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       if (me->https->Server->Result != 416 && me->https->Server->StartPos != 0)
  64          ;
  65       else if (me->https->Server->Result == 416)
  66       {
  67          bool partialHit = false;
  68          if (me->Itm->ExpectedHashes.usable() == true)
  69          {
  70             Hashes resultHashes(me->Itm->ExpectedHashes);
  71             FileFd file(me->Itm->DestFile, FileFd::ReadOnly);
  72             me->https->Server->TotalFileSize = file.FileSize();
  73             me->https->Server->Date = file.ModificationTime();
  74             resultHashes.AddFD(file);
  75             HashStringList const hashList = resultHashes.GetHashStringList();
  76             partialHit = (me->Itm->ExpectedHashes == hashList);
  77          }
  78          else if (me->https->Server->Result == 416 && me->https->Server->TotalFileSize == me->https->File->FileSize())
  79             partialHit = true;
  80
  81          if (partialHit == true)
  82          {
  83             me->https->Server->Result = 200;
  84             me->https->Server->StartPos = me->https->Server->TotalFileSize;
  85             // the actual size is not important for https as curl will deal with it
  86             // by itself and e.g. doesn't bother us with transport-encoding…
  87             me->https->Server->JunkSize = std::numeric_limits<unsigned long long>::max();
  88          }
  89          else
  90             me->https->Server->StartPos = 0;
  91       }
  92       else
  93          me->https->Server->StartPos = 0;
  94
  95       me->Res->LastModified = me->https->Server->Date;
  96       me->Res->Size = me->https->Server->TotalFileSize;
  97       me->Res->ResumePoint = me->https->Server->StartPos;
  98
  99       // we expect valid data, so tell our caller we get the file now
 100       if (me->https->Server->Result >= 200 && me->https->Server->Result < 300)
 101       {
 102          if (me->https->Server->JunkSize == 0 && me->Res->Size != 0 && me->Res->Size > me->Res->ResumePoint)
 103             me->https->URIStart(*me->Res);
 104          if (me->https->Server->AddPartialFileToHashes(*(me->https->File)) == false)
 105             return 0;
 106       }
 107    }
 108    else if (me->https->Server->HeaderLine(line) == false)
 109       return 0;
 110
 111    return size*nmemb;
 112 }
 113
 114 size_t
 115 HttpsMethod::write_data(void *buffer, size_t size, size_t nmemb, void *userp)
 116 {
 117    HttpsMethod *me = static_cast<HttpsMethod *>(userp);
 118    size_t buffer_size = size * nmemb;
 119    // we don't need to count the junk here, just drop anything we get as
 120    // we don't always know how long it would be, e.g. in chunked encoding.
 121    if (me->Server->JunkSize != 0)
 122       return buffer_size;
 123
 124    if(me->File->Write(buffer, buffer_size) != true)
 125       return 0;
 126
 127    if(me->Queue->MaximumSize > 0)
 128    {
 129       unsigned long long const TotalWritten = me->File->Tell();
 130       if (TotalWritten > me->Queue->MaximumSize)
 131       {
 132          me->SetFailReason("MaximumSizeExceeded");
 133          _error->Error("Writing more data than expected (%llu > %llu)",
 134                TotalWritten, me->Queue->MaximumSize);
 135          return 0;
 136       }
 137    }
 138
 139    if (me->Server->GetHashes()->Add((unsigned char const * const)buffer, buffer_size) == false)
 140       return 0;
 141
 142    return buffer_size;
 143 }
 144
 145 // HttpsServerState::HttpsServerState - Constructor                     /*{{{*/
 146 HttpsServerState::HttpsServerState(URI Srv,HttpsMethod * Owner) : ServerState(Srv, Owner), Hash(NULL)
 147 {
 148    TimeOut = _config->FindI("Acquire::https::Timeout",TimeOut);
 149    Reset();
 150 }
 151                                                                         /*}}}*/
 152 bool HttpsServerState::InitHashes(HashStringList const &ExpectedHashes) /*{{{*/
 153 {
 154    delete Hash;
 155    Hash = new Hashes(ExpectedHashes);
 156    return true;
 157 }
 158                                                                         /*}}}*/
 159 APT_PURE Hashes * HttpsServerState::GetHashes()                         /*{{{*/
 160 {
 161    return Hash;
 162 }
 163                                                                         /*}}}*/
 164
 165 void HttpsMethod::SetupProxy()                                          /*{{{*/
 166 {
 167    URI ServerName = Queue->Uri;
 168
 169    // Determine the proxy setting
 170    AutoDetectProxy(ServerName);
 171
 172    // Curl should never read proxy settings from the environment, as
 173    // we determine which proxy to use.  Do this for consistency among
 174    // methods and prevent an environment variable overriding a
 175    // no-proxy ("DIRECT") setting in apt.conf.
