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       else if (Size == 0)
 608          HaveContent = false;
 609       return true;
 610    }
 611
 612    if (stringcasecmp(Tag,"Content-Type:") == 0)
 613    {
 614       HaveContent = true;
 615       return true;
 616    }
 617
 618    if (stringcasecmp(Tag,"Content-Range:") == 0)
 619    {
 620       HaveContent = true;
 621
 622       // §14.16 says 'byte-range-resp-spec' should be a '*' in case of 416
 623       if (Result == 416 && sscanf(Val.c_str(), "bytes */%llu",&Size) == 1)
 624       {
 625          StartPos = 1; // ignore Content-Length, it would override Size
 626          HaveContent = false;
 627       }
 628       else if (sscanf(Val.c_str(),"bytes %llu-%*u/%llu",&StartPos,&Size) != 2)
 629          return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
 630       if ((unsigned long long)StartPos > Size)
 631          return _error->Error(_("This HTTP server has broken range support"));
 632       return true;
 633    }
 634
 635    if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
 636    {
 637       HaveContent = true;
 638       if (stringcasecmp(Val,"chunked") == 0)
 639          Encoding = Chunked;
 640       return true;
 641    }
 642
 643    if (stringcasecmp(Tag,"Connection:") == 0)
 644    {
 645       if (stringcasecmp(Val,"close") == 0)
 646          Persistent = false;
 647       if (stringcasecmp(Val,"keep-alive") == 0)
 648          Persistent = true;
 649       return true;
 650    }
 651
 652    if (stringcasecmp(Tag,"Last-Modified:") == 0)
 653    {
 654       if (RFC1123StrToTime(Val.c_str(), Date) == false)
 655          return _error->Error(_("Unknown date format"));
 656       return true;
 657    }
 658
 659    if (stringcasecmp(Tag,"Location:") == 0)
 660    {
 661       Location = Val;
 662       return true;
 663    }
 664
 665    return true;
 666 }
 667                                                                         /*}}}*/
 668
 669 // HttpMethod::SendReq - Send the HTTP request                          /*{{{*/
 670 // ---------------------------------------------------------------------
 671 /* This places the http request in the outbound buffer */
 672 void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
 673 {
 674    URI Uri = Itm->Uri;
 675
 676    // The HTTP server expects a hostname with a trailing :port
 677    char Buf[1000];
 678    string ProperHost;
 679
 680    if (Uri.Host.find(':') != string::npos)
 681       ProperHost = '[' + Uri.Host + ']';
 682    else
 683       ProperHost = Uri.Host;
 684    if (Uri.Port != 0)
 685    {
 686       sprintf(Buf,":%u",Uri.Port);
 687       ProperHost += Buf;
 688    }
 689
 690    // Just in case.
 691    if (Itm->Uri.length() >= sizeof(Buf))
 692        abort();
 693
 694    /* RFC 2616 §5.1.2 requires absolute URIs for requests to proxies,
 695       but while its a must for all servers to accept absolute URIs,
 696       it is assumed clients will sent an absolute path for non-proxies */
 697    std::string requesturi;
 698    if (Proxy.empty() == true || Proxy.Host.empty())
 699       requesturi = Uri.Path;
 700    else
 701       requesturi = Itm->Uri;
 702
 703    // The "+" is encoded as a workaround for a amazon S3 bug
 704    // see LP bugs #1003633 and #1086997.
 705    requesturi = QuoteString(requesturi, "+~ ");
 706
 707    /* Build the request. No keep-alive is included as it is the default
 708       in 1.1, can cause problems with proxies, and we are an HTTP/1.1
 709       client anyway.
 710       C.f. https://tools.ietf.org/wg/httpbis/trac/ticket/158 */
 711    sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 712            requesturi.c_str(),ProperHost.c_str());
 713
 714    // generate a cache control header (if needed)
 715    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 716    {
 717       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 718    }
 719    else
 720    {
 721       if (Itm->IndexFile == true)
 722       {
 723          sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 724                  _config->FindI("Acquire::http::Max-Age",0));
 725       }
 726       else
 727       {
 728          if (_config->FindB("Acquire::http::No-Store",false) == true)
 729             strcat(Buf,"Cache-Control: no-store\r\n");
 730       }
 731    }
 732
 733    // If we ask for uncompressed files servers might respond with content-
 734    // negotation which lets us end up with compressed files we do not support,
 735    // see 657029, 657560 and co, so if we have no extension on the request
 736    // ask for text only. As a sidenote: If there is nothing to negotate servers
 737    // seem to be nice and ignore it.
