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    /* Build the request. No keep-alive is included as it is the default
 687       in 1.1, can cause problems with proxies, and we are an HTTP/1.1
 688       client anyway.
 689       C.f. https://tools.ietf.org/wg/httpbis/trac/ticket/158 */
 690    // see LP bugs #1003633 and #1086997. The "+" is encoded as a workaround
 691    // for a amazon S3 bug
 692    sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 693            QuoteString(Uri.Path,"+~ ").c_str(),ProperHost.c_str());
 694    // generate a cache control header (if needed)
 695    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 696    {
 697       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 698    }
 699    else
 700    {
 701       if (Itm->IndexFile == true)
 702       {
 703          sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 704                  _config->FindI("Acquire::http::Max-Age",0));
 705       }
 706       else
 707       {
 708          if (_config->FindB("Acquire::http::No-Store",false) == true)
 709             strcat(Buf,"Cache-Control: no-store\r\n");
 710       }
 711    }
 712
 713    // If we ask for uncompressed files servers might respond with content-
 714    // negotation which lets us end up with compressed files we do not support,
 715    // see 657029, 657560 and co, so if we have no extension on the request
 716    // ask for text only. As a sidenote: If there is nothing to negotate servers
 717    // seem to be nice and ignore it.
 718    if (_config->FindB("Acquire::http::SendAccept", true) == true)
 719    {
 720       size_t const filepos = Itm->Uri.find_last_of('/');
 721       string const file = Itm->Uri.substr(filepos + 1);
 722       if (flExtension(file) == file)
 723          strcat(Buf,"Accept: text/*\r\n");
 724    }
 725
 726    string Req = Buf;
 727
 728    // Check for a partial file
 729    struct stat SBuf;
 730    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 731    {
 732       // In this case we send an if-range query with a range header
 733       sprintf(Buf,"Range: bytes=%lli-\r\nIf-Range: %s\r\n",(long long)SBuf.st_size - 1,
 734               TimeRFC1123(SBuf.st_mtime).c_str());
 735       Req += Buf;
 736    }
 737    else
 738    {
 739       if (Itm->LastModified != 0)
 740       {
 741          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 742          Req += Buf;
 743       }
 744    }
 745
 746    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 747       Req += string("Proxy-Authorization: Basic ") +
 748           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 749
 750    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 751    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 752    {
 753       Req += string("Authorization: Basic ") +
 754           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 755    }
 756    Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
 757                 "Debian APT-HTTP/1.3 (" PACKAGE_VERSION ")") + "\r\n\r\n";
 758
 759    if (Debug == true)
 760       cerr << Req << endl;
 761
 762    Out.Read(Req);
 763 }
 764                                                                         /*}}}*/
 765 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 766 // ---------------------------------------------------------------------
 767 /* This runs the select loop over the server FDs, Output file FDs and
 768    stdin. */
 769 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 770 {
 771    // Server has closed the connection
 772    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 773                                ToFile == false))
 774       return false;
 775
 776    fd_set rfds,wfds;
 777    FD_ZERO(&rfds);
 778    FD_ZERO(&wfds);
 779
 780    /* Add the server. We only send more requests if the connection will
 781       be persisting */
 782    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 783        && Srv->Persistent == true)
 784       FD_SET(Srv->ServerFd,&wfds);
 785    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 786       FD_SET(Srv->ServerFd,&rfds);
 787
 788    // Add the file
 789    int FileFD = -1;
 790    if (File != 0)
 791       FileFD = File->Fd();
 792
 793    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 794       FD_SET(FileFD,&wfds);
 795
 796    // Add stdin
 797    if (_config->FindB("Acquire::http::DependOnSTDIN", true) == true)
 798       FD_SET(STDIN_FILENO,&rfds);
 799
 800    // Figure out the max fd
 801    int MaxFd = FileFD;
 802    if (MaxFd < Srv->ServerFd)
 803       MaxFd = Srv->ServerFd;
 804
 805    // Select
 806    struct timeval tv;
 807    tv.tv_sec = TimeOut;
 808    tv.tv_usec = 0;
 809    int Res = 0;
 810    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 811    {
 812       if (errno == EINTR)
 813          return true;
