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 = 10;
  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 == ULLONG_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 = Uri.Host;
 671    if (Uri.Port != 0)
 672    {
 673       sprintf(Buf,":%u",Uri.Port);
 674       ProperHost += Buf;
 675    }
 676
 677    // Just in case.
 678    if (Itm->Uri.length() >= sizeof(Buf))
 679        abort();
 680
 681    /* Build the request. We include a keep-alive header only for non-proxy
 682       requests. This is to tweak old http/1.0 servers that do support keep-alive
 683       but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
 684       will glitch HTTP/1.0 proxies because they do not filter it out and
 685       pass it on, HTTP/1.1 says the connection should default to keep alive
 686       and we expect the proxy to do this */
 687    if (Proxy.empty() == true || Proxy.Host.empty())
 688       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
 689               QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
 690    else
 691    {
 692       /* Generate a cache control header if necessary. We place a max
 693          cache age on index files, optionally set a no-cache directive
 694          and a no-store directive for archives. */
 695       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 696               Itm->Uri.c_str(),ProperHost.c_str());
 697    }
 698    // generate a cache control header (if needed)
 699    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 700    {
 701       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 702    }
 703    else
 704    {
 705       if (Itm->IndexFile == true)
 706       {
 707          sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 708                  _config->FindI("Acquire::http::Max-Age",0));
 709       }
 710       else
 711       {
 712          if (_config->FindB("Acquire::http::No-Store",false) == true)
 713             strcat(Buf,"Cache-Control: no-store\r\n");
 714       }
 715    }
 716
 717    // If we ask for uncompressed files servers might respond with content-
 718    // negotation which lets us end up with compressed files we do not support,
 719    // see 657029, 657560 and co, so if we have no extension on the request
 720    // ask for text only. As a sidenote: If there is nothing to negotate servers
 721    // seem to be nice and ignore it.
 722    if (_config->FindB("Acquire::http::SendAccept", true) == true)
 723    {
 724       size_t const filepos = Itm->Uri.find_last_of('/');
 725       string const file = Itm->Uri.substr(filepos + 1);
 726       if (flExtension(file) == file)
 727          strcat(Buf,"Accept: text/*\r\n");
 728    }
 729
 730    string Req = Buf;
 731
 732    // Check for a partial file
 733    struct stat SBuf;
 734    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 735    {
 736       // In this case we send an if-range query with a range header
 737       sprintf(Buf,"Range: bytes=%lli-\r\nIf-Range: %s\r\n",(long long)SBuf.st_size - 1,
 738               TimeRFC1123(SBuf.st_mtime).c_str());
 739       Req += Buf;
 740    }
 741    else
 742    {
 743       if (Itm->LastModified != 0)
 744       {
 745          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 746          Req += Buf;
 747       }
 748    }
 749
 750    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 751       Req += string("Proxy-Authorization: Basic ") +
 752           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 753
 754    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 755    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 756    {
 757       Req += string("Authorization: Basic ") +
 758           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 759    }
 760    Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
 761                 "Debian APT-HTTP/1.3 ("VERSION")") + "\r\n\r\n";
 762
 763    if (Debug == true)
 764       cerr << Req << endl;
 765
 766    Out.Read(Req);
 767 }
 768                                                                         /*}}}*/
 769 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 770 // ---------------------------------------------------------------------
 771 /* This runs the select loop over the server FDs, Output file FDs and
 772    stdin. */
 773 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 774 {
 775    // Server has closed the connection
 776    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 777                                ToFile == false))
 778       return false;
 779
 780    fd_set rfds,wfds;
 781    FD_ZERO(&rfds);
 782    FD_ZERO(&wfds);
 783
 784    /* Add the server. We only send more requests if the connection will
 785       be persisting */
 786    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 787        && Srv->Persistent == true)
 788       FD_SET(Srv->ServerFd,&wfds);
 789    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 790       FD_SET(Srv->ServerFd,&rfds);
 791
 792    // Add the file
 793    int FileFD = -1;
 794    if (File != 0)
 795       FileFD = File->Fd();
 796
 797    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 798       FD_SET(FileFD,&wfds);
 799
 800    // Add stdin
 801    if (_config->FindB("Acquire::http::DependOnSTDIN", true) == true)
 802       FD_SET(STDIN_FILENO,&rfds);