 176    curl_easy_setopt(curl, CURLOPT_PROXY, "");
 177
 178    // Determine the proxy setting - try https first, fallback to http and use env at last
 179    string UseProxy = _config->Find("Acquire::https::Proxy::" + ServerName.Host,
 180                                    _config->Find("Acquire::http::Proxy::" + ServerName.Host).c_str());
 181
 182    if (UseProxy.empty() == true)
 183       UseProxy = _config->Find("Acquire::https::Proxy", _config->Find("Acquire::http::Proxy").c_str());
 184
 185    // User want to use NO proxy, so nothing to setup
 186    if (UseProxy == "DIRECT")
 187       return;
 188
 189    // Parse no_proxy, a comma (,) separated list of domains we don't want to use
 190    // a proxy for so we stop right here if it is in the list
 191    if (getenv("no_proxy") != 0 && CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 192       return;
 193
 194    if (UseProxy.empty() == true)
 195    {
 196       const char* result = getenv("https_proxy");
 197       // FIXME: Fall back to http_proxy is to remain compatible with
 198       // existing setups and behaviour of apt.conf.  This should be
 199       // deprecated in the future (including apt.conf).  Most other
 200       // programs do not fall back to http proxy settings and neither
 201       // should Apt.
 202       if (result == NULL)
 203          result = getenv("http_proxy");
 204       UseProxy = result == NULL ? "" : result;
 205    }
 206
 207    // Determine what host and port to use based on the proxy settings
 208    if (UseProxy.empty() == false)
 209    {
 210       Proxy = UseProxy;
 211       if (Proxy.Access == "socks5h")
 212          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5_HOSTNAME);
 213       else if (Proxy.Access == "socks5")
 214          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5);
 215       else if (Proxy.Access == "socks4a")
 216          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4A);
 217       else if (Proxy.Access == "socks")
 218          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4);
 219       else
 220          curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_HTTP);
 221
 222       if (Proxy.Port != 1)
 223          curl_easy_setopt(curl, CURLOPT_PROXYPORT, Proxy.Port);
 224       curl_easy_setopt(curl, CURLOPT_PROXY, Proxy.Host.c_str());
 225       if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 226       {
 227          curl_easy_setopt(curl, CURLOPT_PROXYUSERNAME, Proxy.User.c_str());
 228          curl_easy_setopt(curl, CURLOPT_PROXYPASSWORD, Proxy.Password.c_str());
 229       }
 230    }
 231 }                                                                       /*}}}*/
 232 // HttpsMethod::Fetch - Fetch an item                                   /*{{{*/
 233 // ---------------------------------------------------------------------
 234 /* This adds an item to the pipeline. We keep the pipeline at a fixed
 235    depth. */
 236 bool HttpsMethod::Fetch(FetchItem *Itm)
 237 {
 238    struct stat SBuf;
 239    struct curl_slist *headers=NULL;
 240    char curl_errorstr[CURL_ERROR_SIZE];
 241    URI Uri = Itm->Uri;
 242    string remotehost = Uri.Host;
 243
 244    // TODO:
 245    //       - http::Pipeline-Depth
 246    //       - error checking/reporting
 247    //       - more debug options? (CURLOPT_DEBUGFUNCTION?)
