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 <climits>
  46 #include <iostream>
  47 #include <map>
  48
  49 // Internet stuff
  50 #include <netdb.h>
  51
  52 #include "config.h"
  53 #include "connect.h"
  54 #include "rfc2553emu.h"
  55 #include "http.h"
  56
  57 #include <apti18n.h>
  58                                                                         /*}}}*/
  59 using namespace std;
  60
  61 string HttpMethod::FailFile;
  62 int HttpMethod::FailFd = -1;
  63 time_t HttpMethod::FailTime = 0;
  64 unsigned long PipelineDepth = 0;
  65 unsigned long TimeOut = 120;
  66 bool AllowRedirect = false;
  67 bool Debug = false;
  68 URI Proxy;
  69
  70 unsigned long long CircleBuf::BwReadLimit=0;
  71 unsigned long long CircleBuf::BwTickReadData=0;
  72 struct timeval CircleBuf::BwReadTick={0,0};
  73 const unsigned int CircleBuf::BW_HZ=10;
  74
  75 // CircleBuf::CircleBuf - Circular input buffer                         /*{{{*/
  76 // ---------------------------------------------------------------------
  77 /* */
  78 CircleBuf::CircleBuf(unsigned long long Size) : Size(Size), Hash(0)
  79 {
  80    Buf = new unsigned char[Size];
  81    Reset();
  82
  83    CircleBuf::BwReadLimit = _config->FindI("Acquire::http::Dl-Limit",0)*1024;
  84 }
  85                                                                         /*}}}*/
  86 // CircleBuf::Reset - Reset to the default state                        /*{{{*/
  87 // ---------------------------------------------------------------------
  88 /* */
  89 void CircleBuf::Reset()
  90 {
  91    InP = 0;
  92    OutP = 0;
  93    StrPos = 0;
  94    MaxGet = (unsigned long long)-1;
  95    OutQueue = string();
  96    if (Hash != 0)
  97    {
  98       delete Hash;
  99       Hash = new Hashes;
 100    }
 101 };
 102                                                                         /*}}}*/
 103 // CircleBuf::Read - Read from a FD into the circular buffer            /*{{{*/
 104 // ---------------------------------------------------------------------
 105 /* This fills up the buffer with as much data as is in the FD, assuming it
 106    is non-blocking.. */
 107 bool CircleBuf::Read(int Fd)
 108 {
 109    unsigned long long BwReadMax;
 110
 111    while (1)
 112    {
 113       // Woops, buffer is full
 114       if (InP - OutP == Size)
 115          return true;
 116
 117       // what's left to read in this tick
 118       BwReadMax = CircleBuf::BwReadLimit/BW_HZ;
 119
 120       if(CircleBuf::BwReadLimit) {
 121          struct timeval now;
 122          gettimeofday(&now,0);
 123
 124          unsigned long long d = (now.tv_sec-CircleBuf::BwReadTick.tv_sec)*1000000 +
 125             now.tv_usec-CircleBuf::BwReadTick.tv_usec;
 126          if(d > 1000000/BW_HZ) {
 127             CircleBuf::BwReadTick = now;
 128             CircleBuf::BwTickReadData = 0;
 129          }
 130
 131          if(CircleBuf::BwTickReadData >= BwReadMax) {
 132             usleep(1000000/BW_HZ);
 133             return true;
 134          }
 135       }
 136
 137       // Write the buffer segment
 138       ssize_t Res;
 139       if(CircleBuf::BwReadLimit) {
 140          Res = read(Fd,Buf + (InP%Size),
 141                     BwReadMax > LeftRead() ? LeftRead() : BwReadMax);
 142       } else
 143          Res = read(Fd,Buf + (InP%Size),LeftRead());
 144
 145       if(Res > 0 && BwReadLimit > 0)
 146          CircleBuf::BwTickReadData += Res;
 147
 148       if (Res == 0)
 149          return false;
 150       if (Res < 0)
 151       {
 152          if (errno == EAGAIN)
 153             return true;
 154          return false;
 155       }
 156
 157       if (InP == 0)
 158          gettimeofday(&Start,0);
 159       InP += Res;
 160    }
 161 }
 162                                                                         /*}}}*/
 163 // CircleBuf::Read - Put the string into the buffer                     /*{{{*/
 164 // ---------------------------------------------------------------------
 165 /* This will hold the string in and fill the buffer with it as it empties */
 166 bool CircleBuf::Read(string Data)
 167 {
 168    OutQueue += Data;
 169    FillOut();
 170    return true;
 171 }
 172                                                                         /*}}}*/
 173 // CircleBuf::FillOut - Fill the buffer from the output queue           /*{{{*/
 174 // ---------------------------------------------------------------------
 175 /* */
 176 void CircleBuf::FillOut()
 177 {
 178    if (OutQueue.empty() == true)
 179       return;
 180    while (1)
 181    {
 182       // Woops, buffer is full
 183       if (InP - OutP == Size)
 184          return;
 185
 186       // Write the buffer segment
 187       unsigned long long Sz = LeftRead();
 188       if (OutQueue.length() - StrPos < Sz)
 189          Sz = OutQueue.length() - StrPos;
 190       memcpy(Buf + (InP%Size),OutQueue.c_str() + StrPos,Sz);
 191
 192       // Advance
 193       StrPos += Sz;
 194       InP += Sz;
 195       if (OutQueue.length() == StrPos)
 196       {
 197          StrPos = 0;
 198          OutQueue = "";
 199          return;
 200       }
 201    }
 202 }
 203                                                                         /*}}}*/
 204 // CircleBuf::Write - Write from the buffer into a FD                   /*{{{*/
 205 // ---------------------------------------------------------------------
 206 /* This empties the buffer into the FD. */
 207 bool CircleBuf::Write(int Fd)
 208 {
 209    while (1)
 210    {
 211       FillOut();
 212
 213       // Woops, buffer is empty
 214       if (OutP == InP)
 215          return true;
 216
 217       if (OutP == MaxGet)
 218          return true;
 219
 220       // Write the buffer segment
 221       ssize_t Res;
 222       Res = write(Fd,Buf + (OutP%Size),LeftWrite());
 223
 224       if (Res == 0)
 225          return false;
 226       if (Res < 0)
 227       {
 228          if (errno == EAGAIN)
 229             return true;
 230
 231          return false;
 232       }
 233
 234       if (Hash != 0)