 803
 804    // Figure out the max fd
 805    int MaxFd = FileFD;
 806    if (MaxFd < Srv->ServerFd)
 807       MaxFd = Srv->ServerFd;
 808
 809    // Select
 810    struct timeval tv;
 811    tv.tv_sec = TimeOut;
 812    tv.tv_usec = 0;
 813    int Res = 0;
 814    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 815    {
 816       if (errno == EINTR)
 817          return true;
 818       return _error->Errno("select",_("Select failed"));
 819    }
 820
 821    if (Res == 0)
 822    {
 823       _error->Error(_("Connection timed out"));
 824       return ServerDie(Srv);
 825    }
 826
 827    // Handle server IO
 828    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 829    {
 830       errno = 0;
 831       if (Srv->In.Read(Srv->ServerFd) == false)
 832          return ServerDie(Srv);
 833    }
 834
 835    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 836    {
 837       errno = 0;
 838       if (Srv->Out.Write(Srv->ServerFd) == false)
 839          return ServerDie(Srv);
 840    }
 841
 842    // Send data to the file
 843    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 844    {
 845       if (Srv->In.Write(FileFD) == false)
 846          return _error->Errno("write",_("Error writing to output file"));
 847    }
 848
 849    // Handle commands from APT
 850    if (FD_ISSET(STDIN_FILENO,&rfds))
 851    {
 852       if (Run(true) != -1)
 853          exit(100);
 854    }
 855
 856    return true;
 857 }
 858                                                                         /*}}}*/
 859 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 860 // ---------------------------------------------------------------------
 861 /* This takes the current input buffer from the Server FD and writes it
 862    into the file */
 863 bool HttpMethod::Flush(ServerState *Srv)
 864 {
 865    if (File != 0)
 866    {
 867       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 868       // can't be set
 869       if (File->Name() != "/dev/null")
 870          SetNonBlock(File->Fd(),false);
 871       if (Srv->In.WriteSpace() == false)
 872          return true;
 873
 874       while (Srv->In.WriteSpace() == true)
 875       {
 876          if (Srv->In.Write(File->Fd()) == false)
 877             return _error->Errno("write",_("Error writing to file"));
 878          if (Srv->In.IsLimit() == true)
 879             return true;
 880       }
 881
 882       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 883          return true;
 884    }
 885    return false;
 886 }
 887                                                                         /*}}}*/
 888 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 889 // ---------------------------------------------------------------------
 890 /* */
 891 bool HttpMethod::ServerDie(ServerState *Srv)
 892 {
 893    unsigned int LErrno = errno;
 894
 895    // Dump the buffer to the file
 896    if (Srv->State == ServerState::Data)
 897    {
 898       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 899       // can't be set
 900       if (File->Name() != "/dev/null")
 901          SetNonBlock(File->Fd(),false);
 902       while (Srv->In.WriteSpace() == true)
 903       {
 904          if (Srv->In.Write(File->Fd()) == false)
 905             return _error->Errno("write",_("Error writing to the file"));
 906
 907          // Done
 908          if (Srv->In.IsLimit() == true)
 909             return true;
 910       }
 911    }
 912
 913    // See if this is because the server finished the data stream
 914    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 915        Srv->Encoding != ServerState::Closes)
 916    {
 917       Srv->Close();
 918       if (LErrno == 0)
 919          return _error->Error(_("Error reading from server. Remote end closed connection"));
 920       errno = LErrno;
 921       return _error->Errno("read",_("Error reading from server"));
 922    }
 923    else
 924    {
 925       Srv->In.Limit(-1);
 926
 927       // Nothing left in the buffer
 928       if (Srv->In.WriteSpace() == false)
 929          return false;
 930
 931       // We may have got multiple responses back in one packet..
 932       Srv->Close();
 933       return true;
 934    }
 935
 936    return false;
 937 }
 938                                                                         /*}}}*/
 939 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 940 // ---------------------------------------------------------------------
 941 /* We look at the header data we got back from the server and decide what
 942    to do. Returns DealWithHeadersResult (see http.h for details).
 943  */
 944 HttpMethod::DealWithHeadersResult
 945 HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 946 {
 947    // Not Modified
 948    if (Srv->Result == 304)
 949    {
 950       unlink(Queue->DestFile.c_str());
 951       Res.IMSHit = true;
 952       Res.LastModified = Queue->LastModified;
 953       return IMS_HIT;
 954    }
 955
 956    /* Redirect
 957     *
 958     * Note that it is only OK for us to treat all redirection the same
 959     * because we *always* use GET, not other HTTP methods.  There are
 960     * three redirection codes for which it is not appropriate that we
 961     * redirect.  Pass on those codes so the error handling kicks in.
