methods/https.cc

   1 //-*- mode: cpp; mode: fold -*-
   2 // Description                                                          /*{{{*/
   3 // $Id: http.cc,v 1.59 2004/05/08 19:42:35 mdz Exp $
   4 /* ######################################################################
   5
   6    HTTPS Acquire Method - This is the HTTPS acquire method for APT.
   7
   8    It uses libcurl
   9
  10    ##################################################################### */
  11                                                                         /*}}}*/
  12 // Include Files                                                        /*{{{*/
  13 #include <config.h>
  14
  15 #include <apt-pkg/fileutl.h>
  16 #include <apt-pkg/error.h>
  17 #include <apt-pkg/hashes.h>
  18 #include <apt-pkg/netrc.h>
  19 #include <apt-pkg/configuration.h>
  20 #include <apt-pkg/macros.h>
  21 #include <apt-pkg/strutl.h>
  22 #include <apt-pkg/proxy.h>
  23
  24 #include <sys/stat.h>
  25 #include <sys/time.h>
  26 #include <unistd.h>
  27 #include <stdio.h>
  28 #include <iostream>
  29 #include <sstream>
  30 #include <ctype.h>
  31 #include <stdlib.h>
  32
  33 #include "https.h"
  34
  35 #include <apti18n.h>
  36                                                                         /*}}}*/
  37 using namespace std;
  38
  39 struct APT_HIDDEN CURLUserPointer {
  40    HttpsMethod * const https;
  41    HttpsMethod::FetchResult * const Res;
  42    HttpsMethod::FetchItem const * const Itm;
  43    CURLUserPointer(HttpsMethod * const https, HttpsMethod::FetchResult * const Res,
  44          HttpsMethod::FetchItem const * const Itm) : https(https), Res(Res), Itm(Itm) {}
  45 };
  46
  47 size_t
  48 HttpsMethod::parse_header(void *buffer, size_t size, size_t nmemb, void *userp)
  49 {
  50    size_t len = size * nmemb;
  51    CURLUserPointer *me = static_cast<CURLUserPointer *>(userp);
  52    std::string line((char*) buffer, len);
  53    for (--len; len > 0; --len)
  54       if (isspace_ascii(line[len]) == 0)
  55       {
  56          ++len;
  57          break;
  58       }
  59    line.erase(len);
  60
  61    if (line.empty() == true)
  62    {
  63       me->https->Server->JunkSize = 0;
  64       if (me->https->Server->Result != 416 && me->https->Server->StartPos != 0)
  65          ;
  66       else if (me->https->Server->Result == 416)
  67       {
  68          bool partialHit = false;
  69          if (me->Itm->ExpectedHashes.usable() == true)
  70          {
  71             Hashes resultHashes(me->Itm->ExpectedHashes);
  72             FileFd file(me->Itm->DestFile, FileFd::ReadOnly);
  73             me->https->Server->TotalFileSize = file.FileSize();
  74             me->https->Server->Date = file.ModificationTime();
  75             resultHashes.AddFD(file);
  76             HashStringList const hashList = resultHashes.GetHashStringList();
  77             partialHit = (me->Itm->ExpectedHashes == hashList);
  78          }
  79          else if (me->https->Server->Result == 416 && me->https->Server->TotalFileSize == me->https->File->FileSize())
  80             partialHit = true;
  81
  82          if (partialHit == true)
  83          {
  84             me->https->Server->Result = 200;
  85             me->https->Server->StartPos = me->https->Server->TotalFileSize;
  86             // the actual size is not important for https as curl will deal with it
  87             // by itself and e.g. doesn't bother us with transport-encoding…
  88             me->https->Server->JunkSize = std::numeric_limits<unsigned long long>::max();
  89          }
  90          else
  91             me->https->Server->StartPos = 0;
  92       }
  93       else
  94          me->https->Server->StartPos = 0;
  95
  96       me->Res->LastModified = me->https->Server->Date;
  97       me->Res->Size = me->https->Server->TotalFileSize;
  98       me->Res->ResumePoint = me->https->Server->StartPos;
  99
 100       // we expect valid data, so tell our caller we get the file now
 101       if (me->https->Server->Result >= 200 && me->https->Server->Result < 300)
 102       {
 103          if (me->Res->Size != 0 && me->Res->Size > me->Res->ResumePoint)
 104             me->https->URIStart(*me->Res);
 105          if (me->https->Server->AddPartialFileToHashes(*(me->https->File)) == false)
 106             return 0;
 107       }
 108       else
 109          me->https->Server->JunkSize = std::numeric_limits<decltype(me->https->Server->JunkSize)>::max();
 110    }
 111    else if (me->https->Server->HeaderLine(line) == false)
 112       return 0;
 113
 114    return size*nmemb;
 115 }
 116
 117 size_t
 118 HttpsMethod::write_data(void *buffer, size_t size, size_t nmemb, void *userp)
 119 {
 120    HttpsMethod *me = static_cast<HttpsMethod *>(userp);
 121    size_t buffer_size = size * nmemb;
 122    // we don't need to count the junk here, just drop anything we get as
 123    // we don't always know how long it would be, e.g. in chunked encoding.