 235          Hash->Add(Buf + (OutP%Size),Res);
 236
 237       OutP += Res;
 238    }
 239 }
 240                                                                         /*}}}*/
 241 // CircleBuf::WriteTillEl - Write from the buffer to a string           /*{{{*/
 242 // ---------------------------------------------------------------------
 243 /* This copies till the first empty line */
 244 bool CircleBuf::WriteTillEl(string &Data,bool Single)
 245 {
 246    // We cheat and assume it is unneeded to have more than one buffer load
 247    for (unsigned long long I = OutP; I < InP; I++)
 248    {
 249       if (Buf[I%Size] != '\n')
 250          continue;
 251       ++I;
 252
 253       if (Single == false)
 254       {
 255          if (I < InP  && Buf[I%Size] == '\r')
 256             ++I;
 257          if (I >= InP || Buf[I%Size] != '\n')
 258             continue;
 259          ++I;
 260       }
 261
 262       Data = "";
 263       while (OutP < I)
 264       {
 265          unsigned long long Sz = LeftWrite();
 266          if (Sz == 0)
 267             return false;
 268          if (I - OutP < Sz)
 269             Sz = I - OutP;
 270          Data += string((char *)(Buf + (OutP%Size)),Sz);
 271          OutP += Sz;
 272       }
 273       return true;
 274    }
 275    return false;
 276 }
 277                                                                         /*}}}*/
 278 // CircleBuf::Stats - Print out stats information                       /*{{{*/
 279 // ---------------------------------------------------------------------
 280 /* */
 281 void CircleBuf::Stats()
 282 {
 283    if (InP == 0)
 284       return;
 285
 286    struct timeval Stop;
 287    gettimeofday(&Stop,0);
 288 /*   float Diff = Stop.tv_sec - Start.tv_sec +
 289              (float)(Stop.tv_usec - Start.tv_usec)/1000000;
 290    clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
 291 }
 292                                                                         /*}}}*/
 293 CircleBuf::~CircleBuf()
 294 {
 295    delete [] Buf;
 296    delete Hash;
 297 }
 298
 299 // ServerState::ServerState - Constructor                               /*{{{*/
 300 // ---------------------------------------------------------------------
 301 /* */
 302 ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner),
 303                         In(64*1024), Out(4*1024),
 304                         ServerName(Srv)
 305 {
 306    Reset();
 307 }
 308                                                                         /*}}}*/
 309 // ServerState::Open - Open a connection to the server                  /*{{{*/
 310 // ---------------------------------------------------------------------
 311 /* This opens a connection to the server. */
 312 bool ServerState::Open()
 313 {
 314    // Use the already open connection if possible.
 315    if (ServerFd != -1)
 316       return true;
 317
 318    Close();
 319    In.Reset();
 320    Out.Reset();
 321    Persistent = true;
 322
 323    // Determine the proxy setting
 324    string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
 325    if (!SpecificProxy.empty())
 326    {
 327            if (SpecificProxy == "DIRECT")
 328                    Proxy = "";
 329            else
 330                    Proxy = SpecificProxy;
 331    }
 332    else
 333    {
 334            string DefProxy = _config->Find("Acquire::http::Proxy");
 335            if (!DefProxy.empty())
 336            {
 337                    Proxy = DefProxy;
 338            }
 339            else
 340            {
 341                    char* result = getenv("http_proxy");
 342                    Proxy = result ? result : "";
 343            }
 344    }
 345
 346    // Parse no_proxy, a , separated list of domains
 347    if (getenv("no_proxy") != 0)
 348    {
 349       if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 350          Proxy = "";
 351    }
 352
 353    // Determine what host and port to use based on the proxy settings
 354    int Port = 0;
 355    string Host;
 356    if (Proxy.empty() == true || Proxy.Host.empty() == true)
 357    {
 358       if (ServerName.Port != 0)
 359          Port = ServerName.Port;
 360       Host = ServerName.Host;
 361    }
 362    else
 363    {
 364       if (Proxy.Port != 0)
 365          Port = Proxy.Port;
 366       Host = Proxy.Host;
 367    }
 368
 369    // Connect to the remote server
 370    if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
 371       return false;
 372
 373    return true;
 374 }
 375                                                                         /*}}}*/
 376 // ServerState::Close - Close a connection to the server                /*{{{*/
 377 // ---------------------------------------------------------------------
 378 /* */
 379 bool ServerState::Close()
 380 {
 381    close(ServerFd);
 382    ServerFd = -1;
 383    return true;
 384 }
 385                                                                         /*}}}*/
 386 // ServerState::RunHeaders - Get the headers before the data            /*{{{*/
 387 // ---------------------------------------------------------------------
 388 /* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
 389    parse error occurred */
 390 ServerState::RunHeadersResult ServerState::RunHeaders()
 391 {
 392    State = Header;
 393
 394    Owner->Status(_("Waiting for headers"));
 395
 396    Major = 0;
 397    Minor = 0;
 398    Result = 0;
 399    Size = 0;
 400    StartPos = 0;
 401    Encoding = Closes;
 402    HaveContent = false;
 403    time(&Date);
 404
 405    do
 406    {
 407       string Data;
 408       if (In.WriteTillEl(Data) == false)
 409          continue;
 410
 411       if (Debug == true)
 412          clog << Data;
 413
 414       for (string::const_iterator I = Data.begin(); I < Data.end(); ++I)
 415       {
 416          string::const_iterator J = I;
 417          for (; J != Data.end() && *J != '\n' && *J != '\r'; ++J);
 418          if (HeaderLine(string(I,J)) == false)
 419             return RUN_HEADERS_PARSE_ERROR;
 420          I = J;
 421       }
 422
 423       // 100 Continue is a Nop...
