methods/http.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    HTTP Acquire Method - This is the HTTP aquire method for APT.
   7
   8    It uses HTTP/1.1 and many of the fancy options there-in, such as
   9    pipelining, range, if-range and so on.
  10
  11    It is based on a doubly buffered select loop. A groupe of requests are
  12    fed into a single output buffer that is constantly fed out the
  13    socket. This provides ideal pipelining as in many cases all of the
  14    requests will fit into a single packet. The input socket is buffered
  15    the same way and fed into the fd for the file (may be a pipe in future).
  16
  17    This double buffering provides fairly substantial transfer rates,
  18    compared to wget the http method is about 4% faster. Most importantly,
  19    when HTTP is compared with FTP as a protocol the speed difference is
  20    huge. In tests over the internet from two sites to llug (via ATM) this
  21    program got 230k/s sustained http transfer rates. FTP on the other
  22    hand topped out at 170k/s. That combined with the time to setup the
  23    FTP connection makes HTTP a vastly superior protocol.
  24
  25    ##################################################################### */
  26                                                                         /*}}}*/
  27 // Include Files                                                        /*{{{*/
  28 #include <config.h>
  29
  30 #include <apt-pkg/fileutl.h>
  31 #include <apt-pkg/acquire-method.h>
  32 #include <apt-pkg/configuration.h>
  33 #include <apt-pkg/error.h>
  34 #include <apt-pkg/hashes.h>
  35 #include <apt-pkg/netrc.h>
  36
  37 #include <sys/stat.h>
  38 #include <sys/time.h>
  39 #include <utime.h>
  40 #include <unistd.h>
  41 #include <signal.h>
  42 #include <stdio.h>
  43 #include <errno.h>
  44 #include <string.h>
  45 #include <iostream>
  46 #include <map>
  47
  48 // Internet stuff
  49 #include <netdb.h>
  50
  51 #include "config.h"
  52 #include "connect.h"
  53 #include "rfc2553emu.h"
  54 #include "http.h"
  55
  56 #include <apti18n.h>
  57                                                                         /*}}}*/
  58 using namespace std;
  59
  60 string HttpMethod::FailFile;
  61 int HttpMethod::FailFd = -1;
  62 time_t HttpMethod::FailTime = 0;
  63 unsigned long PipelineDepth = 10;
  64 unsigned long TimeOut = 120;
  65 bool AllowRedirect = false;
  66 bool Debug = false;
  67 URI Proxy;
  68
  69 unsigned long long CircleBuf::BwReadLimit=0;
  70 unsigned long long CircleBuf::BwTickReadData=0;
  71 struct timeval CircleBuf::BwReadTick={0,0};
  72 const unsigned int CircleBuf::BW_HZ=10;
  73
  74 // CircleBuf::CircleBuf - Circular input buffer                         /*{{{*/
  75 // ---------------------------------------------------------------------
  76 /* */
  77 CircleBuf::CircleBuf(unsigned long long Size) : Size(Size), Hash(0)
  78 {
  79    Buf = new unsigned char[Size];
  80    Reset();
  81
  82    CircleBuf::BwReadLimit = _config->FindI("Acquire::http::Dl-Limit",0)*1024;
  83 }
  84                                                                         /*}}}*/
  85 // CircleBuf::Reset - Reset to the default state                        /*{{{*/
  86 // ---------------------------------------------------------------------
  87 /* */
  88 void CircleBuf::Reset()
  89 {
  90    InP = 0;
  91    OutP = 0;
  92    StrPos = 0;
  93    MaxGet = (unsigned long long)-1;
  94    OutQueue = string();
  95    if (Hash != 0)
  96    {
  97       delete Hash;
  98       Hash = new Hashes;
  99    }
 100 };
 101                                                                         /*}}}*/
 102 // CircleBuf::Read - Read from a FD into the circular buffer            /*{{{*/
 103 // ---------------------------------------------------------------------
 104 /* This fills up the buffer with as much data as is in the FD, assuming it
 105    is non-blocking.. */
 106 bool CircleBuf::Read(int Fd)
 107 {
 108    unsigned long long BwReadMax;
 109
 110    while (1)
 111    {
 112       // Woops, buffer is full
 113       if (InP - OutP == Size)
 114          return true;
 115
 116       // what's left to read in this tick
 117       BwReadMax = CircleBuf::BwReadLimit/BW_HZ;
 118
 119       if(CircleBuf::BwReadLimit) {
 120          struct timeval now;
 121          gettimeofday(&now,0);
 122
 123          unsigned long long d = (now.tv_sec-CircleBuf::BwReadTick.tv_sec)*1000000 +
 124             now.tv_usec-CircleBuf::BwReadTick.tv_usec;
 125          if(d > 1000000/BW_HZ) {
 126             CircleBuf::BwReadTick = now;
 127             CircleBuf::BwTickReadData = 0;
 128          }
 129
 130          if(CircleBuf::BwTickReadData >= BwReadMax) {
 131             usleep(1000000/BW_HZ);
 132             return true;
 133          }
 134       }
 135
 136       // Write the buffer segment
 137       ssize_t Res;
 138       if(CircleBuf::BwReadLimit) {
 139          Res = read(Fd,Buf + (InP%Size),
 140                     BwReadMax > LeftRead() ? LeftRead() : BwReadMax);
 141       } else
 142          Res = read(Fd,Buf + (InP%Size),LeftRead());
 143
 144       if(Res > 0 && BwReadLimit > 0)
 145          CircleBuf::BwTickReadData += Res;
 146
 147       if (Res == 0)
 148          return false;
 149       if (Res < 0)
 150       {