 814       return _error->Errno("select",_("Select failed"));
 815    }
 816
 817    if (Res == 0)
 818    {
 819       _error->Error(_("Connection timed out"));
 820       return ServerDie(Srv);
 821    }
 822
 823    // Handle server IO
 824    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 825    {
 826       errno = 0;
 827       if (Srv->In.Read(Srv->ServerFd) == false)
 828          return ServerDie(Srv);
 829    }
 830
 831    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 832    {
 833       errno = 0;
 834       if (Srv->Out.Write(Srv->ServerFd) == false)
 835          return ServerDie(Srv);
 836    }
 837
 838    // Send data to the file
 839    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 840    {
 841       if (Srv->In.Write(FileFD) == false)
 842          return _error->Errno("write",_("Error writing to output file"));
 843    }
 844
 845    // Handle commands from APT
 846    if (FD_ISSET(STDIN_FILENO,&rfds))
 847    {
 848       if (Run(true) != -1)
 849          exit(100);
 850    }
 851
 852    return true;
 853 }
 854                                                                         /*}}}*/
 855 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 856 // ---------------------------------------------------------------------
 857 /* This takes the current input buffer from the Server FD and writes it
 858    into the file */
 859 bool HttpMethod::Flush(ServerState *Srv)
 860 {
 861    if (File != 0)
 862    {
 863       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 864       // can't be set
 865       if (File->Name() != "/dev/null")
 866          SetNonBlock(File->Fd(),false);
 867       if (Srv->In.WriteSpace() == false)
 868          return true;
 869
 870       while (Srv->In.WriteSpace() == true)
 871       {
 872          if (Srv->In.Write(File->Fd()) == false)
 873             return _error->Errno("write",_("Error writing to file"));
 874          if (Srv->In.IsLimit() == true)
 875             return true;
 876       }
 877
 878       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 879          return true;
 880    }
 881    return false;
 882 }
 883                                                                         /*}}}*/
 884 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 885 // ---------------------------------------------------------------------
 886 /* */
 887 bool HttpMethod::ServerDie(ServerState *Srv)
 888 {
 889    unsigned int LErrno = errno;
 890
 891    // Dump the buffer to the file
 892    if (Srv->State == ServerState::Data)
 893    {
 894       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 895       // can't be set
 896       if (File->Name() != "/dev/null")
 897          SetNonBlock(File->Fd(),false);
 898       while (Srv->In.WriteSpace() == true)
 899       {
 900          if (Srv->In.Write(File->Fd()) == false)
 901             return _error->Errno("write",_("Error writing to the file"));
 902
 903          // Done
 904          if (Srv->In.IsLimit() == true)
 905             return true;
 906       }
 907    }
 908
 909    // See if this is because the server finished the data stream
 910    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 911        Srv->Encoding != ServerState::Closes)
 912    {
 913       Srv->Close();
 914       if (LErrno == 0)
 915          return _error->Error(_("Error reading from server. Remote end closed connection"));
 916       errno = LErrno;
 917       return _error->Errno("read",_("Error reading from server"));
 918    }
 919    else
 920    {
 921       Srv->In.Limit(-1);
 922
 923       // Nothing left in the buffer
 924       if (Srv->In.WriteSpace() == false)
 925          return false;
 926
 927       // We may have got multiple responses back in one packet..
 928       Srv->Close();
 929       return true;
 930    }
 931
 932    return false;
 933 }
 934                                                                         /*}}}*/
 935 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 936 // ---------------------------------------------------------------------
 937 /* We look at the header data we got back from the server and decide what
 938    to do. Returns DealWithHeadersResult (see http.h for details).
 939  */
 940 HttpMethod::DealWithHeadersResult
 941 HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 942 {
 943    // Not Modified
 944    if (Srv->Result == 304)
 945    {
 946       unlink(Queue->DestFile.c_str());
 947       Res.IMSHit = true;
 948       Res.LastModified = Queue->LastModified;
 949       return IMS_HIT;
 950    }
 951
 952    /* Redirect
 953     *
 954     * Note that it is only OK for us to treat all redirection the same
 955     * because we *always* use GET, not other HTTP methods.  There are
 956     * three redirection codes for which it is not appropriate that we
 957     * redirect.  Pass on those codes so the error handling kicks in.