 424       if (Result == 100)
 425          continue;
 426
 427       // Tidy up the connection persistance state.
 428       if (Encoding == Closes && HaveContent == true)
 429          Persistent = false;
 430
 431       return RUN_HEADERS_OK;
 432    }
 433    while (Owner->Go(false,this) == true);
 434
 435    return RUN_HEADERS_IO_ERROR;
 436 }
 437                                                                         /*}}}*/
 438 // ServerState::RunData - Transfer the data from the socket             /*{{{*/
 439 // ---------------------------------------------------------------------
 440 /* */
 441 bool ServerState::RunData()
 442 {
 443    State = Data;
 444
 445    // Chunked transfer encoding is fun..
 446    if (Encoding == Chunked)
 447    {
 448       while (1)
 449       {
 450          // Grab the block size
 451          bool Last = true;
 452          string Data;
 453          In.Limit(-1);
 454          do
 455          {
 456             if (In.WriteTillEl(Data,true) == true)
 457                break;
 458          }
 459          while ((Last = Owner->Go(false,this)) == true);
 460
 461          if (Last == false)
 462             return false;
 463
 464          // See if we are done
 465          unsigned long long Len = strtoull(Data.c_str(),0,16);
 466          if (Len == 0)
 467          {
 468             In.Limit(-1);
 469
 470             // We have to remove the entity trailer
 471             Last = true;
 472             do
 473             {
 474                if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
 475                   break;
 476             }
 477             while ((Last = Owner->Go(false,this)) == true);
 478             if (Last == false)
 479                return false;
 480             return !_error->PendingError();
 481          }
 482
 483          // Transfer the block
 484          In.Limit(Len);
 485          while (Owner->Go(true,this) == true)
 486             if (In.IsLimit() == true)
 487                break;
 488
 489          // Error
 490          if (In.IsLimit() == false)
 491             return false;
 492
 493          // The server sends an extra new line before the next block specifier..
 494          In.Limit(-1);
 495          Last = true;
 496          do
 497          {
 498             if (In.WriteTillEl(Data,true) == true)
 499                break;
 500          }
 501          while ((Last = Owner->Go(false,this)) == true);
 502          if (Last == false)
 503             return false;
 504       }
 505    }
 506    else
 507    {
 508       /* Closes encoding is used when the server did not specify a size, the
 509          loss of the connection means we are done */
 510       if (Encoding == Closes)
 511          In.Limit(-1);
 512       else
 513          In.Limit(Size - StartPos);
 514
 515       // Just transfer the whole block.
 516       do
 517       {
 518          if (In.IsLimit() == false)
 519             continue;
 520
 521          In.Limit(-1);
 522          return !_error->PendingError();
 523       }
 524       while (Owner->Go(true,this) == true);
 525    }
 526
 527    return Owner->Flush(this) && !_error->PendingError();
 528 }
 529                                                                         /*}}}*/
 530 // ServerState::HeaderLine - Process a header line                      /*{{{*/
 531 // ---------------------------------------------------------------------
 532 /* */
 533 bool ServerState::HeaderLine(string Line)
 534 {
 535    if (Line.empty() == true)
 536       return true;
 537
 538    string::size_type Pos = Line.find(' ');
 539    if (Pos == string::npos || Pos+1 > Line.length())
 540    {
 541       // Blah, some servers use "connection:closes", evil.
 542       Pos = Line.find(':');
 543       if (Pos == string::npos || Pos + 2 > Line.length())
 544          return _error->Error(_("Bad header line"));
 545       Pos++;
 546    }
 547
 548    // Parse off any trailing spaces between the : and the next word.