 151          if (errno == EAGAIN)
 152             return true;
 153          return false;
 154       }
 155
 156       if (InP == 0)
 157          gettimeofday(&Start,0);
 158       InP += Res;
 159    }
 160 }
 161                                                                         /*}}}*/
 162 // CircleBuf::Read - Put the string into the buffer                     /*{{{*/
 163 // ---------------------------------------------------------------------
 164 /* This will hold the string in and fill the buffer with it as it empties */
 165 bool CircleBuf::Read(string Data)
 166 {
 167    OutQueue += Data;
 168    FillOut();
 169    return true;
 170 }
 171                                                                         /*}}}*/
 172 // CircleBuf::FillOut - Fill the buffer from the output queue           /*{{{*/
 173 // ---------------------------------------------------------------------
 174 /* */
 175 void CircleBuf::FillOut()
 176 {
 177    if (OutQueue.empty() == true)
 178       return;
 179    while (1)
 180    {
 181       // Woops, buffer is full
 182       if (InP - OutP == Size)
 183          return;
 184
 185       // Write the buffer segment
 186       unsigned long long Sz = LeftRead();
 187       if (OutQueue.length() - StrPos < Sz)
 188          Sz = OutQueue.length() - StrPos;
 189       memcpy(Buf + (InP%Size),OutQueue.c_str() + StrPos,Sz);
 190
 191       // Advance
 192       StrPos += Sz;
 193       InP += Sz;
 194       if (OutQueue.length() == StrPos)
 195       {
 196          StrPos = 0;
 197          OutQueue = "";
 198          return;
 199       }
 200    }
 201 }
 202                                                                         /*}}}*/
 203 // CircleBuf::Write - Write from the buffer into a FD                   /*{{{*/
 204 // ---------------------------------------------------------------------
 205 /* This empties the buffer into the FD. */
 206 bool CircleBuf::Write(int Fd)
 207 {
 208    while (1)
 209    {
 210       FillOut();
 211
 212       // Woops, buffer is empty
 213       if (OutP == InP)
 214          return true;
 215
 216       if (OutP == MaxGet)
 217          return true;
 218
 219       // Write the buffer segment
 220       ssize_t Res;
 221       Res = write(Fd,Buf + (OutP%Size),LeftWrite());
 222
 223       if (Res == 0)
 224          return false;
 225       if (Res < 0)
 226       {
 227          if (errno == EAGAIN)
 228             return true;
 229
 230          return false;
 231       }
 232
 233       if (Hash != 0)
 234          Hash->Add(Buf + (OutP%Size),Res);
 235
 236       OutP += Res;
 237    }
 238 }
 239                                                                         /*}}}*/
 240 // CircleBuf::WriteTillEl - Write from the buffer to a string           /*{{{*/
 241 // ---------------------------------------------------------------------
 242 /* This copies till the first empty line */
 243 bool CircleBuf::WriteTillEl(string &Data,bool Single)
 244 {
 245    // We cheat and assume it is unneeded to have more than one buffer load
 246    for (unsigned long long I = OutP; I < InP; I++)
 247    {
 248       if (Buf[I%Size] != '\n')
 249          continue;
 250       ++I;
 251
 252       if (Single == false)
 253       {
 254          if (I < InP  && Buf[I%Size] == '\r')
 255             ++I;
 256          if (I >= InP || Buf[I%Size] != '\n')
 257             continue;
 258          ++I;
 259       }
 260
 261       Data = "";
 262       while (OutP < I)
 263       {
 264          unsigned long long Sz = LeftWrite();
 265          if (Sz == 0)
 266             return false;
 267          if (I - OutP < Sz)
 268             Sz = I - OutP;
 269          Data += string((char *)(Buf + (OutP%Size)),Sz);
 270          OutP += Sz;
 271       }
 272       return true;
 273    }
 274    return false;
 275 }
 276                                                                         /*}}}*/
 277 // CircleBuf::Stats - Print out stats information                       /*{{{*/
 278 // ---------------------------------------------------------------------
 279 /* */
 280 void CircleBuf::Stats()
 281 {
 282    if (InP == 0)
 283       return;
 284
 285    struct timeval Stop;
 286    gettimeofday(&Stop,0);
 287 /*   float Diff = Stop.tv_sec - Start.tv_sec +
 288              (float)(Stop.tv_usec - Start.tv_usec)/1000000;
 289    clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
 290 }
 291                                                                         /*}}}*/
 292 CircleBuf::~CircleBuf()
 293 {
 294    delete [] Buf;
 295    delete Hash;
 296 }
 297
 298 // ServerState::ServerState - Constructor                               /*{{{*/
 299 // ---------------------------------------------------------------------
 300 /* */
 301 ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner),
 302                         In(64*1024), Out(4*1024),
 303                         ServerName(Srv)
 304 {
 305    Reset();
 306 }
 307                                                                         /*}}}*/
 308 // ServerState::Open - Open a connection to the server                  /*{{{*/
 309 // ---------------------------------------------------------------------
 310 /* This opens a connection to the server. */
 311 bool ServerState::Open()
 312 {
 313    // Use the already open connection if possible.