 124    if (me->Server->JunkSize != 0)
 125       return buffer_size;
 126
 127    if(me->File->Write(buffer, buffer_size) != true)
 128       return 0;
 129
 130    if(me->Queue->MaximumSize > 0)
 131    {
 132       unsigned long long const TotalWritten = me->File->Tell();
 133       if (TotalWritten > me->Queue->MaximumSize)
 134       {
 135          me->SetFailReason("MaximumSizeExceeded");
 136          _error->Error("Writing more data than expected (%llu > %llu)",
 137                TotalWritten, me->Queue->MaximumSize);
 138          return 0;
 139       }
 140    }
 141
 142    if (me->Server->GetHashes()->Add((unsigned char const * const)buffer, buffer_size) == false)
 143       return 0;
 144
 145    return buffer_size;
 146 }
 147
 148 // HttpsServerState::HttpsServerState - Constructor                     /*{{{*/
 149 HttpsServerState::HttpsServerState(URI Srv,HttpsMethod * Owner) : ServerState(Srv, Owner), Hash(NULL)
 150 {
 151    TimeOut = _config->FindI("Acquire::https::Timeout",TimeOut);
 152    Reset();
 153 }
 154                                                                         /*}}}*/
 155 bool HttpsServerState::InitHashes(HashStringList const &ExpectedHashes) /*{{{*/
 156 {
 157    delete Hash;
 158    Hash = new Hashes(ExpectedHashes);
 159    return true;
 160 }
 161                                                                         /*}}}*/
 162 APT_PURE Hashes * HttpsServerState::GetHashes()                         /*{{{*/
 163 {
 164    return Hash;
 165 }
 166                                                                         /*}}}*/
 167
 168 bool HttpsMethod::SetupProxy()                                          /*{{{*/
 169 {
 170    URI ServerName = Queue->Uri;
 171
 172    // Determine the proxy setting
 173    AutoDetectProxy(ServerName);
 174
 175    // Curl should never read proxy settings from the environment, as
 176    // we determine which proxy to use.  Do this for consistency among
 177    // methods and prevent an environment variable overriding a
 178    // no-proxy ("DIRECT") setting in apt.conf.
 179    curl_easy_setopt(curl, CURLOPT_PROXY, "");
 180
 181    // Determine the proxy setting - try https first, fallback to http and use env at last
 182    string UseProxy = _config->Find("Acquire::https::Proxy::" + ServerName.Host,
 183                                    _config->Find("Acquire::http::Proxy::" + ServerName.Host).c_str());
 184
 185    if (UseProxy.empty() == true)
 186       UseProxy = _config->Find("Acquire::https::Proxy", _config->Find("Acquire::http::Proxy").c_str());
 187
 188    // User want to use NO proxy, so nothing to setup
 189    if (UseProxy == "DIRECT")
 190       return true;
 191
 192    // Parse no_proxy, a comma (,) separated list of domains we don't want to use
 193    // a proxy for so we stop right here if it is in the list
 194    if (getenv("no_proxy") != 0 && CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 195       return true;
 196
 197    if (UseProxy.empty() == true)
 198    {
 199       const char* result = getenv("https_proxy");
 200       // FIXME: Fall back to http_proxy is to remain compatible with
 201       // existing setups and behaviour of apt.conf.  This should be
 202       // deprecated in the future (including apt.conf).  Most other
 203       // programs do not fall back to http proxy settings and neither
 204       // should Apt.