 738    if (_config->FindB("Acquire::http::SendAccept", true) == true)
 739    {
 740       size_t const filepos = Itm->Uri.find_last_of('/');
 741       string const file = Itm->Uri.substr(filepos + 1);
 742       if (flExtension(file) == file)
 743          strcat(Buf,"Accept: text/*\r\n");
 744    }
 745
 746    string Req = Buf;
 747
 748    // Check for a partial file
 749    struct stat SBuf;
 750    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 751    {
 752       // In this case we send an if-range query with a range header
 753       sprintf(Buf,"Range: bytes=%lli-\r\nIf-Range: %s\r\n",(long long)SBuf.st_size - 1,
 754               TimeRFC1123(SBuf.st_mtime).c_str());
 755       Req += Buf;
 756    }
 757    else
 758    {
 759       if (Itm->LastModified != 0)
 760       {
 761          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 762          Req += Buf;
 763       }
 764    }
 765
 766    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 767       Req += string("Proxy-Authorization: Basic ") +
 768           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 769
 770    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 771    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 772    {
 773       Req += string("Authorization: Basic ") +
 774           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 775    }
 776    Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
 777                 "Debian APT-HTTP/1.3 (" PACKAGE_VERSION ")") + "\r\n\r\n";
 778
 779    if (Debug == true)
 780       cerr << Req << endl;
 781
 782    Out.Read(Req);
 783 }
 784                                                                         /*}}}*/
 785 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 786 // ---------------------------------------------------------------------
 787 /* This runs the select loop over the server FDs, Output file FDs and
 788    stdin. */
 789 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 790 {
 791    // Server has closed the connection
 792    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 793                                ToFile == false))
 794       return false;
 795
 796    fd_set rfds,wfds;
 797    FD_ZERO(&rfds);
 798    FD_ZERO(&wfds);
 799
 800    /* Add the server. We only send more requests if the connection will
 801       be persisting */
 802    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 803        && Srv->Persistent == true)
 804       FD_SET(Srv->ServerFd,&wfds);
 805    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 806       FD_SET(Srv->ServerFd,&rfds);
 807
 808    // Add the file
 809    int FileFD = -1;
 810    if (File != 0)
 811       FileFD = File->Fd();
 812
 813    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 814       FD_SET(FileFD,&wfds);
 815
 816    // Add stdin
 817    if (_config->FindB("Acquire::http::DependOnSTDIN", true) == true)
 818       FD_SET(STDIN_FILENO,&rfds);
 819
 820    // Figure out the max fd
 821    int MaxFd = FileFD;
 822    if (MaxFd < Srv->ServerFd)
 823       MaxFd = Srv->ServerFd;
 824
 825    // Select
 826    struct timeval tv;
 827    tv.tv_sec = TimeOut;
 828    tv.tv_usec = 0;
 829    int Res = 0;
 830    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 831    {
 832       if (errno == EINTR)
 833          return true;
 834       return _error->Errno("select",_("Select failed"));
 835    }
 836
 837    if (Res == 0)
 838    {
 839       _error->Error(_("Connection timed out"));
 840       return ServerDie(Srv);
 841    }
 842
 843    // Handle server IO
 844    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 845    {
 846       errno = 0;
 847       if (Srv->In.Read(Srv->ServerFd) == false)
 848          return ServerDie(Srv);
 849    }
 850
 851    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 852    {
 853       errno = 0;
 854       if (Srv->Out.Write(Srv->ServerFd) == false)
 855          return ServerDie(Srv);
 856    }
 857
 858    // Send data to the file
 859    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 860    {
 861       if (Srv->In.Write(FileFD) == false)