 314    if (ServerFd != -1)
 315       return true;
 316
 317    Close();
 318    In.Reset();
 319    Out.Reset();
 320    Persistent = true;
 321
 322    // Determine the proxy setting
 323    string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
 324    if (!SpecificProxy.empty())
 325    {
 326            if (SpecificProxy == "DIRECT")
 327                    Proxy = "";
 328            else
 329                    Proxy = SpecificProxy;
 330    }
 331    else
 332    {
 333            string DefProxy = _config->Find("Acquire::http::Proxy");
 334            if (!DefProxy.empty())
 335            {
 336                    Proxy = DefProxy;
 337            }
 338            else
 339            {
 340                    char* result = getenv("http_proxy");
 341                    Proxy = result ? result : "";
 342            }
 343    }
 344
 345    // Parse no_proxy, a , separated list of domains
 346    if (getenv("no_proxy") != 0)
 347    {
 348       if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 349          Proxy = "";
 350    }
 351
 352    // Determine what host and port to use based on the proxy settings
 353    int Port = 0;
 354    string Host;
 355    if (Proxy.empty() == true || Proxy.Host.empty() == true)
 356    {
 357       if (ServerName.Port != 0)
 358          Port = ServerName.Port;
 359       Host = ServerName.Host;
 360    }
 361    else
 362    {
 363       if (Proxy.Port != 0)
 364          Port = Proxy.Port;
 365       Host = Proxy.Host;
 366    }
 367
 368    // Connect to the remote server
 369    if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
 370       return false;
 371
 372    return true;
 373 }
 374                                                                         /*}}}*/
 375 // ServerState::Close - Close a connection to the server                /*{{{*/
 376 // ---------------------------------------------------------------------
 377 /* */
 378 bool ServerState::Close()
 379 {
 380    close(ServerFd);
 381    ServerFd = -1;
 382    return true;
 383 }
 384                                                                         /*}}}*/
 385 // ServerState::RunHeaders - Get the headers before the data            /*{{{*/
 386 // ---------------------------------------------------------------------
 387 /* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
 388    parse error occurred */
 389 ServerState::RunHeadersResult ServerState::RunHeaders()
 390 {
 391    State = Header;
 392
 393    Owner->Status(_("Waiting for headers"));
 394
 395    Major = 0;
 396    Minor = 0;
 397    Result = 0;
 398    Size = 0;
 399    StartPos = 0;
 400    Encoding = Closes;
 401    HaveContent = false;
 402    time(&Date);
 403
 404    do
 405    {
 406       string Data;
 407       if (In.WriteTillEl(Data) == false)
 408          continue;
 409
 410       if (Debug == true)
 411          clog << Data;
 412
 413       for (string::const_iterator I = Data.begin(); I < Data.end(); ++I)
 414       {
 415          string::const_iterator J = I;
 416          for (; J != Data.end() && *J != '\n' && *J != '\r'; ++J);
 417          if (HeaderLine(string(I,J)) == false)
 418             return RUN_HEADERS_PARSE_ERROR;
 419          I = J;
 420       }
 421
 422       // 100 Continue is a Nop...
 423       if (Result == 100)
 424          continue;
 425
 426       // Tidy up the connection persistance state.
 427       if (Encoding == Closes && HaveContent == true)
 428          Persistent = false;
 429
 430       return RUN_HEADERS_OK;
 431    }
 432    while (Owner->Go(false,this) == true);
 433
 434    return RUN_HEADERS_IO_ERROR;
 435 }
 436                                                                         /*}}}*/
 437 // ServerState::RunData - Transfer the data from the socket             /*{{{*/
 438 // ---------------------------------------------------------------------
 439 /* */
 440 bool ServerState::RunData()
 441 {
 442    State = Data;
 443
 444    // Chunked transfer encoding is fun..
 445    if (Encoding == Chunked)
 446    {
 447       while (1)
 448       {
 449          // Grab the block size
 450          bool Last = true;
 451          string Data;
 452          In.Limit(-1);
 453          do
 454          {
 455             if (In.WriteTillEl(Data,true) == true)
 456                break;
 457          }
 458          while ((Last = Owner->Go(false,this)) == true);
 459
 460          if (Last == false)
 461             return false;
 462
 463          // See if we are done
 464          unsigned long long Len = strtoull(Data.c_str(),0,16);
 465          if (Len == 0)
 466          {
 467             In.Limit(-1);
 468
 469             // We have to remove the entity trailer
 470             Last = true;
 471             do
 472             {
 473                if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
 474                   break;
 475             }
 476             while ((Last = Owner->Go(false,this)) == true);
 477             if (Last == false)
 478                return false;
 479             return !_error->PendingError();
 480          }
 481
 482          // Transfer the block
 483          In.Limit(Len);
 484          while (Owner->Go(true,this) == true)
 485             if (In.IsLimit() == true)
 486                break;
 487
 488          // Error
 489          if (In.IsLimit() == false)
 490             return false;
 491
 492          // The server sends an extra new line before the next block specifier..
 493          In.Limit(-1);
 494          Last = true;
 495          do
 496          {
 497             if (In.WriteTillEl(Data,true) == true)
 498                break;
 499          }
 500          while ((Last = Owner->Go(false,this)) == true);
 501          if (Last == false)
 502             return false;
 503       }
 504    }
 505    else
 506    {
 507       /* Closes encoding is used when the server did not specify a size, the
 508          loss of the connection means we are done */
 509       if (Encoding == Closes)
 510          In.Limit(-1);
 511       else
 512          In.Limit(Size - StartPos);
 513
 514       // Just transfer the whole block.
