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