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 <ctype.h>
  29 #include <stdlib.h>
  30
  31 #include <array>
  32 #include <iostream>
  33 #include <sstream>
  34
  35
  36 #include "https.h"
  37
  38 #include <apti18n.h>
  39                                                                         /*}}}*/
  40 using namespace std;
  41
  42 struct APT_HIDDEN CURLUserPointer {
  43    HttpsMethod * const https;
  44    HttpsMethod::FetchResult * const Res;
  45    HttpsMethod::FetchItem const * const Itm;
  46    CURLUserPointer(HttpsMethod * const https, HttpsMethod::FetchResult * const Res,
  47          HttpsMethod::FetchItem const * const Itm) : https(https), Res(Res), Itm(Itm) {}
  48 };
  49
  50 size_t
  51 HttpsMethod::parse_header(void *buffer, size_t size, size_t nmemb, void *userp)
  52 {
  53    size_t len = size * nmemb;
  54    CURLUserPointer *me = static_cast<CURLUserPointer *>(userp);
  55    std::string line((char*) buffer, len);
  56    for (--len; len > 0; --len)
  57       if (isspace_ascii(line[len]) == 0)
  58       {
  59          ++len;
  60          break;
  61       }
  62    line.erase(len);
  63
  64    if (line.empty() == true)
  65    {
  66       me->https->Server->JunkSize = 0;
  67       if (me->https->Server->Result != 416 && me->https->Server->StartPos != 0)
  68          ;
  69       else if (me->https->Server->Result == 416)
  70       {
  71          bool partialHit = false;
  72          if (me->Itm->ExpectedHashes.usable() == true)
  73          {
  74             Hashes resultHashes(me->Itm->ExpectedHashes);
  75             FileFd file(me->Itm->DestFile, FileFd::ReadOnly);
  76             me->https->Server->TotalFileSize = file.FileSize();
  77             me->https->Server->Date = file.ModificationTime();
  78             resultHashes.AddFD(file);
  79             HashStringList const hashList = resultHashes.GetHashStringList();
  80             partialHit = (me->Itm->ExpectedHashes == hashList);
  81          }
  82          else if (me->https->Server->Result == 416 && me->https->Server->TotalFileSize == me->https->File->FileSize())
  83             partialHit = true;
  84
  85          if (partialHit == true)
  86          {
  87             me->https->Server->Result = 200;
  88             me->https->Server->StartPos = me->https->Server->TotalFileSize;
  89             // the actual size is not important for https as curl will deal with it
  90             // by itself and e.g. doesn't bother us with transport-encoding…
  91             me->https->Server->JunkSize = std::numeric_limits<unsigned long long>::max();
  92          }
  93          else
  94             me->https->Server->StartPos = 0;
  95       }
  96       else
  97          me->https->Server->StartPos = 0;
  98
  99       me->Res->LastModified = me->https->Server->Date;
 100       me->Res->Size = me->https->Server->TotalFileSize;
 101       me->Res->ResumePoint = me->https->Server->StartPos;
 102
 103       // we expect valid data, so tell our caller we get the file now
 104       if (me->https->Server->Result >= 200 && me->https->Server->Result < 300)
 105       {
 106          if (me->Res->Size != 0 && me->Res->Size > me->Res->ResumePoint)
 107             me->https->URIStart(*me->Res);
 108          if (me->https->Server->AddPartialFileToHashes(*(me->https->File)) == false)
 109             return 0;
 110       }
 111       else
 112          me->https->Server->JunkSize = std::numeric_limits<decltype(me->https->Server->JunkSize)>::max();
 113    }
 114    else if (me->https->Server->HeaderLine(line) == false)
 115       return 0;
 116
 117    return size*nmemb;
 118 }
 119
 120 size_t
 121 HttpsMethod::write_data(void *buffer, size_t size, size_t nmemb, void *userp)
 122 {
 123    HttpsMethod *me = static_cast<HttpsMethod *>(userp);
 124    size_t buffer_size = size * nmemb;
 125    // we don't need to count the junk here, just drop anything we get as
 126    // we don't always know how long it would be, e.g. in chunked encoding.