 862          return _error->Errno("write",_("Error writing to output file"));
 863    }
 864
 865    // Handle commands from APT
 866    if (FD_ISSET(STDIN_FILENO,&rfds))
 867    {
 868       if (Run(true) != -1)
 869          exit(100);
 870    }
 871
 872    return true;
 873 }
 874                                                                         /*}}}*/
 875 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 876 // ---------------------------------------------------------------------
 877 /* This takes the current input buffer from the Server FD and writes it
 878    into the file */
 879 bool HttpMethod::Flush(ServerState *Srv)
 880 {
 881    if (File != 0)
 882    {
 883       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 884       // can't be set
 885       if (File->Name() != "/dev/null")
 886          SetNonBlock(File->Fd(),false);
 887       if (Srv->In.WriteSpace() == false)
 888          return true;
 889
 890       while (Srv->In.WriteSpace() == true)
 891       {
 892          if (Srv->In.Write(File->Fd()) == false)
 893             return _error->Errno("write",_("Error writing to file"));
 894          if (Srv->In.IsLimit() == true)
 895             return true;
 896       }
 897
 898       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 899          return true;
 900    }
 901    return false;
 902 }
 903                                                                         /*}}}*/
 904 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 905 // ---------------------------------------------------------------------
 906 /* */
 907 bool HttpMethod::ServerDie(ServerState *Srv)
 908 {
 909    unsigned int LErrno = errno;
 910
 911    // Dump the buffer to the file
 912    if (Srv->State == ServerState::Data)
 913    {
 914       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 915       // can't be set
 916       if (File->Name() != "/dev/null")
 917          SetNonBlock(File->Fd(),false);
 918       while (Srv->In.WriteSpace() == true)
 919       {
 920          if (Srv->In.Write(File->Fd()) == false)
 921             return _error->Errno("write",_("Error writing to the file"));
 922
 923          // Done
 924          if (Srv->In.IsLimit() == true)
 925             return true;
 926       }
 927    }
 928
 929    // See if this is because the server finished the data stream
 930    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 931        Srv->Encoding != ServerState::Closes)
 932    {
 933       Srv->Close();
 934       if (LErrno == 0)
 935          return _error->Error(_("Error reading from server. Remote end closed connection"));
 936       errno = LErrno;
 937       return _error->Errno("read",_("Error reading from server"));
 938    }
 939    else
 940    {
 941       Srv->In.Limit(-1);
 942
 943       // Nothing left in the buffer
 944       if (Srv->In.WriteSpace() == false)
 945          return false;
 946
 947       // We may have got multiple responses back in one packet..
 948       Srv->Close();
 949       return true;
 950    }
 951
 952    return false;
 953 }
 954                                                                         /*}}}*/
 955 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 956 // ---------------------------------------------------------------------
 957 /* We look at the header data we got back from the server and decide what
 958    to do. Returns DealWithHeadersResult (see http.h for details).
 959  */
 960 HttpMethod::DealWithHeadersResult
 961 HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 962 {
 963    // Not Modified
 964    if (Srv->Result == 304)
 965    {
 966       unlink(Queue->DestFile.c_str());
 967       Res.IMSHit = true;
 968       Res.LastModified = Queue->LastModified;
 969       return IMS_HIT;
 970    }
 971
 972    /* Redirect
 973     *
 974     * Note that it is only OK for us to treat all redirection the same
 975     * because we *always* use GET, not other HTTP methods.  There are
 976     * three redirection codes for which it is not appropriate that we
 977     * redirect.  Pass on those codes so the error handling kicks in.