 549    string::size_type Pos2 = Pos;
 550    while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
 551       Pos2++;
 552
 553    string Tag = string(Line,0,Pos);
 554    string Val = string(Line,Pos2);
 555
 556    if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
 557    {
 558       // Evil servers return no version
 559       if (Line[4] == '/')
 560       {
 561          int const elements = sscanf(Line.c_str(),"HTTP/%3u.%3u %3u%359[^\n]",&Major,&Minor,&Result,Code);
 562          if (elements == 3)
 563          {
 564             Code[0] = '\0';
 565             if (Debug == true)
 566                clog << "HTTP server doesn't give Reason-Phrase for " << Result << std::endl;
 567          }
 568          else if (elements != 4)
 569             return _error->Error(_("The HTTP server sent an invalid reply header"));
 570       }
 571       else
 572       {
 573          Major = 0;
 574          Minor = 9;
 575          if (sscanf(Line.c_str(),"HTTP %3u%359[^\n]",&Result,Code) != 2)
 576             return _error->Error(_("The HTTP server sent an invalid reply header"));
 577       }
 578
 579       /* Check the HTTP response header to get the default persistance
 580          state. */
 581       if (Major < 1)
 582          Persistent = false;
 583       else
 584       {
 585          if (Major == 1 && Minor == 0)
 586             Persistent = false;
 587          else
 588             Persistent = true;
 589       }
 590
 591       return true;
 592    }
 593
 594    if (stringcasecmp(Tag,"Content-Length:") == 0)
 595    {
 596       if (Encoding == Closes)
 597          Encoding = Stream;
 598       HaveContent = true;
 599
 600       // The length is already set from the Content-Range header
 601       if (StartPos != 0)
 602          return true;
 603
 604       Size = strtoull(Val.c_str(), NULL, 10);
 605       if (Size >= std::numeric_limits<unsigned long long>::max())
 606          return _error->Errno("HeaderLine", _("The HTTP server sent an invalid Content-Length header"));
 607       return true;
 608    }
 609
 610    if (stringcasecmp(Tag,"Content-Type:") == 0)
 611    {
 612       HaveContent = true;
 613       return true;
 614    }
 615
 616    if (stringcasecmp(Tag,"Content-Range:") == 0)
 617    {
 618       HaveContent = true;
 619
 620       if (sscanf(Val.c_str(),"bytes %llu-%*u/%llu",&StartPos,&Size) != 2)
 621          return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
 622       if ((unsigned long long)StartPos > Size)
 623          return _error->Error(_("This HTTP server has broken range support"));
 624       return true;
 625    }
 626
 627    if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
 628    {
 629       HaveContent = true;
 630       if (stringcasecmp(Val,"chunked") == 0)
 631          Encoding = Chunked;
 632       return true;
 633    }
 634
 635    if (stringcasecmp(Tag,"Connection:") == 0)
 636    {
 637       if (stringcasecmp(Val,"close") == 0)
 638          Persistent = false;
 639       if (stringcasecmp(Val,"keep-alive") == 0)
 640          Persistent = true;
 641       return true;
 642    }
 643
 644    if (stringcasecmp(Tag,"Last-Modified:") == 0)
 645    {
 646       if (RFC1123StrToTime(Val.c_str(), Date) == false)
 647          return _error->Error(_("Unknown date format"));
 648       return true;
 649    }
 650
 651    if (stringcasecmp(Tag,"Location:") == 0)
 652    {
 653       Location = Val;
 654       return true;
 655    }
 656
 657    return true;
 658 }
 659                                                                         /*}}}*/
 660
 661 // HttpMethod::SendReq - Send the HTTP request                          /*{{{*/
 662 // ---------------------------------------------------------------------
 663 /* This places the http request in the outbound buffer */
 664 void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
 665 {
 666    URI Uri = Itm->Uri;
 667
 668    // The HTTP server expects a hostname with a trailing :port
 669    char Buf[1000];
 670    string ProperHost;
 671
 672    if (Uri.Host.find(':') != string::npos)
 673       ProperHost = '[' + Uri.Host + ']';
 674    else
 675       ProperHost = Uri.Host;
 676    if (Uri.Port != 0)
 677    {
 678       sprintf(Buf,":%u",Uri.Port);
 679       ProperHost += Buf;
 680    }
 681
 682    // Just in case.
 683    if (Itm->Uri.length() >= sizeof(Buf))
 684        abort();
 685
 686    /* RFC 2616 §5.1.2 requires absolute URIs for requests to proxies,
 687       but while its a must for all servers to accept absolute URIs,
 688       it is assumed clients will sent an absolute path for non-proxies */
 689    std::string requesturi;
 690    if (Proxy.empty() == true || Proxy.Host.empty())
 691       requesturi = Uri.Path;
 692    else
 693       requesturi = Itm->Uri;
 694
 695    // The "+" is encoded as a workaround for a amazon S3 bug
 696    // see LP bugs #1003633 and #1086997.
 697    requesturi = QuoteString(requesturi, "+~ ");
 698
 699    /* Build the request. No keep-alive is included as it is the default
 700       in 1.1, can cause problems with proxies, and we are an HTTP/1.1
 701       client anyway.
 702       C.f. https://tools.ietf.org/wg/httpbis/trac/ticket/158 */
 703    sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 704            requesturi.c_str(),ProperHost.c_str());
 705
 706    // generate a cache control header (if needed)
 707    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 708    {
 709       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 710    }
 711    else
 712    {
 713       if (Itm->IndexFile == true)
 714       {
 715          sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 716                  _config->FindI("Acquire::http::Max-Age",0));
 717       }
 718       else
 719       {
 720          if (_config->FindB("Acquire::http::No-Store",false) == true)
 721             strcat(Buf,"Cache-Control: no-store\r\n");
 722       }
 723    }
 724
 725    // If we ask for uncompressed files servers might respond with content-
 726    // negotation which lets us end up with compressed files we do not support,
 727    // see 657029, 657560 and co, so if we have no extension on the request
 728    // ask for text only. As a sidenote: If there is nothing to negotate servers
 729    // seem to be nice and ignore it.