 127    if (me->Server->JunkSize != 0)
 128       return buffer_size;
 129
 130    if(me->File->Write(buffer, buffer_size) != true)
 131       return 0;
 132
 133    if(me->Queue->MaximumSize > 0)
 134    {
 135       unsigned long long const TotalWritten = me->File->Tell();
 136       if (TotalWritten > me->Queue->MaximumSize)
 137       {
 138          me->SetFailReason("MaximumSizeExceeded");
 139          _error->Error("Writing more data than expected (%llu > %llu)",
 140                TotalWritten, me->Queue->MaximumSize);
 141          return 0;
 142       }
 143    }
 144
 145    if (me->Server->GetHashes()->Add((unsigned char const * const)buffer, buffer_size) == false)
 146       return 0;
 147
 148    return buffer_size;
 149 }
 150
 151 // HttpsServerState::HttpsServerState - Constructor                     /*{{{*/
 152 HttpsServerState::HttpsServerState(URI Srv,HttpsMethod * Owner) : ServerState(Srv, Owner), Hash(NULL)
 153 {
 154    TimeOut = Owner->ConfigFindI("Timeout", TimeOut);
 155    Reset();
 156 }
 157                                                                         /*}}}*/
 158 bool HttpsServerState::InitHashes(HashStringList const &ExpectedHashes) /*{{{*/
 159 {
 160    delete Hash;
 161    Hash = new Hashes(ExpectedHashes);
 162    return true;
 163 }
 164                                                                         /*}}}*/
 165 APT_PURE Hashes * HttpsServerState::GetHashes()                         /*{{{*/
 166 {
 167    return Hash;
 168 }
 169                                                                         /*}}}*/
 170
 171 bool HttpsMethod::SetupProxy()                                          /*{{{*/
 172 {
 173    URI ServerName = Queue->Uri;
 174
 175    // Determine the proxy setting
 176    AutoDetectProxy(ServerName);
 177
 178    // Curl should never read proxy settings from the environment, as
 179    // we determine which proxy to use.  Do this for consistency among
 180    // methods and prevent an environment variable overriding a
 181    // no-proxy ("DIRECT") setting in apt.conf.
 182    curl_easy_setopt(curl, CURLOPT_PROXY, "");
 183
 184    // Determine the proxy setting - try https first, fallback to http and use env at last
 185    string UseProxy = ConfigFind("Proxy::" + ServerName.Host, "");
 186    if (UseProxy.empty() == true)
 187       UseProxy = ConfigFind("Proxy", "");
 188    // User wants 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 = nullptr;
 200       if (std::find(methodNames.begin(), methodNames.end(), "https") != methodNames.end())
 201          result = getenv("https_proxy");
 202       // FIXME: Fall back to http_proxy is to remain compatible with
 203       // existing setups and behaviour of apt.conf.  This should be
 204       // deprecated in the future (including apt.conf).  Most other
 205       // programs do not fall back to http proxy settings and neither
 206       // should Apt.
 207       if (result == nullptr && std::find(methodNames.begin(), methodNames.end(), "http") != methodNames.end())
 208          result = getenv("http_proxy");
 209       UseProxy = result == nullptr ? "" : result;
 210    }
 211
 212    // Determine what host and port to use based on the proxy settings
 213    if (UseProxy.empty() == false)
 214    {
 215       Proxy = UseProxy;
 216       AddProxyAuth(Proxy, ServerName);
 217
 218       if (Proxy.Access == "socks5h")
 219          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5_HOSTNAME);
 220       else if (Proxy.Access == "socks5")
 221          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5);
 222       else if (Proxy.Access == "socks4a")
 223          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4A);
 224       else if (Proxy.Access == "socks")
 225          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4);
 226       else if (Proxy.Access == "http" || Proxy.Access == "https")
 227          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_HTTP);
 228       else
 229          return false;
 230
 231       if (Proxy.Port != 1)
 232          curl_easy_setopt(curl, CURLOPT_PROXYPORT, Proxy.Port);
 233       curl_easy_setopt(curl, CURLOPT_PROXY, Proxy.Host.c_str());
 234       if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 235       {
 236          curl_easy_setopt(curl, CURLOPT_PROXYUSERNAME, Proxy.User.c_str());
 237          curl_easy_setopt(curl, CURLOPT_PROXYPASSWORD, Proxy.Password.c_str());
 238       }
 239    }
 240    return true;
 241 }                                                                       /*}}}*/
 242 // HttpsMethod::Fetch - Fetch an item                                   /*{{{*/
 243 // ---------------------------------------------------------------------
 244 /* This adds an item to the pipeline. We keep the pipeline at a fixed
 245    depth. */
 246 bool HttpsMethod::Fetch(FetchItem *Itm)
 247 {
 248    struct stat SBuf;
 249    struct curl_slist *headers=NULL;
 250    char curl_errorstr[CURL_ERROR_SIZE];
 251    URI Uri = Itm->Uri;
 252    setPostfixForMethodNames(Uri.Host.c_str());
 253    AllowRedirect = ConfigFindB("AllowRedirect", true);
 254    Debug = DebugEnabled();
 255
 256    // TODO:
 257    //       - http::Pipeline-Depth
 258    //       - error checking/reporting
 259    //       - more debug options? (CURLOPT_DEBUGFUNCTION?)
