *parse_start, int len)
 341 {
 342     const char *blk_start, *blk_end;
 343
 344     if (!httpMsgIsolateHeaders(&parse_start, len, &blk_start, &blk_end))
 345         return 0;
 346
 347     int result = header.parse(blk_start, blk_end);
 348
 349     if (result)
 350         hdrCacheInit();
 351
 352     return result;
 353 }
 354
 355 /* swaps out request using httpRequestPack */
 356 void
 357 HttpRequest::swapOut(StoreEntry * e)
 358 {
 359     Packer p;
 360     assert(e);
 361     packerToStoreInit(&p, e);
 362     pack(&p);
 363     packerClean(&p);
 364 }
 365
 366 /* packs request-line and headers, appends <crlf> terminator */
 367 void
 368 HttpRequest::pack(Packer * p)
 369 {
 370     assert(p);
 371     /* pack request-line */
 372     packerPrintf(p, SQUIDSBUFPH " " SQUIDSTRINGPH " HTTP/%d.%d\r\n",
 373                  SQUIDSBUFPRINT(method.image()), SQUIDSTRINGPRINT(urlpath),
 374                  http_ver.major, http_ver.minor);
 375     /* headers */
 376     header.packInto(p);
 377     /* trailer */
 378     packerAppend(p, "\r\n", 2);
 379 }
 380
 381 /*
 382  * A wrapper for debugObj()
 383  */
 384 void
 385 httpRequestPack(void *obj, Packer *p)
 386 {
 387     HttpRequest *request = static_cast<HttpRequest*>(obj);
 388     request->pack(p);
 389 }
 390
 391 /* returns the length of request line + headers + crlf */
 392 int
 393 HttpRequest::prefixLen()
 394 {
 395     return method.image().length() + 1 +
 396            urlpath.size() + 1 +
 397            4 + 1 + 3 + 2 +
 398            header.len + 2;
 399 }
 400
 401 /* sync this routine when you update HttpRequest struct */
 402 void
 403 HttpRequest::hdrCacheInit()
 404 {
 405     HttpMsg::hdrCacheInit();
 406
 407     assert(!range);
 408     range = header.getRange();
 409 }
 410
 411 #if ICAP_CLIENT
 412 Adaptation::Icap::History::Pointer
 413 HttpRequest::icapHistory() const
 414 {
 415     if (!icapHistory_) {
 416         if (Log::TheConfig.hasIcapToken || IcapLogfileStatus == LOG_ENABLE) {
 417             icapHistory_ = new Adaptation::Icap::History();
 418             debugs(93,4, HERE << "made " << icapHistory_ << " for " << this);
 419         }
 420     }
 421
 422     return icapHistory_;
 423 }
 424 #endif
 425
 426 #if USE_ADAPTATION
 427 Adaptation::History::Pointer
 428 HttpRequest::adaptHistory(bool createIfNone) const
 429 {
 430     if (!adaptHistory_ && createIfNone) {
 431         adaptHistory_ = new Adaptation::History();
 432         debugs(93,4, HERE << "made " << adaptHistory_ << " for " << this);
 433     }
 434
 435     return adaptHistory_;
 436 }
 437
 438 Adaptation::History::Pointer
 439 HttpRequest::adaptLogHistory() const
 440 {
 441     return HttpRequest::adaptHistory(Log::TheConfig.hasAdaptToken);
 442 }
 443
 444 void
 445 HttpRequest::adaptHistoryImport(const HttpRequest &them)
 446 {
 447     if (!adaptHistory_) {
 448         adaptHistory_ = them.adaptHistory_; // may be nil
 449     } else {
 450         // check that histories did not diverge
 451         Must(!them.adaptHistory_ || them.adaptHistory_ == adaptHistory_);
 452     }
 453 }
 454
 455 #endif
 456
 457 bool
 458 HttpRequest::multipartRangeRequest() const
 459 {
 460     return (range && range->specs.size() > 1);
 461 }
 462
 463 bool
 464 HttpRequest::bodyNibbled() const
 465 {