 958     */
 959    if (AllowRedirect
 960        && (Srv->Result > 300 && Srv->Result < 400)
 961        && (Srv->Result != 300       // Multiple Choices
 962            && Srv->Result != 304    // Not Modified
 963            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 964    {
 965       if (Srv->Location.empty() == true);
 966       else if (Srv->Location[0] == '/' && Queue->Uri.empty() == false)
 967       {
 968          URI Uri = Queue->Uri;
 969          if (Uri.Host.empty() == false)
 970             NextURI = URI::SiteOnly(Uri);
 971          else
 972             NextURI.clear();
 973          NextURI.append(DeQuoteString(Srv->Location));
 974          return TRY_AGAIN_OR_REDIRECT;
 975       }
 976       else
 977       {
 978          NextURI = DeQuoteString(Srv->Location);
 979          URI tmpURI = NextURI;
 980          // Do not allow a redirection to switch protocol
 981          if (tmpURI.Access == "http")
 982             return TRY_AGAIN_OR_REDIRECT;
 983       }
 984       /* else pass through for error message */
 985    }
 986
 987    /* We have a reply we dont handle. This should indicate a perm server
 988       failure */
 989    if (Srv->Result < 200 || Srv->Result >= 300)
 990    {
 991       char err[255];
 992       snprintf(err,sizeof(err)-1,"HttpError%i",Srv->Result);
 993       SetFailReason(err);
 994       _error->Error("%u %s",Srv->Result,Srv->Code);
 995       if (Srv->HaveContent == true)
 996          return ERROR_WITH_CONTENT_PAGE;
 997       return ERROR_UNRECOVERABLE;
 998    }
 999
1000    // This is some sort of 2xx 'data follows' reply
1001    Res.LastModified = Srv->Date;
1002    Res.Size = Srv->Size;
1003
1004    // Open the file
1005    delete File;
1006    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
1007    if (_error->PendingError() == true)
1008       return ERROR_NOT_FROM_SERVER;
1009
1010    FailFile = Queue->DestFile;
1011    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
1012    FailFd = File->Fd();
1013    FailTime = Srv->Date;
1014
1015    delete Srv->In.Hash;
1016    Srv->In.Hash = new Hashes;
1017
1018    // Set the expected size and read file for the hashes
1019    if (Srv->StartPos >= 0)
1020    {
1021       Res.ResumePoint = Srv->StartPos;
1022       File->Truncate(Srv->StartPos);
1023
1024       if (Srv->In.Hash->AddFD(*File,Srv->StartPos) == false)
1025       {
1026          _error->Errno("read",_("Problem hashing file"));
1027          return ERROR_NOT_FROM_SERVER;
1028       }
1029    }
1030
1031    SetNonBlock(File->Fd(),true);
1032    return FILE_IS_OPEN;
1033 }
1034                                                                         /*}}}*/
1035 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1036 // ---------------------------------------------------------------------
1037 /* This closes and timestamps the open file. This is neccessary to get
1038    resume behavoir on user abort */
1039 void HttpMethod::SigTerm(int)
1040 {
1041    if (FailFd == -1)
1042       _exit(100);
1043    close(FailFd);
1044
1045    // Timestamp
1046    struct utimbuf UBuf;
1047    UBuf.actime = FailTime;
1048    UBuf.modtime = FailTime;
1049    utime(FailFile.c_str(),&UBuf);
1050
1051    _exit(100);
1052 }
1053                                                                         /*}}}*/
1054 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1055 // ---------------------------------------------------------------------
1056 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1057    depth. */
1058 bool HttpMethod::Fetch(FetchItem *)
1059 {
1060    if (Server == 0)
1061       return true;
1062
1063    // Queue the requests
1064    int Depth = -1;
1065    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1066         I = I->Next, Depth++)
1067    {
1068       // If pipelining is disabled, we only queue 1 request
1069       if (Server->Pipeline == false && Depth >= 0)
1070          break;
1071
1072       // Make sure we stick with the same server
1073       if (Server->Comp(I->Uri) == false)
1074          break;
1075       if (QueueBack == I)
1076       {
1077          QueueBack = I->Next;
1078          SendReq(I,Server->Out);
1079          continue;
1080       }
1081    }
1082
1083    return true;
1084 };
1085                                                                         /*}}}*/
1086 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1087 // ---------------------------------------------------------------------
1088 /* We stash the desired pipeline depth */
1089 bool HttpMethod::Configuration(string Message)
1090 {
1091    if (pkgAcqMethod::Configuration(Message) == false)
1092       return false;
1093
1094    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1095    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1096    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1097                                   PipelineDepth);