 205       if (result == NULL)
 206          result = getenv("http_proxy");
 207       UseProxy = result == NULL ? "" : result;
 208    }
 209
 210    // Determine what host and port to use based on the proxy settings
 211    if (UseProxy.empty() == false)
 212    {
 213       Proxy = UseProxy;
 214       if (Proxy.Access == "socks5h")
 215          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5_HOSTNAME);
 216       else if (Proxy.Access == "socks5")
 217          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5);
 218       else if (Proxy.Access == "socks4a")
 219          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4A);
 220       else if (Proxy.Access == "socks")
 221          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4);
 222       else if (Proxy.Access == "http" || Proxy.Access == "https")
 223          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_HTTP);
 224       else
 225          return false;
 226
 227       if (Proxy.Port != 1)
 228          curl_easy_setopt(curl, CURLOPT_PROXYPORT, Proxy.Port);
 229       curl_easy_setopt(curl, CURLOPT_PROXY, Proxy.Host.c_str());
 230       if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 231       {
 232          curl_easy_setopt(curl, CURLOPT_PROXYUSERNAME, Proxy.User.c_str());
 233          curl_easy_setopt(curl, CURLOPT_PROXYPASSWORD, Proxy.Password.c_str());
 234       }
 235    }
 236    return true;
 237 }                                                                       /*}}}*/
 238 // HttpsMethod::Fetch - Fetch an item                                   /*{{{*/
 239 // ---------------------------------------------------------------------
 240 /* This adds an item to the pipeline. We keep the pipeline at a fixed
 241    depth. */
 242 bool HttpsMethod::Fetch(FetchItem *Itm)
 243 {
 244    struct stat SBuf;
 245    struct curl_slist *headers=NULL;
 246    char curl_errorstr[CURL_ERROR_SIZE];
 247    URI Uri = Itm->Uri;
 248    string remotehost = Uri.Host;
 249
 250    // TODO:
 251    //       - http::Pipeline-Depth
 252    //       - error checking/reporting
 253    //       - more debug options? (CURLOPT_DEBUGFUNCTION?)
 254
 255    curl_easy_reset(curl);
 256    if (SetupProxy() == false)
 257       return _error->Error("Unsupported proxy configured: %s", URI::SiteOnly(Proxy).c_str());
 258
 259    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 260
 261    FetchResult Res;
 262    CURLUserPointer userp(this, &Res, Itm);
 263    // callbacks
 264    curl_easy_setopt(curl, CURLOPT_URL, static_cast<string>(Uri).c_str());
 265    curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION, parse_header);
 266    curl_easy_setopt(curl, CURLOPT_WRITEHEADER, &userp);
 267    curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_data);
 268    curl_easy_setopt(curl, CURLOPT_WRITEDATA, this);
 269    // options
 270    curl_easy_setopt(curl, CURLOPT_NOPROGRESS, true);
 271    curl_easy_setopt(curl, CURLOPT_FILETIME, true);
 272    curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 0);
 273    // only allow curl to handle https, not the other stuff it supports
 274    curl_easy_setopt(curl, CURLOPT_PROTOCOLS, CURLPROTO_HTTPS);
 275    curl_easy_setopt(curl, CURLOPT_REDIR_PROTOCOLS, CURLPROTO_HTTPS);
 276
 277    // SSL parameters are set by default to the common (non mirror-specific) value
 278    // if available (or a default one) and gets overload by mirror-specific ones.
 279
 280    // File containing the list of trusted CA.