 730    if (_config->FindB("Acquire::http::SendAccept", true) == true)
 731    {
 732       size_t const filepos = Itm->Uri.find_last_of('/');
 733       string const file = Itm->Uri.substr(filepos + 1);
 734       if (flExtension(file) == file)
 735          strcat(Buf,"Accept: text/*\r\n");
 736    }
 737
 738    string Req = Buf;
 739
 740    // Check for a partial file
 741    struct stat SBuf;
 742    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 743    {
 744       // In this case we send an if-range query with a range header
 745       sprintf(Buf,"Range: bytes=%lli-\r\nIf-Range: %s\r\n",(long long)SBuf.st_size - 1,
 746               TimeRFC1123(SBuf.st_mtime).c_str());
 747       Req += Buf;
 748    }
 749    else
 750    {
 751       if (Itm->LastModified != 0)
 752       {
 753          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 754          Req += Buf;
 755       }
 756    }
 757
 758    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 759       Req += string("Proxy-Authorization: Basic ") +
 760           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 761
 762    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 763    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 764    {
 765       Req += string("Authorization: Basic ") +
 766           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 767    }
 768    Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
 769                 "Debian APT-HTTP/1.3 (" PACKAGE_VERSION ")") + "\r\n\r\n";
 770
 771    if (Debug == true)
 772       cerr << Req << endl;
 773
 774    Out.Read(Req);
 775 }
 776                                                                         /*}}}*/
 777 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 778 // ---------------------------------------------------------------------
 779 /* This runs the select loop over the server FDs, Output file FDs and
 780    stdin. */
 781 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 782 {
 783    // Server has closed the connection
 784    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 785                                ToFile == false))
 786       return false;
 787
 788    fd_set rfds,wfds;
 789    FD_ZERO(&rfds);
 790    FD_ZERO(&wfds);
 791
 792    /* Add the server. We only send more requests if the connection will
 793       be persisting */
 794    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 795        && Srv->Persistent == true)
 796       FD_SET(Srv->ServerFd,&wfds);
 797    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 798       FD_SET(Srv->ServerFd,&rfds);
 799
 800    // Add the file
 801    int FileFD = -1;
 802    if (File != 0)
 803       FileFD = File->Fd();
 804
 805    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 806       FD_SET(FileFD,&wfds);
 807
 808    // Add stdin
 809    if (_config->FindB("Acquire::http::DependOnSTDIN", true) == true)
 810       FD_SET(STDIN_FILENO,&rfds);
 811
 812    // Figure out the max fd
 813    int MaxFd = FileFD;
 814    if (MaxFd < Srv->ServerFd)
 815       MaxFd = Srv->ServerFd;
 816
 817    // Select
 818    struct timeval tv;
 819    tv.tv_sec = TimeOut;
 820    tv.tv_usec = 0;
 821    int Res = 0;
 822    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 823    {
 824       if (errno == EINTR)
 825          return true;
 826       return _error->Errno("select",_("Select failed"));
 827    }
 828
 829    if (Res == 0)
 830    {
 831       _error->Error(_("Connection timed out"));
 832       return ServerDie(Srv);
 833    }
 834
 835    // Handle server IO
 836    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 837    {
 838       errno = 0;
 839       if (Srv->In.Read(Srv->ServerFd) == false)
 840          return ServerDie(Srv);
 841    }
 842
 843    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 844    {
 845       errno = 0;
 846       if (Srv->Out.Write(Srv->ServerFd) == false)
 847          return ServerDie(Srv);
 848    }
 849
 850    // Send data to the file
 851    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 852    {
 853       if (Srv->In.Write(FileFD) == false)
 854          return _error->Errno("write",_("Error writing to output file"));
 855    }
 856
 857    // Handle commands from APT
 858    if (FD_ISSET(STDIN_FILENO,&rfds))
 859    {
 860       if (Run(true) != -1)
 861          exit(100);
 862    }
 863
 864    return true;
 865 }
 866                                                                         /*}}}*/
 867 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 868 // ---------------------------------------------------------------------
 869 /* This takes the current input buffer from the Server FD and writes it
 870    into the file */
 871 bool HttpMethod::Flush(ServerState *Srv)
 872 {
 873    if (File != 0)
 874    {
 875       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 876       // can't be set
 877       if (File->Name() != "/dev/null")
 878          SetNonBlock(File->Fd(),false);
 879       if (Srv->In.WriteSpace() == false)
 880          return true;
 881
 882       while (Srv->In.WriteSpace() == true)
 883       {
 884          if (Srv->In.Write(File->Fd()) == false)
 885             return _error->Errno("write",_("Error writing to file"));
 886          if (Srv->In.IsLimit() == true)
 887             return true;
 888       }
 889
 890       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 891          return true;
 892    }
 893    return false;
 894 }
 895                                                                         /*}}}*/
 896 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 897 // ---------------------------------------------------------------------
 898 /* */
 899 bool HttpMethod::ServerDie(ServerState *Srv)
 900 {
 901    unsigned int LErrno = errno;
 902
 903    // Dump the buffer to the file
 904    if (Srv->State == ServerState::Data)
 905    {
 906       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 907       // can't be set
 908       if (File->Name() != "/dev/null")
 909          SetNonBlock(File->Fd(),false);
 910       while (Srv->In.WriteSpace() == true)
 911       {
 912          if (Srv->In.Write(File->Fd()) == false)
 913             return _error->Errno("write",_("Error writing to the file"));
 914
 915          // Done
 916          if (Srv->In.IsLimit() == true)
 917             return true;
 918       }
 919    }
 920
 921    // See if this is because the server finished the data stream
 922    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 923        Srv->Encoding != ServerState::Closes)
 924    {
 925       Srv->Close();
 926       if (LErrno == 0)
 927          return _error->Error(_("Error reading from server. Remote end closed connection"));
 928       errno = LErrno;
 929       return _error->Errno("read",_("Error reading from server"));
 930    }
 931    else
 932    {
 933       Srv->In.Limit(-1);
 934
 935       // Nothing left in the buffer
 936       if (Srv->In.WriteSpace() == false)
 937          return false;
 938
 939       // We may have got multiple responses back in one packet..