 962     */
 963    if (AllowRedirect
 964        && (Srv->Result > 300 && Srv->Result < 400)
 965        && (Srv->Result != 300       // Multiple Choices
 966            && Srv->Result != 304    // Not Modified
 967            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 968    {
 969       if (Srv->Location.empty() == true);
 970       else if (Srv->Location[0] == '/' && Queue->Uri.empty() == false)
 971       {
 972          URI Uri = Queue->Uri;
 973          if (Uri.Host.empty() == false)
 974          {
 975             if (Uri.Port != 0)
 976                strprintf(NextURI, "http://%s:%u", Uri.Host.c_str(), Uri.Port);
 977             else
 978                NextURI = "http://" + Uri.Host;
 979          }
 980          else
 981             NextURI.clear();
 982          NextURI.append(DeQuoteString(Srv->Location));
 983          return TRY_AGAIN_OR_REDIRECT;
 984       }
 985       else
 986       {
 987          NextURI = DeQuoteString(Srv->Location);
 988          return TRY_AGAIN_OR_REDIRECT;
 989       }
 990       /* else pass through for error message */
 991    }
 992
 993    /* We have a reply we dont handle. This should indicate a perm server
 994       failure */
 995    if (Srv->Result < 200 || Srv->Result >= 300)
 996    {
 997       char err[255];
 998       snprintf(err,sizeof(err)-1,"HttpError%i",Srv->Result);
 999       SetFailReason(err);
1000       _error->Error("%u %s",Srv->Result,Srv->Code);
1001       if (Srv->HaveContent == true)
1002          return ERROR_WITH_CONTENT_PAGE;
1003       return ERROR_UNRECOVERABLE;
1004    }
1005
1006    // This is some sort of 2xx 'data follows' reply
1007    Res.LastModified = Srv->Date;
1008    Res.Size = Srv->Size;
1009
1010    // Open the file
1011    delete File;
1012    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
1013    if (_error->PendingError() == true)
1014       return ERROR_NOT_FROM_SERVER;
1015
1016    FailFile = Queue->DestFile;
1017    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
1018    FailFd = File->Fd();
1019    FailTime = Srv->Date;
1020
1021    delete Srv->In.Hash;
1022    Srv->In.Hash = new Hashes;
1023
1024    // Set the expected size and read file for the hashes
1025    if (Srv->StartPos >= 0)
1026    {
1027       Res.ResumePoint = Srv->StartPos;
1028       File->Truncate(Srv->StartPos);
1029
1030       if (Srv->In.Hash->AddFD(*File,Srv->StartPos) == false)
1031       {
1032          _error->Errno("read",_("Problem hashing file"));
1033          return ERROR_NOT_FROM_SERVER;
1034       }
1035    }
1036
1037    SetNonBlock(File->Fd(),true);
1038    return FILE_IS_OPEN;
1039 }
1040                                                                         /*}}}*/
1041 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1042 // ---------------------------------------------------------------------
1043 /* This closes and timestamps the open file. This is neccessary to get
1044    resume behavoir on user abort */
1045 void HttpMethod::SigTerm(int)
1046 {
1047    if (FailFd == -1)
1048       _exit(100);
1049    close(FailFd);
1050
1051    // Timestamp
1052    struct utimbuf UBuf;
1053    UBuf.actime = FailTime;
1054    UBuf.modtime = FailTime;
1055    utime(FailFile.c_str(),&UBuf);
1056
1057    _exit(100);
1058 }
1059                                                                         /*}}}*/
1060 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1061 // ---------------------------------------------------------------------
1062 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1063    depth. */
1064 bool HttpMethod::Fetch(FetchItem *)
1065 {
1066    if (Server == 0)
1067       return true;
1068
1069    // Queue the requests
1070    int Depth = -1;
1071    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1072         I = I->Next, Depth++)
1073    {
1074       // If pipelining is disabled, we only queue 1 request
1075       if (Server->Pipeline == false && Depth >= 0)
1076          break;
1077
1078       // Make sure we stick with the same server
1079       if (Server->Comp(I->Uri) == false)
1080          break;
1081       if (QueueBack == I)
1082       {
1083          QueueBack = I->Next;
1084          SendReq(I,Server->Out);
1085          continue;
1086       }
1087    }
1088
1089    return true;
