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