 260    {
 261       auto const plus = Binary.find('+');
 262       if (plus != std::string::npos)
 263          Uri.Access = Binary.substr(plus + 1);
 264    }
 265
 266    curl_easy_reset(curl);
 267    if (SetupProxy() == false)
 268       return _error->Error("Unsupported proxy configured: %s", URI::SiteOnly(Proxy).c_str());
 269
 270    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 271
 272    // The "+" is encoded as a workaround for a amazon S3 bug
 273    // see LP bugs #1003633 and #1086997. (taken from http method)
 274    Uri.Path = QuoteString(Uri.Path, "+~ ");
 275
 276    FetchResult Res;
 277    CURLUserPointer userp(this, &Res, Itm);
 278    // callbacks
 279    curl_easy_setopt(curl, CURLOPT_URL, static_cast<string>(Uri).c_str());
 280    curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION, parse_header);
 281    curl_easy_setopt(curl, CURLOPT_WRITEHEADER, &userp);
 282    curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_data);
 283    curl_easy_setopt(curl, CURLOPT_WRITEDATA, this);
 284    // options
 285    curl_easy_setopt(curl, CURLOPT_NOPROGRESS, true);
 286    curl_easy_setopt(curl, CURLOPT_FILETIME, true);
 287    curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 0);
 288
 289    if (std::find(methodNames.begin(), methodNames.end(), "https") != methodNames.end())
 290    {
 291       curl_easy_setopt(curl, CURLOPT_PROTOCOLS, CURLPROTO_HTTPS);
 292       curl_easy_setopt(curl, CURLOPT_REDIR_PROTOCOLS, CURLPROTO_HTTPS);
 293
 294       // File containing the list of trusted CA.
 295       std::string const cainfo = ConfigFind("CaInfo", "");
 296       if(cainfo.empty() == false)
 297          curl_easy_setopt(curl, CURLOPT_CAINFO, cainfo.c_str());
 298       // Check server certificate against previous CA list ...
 299       curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, ConfigFindB("Verify-Peer", true) ? 1 : 0);
 300       // ... and hostname against cert CN or subjectAltName
 301       curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, ConfigFindB("Verify-Host", true) ? 2 : 0);
 302       // Also enforce issuer of server certificate using its cert
 303       std::string const issuercert = ConfigFind("IssuerCert", "");
 304       if(issuercert.empty() == false)
 305          curl_easy_setopt(curl, CURLOPT_ISSUERCERT, issuercert.c_str());
 306       // For client authentication, certificate file ...
 307       std::string const pem = ConfigFind("SslCert", "");
 308       if(pem.empty() == false)
 309          curl_easy_setopt(curl, CURLOPT_SSLCERT, pem.c_str());
 310       // ... and associated key.
 311       std::string const key = ConfigFind("SslKey", "");
 312       if(key.empty() == false)
 313          curl_easy_setopt(curl, CURLOPT_SSLKEY, key.c_str());
 314       // Allow forcing SSL version to SSLv3 or TLSv1
 315       long final_version = CURL_SSLVERSION_DEFAULT;
 316       std::string const sslversion = ConfigFind("SslForceVersion", "");
 317       if(sslversion == "TLSv1")
 318          final_version = CURL_SSLVERSION_TLSv1;
 319       else if(sslversion == "TLSv1.0")
 320          final_version = CURL_SSLVERSION_TLSv1_0;
 321       else if(sslversion == "TLSv1.1")
 322          final_version = CURL_SSLVERSION_TLSv1_1;
 323       else if(sslversion == "TLSv1.2")
 324          final_version = CURL_SSLVERSION_TLSv1_2;
 325       else if(sslversion == "SSLv3")
 326          final_version = CURL_SSLVERSION_SSLv3;
 327       curl_easy_setopt(curl, CURLOPT_SSLVERSION, final_version);
 328       // CRL file
 329       std::string const crlfile = ConfigFind("CrlFile", "");
 330       if(crlfile.empty() == false)
 331          curl_easy_setopt(curl, CURLOPT_CRLFILE, crlfile.c_str());