 978     */
 979    if (AllowRedirect
 980        && (Srv->Result > 300 && Srv->Result < 400)
 981        && (Srv->Result != 300       // Multiple Choices
 982            && Srv->Result != 304    // Not Modified
 983            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 984    {
 985       if (Srv->Location.empty() == true);
 986       else if (Srv->Location[0] == '/' && Queue->Uri.empty() == false)
 987       {
 988          URI Uri = Queue->Uri;
 989          if (Uri.Host.empty() == false)
 990             NextURI = URI::SiteOnly(Uri);
 991          else
 992             NextURI.clear();
 993          NextURI.append(DeQuoteString(Srv->Location));
 994          return TRY_AGAIN_OR_REDIRECT;
 995       }
 996       else
 997       {
 998          NextURI = DeQuoteString(Srv->Location);
 999          URI tmpURI = NextURI;
1000          // Do not allow a redirection to switch protocol
1001          if (tmpURI.Access == "http")
1002             return TRY_AGAIN_OR_REDIRECT;
1003       }
1004       /* else pass through for error message */
1005    }
1006    // retry after an invalid range response without partial data
1007    else if (Srv->Result == 416 && FileExists(Queue->DestFile) == true &&
1008          unlink(Queue->DestFile.c_str()) == 0)
1009    {
1010       NextURI = Queue->Uri;
1011       return TRY_AGAIN_OR_REDIRECT;
1012    }
1013
1014    /* We have a reply we dont handle. This should indicate a perm server
1015       failure */
1016    if (Srv->Result < 200 || Srv->Result >= 300)
1017    {
1018       char err[255];
1019       snprintf(err,sizeof(err)-1,"HttpError%i",Srv->Result);
1020       SetFailReason(err);
1021       _error->Error("%u %s",Srv->Result,Srv->Code);
1022       if (Srv->HaveContent == true)
1023          return ERROR_WITH_CONTENT_PAGE;
1024       return ERROR_UNRECOVERABLE;
1025    }
1026
1027    // This is some sort of 2xx 'data follows' reply
1028    Res.LastModified = Srv->Date;
1029    Res.Size = Srv->Size;
1030
1031    // Open the file
1032    delete File;
1033    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
1034    if (_error->PendingError() == true)
1035       return ERROR_NOT_FROM_SERVER;
1036
1037    FailFile = Queue->DestFile;
1038    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
1039    FailFd = File->Fd();
1040    FailTime = Srv->Date;
1041
1042    delete Srv->In.Hash;
1043    Srv->In.Hash = new Hashes;
1044
1045    // Set the expected size and read file for the hashes
1046    if (Srv->StartPos >= 0)
1047    {
1048       Res.ResumePoint = Srv->StartPos;
1049       File->Truncate(Srv->StartPos);
1050
1051       if (Srv->In.Hash->AddFD(*File,Srv->StartPos) == false)
1052       {
1053          _error->Errno("read",_("Problem hashing file"));
1054          return ERROR_NOT_FROM_SERVER;
1055       }
1056    }
1057
1058    SetNonBlock(File->Fd(),true);
1059    return FILE_IS_OPEN;
1060 }
1061                                                                         /*}}}*/
1062 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1063 // ---------------------------------------------------------------------
1064 /* This closes and timestamps the open file. This is neccessary to get
1065    resume behavoir on user abort */
1066 void HttpMethod::SigTerm(int)
1067 {
1068    if (FailFd == -1)
1069       _exit(100);
1070    close(FailFd);
1071
1072    // Timestamp
1073    struct utimbuf UBuf;
1074    UBuf.actime = FailTime;
1075    UBuf.modtime = FailTime;
1076    utime(FailFile.c_str(),&UBuf);
1077
1078    _exit(100);
1079 }
1080                                                                         /*}}}*/
1081 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1082 // ---------------------------------------------------------------------
1083 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1084    depth. */
1085 bool HttpMethod::Fetch(FetchItem *)
1086 {
1087    if (Server == 0)
1088       return true;
1089
1090    // Queue the requests
1091    int Depth = -1;
1092    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1093         I = I->Next, Depth++)
1094    {
1095       // If pipelining is disabled, we only queue 1 request
1096       if (Server->Pipeline == false && Depth >= 0)
1097          break;
1098
1099       // Make sure we stick with the same server
1100       if (Server->Comp(I->Uri) == false)
1101          break;
1102       if (QueueBack == I)
1103       {
1104          QueueBack = I->Next;
1105          SendReq(I,Server->Out);
1106          continue;
1107       }
1108    }
1109
1110    return true;
1111 };
1112                                                                         /*}}}*/
1113 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1114 // ---------------------------------------------------------------------
1115 /* We stash the desired pipeline depth */
1116 bool HttpMethod::Configuration(string Message)
1117 {
1118    if (pkgAcqMethod::Configuration(Message) == false)
1119       return false;
1120