 940       Srv->Close();
 941       return true;
 942    }
 943
 944    return false;
 945 }
 946                                                                         /*}}}*/
 947 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 948 // ---------------------------------------------------------------------
 949 /* We look at the header data we got back from the server and decide what
 950    to do. Returns DealWithHeadersResult (see http.h for details).
 951  */
 952 HttpMethod::DealWithHeadersResult
 953 HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 954 {
 955    // Not Modified
 956    if (Srv->Result == 304)
 957    {
 958       unlink(Queue->DestFile.c_str());
 959       Res.IMSHit = true;
 960       Res.LastModified = Queue->LastModified;
 961       return IMS_HIT;
 962    }
 963
 964    /* Redirect
 965     *
 966     * Note that it is only OK for us to treat all redirection the same
 967     * because we *always* use GET, not other HTTP methods.  There are
 968     * three redirection codes for which it is not appropriate that we
 969     * redirect.  Pass on those codes so the error handling kicks in.
 970     */
 971    if (AllowRedirect
 972        && (Srv->Result > 300 && Srv->Result < 400)
 973        && (Srv->Result != 300       // Multiple Choices
 974            && Srv->Result != 304    // Not Modified
 975            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 976    {
 977       if (Srv->Location.empty() == true);
 978       else if (Srv->Location[0] == '/' && Queue->Uri.empty() == false)
 979       {
 980          URI Uri = Queue->Uri;
 981          if (Uri.Host.empty() == false)
 982             NextURI = URI::SiteOnly(Uri);
 983          else
 984             NextURI.clear();
 985          NextURI.append(DeQuoteString(Srv->Location));
 986          return TRY_AGAIN_OR_REDIRECT;
 987       }
 988       else
 989       {
 990          NextURI = DeQuoteString(Srv->Location);
 991          URI tmpURI = NextURI;
 992          // Do not allow a redirection to switch protocol
 993          if (tmpURI.Access == "http")
 994             return TRY_AGAIN_OR_REDIRECT;
 995       }
 996       /* else pass through for error message */
 997    }
 998
 999    /* We have a reply we dont handle. This should indicate a perm server
1000       failure */
1001    if (Srv->Result < 200 || Srv->Result >= 300)
1002    {
1003       char err[255];
1004       snprintf(err,sizeof(err)-1,"HttpError%i",Srv->Result);
1005       SetFailReason(err);
1006       _error->Error("%u %s",Srv->Result,Srv->Code);
1007       if (Srv->HaveContent == true)
1008          return ERROR_WITH_CONTENT_PAGE;
1009       return ERROR_UNRECOVERABLE;
1010    }
1011
1012    // This is some sort of 2xx 'data follows' reply
1013    Res.LastModified = Srv->Date;
1014    Res.Size = Srv->Size;
1015
1016    // Open the file
1017    delete File;
1018    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
1019    if (_error->PendingError() == true)
1020       return ERROR_NOT_FROM_SERVER;
1021
1022    FailFile = Queue->DestFile;
1023    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
1024    FailFd = File->Fd();
1025    FailTime = Srv->Date;
1026
1027    delete Srv->In.Hash;
1028    Srv->In.Hash = new Hashes;
1029
1030    // Set the expected size and read file for the hashes
1031    if (Srv->StartPos >= 0)
1032    {
1033       Res.ResumePoint = Srv->StartPos;
1034       File->Truncate(Srv->StartPos);
1035
1036       if (Srv->In.Hash->AddFD(*File,Srv->StartPos) == false)
1037       {
1038          _error->Errno("read",_("Problem hashing file"));
1039          return ERROR_NOT_FROM_SERVER;
1040       }
1041    }
1042
1043    SetNonBlock(File->Fd(),true);
1044    return FILE_IS_OPEN;
1045 }
1046                                                                         /*}}}*/
1047 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1048 // ---------------------------------------------------------------------
1049 /* This closes and timestamps the open file. This is neccessary to get
1050    resume behavoir on user abort */
1051 void HttpMethod::SigTerm(int)
1052 {
1053    if (FailFd == -1)
1054       _exit(100);
1055    close(FailFd);
1056
1057    // Timestamp
1058    struct utimbuf UBuf;
1059    UBuf.actime = FailTime;
1060    UBuf.modtime = FailTime;
1061    utime(FailFile.c_str(),&UBuf);
1062
1063    _exit(100);
1064 }
1065                                                                         /*}}}*/
1066 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1067 // ---------------------------------------------------------------------
1068 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1069    depth. */
1070 bool HttpMethod::Fetch(FetchItem *)
1071 {
1072    if (Server == 0)
1073       return true;
1074
1075    // Queue the requests
1076    int Depth = -1;