 332    }
 333    else
 334    {
 335       curl_easy_setopt(curl, CURLOPT_PROTOCOLS, CURLPROTO_HTTP);
 336       curl_easy_setopt(curl, CURLOPT_REDIR_PROTOCOLS, CURLPROTO_HTTP);
 337    }
 338    // cache-control
 339    if(ConfigFindB("No-Cache", false) == false)
 340    {
 341       // cache enabled
 342       if (ConfigFindB("No-Store", false) == true)
 343          headers = curl_slist_append(headers,"Cache-Control: no-store");
 344       std::string ss;
 345       strprintf(ss, "Cache-Control: max-age=%u", ConfigFindI("Max-Age", 0));
 346       headers = curl_slist_append(headers, ss.c_str());
 347    } else {
 348       // cache disabled by user
 349       headers = curl_slist_append(headers, "Cache-Control: no-cache");
 350       headers = curl_slist_append(headers, "Pragma: no-cache");
 351    }
 352    curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 353    // speed limit
 354    int const dlLimit = ConfigFindI("Dl-Limit", 0) * 1024;
 355    if (dlLimit > 0)
 356       curl_easy_setopt(curl, CURLOPT_MAX_RECV_SPEED_LARGE, dlLimit);
 357
 358    // set header
 359    curl_easy_setopt(curl, CURLOPT_USERAGENT, ConfigFind("User-Agent", "Debian APT-CURL/1.0 (" PACKAGE_VERSION ")").c_str());
 360
 361    // set timeout
 362    int const timeout = ConfigFindI("Timeout", 120);
 363    curl_easy_setopt(curl, CURLOPT_CONNECTTIMEOUT, timeout);
 364    //set really low lowspeed timeout (see #497983)
 365    curl_easy_setopt(curl, CURLOPT_LOW_SPEED_LIMIT, DL_MIN_SPEED);
 366    curl_easy_setopt(curl, CURLOPT_LOW_SPEED_TIME, timeout);
 367
 368    // debug
 369    if (Debug == true)
 370       curl_easy_setopt(curl, CURLOPT_VERBOSE, true);
 371
 372    // error handling
 373    curl_errorstr[0] = '\0';
 374    curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 375
 376    // If we ask for uncompressed files servers might respond with content-
 377    // negotiation which lets us end up with compressed files we do not support,
 378    // see 657029, 657560 and co, so if we have no extension on the request
 379    // ask for text only. As a sidenote: If there is nothing to negotate servers
 380    // seem to be nice and ignore it.
 381    if (ConfigFindB("SendAccept", true))
 382    {
 383       size_t const filepos = Itm->Uri.find_last_of('/');
 384       string const file = Itm->Uri.substr(filepos + 1);
 385       if (flExtension(file) == file)
 386          headers = curl_slist_append(headers, "Accept: text/*");
 387    }
 388
 389    // go for it - if the file exists, append on it
 390    File = new FileFd(Itm->DestFile, FileFd::WriteAny);
 391    if (Server == nullptr || Server->Comp(Itm->Uri) == false)
 392       Server = CreateServerState(Itm->Uri);
 393    else
 394       Server->Reset(false);
 395
 396    // if we have the file send an if-range query with a range header
 397    if (Server->RangesAllowed && stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 398    {
 399       std::string Buf;
 400       strprintf(Buf, "Range: bytes=%lli-", (long long) SBuf.st_size);
 401       headers = curl_slist_append(headers, Buf.c_str());
 402       strprintf(Buf, "If-Range: %s", TimeRFC1123(SBuf.st_mtime, false).c_str());
 403       headers = curl_slist_append(headers, Buf.c_str());
 404    }
 405    else if(Itm->LastModified > 0)
 406    {
 407       curl_easy_setopt(curl, CURLOPT_TIMECONDITION, CURL_TIMECOND_IFMODSINCE);
 408       curl_easy_setopt(curl, CURLOPT_TIMEVALUE, Itm->LastModified);
 409    }
 410
 411    if (Server->InitHashes(Itm->ExpectedHashes) == false)
 412       return false;
 413
 414    // keep apt updated
 415    Res.Filename = Itm->DestFile;
 416
 417    // get it!