 281    string cainfo = _config->Find("Acquire::https::CaInfo","");
 282    string knob = "Acquire::https::"+remotehost+"::CaInfo";
 283    cainfo = _config->Find(knob.c_str(),cainfo.c_str());
 284    if(cainfo.empty() == false)
 285       curl_easy_setopt(curl, CURLOPT_CAINFO,cainfo.c_str());
 286
 287    // Check server certificate against previous CA list ...
 288    bool peer_verify = _config->FindB("Acquire::https::Verify-Peer",true);
 289    knob = "Acquire::https::" + remotehost + "::Verify-Peer";
 290    peer_verify = _config->FindB(knob.c_str(), peer_verify);
 291    curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, peer_verify);
 292
 293    // ... and hostname against cert CN or subjectAltName
 294    bool verify = _config->FindB("Acquire::https::Verify-Host",true);
 295    knob = "Acquire::https::"+remotehost+"::Verify-Host";
 296    verify = _config->FindB(knob.c_str(),verify);
 297    int const default_verify = (verify == true) ? 2 : 0;
 298    curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, default_verify);
 299
 300    // Also enforce issuer of server certificate using its cert
 301    string issuercert = _config->Find("Acquire::https::IssuerCert","");
 302    knob = "Acquire::https::"+remotehost+"::IssuerCert";
 303    issuercert = _config->Find(knob.c_str(),issuercert.c_str());
 304    if(issuercert.empty() == false)
 305       curl_easy_setopt(curl, CURLOPT_ISSUERCERT,issuercert.c_str());
 306
 307    // For client authentication, certificate file ...
 308    string pem = _config->Find("Acquire::https::SslCert","");
 309    knob = "Acquire::https::"+remotehost+"::SslCert";
 310    pem = _config->Find(knob.c_str(),pem.c_str());
 311    if(pem.empty() == false)
 312       curl_easy_setopt(curl, CURLOPT_SSLCERT, pem.c_str());
 313
 314    // ... and associated key.
 315    string key = _config->Find("Acquire::https::SslKey","");
 316    knob = "Acquire::https::"+remotehost+"::SslKey";
 317    key = _config->Find(knob.c_str(),key.c_str());
 318    if(key.empty() == false)
 319       curl_easy_setopt(curl, CURLOPT_SSLKEY, key.c_str());
 320
 321    // Allow forcing SSL version to SSLv3 or TLSv1 (SSLv2 is not
 322    // supported by GnuTLS).
 323    long final_version = CURL_SSLVERSION_DEFAULT;
 324    string sslversion = _config->Find("Acquire::https::SslForceVersion","");
 325    knob = "Acquire::https::"+remotehost+"::SslForceVersion";
 326    sslversion = _config->Find(knob.c_str(),sslversion.c_str());
 327    if(sslversion == "TLSv1")
 328      final_version = CURL_SSLVERSION_TLSv1;
 329    else if(sslversion == "SSLv3")
 330      final_version = CURL_SSLVERSION_SSLv3;
 331    curl_easy_setopt(curl, CURLOPT_SSLVERSION, final_version);
 332
 333    // CRL file
 334    string crlfile = _config->Find("Acquire::https::CrlFile","");
 335    knob = "Acquire::https::"+remotehost+"::CrlFile";
 336    crlfile = _config->Find(knob.c_str(),crlfile.c_str());
 337    if(crlfile.empty() == false)
 338       curl_easy_setopt(curl, CURLOPT_CRLFILE, crlfile.c_str());
 339
 340    // cache-control
 341    if(_config->FindB("Acquire::https::No-Cache",
 342         _config->FindB("Acquire::http::No-Cache",false)) == false)
 343    {
 344       // cache enabled
 345       if (_config->FindB("Acquire::https::No-Store",
 346                 _config->FindB("Acquire::http::No-Store",false)) == true)
 347          headers = curl_slist_append(headers,"Cache-Control: no-store");
 348       stringstream ss;