 466     return body_pipe != NULL && body_pipe->consumedSize() > 0;
 467 }
 468
 469 void
 470 HttpRequest::detailError(err_type aType, int aDetail)
 471 {
 472     if (errType || errDetail)
 473         debugs(11, 5, HERE << "old error details: " << errType << '/' << errDetail);
 474     debugs(11, 5, HERE << "current error details: " << aType << '/' << aDetail);
 475     // checking type and detail separately may cause inconsistency, but
 476     // may result in more details available if they only become available later
 477     if (!errType)
 478         errType = aType;
 479     if (!errDetail)
 480         errDetail = aDetail;
 481 }
 482
 483 void
 484 HttpRequest::clearError()
 485 {
 486     debugs(11, 7, HERE << "old error details: " << errType << '/' << errDetail);
 487     errType = ERR_NONE;
 488     errDetail = ERR_DETAIL_NONE;
 489 }
 490
 491 const char *HttpRequest::packableURI(bool full_uri) const
 492 {
 493     if (full_uri)
 494         return urlCanonical((HttpRequest*)this);
 495
 496     if (urlpath.size())
 497         return urlpath.termedBuf();
 498
 499     return "/";
 500 }
 501
 502 void HttpRequest::packFirstLineInto(Packer * p, bool full_uri) const
 503 {
 504     // form HTTP request-line
 505     packerPrintf(p, SQUIDSBUFPH " %s HTTP/%d.%d\r\n",
 506                  SQUIDSBUFPRINT(method.image()),
 507                  packableURI(full_uri),
 508                  http_ver.major, http_ver.minor);
 509 }
 510
 511 /*
 512  * Indicate whether or not we would expect an entity-body
 513  * along with this request
 514  */
 515 bool
 516 HttpRequest::expectingBody(const HttpRequestMethod& unused, int64_t& theSize) const
 517 {
 518     bool expectBody = false;
 519
 520     /*
 521      * Note: Checks for message validity is in clientIsContentLengthValid().
 522      * this just checks if a entity-body is expected based on HTTP message syntax
 523      */
 524     if (header.chunked()) {
 525         expectBody = true;
 526         theSize = -1;
 527     } else if (content_length >= 0) {
 528         expectBody = true;
 529         theSize = content_length;
 530     } else {
 531         expectBody = false;
 532         // theSize undefined
 533     }
 534
 535     return expectBody;
 536 }
 537
 538 /*
 539  * Create a Request from a URL and METHOD.
 540  *
 541  * If the METHOD is CONNECT, then a host:port pair is looked for instead of a URL.
 542  * If the request cannot be created cleanly, NULL is returned
 543  */
 544 HttpRequest *
 545 HttpRequest::CreateFromUrlAndMethod(char * url, const HttpRequestMethod& method)
 546 {
 547     return urlParse(method, url, NULL);
 548 }
 549
 550 /*
 551  * Create a Request from a URL.
 552  *
 553  * If the request cannot be created cleanly, NULL is returned
 554  */
 555 HttpRequest *
 556 HttpRequest::CreateFromUrl(char * url)
 557 {
 558     return urlParse(Http::METHOD_GET, url, NULL);
 559 }
 560
 561 /**
 562  * Are responses to this request possible cacheable ?
 563  * If false then no matter what the response must not be cached.
 564  */
 565 bool
 566 HttpRequest::maybeCacheable()
 567 {
 568     // Intercepted request with Host: header which cannot be trusted.
 569     // Because it failed verification, or someone bypassed the security tests
 570     // we cannot cache the reponse for sharing between clients.