 418    CURLcode success = curl_easy_perform(curl);
 419
 420    // If the server returns 200 OK but the If-Modified-Since condition is not
 421    // met, CURLINFO_CONDITION_UNMET will be set to 1
 422    long curl_condition_unmet = 0;
 423    curl_easy_getinfo(curl, CURLINFO_CONDITION_UNMET, &curl_condition_unmet);
 424    if (curl_condition_unmet == 1)
 425       Server->Result = 304;
 426
 427    File->Close();
 428    curl_slist_free_all(headers);
 429
 430    // cleanup
 431    if (success != CURLE_OK)
 432    {
 433 #pragma GCC diagnostic push
 434 #pragma GCC diagnostic ignored "-Wswitch"
 435       switch (success)
 436       {
 437          case CURLE_COULDNT_RESOLVE_PROXY:
 438          case CURLE_COULDNT_RESOLVE_HOST:
 439             SetFailReason("ResolveFailure");
 440             break;
 441          case CURLE_COULDNT_CONNECT:
 442             SetFailReason("ConnectionRefused");
 443             break;
 444          case CURLE_OPERATION_TIMEDOUT:
 445             SetFailReason("Timeout");
 446             break;
 447       }
 448 #pragma GCC diagnostic pop
 449       // only take curls technical errors if we haven't our own
 450       // (e.g. for the maximum size limit we have and curls can be confusing)
 451       if (_error->PendingError() == false)
 452          _error->Error("%s", curl_errorstr);
 453       else
 454          _error->Warning("curl: %s", curl_errorstr);
 455       return false;
 456    }
 457
 458    switch (DealWithHeaders(Res))
 459    {
 460       case ServerMethod::IMS_HIT:
 461          URIDone(Res);
 462          break;
 463
 464       case ServerMethod::ERROR_WITH_CONTENT_PAGE:
 465          // unlink, no need keep 401/404 page content in partial/
 466          RemoveFile(Binary.c_str(), File->Name());
 467       case ServerMethod::ERROR_UNRECOVERABLE:
 468       case ServerMethod::ERROR_NOT_FROM_SERVER:
 469          return false;
 470
 471       case ServerMethod::TRY_AGAIN_OR_REDIRECT:
 472          Redirect(NextURI);
 473          break;
 474
 475       case ServerMethod::FILE_IS_OPEN:
 476          struct stat resultStat;
 477          if (unlikely(stat(File->Name().c_str(), &resultStat) != 0))
 478          {
 479             _error->Errno("stat", "Unable to access file %s", File->Name().c_str());
 480             return false;
 481          }
 482          Res.Size = resultStat.st_size;
 483
 484          // Timestamp
 485          curl_easy_getinfo(curl, CURLINFO_FILETIME, &Res.LastModified);
 486          if (Res.LastModified != -1)
 487          {
 488             struct timeval times[2];
 489             times[0].tv_sec = Res.LastModified;
 490             times[1].tv_sec = Res.LastModified;
 491             times[0].tv_usec = times[1].tv_usec = 0;
 492             utimes(File->Name().c_str(), times);
 493          }
 494          else
 495             Res.LastModified = resultStat.st_mtime;
 496
 497          // take hashes
 498          Res.TakeHashes(*(Server->GetHashes()));
 499
 500          // keep apt updated
 501          URIDone(Res);
 502          break;
 503    }
 504
 505    delete File;
 506    return true;
 507 }
 508                                                                         /*}}}*/
 509 std::unique_ptr<ServerState> HttpsMethod::CreateServerState(URI const &uri)/*{{{*/
 510 {
 511    return std::unique_ptr<ServerState>(new HttpsServerState(uri, this));
 512 }
 513                                                                         /*}}}*/
 514 HttpsMethod::HttpsMethod(std::string &&pProg) : ServerMethod(std::move(pProg),"1.2",Pipeline | SendConfig)/*{{{*/
 515 {
 516    auto addName = std::inserter(methodNames, methodNames.begin());
 517    addName = "http";
 518    auto const plus = Binary.find('+');
 519    if (plus != std::string::npos)
 520    {
 521       addName = Binary.substr(plus + 1);
 522       auto base = Binary.substr(0, plus);
 523       if (base != "https")
 524          addName = base;
 525    }
 526    if (std::find(methodNames.begin(), methodNames.end(), "https") != methodNames.end())
 527       curl_global_init(CURL_GLOBAL_SSL);
 528    else
 529       curl_global_init(CURL_GLOBAL_NOTHING);
 530    curl = curl_easy_init();
 531 }
 532                                                                         /*}}}*/
 533 HttpsMethod::~HttpsMethod()                                             /*{{{*/
 534 {
 535    curl_easy_cleanup(curl);
 536 }
 537                                                                         /*}}}*/
 538 int main(int, const char *argv[])                                       /*{{{*/
 539 {
 540    std::string Binary = flNotDir(argv[0]);
 541    if (Binary.find('+') == std::string::npos && Binary != "https")
 542       Binary.append("+https");
 543    return HttpsMethod(std::move(Binary)).Run();
 544 }
 545                                                                         /*}}}*/