 248
 249    curl_easy_reset(curl);
 250    SetupProxy();
 251
 252    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 253
 254    FetchResult Res;
 255    CURLUserPointer userp(this, &Res, Itm);
 256    // callbacks
 257    curl_easy_setopt(curl, CURLOPT_URL, static_cast<string>(Uri).c_str());
 258    curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION, parse_header);
 259    curl_easy_setopt(curl, CURLOPT_WRITEHEADER, &userp);
 260    curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, write_data);
 261    curl_easy_setopt(curl, CURLOPT_WRITEDATA, this);
 262    // options
 263    curl_easy_setopt(curl, CURLOPT_NOPROGRESS, true);
 264    curl_easy_setopt(curl, CURLOPT_FILETIME, true);
 265    // only allow curl to handle https, not the other stuff it supports
 266    curl_easy_setopt(curl, CURLOPT_PROTOCOLS, CURLPROTO_HTTPS);
 267    curl_easy_setopt(curl, CURLOPT_REDIR_PROTOCOLS, CURLPROTO_HTTPS);
 268
 269    // SSL parameters are set by default to the common (non mirror-specific) value
 270    // if available (or a default one) and gets overload by mirror-specific ones.
 271
 272    // File containing the list of trusted CA.
 273    string cainfo = _config->Find("Acquire::https::CaInfo","");
 274    string knob = "Acquire::https::"+remotehost+"::CaInfo";
 275    cainfo = _config->Find(knob.c_str(),cainfo.c_str());
 276    if(cainfo.empty() == false)
 277       curl_easy_setopt(curl, CURLOPT_CAINFO,cainfo.c_str());
 278
 279    // Check server certificate against previous CA list ...
 280    bool peer_verify = _config->FindB("Acquire::https::Verify-Peer",true);
 281    knob = "Acquire::https::" + remotehost + "::Verify-Peer";
 282    peer_verify = _config->FindB(knob.c_str(), peer_verify);
 283    curl_easy_setopt(curl, CURLOPT_SSL_VERIFYPEER, peer_verify);
 284
 285    // ... and hostname against cert CN or subjectAltName
 286    bool verify = _config->FindB("Acquire::https::Verify-Host",true);
 287    knob = "Acquire::https::"+remotehost+"::Verify-Host";
 288    verify = _config->FindB(knob.c_str(),verify);
 289    int const default_verify = (verify == true) ? 2 : 0;
 290    curl_easy_setopt(curl, CURLOPT_SSL_VERIFYHOST, default_verify);
 291
 292    // Also enforce issuer of server certificate using its cert
 293    string issuercert = _config->Find("Acquire::https::IssuerCert","");
 294    knob = "Acquire::https::"+remotehost+"::IssuerCert";
 295    issuercert = _config->Find(knob.c_str(),issuercert.c_str());
 296    if(issuercert.empty() == false)
 297       curl_easy_setopt(curl, CURLOPT_ISSUERCERT,issuercert.c_str());
 298
 299    // For client authentication, certificate file ...
 300    string pem = _config->Find("Acquire::https::SslCert","");
 301    knob = "Acquire::https::"+remotehost+"::SslCert";
 302    pem = _config->Find(knob.c_str(),pem.c_str());
 303    if(pem.empty() == false)
 304       curl_easy_setopt(curl, CURLOPT_SSLCERT, pem.c_str());
 305
 306    // ... and associated key.
 307    string key = _config->Find("Acquire::https::SslKey","");
 308    knob = "Acquire::https::"+remotehost+"::SslKey";
 309    key = _config->Find(knob.c_str(),key.c_str());
 310    if(key.empty() == false)
 311       curl_easy_setopt(curl, CURLOPT_SSLKEY, key.c_str());
 312
 313    // Allow forcing SSL version to SSLv3 or TLSv1 (SSLv2 is not
 314    // supported by GnuTLS).