 571     // TODO: update cache to store for particular clients only (going to same Host: and destination IP)
 572     if (!flags.hostVerified && (flags.intercepted || flags.interceptTproxy))
 573         return false;
 574
 575     switch (url.getScheme()) {
 576     case AnyP::PROTO_HTTP:
 577     case AnyP::PROTO_HTTPS:
 578         if (!method.respMaybeCacheable())
 579             return false;
 580
 581         // XXX: this would seem the correct place to detect request cache-controls
 582         //      no-store, private and related which block cacheability
 583         break;
 584
 585     case AnyP::PROTO_GOPHER:
 586         if (!gopherCachable(this))
 587             return false;
 588         break;
 589
 590     case AnyP::PROTO_CACHE_OBJECT:
 591         return false;
 592
 593         //case AnyP::PROTO_FTP:
 594     default:
 595         break;
 596     }
 597
 598     return true;
 599 }
 600
 601 bool
 602 HttpRequest::conditional() const
 603 {
 604     return flags.ims ||
 605            header.has(HDR_IF_MATCH) ||
 606            header.has(HDR_IF_NONE_MATCH);
 607 }
 608
 609 void
 610 HttpRequest::recordLookup(const DnsLookupDetails &dns)
 611 {
 612     if (dns.wait >= 0) { // known delay
 613         if (dnsWait >= 0) // have recorded DNS wait before
 614             dnsWait += dns.wait;
 615         else
 616             dnsWait = dns.wait;
 617     }
 618 }
 619
 620 int64_t
 621 HttpRequest::getRangeOffsetLimit()
 622 {
 623     /* -2 is the starting value of rangeOffsetLimit.
 624      * If it is -2, that means we haven't checked it yet.
 625      *  Otherwise, return the current value */
 626     if (rangeOffsetLimit != -2)
 627         return rangeOffsetLimit;
 628
 629     rangeOffsetLimit = 0; // default value for rangeOffsetLimit
 630
 631     ACLFilledChecklist ch(NULL, this, NULL);
 632     ch.src_addr = client_addr;
 633     ch.my_addr =  my_addr;
 634
 635     for (AclSizeLimit *l = Config.rangeOffsetLimit; l; l = l -> next) {
 636         /* if there is no ACL list or if the ACLs listed match use this limit value */
 637         if (!l->aclList || ch.fastCheck(l->aclList) == ACCESS_ALLOWED) {
 638             debugs(58, 4, HERE << "rangeOffsetLimit=" << rangeOffsetLimit);
 639             rangeOffsetLimit = l->size; // may be -1
 640             break;
 641         }
 642     }
 643
 644     return rangeOffsetLimit;
 645 }
 646
 647 void
 648 HttpRequest::ignoreRange(const char *reason)
 649 {
 650     if (range) {
 651         debugs(73, 3, static_cast<void*>(range) << " for " << reason);
 652         delete range;
 653         range = NULL;
 654     }
 655     // Some callers also reset isRanged but it may not be safe for all callers:
 656     // isRanged is used to determine whether a weak ETag comparison is allowed,
 657     // and that check should not ignore the Range header if it was present.
 658     // TODO: Some callers also delete HDR_RANGE, HDR_REQUEST_RANGE. Should we?
 659 }
 660
 661 bool
 662 HttpRequest::canHandle1xx() const
 663 {
 664     // old clients do not support 1xx unless they sent Expect: 100-continue
 665     // (we reject all other HDR_EXPECT values so just check for HDR_EXPECT)
 666     if (http_ver <= Http::ProtocolVersion(1,0) && !header.has(HDR_EXPECT))
 667         return false;
 668
 669     // others must support 1xx control messages
 670     return true;
 671 }
 672
 673 ConnStateData *
 674 HttpRequest::pinnedConnection()
 675 {
 676     if (clientConnectionManager.valid() && clientConnectionManager->pinning.pinned)
 677         return clientConnectionManager.get();
 678     return NULL;
 679 }
 680
 681 const char *
 682 HttpRequest::storeId()
 683 {
 684     if (store_id.size() != 0) {
 685         debugs(73, 3, "sent back store_id:" << store_id);
 686
 687         return store_id.termedBuf();
 688     }
 689     debugs(73, 3, "sent back canonicalUrl:" << urlCanonical(this) );
 690
 691     return urlCanonical(this);
 692 }