 349       ioprintf(ss, "Cache-Control: max-age=%u", _config->FindI("Acquire::https::Max-Age",
 350                 _config->FindI("Acquire::http::Max-Age",0)));
 351       headers = curl_slist_append(headers, ss.str().c_str());
 352    } else {
 353       // cache disabled by user
 354       headers = curl_slist_append(headers, "Cache-Control: no-cache");
 355       headers = curl_slist_append(headers, "Pragma: no-cache");
 356    }
 357    curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 358
 359    // speed limit
 360    int const dlLimit = _config->FindI("Acquire::https::Dl-Limit",
 361                 _config->FindI("Acquire::http::Dl-Limit",0))*1024;
 362    if (dlLimit > 0)
 363       curl_easy_setopt(curl, CURLOPT_MAX_RECV_SPEED_LARGE, dlLimit);
 364
 365    // set header
 366    curl_easy_setopt(curl, CURLOPT_USERAGENT,
 367         _config->Find("Acquire::https::User-Agent",
 368                 _config->Find("Acquire::http::User-Agent",
 369                         "Debian APT-CURL/1.0 (" PACKAGE_VERSION ")").c_str()).c_str());
 370
 371    // set timeout
 372    int const timeout = _config->FindI("Acquire::https::Timeout",
 373                 _config->FindI("Acquire::http::Timeout",120));
 374    curl_easy_setopt(curl, CURLOPT_CONNECTTIMEOUT, timeout);
 375    //set really low lowspeed timeout (see #497983)
 376    curl_easy_setopt(curl, CURLOPT_LOW_SPEED_LIMIT, DL_MIN_SPEED);
 377    curl_easy_setopt(curl, CURLOPT_LOW_SPEED_TIME, timeout);
 378
 379    // debug
 380    if (Debug == true)
 381       curl_easy_setopt(curl, CURLOPT_VERBOSE, true);
 382
 383    // error handling
 384    curl_errorstr[0] = '\0';
 385    curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 386
 387    // If we ask for uncompressed files servers might respond with content-
 388    // negotiation which lets us end up with compressed files we do not support,
 389    // see 657029, 657560 and co, so if we have no extension on the request
 390    // ask for text only. As a sidenote: If there is nothing to negotate servers
 391    // seem to be nice and ignore it.
 392    if (_config->FindB("Acquire::https::SendAccept", _config->FindB("Acquire::http::SendAccept", true)) == true)
 393    {
 394       size_t const filepos = Itm->Uri.find_last_of('/');
 395       string const file = Itm->Uri.substr(filepos + 1);
 396       if (flExtension(file) == file)
 397          headers = curl_slist_append(headers, "Accept: text/*");
 398    }
 399
 400    // if we have the file send an if-range query with a range header
 401    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 402    {
 403       std::string Buf;
 404       strprintf(Buf, "Range: bytes=%lli-", (long long) SBuf.st_size);
 405       headers = curl_slist_append(headers, Buf.c_str());
 406       strprintf(Buf, "If-Range: %s", TimeRFC1123(SBuf.st_mtime, false).c_str());
 407       headers = curl_slist_append(headers, Buf.c_str());
 408    }
 409    else if(Itm->LastModified > 0)
 410    {
 411       curl_easy_setopt(curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_IFMODSINCE);
 412       curl_easy_setopt(curl, CURLOPT_TIMEVALUE, Itm->LastModified);
 413    }
 414
 415    // go for it - if the file exists, append on it
 416    File = new FileFd(Itm->DestFile, FileFd::WriteAny);
 417    Server = CreateServerState(Itm->Uri);
 418    if (Server->InitHashes(Itm->ExpectedHashes) == false)
 419       return false;
 420
 421    // keep apt updated
 422    Res.Filename = Itm->DestFile;
 423
 424    // get it!
