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