1098    Debug = _config->FindB("Debug::Acquire::http",false);
1099    AutoDetectProxyCmd = _config->Find("Acquire::http::ProxyAutoDetect");
1100
1101    // Get the proxy to use
1102    AutoDetectProxy();
1103
1104    return true;
1105 }
1106                                                                         /*}}}*/
1107 // HttpMethod::Loop - Main loop                                         /*{{{*/
1108 // ---------------------------------------------------------------------
1109 /* */
1110 int HttpMethod::Loop()
1111 {
1112    typedef vector<string> StringVector;
1113    typedef vector<string>::iterator StringVectorIterator;
1114    map<string, StringVector> Redirected;
1115
1116    signal(SIGTERM,SigTerm);
1117    signal(SIGINT,SigTerm);
1118
1119    Server = 0;
1120
1121    int FailCounter = 0;
1122    while (1)
1123    {
1124       // We have no commands, wait for some to arrive
1125       if (Queue == 0)
1126       {
1127          if (WaitFd(STDIN_FILENO) == false)
1128             return 0;
1129       }
1130
1131       /* Run messages, we can accept 0 (no message) if we didn't
1132          do a WaitFd above.. Otherwise the FD is closed. */
1133       int Result = Run(true);
1134       if (Result != -1 && (Result != 0 || Queue == 0))
1135       {
1136          if(FailReason.empty() == false ||
1137             _config->FindB("Acquire::http::DependOnSTDIN", true) == true)
1138             return 100;
1139          else
1140             return 0;
1141       }
1142
1143       if (Queue == 0)
1144          continue;
1145
1146       // Connect to the server
1147       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1148       {
1149          delete Server;
1150          Server = new ServerState(Queue->Uri,this);
1151       }
1152       /* If the server has explicitly said this is the last connection
1153          then we pre-emptively shut down the pipeline and tear down
1154          the connection. This will speed up HTTP/1.0 servers a tad
1155          since we don't have to wait for the close sequence to
1156          complete */
1157       if (Server->Persistent == false)
1158          Server->Close();
1159
1160       // Reset the pipeline
1161       if (Server->ServerFd == -1)
1162          QueueBack = Queue;
1163
1164       // Connnect to the host
1165       if (Server->Open() == false)
1166       {
1167          Fail(true);
1168          delete Server;
1169          Server = 0;
1170          continue;
1171       }
1172
1173       // Fill the pipeline.
1174       Fetch(0);
1175
1176       // Fetch the next URL header data from the server.
1177       switch (Server->RunHeaders())
1178       {
1179          case ServerState::RUN_HEADERS_OK:
1180          break;
1181
1182          // The header data is bad
1183          case ServerState::RUN_HEADERS_PARSE_ERROR:
1184          {
1185             _error->Error(_("Bad header data"));
1186             Fail(true);
1187             RotateDNS();
1188             continue;
1189          }
1190
1191          // The server closed a connection during the header get..
1192          default:
1193          case ServerState::RUN_HEADERS_IO_ERROR:
1194          {
1195             FailCounter++;
1196             _error->Discard();
1197             Server->Close();
1198             Server->Pipeline = false;
1199
1200             if (FailCounter >= 2)
1201             {
1202                Fail(_("Connection failed"),true);
1203                FailCounter = 0;
1204             }
1205
1206             RotateDNS();
1207             continue;
1208          }
1209       };
1210
1211       // Decide what to do.
1212       FetchResult Res;
1213       Res.Filename = Queue->DestFile;
1214       switch (DealWithHeaders(Res,Server))
1215       {
1216          // Ok, the file is Open
1217          case FILE_IS_OPEN:
1218          {
1219             URIStart(Res);
1220
1221             // Run the data
1222             bool Result =  Server->RunData();
1223
1224             /* If the server is sending back sizeless responses then fill in
1225                the size now */
1226             if (Res.Size == 0)
1227                Res.Size = File->Size();
1228
1229             // Close the file, destroy the FD object and timestamp it
1230             FailFd = -1;
1231             delete File;
1232             File = 0;
1233
1234             // Timestamp
1235             struct utimbuf UBuf;
1236             time(&UBuf.actime);
1237             UBuf.actime = Server->Date;
1238             UBuf.modtime = Server->Date;
1239             utime(Queue->DestFile.c_str(),&UBuf);
1240
1241             // Send status to APT
1242             if (Result == true)
1243             {
1244                Res.TakeHashes(*Server->In.Hash);
1245                URIDone(Res);
1246             }
1247             else
1248             {
1249                if (Server->ServerFd == -1)
1250                {
1251                   FailCounter++;