 425    CURLcode success = curl_easy_perform(curl);
 426
 427    // If the server returns 200 OK but the If-Modified-Since condition is not
 428    // met, CURLINFO_CONDITION_UNMET will be set to 1
 429    long curl_condition_unmet = 0;
 430    curl_easy_getinfo(curl, CURLINFO_CONDITION_UNMET, &curl_condition_unmet);
 431    if (curl_condition_unmet == 1)
 432       Server->Result = 304;
 433
 434    File->Close();
 435    curl_slist_free_all(headers);
 436
 437    // cleanup
 438    if (success != CURLE_OK)
 439    {
 440 #pragma GCC diagnostic push
 441 #pragma GCC diagnostic ignored "-Wswitch"
 442       switch (success)
 443       {
 444          case CURLE_COULDNT_RESOLVE_PROXY:
 445          case CURLE_COULDNT_RESOLVE_HOST:
 446             SetFailReason("ResolveFailure");
 447             break;
 448          case CURLE_COULDNT_CONNECT:
 449             SetFailReason("ConnectionRefused");
 450             break;
 451          case CURLE_OPERATION_TIMEDOUT:
 452             SetFailReason("Timeout");
 453             break;
 454       }
 455 #pragma GCC diagnostic pop
 456       // only take curls technical errors if we haven't our own
 457       // (e.g. for the maximum size limit we have and curls can be confusing)
 458       if (_error->PendingError() == false)
 459          _error->Error("%s", curl_errorstr);
 460       else
 461          _error->Warning("curl: %s", curl_errorstr);
 462       return false;
 463    }
 464
 465    switch (DealWithHeaders(Res))
 466    {
 467       case ServerMethod::IMS_HIT:
 468          URIDone(Res);
 469          break;
 470
 471       case ServerMethod::ERROR_WITH_CONTENT_PAGE:
 472          // unlink, no need keep 401/404 page content in partial/
 473          RemoveFile(Binary.c_str(), File->Name());
 474       case ServerMethod::ERROR_UNRECOVERABLE:
 475       case ServerMethod::ERROR_NOT_FROM_SERVER:
 476          return false;
 477
 478       case ServerMethod::TRY_AGAIN_OR_REDIRECT:
 479          Redirect(NextURI);
 480          break;
 481
 482       case ServerMethod::FILE_IS_OPEN:
 483          struct stat resultStat;
 484          if (unlikely(stat(File->Name().c_str(), &resultStat) != 0))
 485          {
 486             _error->Errno("stat", "Unable to access file %s", File->Name().c_str());
 487             return false;
 488          }
 489          Res.Size = resultStat.st_size;
 490
 491          // Timestamp
 492          curl_easy_getinfo(curl, CURLINFO_FILETIME, &Res.LastModified);
 493          if (Res.LastModified != -1)
 494          {
 495             struct timeval times[2];
 496             times[0].tv_sec = Res.LastModified;
 497             times[1].tv_sec = Res.LastModified;
 498             times[0].tv_usec = times[1].tv_usec = 0;
 499             utimes(File->Name().c_str(), times);
 500          }
 501          else
 502             Res.LastModified = resultStat.st_mtime;
 503
 504          // take hashes
 505          Res.TakeHashes(*(Server->GetHashes()));
 506
 507          // keep apt updated
 508          URIDone(Res);
 509          break;
 510    }
 511
 512    delete File;
 513    return true;
 514 }
 515                                                                         /*}}}*/
 516 // HttpsMethod::Configuration - Handle a configuration message          /*{{{*/
 517 bool HttpsMethod::Configuration(string Message)
 518 {
 519    if (ServerMethod::Configuration(Message) == false)
 520       return false;
 521
 522    AllowRedirect = _config->FindB("Acquire::https::AllowRedirect",
 523         _config->FindB("Acquire::http::AllowRedirect", true));
 524    Debug = _config->FindB("Debug::Acquire::https",false);
 525
 526    return true;
 527 }
 528                                                                         /*}}}*/
 529 std::unique_ptr<ServerState> HttpsMethod::CreateServerState(URI const &uri)/*{{{*/
 530 {
 531    return std::unique_ptr<ServerState>(new HttpsServerState(uri, this));
 532 }
 533                                                                         /*}}}*/
 534
 535 int main()
 536 {
 537    return HttpsMethod().Run();
 538 }
 539