 515       do
 516       {
 517          if (In.IsLimit() == false)
 518             continue;
 519
 520          In.Limit(-1);
 521          return !_error->PendingError();
 522       }
 523       while (Owner->Go(true,this) == true);
 524    }
 525
 526    return Owner->Flush(this) && !_error->PendingError();
 527 }
 528                                                                         /*}}}*/
 529 // ServerState::HeaderLine - Process a header line                      /*{{{*/
 530 // ---------------------------------------------------------------------
 531 /* */
 532 bool ServerState::HeaderLine(string Line)
 533 {
 534    if (Line.empty() == true)
 535       return true;
 536
 537    string::size_type Pos = Line.find(' ');
 538    if (Pos == string::npos || Pos+1 > Line.length())
 539    {
 540       // Blah, some servers use "connection:closes", evil.
 541       Pos = Line.find(':');
 542       if (Pos == string::npos || Pos + 2 > Line.length())
 543          return _error->Error(_("Bad header line"));
 544       Pos++;
 545    }
 546
 547    // Parse off any trailing spaces between the : and the next word.
 548    string::size_type Pos2 = Pos;
 549    while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
 550       Pos2++;
 551
 552    string Tag = string(Line,0,Pos);
 553    string Val = string(Line,Pos2);
 554
 555    if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
 556    {
 557       // Evil servers return no version
 558       if (Line[4] == '/')
 559       {
 560          int const elements = sscanf(Line.c_str(),"HTTP/%u.%u %u%[^\n]",&Major,&Minor,&Result,Code);
 561          if (elements == 3)
 562          {
 563             Code[0] = '\0';
 564             if (Debug == true)
 565                clog << "HTTP server doesn't give Reason-Phrase for " << Result << std::endl;
 566          }
 567          else if (elements != 4)
 568             return _error->Error(_("The HTTP server sent an invalid reply header"));
 569       }
 570       else
 571       {
 572          Major = 0;
 573          Minor = 9;
 574          if (sscanf(Line.c_str(),"HTTP %u%[^\n]",&Result,Code) != 2)
 575             return _error->Error(_("The HTTP server sent an invalid reply header"));
 576       }
 577
 578       /* Check the HTTP response header to get the default persistance
 579          state. */
 580       if (Major < 1)
 581          Persistent = false;
 582       else
 583       {
 584          if (Major == 1 && Minor <= 0)
 585             Persistent = false;
 586          else
 587             Persistent = true;
 588       }
 589
 590       return true;
 591    }
 592
 593    if (stringcasecmp(Tag,"Content-Length:") == 0)
 594    {
 595       if (Encoding == Closes)
 596          Encoding = Stream;
 597       HaveContent = true;
 598
 599       // The length is already set from the Content-Range header
 600       if (StartPos != 0)
 601          return true;
 602
 603       if (sscanf(Val.c_str(),"%llu",&Size) != 1)
 604          return _error->Error(_("The HTTP server sent an invalid Content-Length header"));
 605       return true;
 606    }
 607
 608    if (stringcasecmp(Tag,"Content-Type:") == 0)
 609    {
 610       HaveContent = true;
 611       return true;
 612    }
 613
 614    if (stringcasecmp(Tag,"Content-Range:") == 0)
 615    {
 616       HaveContent = true;
 617
 618       if (sscanf(Val.c_str(),"bytes %llu-%*u/%llu",&StartPos,&Size) != 2)
 619          return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
 620       if ((unsigned long long)StartPos > Size)
 621          return _error->Error(_("This HTTP server has broken range support"));
 622       return true;
 623    }
 624
 625    if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
 626    {
 627       HaveContent = true;
 628       if (stringcasecmp(Val,"chunked") == 0)
 629          Encoding = Chunked;
 630       return true;
 631    }
 632
 633    if (stringcasecmp(Tag,"Connection:") == 0)
 634    {
 635       if (stringcasecmp(Val,"close") == 0)
 636          Persistent = false;
 637       if (stringcasecmp(Val,"keep-alive") == 0)
 638          Persistent = true;
 639       return true;
 640    }
 641
 642    if (stringcasecmp(Tag,"Last-Modified:") == 0)
 643    {
 644       if (RFC1123StrToTime(Val.c_str(), Date) == false)
 645          return _error->Error(_("Unknown date format"));
 646       return true;
 647    }
 648
 649    if (stringcasecmp(Tag,"Location:") == 0)
 650    {
 651       Location = Val;
 652       return true;
 653    }
 654
 655    return true;
 656 }
 657                                                                         /*}}}*/
 658
 659 // HttpMethod::SendReq - Send the HTTP request                          /*{{{*/
 660 // ---------------------------------------------------------------------
 661 /* This places the http request in the outbound buffer */
 662 void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
 663 {
 664    URI Uri = Itm->Uri;
 665
 666    // The HTTP server expects a hostname with a trailing :port
 667    char Buf[1000];
 668    string ProperHost = Uri.Host;
 669    if (Uri.Port != 0)
 670    {
 671       sprintf(Buf,":%u",Uri.Port);
 672       ProperHost += Buf;
 673    }
 674
 675    // Just in case.