 315    long final_version = CURL_SSLVERSION_DEFAULT;
 316    string sslversion = _config->Find("Acquire::https::SslForceVersion","");
 317    knob = "Acquire::https::"+remotehost+"::SslForceVersion";
 318    sslversion = _config->Find(knob.c_str(),sslversion.c_str());
 319    if(sslversion == "TLSv1")
 320      final_version = CURL_SSLVERSION_TLSv1;
 321    else if(sslversion == "SSLv3")
 322      final_version = CURL_SSLVERSION_SSLv3;
 323    curl_easy_setopt(curl, CURLOPT_SSLVERSION, final_version);
 324
 325    // CRL file
 326    string crlfile = _config->Find("Acquire::https::CrlFile","");
 327    knob = "Acquire::https::"+remotehost+"::CrlFile";
 328    crlfile = _config->Find(knob.c_str(),crlfile.c_str());
 329    if(crlfile.empty() == false)
 330       curl_easy_setopt(curl, CURLOPT_CRLFILE, crlfile.c_str());
 331
 332    // cache-control
 333    if(_config->FindB("Acquire::https::No-Cache",
 334         _config->FindB("Acquire::http::No-Cache",false)) == false)
 335    {
 336       // cache enabled
 337       if (_config->FindB("Acquire::https::No-Store",
 338                 _config->FindB("Acquire::http::No-Store",false)) == true)
 339          headers = curl_slist_append(headers,"Cache-Control: no-store");
 340       stringstream ss;
 341       ioprintf(ss, "Cache-Control: max-age=%u", _config->FindI("Acquire::https::Max-Age",
 342                 _config->FindI("Acquire::http::Max-Age",0)));
 343       headers = curl_slist_append(headers, ss.str().c_str());
 344    } else {
 345       // cache disabled by user
 346       headers = curl_slist_append(headers, "Cache-Control: no-cache");
 347       headers = curl_slist_append(headers, "Pragma: no-cache");
 348    }
 349    curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
 350
 351    // speed limit
 352    int const dlLimit = _config->FindI("Acquire::https::Dl-Limit",
 353                 _config->FindI("Acquire::http::Dl-Limit",0))*1024;
 354    if (dlLimit > 0)
 355       curl_easy_setopt(curl, CURLOPT_MAX_RECV_SPEED_LARGE, dlLimit);
 356
 357    // set header
 358    curl_easy_setopt(curl, CURLOPT_USERAGENT,
 359         _config->Find("Acquire::https::User-Agent",
 360                 _config->Find("Acquire::http::User-Agent",
 361                         "Debian APT-CURL/1.0 (" PACKAGE_VERSION ")").c_str()).c_str());
 362
 363    // set timeout
 364    int const timeout = _config->FindI("Acquire::https::Timeout",
 365                 _config->FindI("Acquire::http::Timeout",120));
 366    curl_easy_setopt(curl, CURLOPT_CONNECTTIMEOUT, timeout);
 367    //set really low lowspeed timeout (see #497983)
 368    curl_easy_setopt(curl, CURLOPT_LOW_SPEED_LIMIT, DL_MIN_SPEED);
 369    curl_easy_setopt(curl, CURLOPT_LOW_SPEED_TIME, timeout);
 370
 371    // set redirect options and default to 10 redirects
 372    curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, AllowRedirect);
 373    curl_easy_setopt(curl, CURLOPT_MAXREDIRS, 10);
 374
 375    // debug
 376    if (Debug == true)
 377       curl_easy_setopt(curl, CURLOPT_VERBOSE, true);
 378
 379    // error handling
 380    curl_errorstr[0] = '\0';
 381    curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_errorstr);
 382
 383    // If we ask for uncompressed files servers might respond with content-
 384    // negotiation which lets us end up with compressed files we do not support,
 385    // see 657029, 657560 and co, so if we have no extension on the request
 386    // ask for text only. As a sidenote: If there is nothing to negotate servers
 387    // seem to be nice and ignore it.
 388    if (_config->FindB("Acquire::https::SendAccept", _config->FindB("Acquire::http::SendAccept", true)) == true)
 389    {
 390       size_t const filepos = Itm->Uri.find_last_of('/');
 391       string const file = Itm->Uri.substr(filepos + 1);
 392       if (flExtension(file) == file)
 393          headers = curl_slist_append(headers, "Accept: text/*");
 394    }
 395
 396    // if we have the file send an if-range query with a range header
 397    if (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    // go for it - if the file exists, append on it
 412    File = new FileFd(Itm->DestFile, FileFd::WriteAny);
 413    Server = CreateServerState(Itm->Uri);
 414    if (Server->InitHashes(Itm->ExpectedHashes) == false)
 415       return false;
 416
 417    // keep apt updated
 418    Res.Filename = Itm->DestFile;
 419
 420    // get it!