1090 };
1091                                                                         /*}}}*/
1092 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1093 // ---------------------------------------------------------------------
1094 /* We stash the desired pipeline depth */
1095 bool HttpMethod::Configuration(string Message)
1096 {
1097    if (pkgAcqMethod::Configuration(Message) == false)
1098       return false;
1099
1100    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1101    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1102    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1103                                   PipelineDepth);
1104    Debug = _config->FindB("Debug::Acquire::http",false);
1105    AutoDetectProxyCmd = _config->Find("Acquire::http::ProxyAutoDetect");
1106
1107    // Get the proxy to use
1108    AutoDetectProxy();
1109
1110    return true;
1111 }
1112                                                                         /*}}}*/
1113 // HttpMethod::Loop - Main loop                                         /*{{{*/
1114 // ---------------------------------------------------------------------
1115 /* */
1116 int HttpMethod::Loop()
1117 {
1118    typedef vector<string> StringVector;
1119    typedef vector<string>::iterator StringVectorIterator;
1120    map<string, StringVector> Redirected;
1121
1122    signal(SIGTERM,SigTerm);
1123    signal(SIGINT,SigTerm);
1124
1125    Server = 0;
1126
1127    int FailCounter = 0;
1128    while (1)
1129    {
1130       // We have no commands, wait for some to arrive
1131       if (Queue == 0)
1132       {
1133          if (WaitFd(STDIN_FILENO) == false)
1134             return 0;
1135       }
1136
1137       /* Run messages, we can accept 0 (no message) if we didn't
1138          do a WaitFd above.. Otherwise the FD is closed. */
1139       int Result = Run(true);
1140       if (Result != -1 && (Result != 0 || Queue == 0))
1141       {
1142          if(FailReason.empty() == false ||
1143             _config->FindB("Acquire::http::DependOnSTDIN", true) == true)
1144             return 100;
1145          else
1146             return 0;
1147       }
1148
1149       if (Queue == 0)
1150          continue;
1151
1152       // Connect to the server
1153       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1154       {
1155          delete Server;
1156          Server = new ServerState(Queue->Uri,this);
1157       }
1158       /* If the server has explicitly said this is the last connection
1159          then we pre-emptively shut down the pipeline and tear down
1160          the connection. This will speed up HTTP/1.0 servers a tad
1161          since we don't have to wait for the close sequence to
1162          complete */
1163       if (Server->Persistent == false)
1164          Server->Close();
1165
1166       // Reset the pipeline
1167       if (Server->ServerFd == -1)
1168          QueueBack = Queue;
1169
1170       // Connnect to the host
1171       if (Server->Open() == false)
1172       {
1173          Fail(true);
1174          delete Server;
1175          Server = 0;
1176          continue;
1177       }
1178
1179       // Fill the pipeline.
1180       Fetch(0);
1181
1182       // Fetch the next URL header data from the server.
1183       switch (Server->RunHeaders())
1184       {
1185          case ServerState::RUN_HEADERS_OK:
1186          break;
1187
1188          // The header data is bad
1189          case ServerState::RUN_HEADERS_PARSE_ERROR:
1190          {
1191             _error->Error(_("Bad header data"));
1192             Fail(true);
1193             RotateDNS();
1194             continue;
1195          }
1196
1197          // The server closed a connection during the header get..
1198          default:
1199          case ServerState::RUN_HEADERS_IO_ERROR:
1200          {
1201             FailCounter++;
1202             _error->Discard();
1203             Server->Close();
1204             Server->Pipeline = false;
1205
1206             if (FailCounter >= 2)
1207             {
1208                Fail(_("Connection failed"),true);
1209                FailCounter = 0;
1210             }
1211
1212             RotateDNS();
1213             continue;
1214          }
1215       };
1216
1217       // Decide what to do.