 676    if (Itm->Uri.length() >= sizeof(Buf))
 677        abort();
 678
 679    /* Build the request. We include a keep-alive header only for non-proxy
 680       requests. This is to tweak old http/1.0 servers that do support keep-alive
 681       but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
 682       will glitch HTTP/1.0 proxies because they do not filter it out and
 683       pass it on, HTTP/1.1 says the connection should default to keep alive
 684       and we expect the proxy to do this */
 685    if (Proxy.empty() == true || Proxy.Host.empty())
 686       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
 687               QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
 688    else
 689    {
 690       /* Generate a cache control header if necessary. We place a max
 691          cache age on index files, optionally set a no-cache directive
 692          and a no-store directive for archives. */
 693       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 694               Itm->Uri.c_str(),ProperHost.c_str());
 695    }
 696    // generate a cache control header (if needed)
 697    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 698    {
 699       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 700    }
 701    else
 702    {
 703       if (Itm->IndexFile == true)
 704       {
 705          sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 706                  _config->FindI("Acquire::http::Max-Age",0));
 707       }
 708       else
 709       {
 710          if (_config->FindB("Acquire::http::No-Store",false) == true)
 711             strcat(Buf,"Cache-Control: no-store\r\n");
 712       }
 713    }
 714
 715    // If we ask for uncompressed files servers might respond with content-
 716    // negotation which lets us end up with compressed files we do not support,
 717    // see 657029, 657560 and co, so if we have no extension on the request
 718    // ask for text only. As a sidenote: If there is nothing to negotate servers
 719    // seem to be nice and ignore it.
 720    if (_config->FindB("Acquire::http::SendAccept", true) == true)
 721    {
 722       size_t const filepos = Itm->Uri.find_last_of('/');
 723       string const file = Itm->Uri.substr(filepos + 1);
 724       if (flExtension(file) == file)
 725          strcat(Buf,"Accept: text/*\r\n");
 726    }
 727
 728    string Req = Buf;
 729
 730    // Check for a partial file
 731    struct stat SBuf;
 732    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 733    {
 734       // In this case we send an if-range query with a range header
 735       sprintf(Buf,"Range: bytes=%lli-\r\nIf-Range: %s\r\n",(long long)SBuf.st_size - 1,
 736               TimeRFC1123(SBuf.st_mtime).c_str());
 737       Req += Buf;
 738    }
 739    else
 740    {
 741       if (Itm->LastModified != 0)
 742       {
 743          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 744          Req += Buf;
 745       }
 746    }
 747
 748    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 749       Req += string("Proxy-Authorization: Basic ") +
 750           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 751
 752    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 753    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 754    {
 755       Req += string("Authorization: Basic ") +
 756           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 757    }
 758    Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
 759                 "Debian APT-HTTP/1.3 ("VERSION")") + "\r\n\r\n";
 760
 761    if (Debug == true)
 762       cerr << Req << endl;
 763
 764    Out.Read(Req);
 765 }
 766                                                                         /*}}}*/
 767 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 768 // ---------------------------------------------------------------------
 769 /* This runs the select loop over the server FDs, Output file FDs and
 770    stdin. */
 771 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 772 {
 773    // Server has closed the connection
 774    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 775                                ToFile == false))
 776       return false;
 777
 778    fd_set rfds,wfds;
 779    FD_ZERO(&rfds);
 780    FD_ZERO(&wfds);
 781
 782    /* Add the server. We only send more requests if the connection will
 783       be persisting */
 784    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 785        && Srv->Persistent == true)
 786       FD_SET(Srv->ServerFd,&wfds);
 787    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 788       FD_SET(Srv->ServerFd,&rfds);
 789
 790    // Add the file
 791    int FileFD = -1;
 792    if (File != 0)
 793       FileFD = File->Fd();
 794
 795    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 796       FD_SET(FileFD,&wfds);
 797
 798    // Add stdin
 799    if (_config->FindB("Acquire::http::DependOnSTDIN", true) == true)
 800       FD_SET(STDIN_FILENO,&rfds);
 801
 802    // Figure out the max fd
 803    int MaxFd = FileFD;
 804    if (MaxFd < Srv->ServerFd)
 805       MaxFd = Srv->ServerFd;
 806
 807    // Select
 808    struct timeval tv;
 809    tv.tv_sec = TimeOut;
 810    tv.tv_usec = 0;
 811    int Res = 0;
 812    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 813    {
 814       if (errno == EINTR)
 815          return true;
 816       return _error->Errno("select",_("Select failed"));
 817    }
 818
 819    if (Res == 0)
 820    {
 821       _error->Error(_("Connection timed out"));
 822       return ServerDie(Srv);
 823    }
 824
 825    // Handle server IO
 826    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 827    {
 828       errno = 0;
 829       if (Srv->In.Read(Srv->ServerFd) == false)
 830          return ServerDie(Srv);
 831    }
 832
 833    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 834    {
 835       errno = 0;
 836       if (Srv->Out.Write(Srv->ServerFd) == false)
 837          return ServerDie(Srv);
 838    }
 839
 840    // Send data to the file
 841    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 842    {
 843       if (Srv->In.Write(FileFD) == false)
 844          return _error->Errno("write",_("Error writing to output file"));
 845    }
 846
 847    // Handle commands from APT
 848    if (FD_ISSET(STDIN_FILENO,&rfds))
 849    {
 850       if (Run(true) != -1)
 851          exit(100);
 852    }
 853
 854    return true;
 855 }
 856                                                                         /*}}}*/
 857 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 858 // ---------------------------------------------------------------------
 859 /* This takes the current input buffer from the Server FD and writes it
 860    into the file */
 861 bool HttpMethod::Flush(ServerState *Srv)
 862 {
 863    if (File != 0)
 864    {
 865       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 866       // can't be set
 867       if (File->Name() != "/dev/null")
 868          SetNonBlock(File->Fd(),false);
 869       if (Srv->In.WriteSpace() == false)
 870          return true;
 871
 872       while (Srv->In.WriteSpace() == true)
 873       {
 874          if (Srv->In.Write(File->Fd()) == false)
 875             return _error->Errno("write",_("Error writing to file"));
 876          if (Srv->In.IsLimit() == true)
 877             return true;
 878       }
 879
 880       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 881          return true;
 882    }
 883    return false;
 884 }
 885                                                                         /*}}}*/
 886 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 887 // ---------------------------------------------------------------------
 888 /* */
 889 bool HttpMethod::ServerDie(ServerState *Srv)
 890 {
 891    unsigned int LErrno = errno;
 892
 893    // Dump the buffer to the file
 894    if (Srv->State == ServerState::Data)
 895    {
 896       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 897       // can't be set
 898       if (File->Name() != "/dev/null")
 899          SetNonBlock(File->Fd(),false);
 900       while (Srv->In.WriteSpace() == true)
 901       {
 902          if (Srv->In.Write(File->Fd()) == false)
 903             return _error->Errno("write",_("Error writing to the file"));
 904
 905          // Done
 906          if (Srv->In.IsLimit() == true)
 907             return true;
 908       }
 909    }
 910
 911    // See if this is because the server finished the data stream
 912    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 913        Srv->Encoding != ServerState::Closes)
 914    {
 915       Srv->Close();
 916       if (LErrno == 0)
 917          return _error->Error(_("Error reading from server. Remote end closed connection"));
 918       errno = LErrno;
 919       return _error->Errno("read",_("Error reading from server"));
 920    }
 921    else
 922    {
 923       Srv->In.Limit(-1);
 924
 925       // Nothing left in the buffer
 926       if (Srv->In.WriteSpace() == false)
 927          return false;
 928
 929       // We may have got multiple responses back in one packet..