1077    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1078         I = I->Next, Depth++)
1079    {
1080       // If pipelining is disabled, we only queue 1 request
1081       if (Server->Pipeline == false && Depth >= 0)
1082          break;
1083
1084       // Make sure we stick with the same server
1085       if (Server->Comp(I->Uri) == false)
1086          break;
1087       if (QueueBack == I)
1088       {
1089          QueueBack = I->Next;
1090          SendReq(I,Server->Out);
1091          continue;
1092       }
1093    }
1094
1095    return true;
1096 };
1097                                                                         /*}}}*/
1098 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1099 // ---------------------------------------------------------------------
1100 /* We stash the desired pipeline depth */
1101 bool HttpMethod::Configuration(string Message)
1102 {
1103    if (pkgAcqMethod::Configuration(Message) == false)
1104       return false;
1105
1106    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1107    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1108    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1109                                   PipelineDepth);
1110    Debug = _config->FindB("Debug::Acquire::http",false);
1111    AutoDetectProxyCmd = _config->Find("Acquire::http::ProxyAutoDetect");
1112
1113    // Get the proxy to use
1114    AutoDetectProxy();
1115
1116    return true;
1117 }
1118                                                                         /*}}}*/
1119 // HttpMethod::Loop - Main loop                                         /*{{{*/
1120 // ---------------------------------------------------------------------
1121 /* */
1122 int HttpMethod::Loop()
1123 {
1124    typedef vector<string> StringVector;
1125    typedef vector<string>::iterator StringVectorIterator;
1126    map<string, StringVector> Redirected;
1127
1128    signal(SIGTERM,SigTerm);
1129    signal(SIGINT,SigTerm);
1130
1131    Server = 0;
1132
1133    int FailCounter = 0;
1134    while (1)
1135    {
1136       // We have no commands, wait for some to arrive
1137       if (Queue == 0)
1138       {
1139          if (WaitFd(STDIN_FILENO) == false)
1140             return 0;
1141       }
1142
1143       /* Run messages, we can accept 0 (no message) if we didn't
1144          do a WaitFd above.. Otherwise the FD is closed. */
1145       int Result = Run(true);
1146       if (Result != -1 && (Result != 0 || Queue == 0))
1147       {
1148          if(FailReason.empty() == false ||
1149             _config->FindB("Acquire::http::DependOnSTDIN", true) == true)
1150             return 100;
1151          else
1152             return 0;
1153       }
1154
1155       if (Queue == 0)
1156          continue;
1157
1158       // Connect to the server
1159       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1160       {
1161          delete Server;
1162          Server = new ServerState(Queue->Uri,this);
1163       }
1164       /* If the server has explicitly said this is the last connection
1165          then we pre-emptively shut down the pipeline and tear down
1166          the connection. This will speed up HTTP/1.0 servers a tad
1167          since we don't have to wait for the close sequence to
1168          complete */
1169       if (Server->Persistent == false)
1170          Server->Close();
1171
1172       // Reset the pipeline
1173       if (Server->ServerFd == -1)
1174          QueueBack = Queue;
1175
1176       // Connnect to the host
1177       if (Server->Open() == false)
1178       {
1179          Fail(true);
1180          delete Server;
1181          Server = 0;
1182          continue;
1183       }
1184
1185       // Fill the pipeline.
1186       Fetch(0);
1187
1188       // Fetch the next URL header data from the server.
1189       switch (Server->RunHeaders())
1190       {
1191          case ServerState::RUN_HEADERS_OK:
1192          break;
1193
1194          // The header data is bad
1195          case ServerState::RUN_HEADERS_PARSE_ERROR:
1196          {
1197             _error->Error(_("Bad header data"));
1198             Fail(true);
1199             RotateDNS();
1200             continue;
1201          }
1202
1203          // The server closed a connection during the header get..
1204          default:
1205          case ServerState::RUN_HEADERS_IO_ERROR:
1206          {
1207             FailCounter++;
1208             _error->Discard();
1209             Server->Close();
1210             Server->Pipeline = false;
1211
1212             if (FailCounter >= 2)
1213             {
1214                Fail(_("Connection failed"),true);
1215                FailCounter = 0;
1216             }
1217
1218             RotateDNS();
1219             continue;
1220          }
1221       };
1222
1223       // Decide what to do.