1218       FetchResult Res;
1219       Res.Filename = Queue->DestFile;
1220       switch (DealWithHeaders(Res,Server))
1221       {
1222          // Ok, the file is Open
1223          case FILE_IS_OPEN:
1224          {
1225             URIStart(Res);
1226
1227             // Run the data
1228             bool Result =  Server->RunData();
1229
1230             /* If the server is sending back sizeless responses then fill in
1231                the size now */
1232             if (Res.Size == 0)
1233                Res.Size = File->Size();
1234
1235             // Close the file, destroy the FD object and timestamp it
1236             FailFd = -1;
1237             delete File;
1238             File = 0;
1239
1240             // Timestamp
1241             struct utimbuf UBuf;
1242             time(&UBuf.actime);
1243             UBuf.actime = Server->Date;
1244             UBuf.modtime = Server->Date;
1245             utime(Queue->DestFile.c_str(),&UBuf);
1246
1247             // Send status to APT
1248             if (Result == true)
1249             {
1250                Res.TakeHashes(*Server->In.Hash);
1251                URIDone(Res);
1252             }
1253             else
1254             {
1255                if (Server->ServerFd == -1)
1256                {
1257                   FailCounter++;
1258                   _error->Discard();
1259                   Server->Close();
1260
1261                   if (FailCounter >= 2)
1262                   {
1263                      Fail(_("Connection failed"),true);
1264                      FailCounter = 0;
1265                   }
1266
1267                   QueueBack = Queue;
1268                }
1269                else
1270                   Fail(true);
1271             }
1272             break;
1273          }
1274
1275          // IMS hit
1276          case IMS_HIT:
1277          {
1278             URIDone(Res);
1279             break;
1280          }
1281
1282          // Hard server error, not found or something
1283          case ERROR_UNRECOVERABLE:
1284          {
1285             Fail();
1286             break;
1287          }
1288
1289          // Hard internal error, kill the connection and fail
1290          case ERROR_NOT_FROM_SERVER:
1291          {
1292             delete File;
1293             File = 0;
1294
1295             Fail();
1296             RotateDNS();
1297             Server->Close();
1298             break;
1299          }
1300
1301          // We need to flush the data, the header is like a 404 w/ error text
1302          case ERROR_WITH_CONTENT_PAGE:
1303          {
1304             Fail();
1305
1306             // Send to content to dev/null
1307             File = new FileFd("/dev/null",FileFd::WriteExists);
1308             Server->RunData();
1309             delete File;
1310             File = 0;
1311             break;
1312          }
1313
1314          // Try again with a new URL
1315          case TRY_AGAIN_OR_REDIRECT:
1316          {
1317             // Clear rest of response if there is content
1318             if (Server->HaveContent)
1319             {
1320                File = new FileFd("/dev/null",FileFd::WriteExists);
1321                Server->RunData();
1322                delete File;
1323                File = 0;
1324             }
1325
1326             /* Detect redirect loops.  No more redirects are allowed
1327                after the same URI is seen twice in a queue item. */
1328             StringVector &R = Redirected[Queue->DestFile];
1329             bool StopRedirects = false;
1330             if (R.empty() == true)
1331                R.push_back(Queue->Uri);
1332             else if (R[0] == "STOP" || R.size() > 10)
1333                StopRedirects = true;
1334             else
1335             {
1336                for (StringVectorIterator I = R.begin(); I != R.end(); ++I)
1337                   if (Queue->Uri == *I)
1338                   {
1339                      R[0] = "STOP";
1340                      break;
1341                   }
1342
1343                R.push_back(Queue->Uri);
1344             }
1345
1346             if (StopRedirects == false)
1347                Redirect(NextURI);
1348             else
1349                Fail();
1350
1351             break;
1352          }
1353
1354          default:
1355          Fail(_("Internal error"));
1356          break;
1357       }
1358
1359       FailCounter = 0;
1360    }
1361
1362    return 0;
1363 }
1364                                                                         /*}}}*/
1365 // HttpMethod::AutoDetectProxy - auto detect proxy                      /*{{{*/
1366 // ---------------------------------------------------------------------
1367 /* */
1368 bool HttpMethod::AutoDetectProxy()
1369 {
1370    if (AutoDetectProxyCmd.empty())
1371       return true;
1372
1373    if (Debug)
1374       clog << "Using auto proxy detect command: " << AutoDetectProxyCmd << endl;
1375
1376    int Pipes[2] = {-1,-1};
1377    if (pipe(Pipes) != 0)
1378       return _error->Errno("pipe", "Failed to create Pipe");
1379
1380    pid_t Process = ExecFork();
1381    if (Process == 0)
1382    {
1383       close(Pipes[0]);
1384       dup2(Pipes[1],STDOUT_FILENO);
1385       SetCloseExec(STDOUT_FILENO,false);
1386
1387       const char *Args[2];
1388       Args[0] = AutoDetectProxyCmd.c_str();
1389       Args[1] = 0;
1390       execv(Args[0],(char **)Args);
1391       cerr << "Failed to exec method " << Args[0] << endl;
1392       _exit(100);
1393    }
1394    char buf[512];
1395    int InFd = Pipes[0];
1396    close(Pipes[1]);
1397    int res = read(InFd, buf, sizeof(buf));
1398    ExecWait(Process, "ProxyAutoDetect", true);
1399
1400    if (res < 0)
1401       return _error->Errno("read", "Failed to read");
1402    if (res == 0)
1403       return _error->Warning("ProxyAutoDetect returned no data");
1404
1405    // add trailing \0
1406    buf[res] = 0;
1407
1408    if (Debug)
1409       clog << "auto detect command returned: '" << buf << "'" << endl;
1410
1411    if (strstr(buf, "http://") == buf)
1412       _config->Set("Acquire::http::proxy", _strstrip(buf));
1413
1414    return true;
1415 }
1416                                                                         /*}}}*/
1417
1418