 930       Srv->Close();
 931       return true;
 932    }
 933
 934    return false;
 935 }
 936                                                                         /*}}}*/
 937 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 938 // ---------------------------------------------------------------------
 939 /* We look at the header data we got back from the server and decide what
 940    to do. Returns DealWithHeadersResult (see http.h for details).
 941  */
 942 HttpMethod::DealWithHeadersResult
 943 HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 944 {
 945    // Not Modified
 946    if (Srv->Result == 304)
 947    {
 948       unlink(Queue->DestFile.c_str());
 949       Res.IMSHit = true;
 950       Res.LastModified = Queue->LastModified;
 951       return IMS_HIT;
 952    }
 953
 954    /* Redirect
 955     *
 956     * Note that it is only OK for us to treat all redirection the same
 957     * because we *always* use GET, not other HTTP methods.  There are
 958     * three redirection codes for which it is not appropriate that we
 959     * redirect.  Pass on those codes so the error handling kicks in.
 960     */
 961    if (AllowRedirect
 962        && (Srv->Result > 300 && Srv->Result < 400)
 963        && (Srv->Result != 300       // Multiple Choices
 964            && Srv->Result != 304    // Not Modified
 965            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 966    {
 967       if (Srv->Location.empty() == true);
 968       else if (Srv->Location[0] == '/' && Queue->Uri.empty() == false)
 969       {
 970          URI Uri = Queue->Uri;
 971          if (Uri.Host.empty() == false)
 972          {
 973             if (Uri.Port != 0)
 974                strprintf(NextURI, "http://%s:%u", Uri.Host.c_str(), Uri.Port);
 975             else
 976                NextURI = "http://" + Uri.Host;
 977          }
 978          else
 979             NextURI.clear();
 980          NextURI.append(DeQuoteString(Srv->Location));
 981          return TRY_AGAIN_OR_REDIRECT;
 982       }
 983       else
 984       {
 985          NextURI = DeQuoteString(Srv->Location);
 986          return TRY_AGAIN_OR_REDIRECT;
 987       }
 988       /* else pass through for error message */
 989    }
 990
 991    /* We have a reply we dont handle. This should indicate a perm server
 992       failure */
 993    if (Srv->Result < 200 || Srv->Result >= 300)
 994    {
 995       char err[255];
 996       snprintf(err,sizeof(err)-1,"HttpError%i",Srv->Result);
 997       SetFailReason(err);
 998       _error->Error("%u %s",Srv->Result,Srv->Code);
 999       if (Srv->HaveContent == true)
1000          return ERROR_WITH_CONTENT_PAGE;
1001       return ERROR_UNRECOVERABLE;
1002    }
1003
1004    // This is some sort of 2xx 'data follows' reply
1005    Res.LastModified = Srv->Date;
1006    Res.Size = Srv->Size;
1007
1008    // Open the file
1009    delete File;
1010    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
1011    if (_error->PendingError() == true)
1012       return ERROR_NOT_FROM_SERVER;
1013
1014    FailFile = Queue->DestFile;
1015    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
1016    FailFd = File->Fd();
1017    FailTime = Srv->Date;
1018
1019    delete Srv->In.Hash;
1020    Srv->In.Hash = new Hashes;
1021
1022    // Set the expected size and read file for the hashes
1023    if (Srv->StartPos >= 0)
1024    {
1025       Res.ResumePoint = Srv->StartPos;
1026       File->Truncate(Srv->StartPos);
1027
1028       if (Srv->In.Hash->AddFD(*File,Srv->StartPos) == false)
1029       {
1030          _error->Errno("read",_("Problem hashing file"));
1031          return ERROR_NOT_FROM_SERVER;
1032       }
1033    }
1034
1035    SetNonBlock(File->Fd(),true);
1036    return FILE_IS_OPEN;
1037 }
1038                                                                         /*}}}*/
1039 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1040 // ---------------------------------------------------------------------
1041 /* This closes and timestamps the open file. This is neccessary to get
1042    resume behavoir on user abort */
1043 void HttpMethod::SigTerm(int)
1044 {
1045    if (FailFd == -1)
1046       _exit(100);
1047    close(FailFd);
1048
1049    // Timestamp
1050    struct utimbuf UBuf;
1051    UBuf.actime = FailTime;
1052    UBuf.modtime = FailTime;
1053    utime(FailFile.c_str(),&UBuf);
1054
1055    _exit(100);
1056 }
1057                                                                         /*}}}*/
1058 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1059 // ---------------------------------------------------------------------
1060 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1061    depth. */
1062 bool HttpMethod::Fetch(FetchItem *)
1063 {
1064    if (Server == 0)
1065       return true;
1066
1067    // Queue the requests
