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