 421    CURLcode success = curl_easy_perform(curl);
 422
 423    // If the server returns 200 OK but the If-Modified-Since condition is not
 424    // met, CURLINFO_CONDITION_UNMET will be set to 1
 425    long curl_condition_unmet = 0;
 426    curl_easy_getinfo(curl, CURLINFO_CONDITION_UNMET, &curl_condition_unmet);
 427
 428    File->Close();
 429    curl_slist_free_all(headers);
 430
 431    // cleanup
 432    if (success != CURLE_OK)
 433    {
 434 #pragma GCC diagnostic push
 435 #pragma GCC diagnostic ignored "-Wswitch"
 436       switch (success)
 437       {
 438          case CURLE_COULDNT_RESOLVE_PROXY:
 439          case CURLE_COULDNT_RESOLVE_HOST:
 440             SetFailReason("ResolveFailure");
 441             break;
 442          case CURLE_COULDNT_CONNECT:
 443             SetFailReason("ConnectionRefused");
 444             break;
 445          case CURLE_OPERATION_TIMEDOUT:
 446             SetFailReason("Timeout");
 447             break;
 448       }
 449 #pragma GCC diagnostic pop
 450       // only take curls technical errors if we haven't our own
 451       // (e.g. for the maximum size limit we have and curls can be confusing)
 452       if (_error->PendingError() == false)
 453          _error->Error("%s", curl_errorstr);
 454       else
 455          _error->Warning("curl: %s", curl_errorstr);
 456       return false;
 457    }
 458
 459    // server says file not modified
 460    if (Server->Result == 304 || curl_condition_unmet == 1)
 461    {
 462       RemoveFile("https", File->Name());
 463       Res.IMSHit = true;
 464       Res.LastModified = Itm->LastModified;
 465       Res.Size = 0;
 466       URIDone(Res);
 467       return true;
 468    }
 469    Res.IMSHit = false;
 470
 471    if (Server->Result != 200 && // OK
 472          Server->Result != 206 && // Partial
 473          Server->Result != 416) // invalid Range
 474    {
 475       char err[255];
 476       snprintf(err, sizeof(err) - 1, "HttpError%i", Server->Result);
 477       SetFailReason(err);
 478       _error->Error("%i %s", Server->Result, Server->Code);
 479       // unlink, no need keep 401/404 page content in partial/
 480       RemoveFile("https", File->Name());
 481       return false;
 482    }
 483
 484    // invalid range-request
 485    if (Server->Result == 416)
 486    {
 487       RemoveFile("https", File->Name());
 488       delete File;
 489       Redirect(Itm->Uri);
 490       return true;
 491    }
 492
 493    struct stat resultStat;
 494    if (unlikely(stat(File->Name().c_str(), &resultStat) != 0))
 495    {
 496       _error->Errno("stat", "Unable to access file %s", File->Name().c_str());
 497       return false;
 498    }
 499    Res.Size = resultStat.st_size;
 500
 501    // Timestamp
 502    curl_easy_getinfo(curl, CURLINFO_FILETIME, &Res.LastModified);
 503    if (Res.LastModified != -1)
 504    {
 505       struct timeval times[2];
 506       times[0].tv_sec = Res.LastModified;
 507       times[1].tv_sec = Res.LastModified;
 508       times[0].tv_usec = times[1].tv_usec = 0;
 509       utimes(File->Name().c_str(), times);
 510    }
 511    else
 512       Res.LastModified = resultStat.st_mtime;
 513
 514    // take hashes
 515    Res.TakeHashes(*(Server->GetHashes()));
 516
 517    // keep apt updated
 518    URIDone(Res);
 519
 520    // cleanup
 521    delete File;
 522
 523    return true;
 524 }
 525                                                                         /*}}}*/
 526 // HttpsMethod::Configuration - Handle a configuration message          /*{{{*/
 527 bool HttpsMethod::Configuration(string Message)
 528 {
 529    if (ServerMethod::Configuration(Message) == false)
 530       return false;
 531
 532    AllowRedirect = _config->FindB("Acquire::https::AllowRedirect",
 533         _config->FindB("Acquire::http::AllowRedirect", true));
 534    Debug = _config->FindB("Debug::Acquire::https",false);
 535
 536    return true;
 537 }
 538                                                                         /*}}}*/
 539 std::unique_ptr<ServerState> HttpsMethod::CreateServerState(URI const &uri)/*{{{*/
 540 {
 541    return std::unique_ptr<ServerState>(new HttpsServerState(uri, this));
 542 }
 543                                                                         /*}}}*/
 544
 545 int main()
 546 {
 547    return HttpsMethod().Run();
 548 }
 549