1068    int Depth = -1;
1069    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1070         I = I->Next, Depth++)
1071    {
1072       // If pipelining is disabled, we only queue 1 request
1073       if (Server->Pipeline == false && Depth >= 0)
1074          break;
1075
1076       // Make sure we stick with the same server
1077       if (Server->Comp(I->Uri) == false)
1078          break;
1079       if (QueueBack == I)
1080       {
1081          QueueBack = I->Next;
1082          SendReq(I,Server->Out);
1083          continue;
1084       }
1085    }
1086
1087    return true;
1088 };
1089                                                                         /*}}}*/
1090 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1091 // ---------------------------------------------------------------------
1092 /* We stash the desired pipeline depth */
1093 bool HttpMethod::Configuration(string Message)
1094 {
1095    if (pkgAcqMethod::Configuration(Message) == false)
1096       return false;
1097
1098    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1099    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1100    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1101                                   PipelineDepth);
1102    Debug = _config->FindB("Debug::Acquire::http",false);
1103    AutoDetectProxyCmd = _config->Find("Acquire::http::ProxyAutoDetect");
1104
1105    // Get the proxy to use
1106    AutoDetectProxy();
1107
1108    return true;
1109 }
1110                                                                         /*}}}*/
1111 // HttpMethod::Loop - Main loop                                         /*{{{*/
1112 // ---------------------------------------------------------------------
1113 /* */
1114 int HttpMethod::Loop()
1115 {
1116    typedef vector<string> StringVector;
1117    typedef vector<string>::iterator StringVectorIterator;
1118    map<string, StringVector> Redirected;
1119
1120    signal(SIGTERM,SigTerm);
1121    signal(SIGINT,SigTerm);
1122
1123    Server = 0;
1124
1125    int FailCounter = 0;
1126    while (1)
1127    {
1128       // We have no commands, wait for some to arrive
1129       if (Queue == 0)
1130       {
1131          if (WaitFd(STDIN_FILENO) == false)
1132             return 0;
1133       }
1134
1135       /* Run messages, we can accept 0 (no message) if we didn't
1136          do a WaitFd above.. Otherwise the FD is closed. */
1137       int Result = Run(true);
1138       if (Result != -1 && (Result != 0 || Queue == 0))
1139       {
1140          if(FailReason.empty() == false ||
1141             _config->FindB("Acquire::http::DependOnSTDIN", true) == true)
1142             return 100;
1143          else
1144             return 0;
1145       }
1146
1147       if (Queue == 0)
1148          continue;
1149
1150       // Connect to the server
1151       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1152       {
1153          delete Server;
1154          Server = new ServerState(Queue->Uri,this);
1155       }
1156       /* If the server has explicitly said this is the last connection
1157          then we pre-emptively shut down the pipeline and tear down
1158          the connection. This will speed up HTTP/1.0 servers a tad
1159          since we don't have to wait for the close sequence to
1160          complete */
1161       if (Server->Persistent == false)
1162          Server->Close();
1163
1164       // Reset the pipeline
1165       if (Server->ServerFd == -1)
1166          QueueBack = Queue;
1167
1168       // Connnect to the host
1169       if (Server->Open() == false)
1170       {
1171          Fail(true);
1172          delete Server;
1173          Server = 0;
1174          continue;
1175       }
1176
1177       // Fill the pipeline.
1178       Fetch(0);
1179
1180       // Fetch the next URL header data from the server.
1181       switch (Server->RunHeaders())
1182       {
1183          case ServerState::RUN_HEADERS_OK:
1184          break;
1185
1186          // The header data is bad
1187          case ServerState::RUN_HEADERS_PARSE_ERROR:
1188          {
1189             _error->Error(_("Bad header data"));
1190             Fail(true);
1191             RotateDNS();
1192             continue;
1193          }
1194
1195          // The server closed a connection during the header get..
1196          default:
1197          case ServerState::RUN_HEADERS_IO_ERROR:
1198          {
1199             FailCounter++;
1200             _error->Discard();
1201             Server->Close();
1202             Server->Pipeline = false;
1203
1204             if (FailCounter >= 2)
1205             {
1206                Fail(_("Connection failed"),true);
1207                FailCounter = 0;
1208             }
1209
1210             RotateDNS();
1211             continue;
1212          }
1213       };
1214
1215       // Decide what to do.