1121    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1122    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1123    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1124                                   PipelineDepth);
1125    Debug = _config->FindB("Debug::Acquire::http",false);
1126    AutoDetectProxyCmd = _config->Find("Acquire::http::ProxyAutoDetect");
1127
1128    // Get the proxy to use
1129    AutoDetectProxy();
1130
1131    return true;
1132 }
1133                                                                         /*}}}*/
1134 // HttpMethod::Loop - Main loop                                         /*{{{*/
1135 // ---------------------------------------------------------------------
1136 /* */
1137 int HttpMethod::Loop()
1138 {
1139    typedef vector<string> StringVector;
1140    typedef vector<string>::iterator StringVectorIterator;
1141    map<string, StringVector> Redirected;
1142
1143    signal(SIGTERM,SigTerm);
1144    signal(SIGINT,SigTerm);
1145
1146    Server = 0;
1147
1148    int FailCounter = 0;
1149    while (1)
1150    {
1151       // We have no commands, wait for some to arrive
1152       if (Queue == 0)
1153       {
1154          if (WaitFd(STDIN_FILENO) == false)
1155             return 0;
1156       }
1157
1158       /* Run messages, we can accept 0 (no message) if we didn't
1159          do a WaitFd above.. Otherwise the FD is closed. */
1160       int Result = Run(true);
1161       if (Result != -1 && (Result != 0 || Queue == 0))
1162       {
1163          if(FailReason.empty() == false ||
1164             _config->FindB("Acquire::http::DependOnSTDIN", true) == true)
1165             return 100;
1166          else
1167             return 0;
1168       }
1169
1170       if (Queue == 0)
1171          continue;
1172
1173       // Connect to the server
1174       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1175       {
1176          delete Server;
1177          Server = new ServerState(Queue->Uri,this);
1178       }
1179       /* If the server has explicitly said this is the last connection
1180          then we pre-emptively shut down the pipeline and tear down
1181          the connection. This will speed up HTTP/1.0 servers a tad
1182          since we don't have to wait for the close sequence to
1183          complete */
1184       if (Server->Persistent == false)
1185          Server->Close();
1186
1187       // Reset the pipeline
1188       if (Server->ServerFd == -1)
1189          QueueBack = Queue;
1190
1191       // Connnect to the host
1192       if (Server->Open() == false)
1193       {
1194          Fail(true);
1195          delete Server;
1196          Server = 0;
1197          continue;
1198       }
1199
1200       // Fill the pipeline.
1201       Fetch(0);
1202
1203       // Fetch the next URL header data from the server.
1204       switch (Server->RunHeaders())
1205       {
1206          case ServerState::RUN_HEADERS_OK:
1207          break;
1208
1209          // The header data is bad
1210          case ServerState::RUN_HEADERS_PARSE_ERROR:
1211          {
1212             _error->Error(_("Bad header data"));
1213             Fail(true);
1214             RotateDNS();
1215             continue;
1216          }
1217
1218          // The server closed a connection during the header get..
1219          default:
1220          case ServerState::RUN_HEADERS_IO_ERROR:
1221          {
1222             FailCounter++;
1223             _error->Discard();
1224             Server->Close();
1225             Server->Pipeline = false;
1226
1227             if (FailCounter >= 2)
1228             {
1229                Fail(_("Connection failed"),true);
1230                FailCounter = 0;
1231             }
1232
1233             RotateDNS();
1234             continue;
1235          }
1236       };
1237
1238       // Decide what to do.
1239       FetchResult Res;
1240       Res.Filename = Queue->DestFile;
1241       switch (DealWithHeaders(Res,Server))
1242       {
1243          // Ok, the file is Open
1244          case FILE_IS_OPEN:
1245          {
1246             URIStart(Res);
1247
1248             // Run the data
1249             bool Result =  Server->RunData();
1250
1251             /* If the server is sending back sizeless responses then fill in
1252                the size now */
1253             if (Res.Size == 0)
1254                Res.Size = File->Size();
1255
1256             // Close the file, destroy the FD object and timestamp it
1257             FailFd = -1;
1258             delete File;
1259             File = 0;
1260
1261             // Timestamp
1262             struct utimbuf UBuf;
1263             time(&UBuf.actime);
1264             UBuf.actime = Server->Date;
1265             UBuf.modtime = Server->Date;
1266             utime(Queue->DestFile.c_str(),&UBuf);
1267
1268             // Send status to APT
1269             if (Result == true)
1270             {
1271                Res.TakeHashes(*Server->In.Hash);
1272                URIDone(Res);
1273             }
1274             else
1275             {