1224       FetchResult Res;
1225       Res.Filename = Queue->DestFile;
1226       switch (DealWithHeaders(Res,Server))
1227       {
1228          // Ok, the file is Open
1229          case FILE_IS_OPEN:
1230          {
1231             URIStart(Res);
1232
1233             // Run the data
1234             bool Result =  Server->RunData();
1235
1236             /* If the server is sending back sizeless responses then fill in
1237                the size now */
1238             if (Res.Size == 0)
1239                Res.Size = File->Size();
1240
1241             // Close the file, destroy the FD object and timestamp it
1242             FailFd = -1;
1243             delete File;
1244             File = 0;
1245
1246             // Timestamp
1247             struct utimbuf UBuf;
1248             time(&UBuf.actime);
1249             UBuf.actime = Server->Date;
1250             UBuf.modtime = Server->Date;
1251             utime(Queue->DestFile.c_str(),&UBuf);
1252
1253             // Send status to APT
1254             if (Result == true)
1255             {
1256                Res.TakeHashes(*Server->In.Hash);
1257                URIDone(Res);
1258             }
1259             else
1260             {
1261                if (Server->ServerFd == -1)
1262                {
1263                   FailCounter++;
1264                   _error->Discard();
1265                   Server->Close();
1266
1267                   if (FailCounter >= 2)
1268                   {
1269                      Fail(_("Connection failed"),true);
1270                      FailCounter = 0;
1271                   }
1272
1273                   QueueBack = Queue;
1274                }
1275                else
1276                   Fail(true);
1277             }
1278             break;
1279          }
1280
1281          // IMS hit
1282          case IMS_HIT:
1283          {
1284             URIDone(Res);
1285             break;
1286          }
1287
1288          // Hard server error, not found or something
1289          case ERROR_UNRECOVERABLE:
1290          {
1291             Fail();
1292             break;
1293          }
1294
1295          // Hard internal error, kill the connection and fail
1296          case ERROR_NOT_FROM_SERVER:
1297          {
1298             delete File;
1299             File = 0;
1300
1301             Fail();
1302             RotateDNS();
1303             Server->Close();
1304             break;
1305          }
1306
1307          // We need to flush the data, the header is like a 404 w/ error text
1308          case ERROR_WITH_CONTENT_PAGE:
1309          {
1310             Fail();
1311
1312             // Send to content to dev/null
1313             File = new FileFd("/dev/null",FileFd::WriteExists);
1314             Server->RunData();
1315             delete File;
1316             File = 0;
1317             break;
1318          }
1319
1320          // Try again with a new URL
1321          case TRY_AGAIN_OR_REDIRECT:
1322          {
1323             // Clear rest of response if there is content
1324             if (Server->HaveContent)
1325             {
1326                File = new FileFd("/dev/null",FileFd::WriteExists);
1327                Server->RunData();
1328                delete File;
1329                File = 0;
1330             }
1331
1332             /* Detect redirect loops.  No more redirects are allowed
1333                after the same URI is seen twice in a queue item. */
1334             StringVector &R = Redirected[Queue->DestFile];
1335             bool StopRedirects = false;
1336             if (R.empty() == true)
1337                R.push_back(Queue->Uri);
1338             else if (R[0] == "STOP" || R.size() > 10)
1339                StopRedirects = true;
1340             else
1341             {
1342                for (StringVectorIterator I = R.begin(); I != R.end(); ++I)
1343                   if (Queue->Uri == *I)
1344                   {
1345                      R[0] = "STOP";
1346                      break;
1347                   }
1348
1349                R.push_back(Queue->Uri);
1350             }
1351
1352             if (StopRedirects == false)
1353                Redirect(NextURI);
1354             else
1355                Fail();
1356
1357             break;
1358          }
1359
1360          default:
1361          Fail(_("Internal error"));
1362          break;
1363       }
1364
1365       FailCounter = 0;
1366    }
1367
1368    return 0;
1369 }
1370                                                                         /*}}}*/
1371 // HttpMethod::AutoDetectProxy - auto detect proxy                      /*{{{*/
1372 // ---------------------------------------------------------------------
1373 /* */
1374 bool HttpMethod::AutoDetectProxy()
1375 {
1376    if (AutoDetectProxyCmd.empty())
1377       return true;
1378
1379    if (Debug)
1380       clog << "Using auto proxy detect command: " << AutoDetectProxyCmd << endl;
1381
1382    int Pipes[2] = {-1,-1};
1383    if (pipe(Pipes) != 0)
1384       return _error->Errno("pipe", "Failed to create Pipe");
1385
1386    pid_t Process = ExecFork();
1387    if (Process == 0)
1388    {
1389       close(Pipes[0]);
1390       dup2(Pipes[1],STDOUT_FILENO);
1391       SetCloseExec(STDOUT_FILENO,false);
1392
1393       const char *Args[2];
1394       Args[0] = AutoDetectProxyCmd.c_str();
1395       Args[1] = 0;
1396       execv(Args[0],(char **)Args);
1397       cerr << "Failed to exec method " << Args[0] << endl;
1398       _exit(100);
1399    }
1400    char buf[512];
1401    int InFd = Pipes[0];
1402    close(Pipes[1]);
1403    int res = read(InFd, buf, sizeof(buf));
1404    ExecWait(Process, "ProxyAutoDetect", true);
1405
1406    if (res < 0)
1407       return _error->Errno("read", "Failed to read");
1408    if (res == 0)
1409       return _error->Warning("ProxyAutoDetect returned no data");
1410
1411    // add trailing \0
1412    buf[res] = 0;
1413
1414    if (Debug)
1415       clog << "auto detect command returned: '" << buf << "'" << endl;
1416
1417    if (strstr(buf, "http://") == buf)
1418       _config->Set("Acquire::http::proxy", _strstrip(buf));
1419
1420    return true;
1421 }
1422                                                                         /*}}}*/
1423
1424