1252                   _error->Discard();
1253                   Server->Close();
1254
1255                   if (FailCounter >= 2)
1256                   {
1257                      Fail(_("Connection failed"),true);
1258                      FailCounter = 0;
1259                   }
1260
1261                   QueueBack = Queue;
1262                }
1263                else
1264                   Fail(true);
1265             }
1266             break;
1267          }
1268
1269          // IMS hit
1270          case IMS_HIT:
1271          {
1272             URIDone(Res);
1273             break;
1274          }
1275
1276          // Hard server error, not found or something
1277          case ERROR_UNRECOVERABLE:
1278          {
1279             Fail();
1280             break;
1281          }
1282
1283          // Hard internal error, kill the connection and fail
1284          case ERROR_NOT_FROM_SERVER:
1285          {
1286             delete File;
1287             File = 0;
1288
1289             Fail();
1290             RotateDNS();
1291             Server->Close();
1292             break;
1293          }
1294
1295          // We need to flush the data, the header is like a 404 w/ error text
1296          case ERROR_WITH_CONTENT_PAGE:
1297          {
1298             Fail();
1299
1300             // Send to content to dev/null
1301             File = new FileFd("/dev/null",FileFd::WriteExists);
1302             Server->RunData();
1303             delete File;
1304             File = 0;
1305             break;
1306          }
1307
1308          // Try again with a new URL
1309          case TRY_AGAIN_OR_REDIRECT:
1310          {
1311             // Clear rest of response if there is content
1312             if (Server->HaveContent)
1313             {
1314                File = new FileFd("/dev/null",FileFd::WriteExists);
1315                Server->RunData();
1316                delete File;
1317                File = 0;
1318             }
1319
1320             /* Detect redirect loops.  No more redirects are allowed
1321                after the same URI is seen twice in a queue item. */
1322             StringVector &R = Redirected[Queue->DestFile];
1323             bool StopRedirects = false;
1324             if (R.empty() == true)
1325                R.push_back(Queue->Uri);
1326             else if (R[0] == "STOP" || R.size() > 10)
1327                StopRedirects = true;
1328             else
1329             {
1330                for (StringVectorIterator I = R.begin(); I != R.end(); ++I)
1331                   if (Queue->Uri == *I)
1332                   {
1333                      R[0] = "STOP";
1334                      break;
1335                   }
1336
1337                R.push_back(Queue->Uri);
1338             }
1339
1340             if (StopRedirects == false)
1341                Redirect(NextURI);
1342             else
1343                Fail();
1344
1345             break;
1346          }
1347
1348          default:
1349          Fail(_("Internal error"));
1350          break;
1351       }
1352
1353       FailCounter = 0;
1354    }
1355
1356    return 0;
1357 }
1358                                                                         /*}}}*/
1359 // HttpMethod::AutoDetectProxy - auto detect proxy                      /*{{{*/
1360 // ---------------------------------------------------------------------
1361 /* */
1362 bool HttpMethod::AutoDetectProxy()
1363 {
1364    if (AutoDetectProxyCmd.empty())
1365       return true;
1366
1367    if (Debug)
1368       clog << "Using auto proxy detect command: " << AutoDetectProxyCmd << endl;
1369
1370    int Pipes[2] = {-1,-1};
1371    if (pipe(Pipes) != 0)
1372       return _error->Errno("pipe", "Failed to create Pipe");
1373
1374    pid_t Process = ExecFork();
1375    if (Process == 0)
1376    {
1377       close(Pipes[0]);
1378       dup2(Pipes[1],STDOUT_FILENO);
1379       SetCloseExec(STDOUT_FILENO,false);
1380
1381       const char *Args[2];
1382       Args[0] = AutoDetectProxyCmd.c_str();
1383       Args[1] = 0;
1384       execv(Args[0],(char **)Args);
1385       cerr << "Failed to exec method " << Args[0] << endl;
1386       _exit(100);
1387    }
1388    char buf[512];
1389    int InFd = Pipes[0];
1390    close(Pipes[1]);
1391    int res = read(InFd, buf, sizeof(buf));
1392    ExecWait(Process, "ProxyAutoDetect", true);
1393
1394    if (res < 0)
1395       return _error->Errno("read", "Failed to read");
1396    if (res == 0)
1397       return _error->Warning("ProxyAutoDetect returned no data");
1398
1399    // add trailing \0
1400    buf[res] = 0;
1401
1402    if (Debug)
1403       clog << "auto detect command returned: '" << buf << "'" << endl;
1404
1405    if (strstr(buf, "http://") == buf)
1406       _config->Set("Acquire::http::proxy", _strstrip(buf));
1407
1408    return true;
1409 }
1410                                                                         /*}}}*/
1411
1412