1216       FetchResult Res;
1217       Res.Filename = Queue->DestFile;
1218       switch (DealWithHeaders(Res,Server))
1219       {
1220          // Ok, the file is Open
1221          case FILE_IS_OPEN:
1222          {
1223             URIStart(Res);
1224
1225             // Run the data
1226             bool Result =  Server->RunData();
1227
1228             /* If the server is sending back sizeless responses then fill in
1229                the size now */
1230             if (Res.Size == 0)
1231                Res.Size = File->Size();
1232
1233             // Close the file, destroy the FD object and timestamp it
1234             FailFd = -1;
1235             delete File;
1236             File = 0;
1237
1238             // Timestamp
1239             struct utimbuf UBuf;
1240             time(&UBuf.actime);
1241             UBuf.actime = Server->Date;
1242             UBuf.modtime = Server->Date;
1243             utime(Queue->DestFile.c_str(),&UBuf);
1244
1245             // Send status to APT
1246             if (Result == true)
1247             {
1248                Res.TakeHashes(*Server->In.Hash);
1249                URIDone(Res);
1250             }
1251             else
1252             {
1253                if (Server->ServerFd == -1)
1254                {
1255                   FailCounter++;
1256                   _error->Discard();
1257                   Server->Close();
1258
1259                   if (FailCounter >= 2)
1260                   {
1261                      Fail(_("Connection failed"),true);
1262                      FailCounter = 0;
1263                   }
1264
1265                   QueueBack = Queue;
1266                }
1267                else
1268                   Fail(true);
1269             }
1270             break;
1271          }
1272
1273          // IMS hit
1274          case IMS_HIT:
1275          {
1276             URIDone(Res);
1277             break;
1278          }
1279
1280          // Hard server error, not found or something
1281          case ERROR_UNRECOVERABLE:
1282          {
1283             Fail();
1284             break;
1285          }
1286
1287          // Hard internal error, kill the connection and fail
1288          case ERROR_NOT_FROM_SERVER:
1289          {
1290             delete File;
1291             File = 0;
1292
1293             Fail();
1294             RotateDNS();
1295             Server->Close();
1296             break;
1297          }
1298
1299          // We need to flush the data, the header is like a 404 w/ error text
1300          case ERROR_WITH_CONTENT_PAGE:
1301          {
1302             Fail();
1303
1304             // Send to content to dev/null
1305             File = new FileFd("/dev/null",FileFd::WriteExists);
1306             Server->RunData();
1307             delete File;
1308             File = 0;
1309             break;
1310          }
1311
1312          // Try again with a new URL
1313          case TRY_AGAIN_OR_REDIRECT:
1314          {
1315             // Clear rest of response if there is content
1316             if (Server->HaveContent)
1317             {
1318                File = new FileFd("/dev/null",FileFd::WriteExists);
1319                Server->RunData();
1320                delete File;
1321                File = 0;
1322             }
1323
1324             /* Detect redirect loops.  No more redirects are allowed
1325                after the same URI is seen twice in a queue item. */
1326             StringVector &R = Redirected[Queue->DestFile];
1327             bool StopRedirects = false;
1328             if (R.size() == 0)
1329                R.push_back(Queue->Uri);
1330             else if (R[0] == "STOP" || R.size() > 10)
1331                StopRedirects = true;
1332             else
1333             {
1334                for (StringVectorIterator I = R.begin(); I != R.end(); ++I)
1335                   if (Queue->Uri == *I)
1336                   {
1337                      R[0] = "STOP";
1338                      break;
1339                   }
1340
1341                R.push_back(Queue->Uri);
1342             }
1343
1344             if (StopRedirects == false)
1345                Redirect(NextURI);
1346             else
1347                Fail();
1348
1349             break;
1350          }
1351
1352          default:
1353          Fail(_("Internal error"));
1354          break;
1355       }
1356
1357       FailCounter = 0;
1358    }
1359
1360    return 0;
1361 }
1362                                                                         /*}}}*/
1363 // HttpMethod::AutoDetectProxy - auto detect proxy                      /*{{{*/
1364 // ---------------------------------------------------------------------
1365 /* */
1366 bool HttpMethod::AutoDetectProxy()
1367 {
1368    if (AutoDetectProxyCmd.empty())
1369       return true;
1370
1371    if (Debug)
1372       clog << "Using auto proxy detect command: " << AutoDetectProxyCmd << endl;
1373
1374    int Pipes[2] = {-1,-1};
1375    if (pipe(Pipes) != 0)
1376       return _error->Errno("pipe", "Failed to create Pipe");
1377
1378    pid_t Process = ExecFork();
1379    if (Process == 0)
1380    {
1381       close(Pipes[0]);
1382       dup2(Pipes[1],STDOUT_FILENO);
1383       SetCloseExec(STDOUT_FILENO,false);
1384
1385       const char *Args[2];
1386       Args[0] = AutoDetectProxyCmd.c_str();
1387       Args[1] = 0;
1388       execv(Args[0],(char **)Args);
1389       cerr << "Failed to exec method " << Args[0] << endl;
1390       _exit(100);
1391    }
1392    char buf[512];
1393    int InFd = Pipes[0];
1394    close(Pipes[1]);
1395    int res = read(InFd, buf, sizeof(buf));
1396    ExecWait(Process, "ProxyAutoDetect", true);
1397
1398    if (res < 0)
1399       return _error->Errno("read", "Failed to read");
1400    if (res == 0)
1401       return _error->Warning("ProxyAutoDetect returned no data");
1402
1403    // add trailing \0
1404    buf[res] = 0;
1405
1406    if (Debug)
1407       clog << "auto detect command returned: '" << buf << "'" << endl;
1408
1409    if (strstr(buf, "http://") == buf)
1410       _config->Set("Acquire::http::proxy", _strstrip(buf));
1411
1412    return true;
1413 }
1414                                                                         /*}}}*/
1415
1416