1276                if (Server->ServerFd == -1)
1277                {
1278                   FailCounter++;
1279                   _error->Discard();
1280                   Server->Close();
1281
1282                   if (FailCounter >= 2)
1283                   {
1284                      Fail(_("Connection failed"),true);
1285                      FailCounter = 0;
1286                   }
1287
1288                   QueueBack = Queue;
1289                }
1290                else
1291                   Fail(true);
1292             }
1293             break;
1294          }
1295
1296          // IMS hit
1297          case IMS_HIT:
1298          {
1299             URIDone(Res);
1300             break;
1301          }
1302
1303          // Hard server error, not found or something
1304          case ERROR_UNRECOVERABLE:
1305          {
1306             Fail();
1307             break;
1308          }
1309
1310          // Hard internal error, kill the connection and fail
1311          case ERROR_NOT_FROM_SERVER:
1312          {
1313             delete File;
1314             File = 0;
1315
1316             Fail();
1317             RotateDNS();
1318             Server->Close();
1319             break;
1320          }
1321
1322          // We need to flush the data, the header is like a 404 w/ error text
1323          case ERROR_WITH_CONTENT_PAGE:
1324          {
1325             Fail();
1326
1327             // Send to content to dev/null
1328             File = new FileFd("/dev/null",FileFd::WriteExists);
1329             Server->RunData();
1330             delete File;
1331             File = 0;
1332             break;
1333          }
1334
1335          // Try again with a new URL
1336          case TRY_AGAIN_OR_REDIRECT:
1337          {
1338             // Clear rest of response if there is content
1339             if (Server->HaveContent)
1340             {
1341                File = new FileFd("/dev/null",FileFd::WriteExists);
1342                Server->RunData();
1343                delete File;
1344                File = 0;
1345             }
1346
1347             /* Detect redirect loops.  No more redirects are allowed
1348                after the same URI is seen twice in a queue item. */
1349             StringVector &R = Redirected[Queue->DestFile];
1350             bool StopRedirects = false;
1351             if (R.empty() == true)
1352                R.push_back(Queue->Uri);
1353             else if (R[0] == "STOP" || R.size() > 10)
1354                StopRedirects = true;
1355             else
1356             {
1357                for (StringVectorIterator I = R.begin(); I != R.end(); ++I)
1358                   if (Queue->Uri == *I)
1359                   {
1360                      R[0] = "STOP";
1361                      break;
1362                   }
1363
1364                R.push_back(Queue->Uri);
1365             }
1366
1367             if (StopRedirects == false)
1368                Redirect(NextURI);
1369             else
1370                Fail();
1371
1372             break;
1373          }
1374
1375          default:
1376          Fail(_("Internal error"));
1377          break;
1378       }
1379
1380       FailCounter = 0;
1381    }
1382
1383    return 0;
1384 }
1385                                                                         /*}}}*/
1386 // HttpMethod::AutoDetectProxy - auto detect proxy                      /*{{{*/
1387 // ---------------------------------------------------------------------
1388 /* */
1389 bool HttpMethod::AutoDetectProxy()
1390 {
1391    if (AutoDetectProxyCmd.empty())
1392       return true;
1393
1394    if (Debug)
1395       clog << "Using auto proxy detect command: " << AutoDetectProxyCmd << endl;
1396
1397    int Pipes[2] = {-1,-1};
1398    if (pipe(Pipes) != 0)
1399       return _error->Errno("pipe", "Failed to create Pipe");
1400
1401    pid_t Process = ExecFork();
1402    if (Process == 0)
1403    {
1404       close(Pipes[0]);
1405       dup2(Pipes[1],STDOUT_FILENO);
1406       SetCloseExec(STDOUT_FILENO,false);
1407
1408       const char *Args[2];
1409       Args[0] = AutoDetectProxyCmd.c_str();
1410       Args[1] = 0;
1411       execv(Args[0],(char **)Args);
1412       cerr << "Failed to exec method " << Args[0] << endl;
1413       _exit(100);
1414    }
1415    char buf[512];
1416    int InFd = Pipes[0];
1417    close(Pipes[1]);
1418    int res = read(InFd, buf, sizeof(buf)-1);
1419    ExecWait(Process, "ProxyAutoDetect", true);
1420
1421    if (res < 0)
1422       return _error->Errno("read", "Failed to read");
1423    if (res == 0)
1424       return _error->Warning("ProxyAutoDetect returned no data");
1425
1426    // add trailing \0
1427    buf[res] = 0;
1428
1429    if (Debug)
1430       clog << "auto detect command returned: '" << buf << "'" << endl;
1431
1432    if (strstr(buf, "http://") == buf)
1433       _config->Set("Acquire::http::proxy", _strstrip(buf));
1434
1435    return true;
1436 }
1437                                                                         /*}}}*/
1438
1439