methods/http.cc

   1 // -*- mode: cpp; mode: fold -*-
   2 // Description                                                          /*{{{*/
   3 // $Id: http.cc,v 1.59 2004/05/08 19:42:35 mdz Exp $
   4 /* ######################################################################
   5
   6    HTTP Acquire Method - This is the HTTP aquire method for APT.
   7
   8    It uses HTTP/1.1 and many of the fancy options there-in, such as
   9    pipelining, range, if-range and so on.
  10
  11    It is based on a doubly buffered select loop. A groupe of requests are
  12    fed into a single output buffer that is constantly fed out the
  13    socket. This provides ideal pipelining as in many cases all of the
  14    requests will fit into a single packet. The input socket is buffered
  15    the same way and fed into the fd for the file (may be a pipe in future).
  16
  17    This double buffering provides fairly substantial transfer rates,
  18    compared to wget the http method is about 4% faster. Most importantly,
  19    when HTTP is compared with FTP as a protocol the speed difference is
  20    huge. In tests over the internet from two sites to llug (via ATM) this
  21    program got 230k/s sustained http transfer rates. FTP on the other
  22    hand topped out at 170k/s. That combined with the time to setup the
  23    FTP connection makes HTTP a vastly superior protocol.
  24
  25    ##################################################################### */
  26                                                                         /*}}}*/
  27 // Include Files                                                        /*{{{*/
  28 #include <config.h>
  29
  30 #include <apt-pkg/fileutl.h>
  31 #include <apt-pkg/acquire-method.h>
  32 #include <apt-pkg/configuration.h>
  33 #include <apt-pkg/error.h>
  34 #include <apt-pkg/hashes.h>
  35 #include <apt-pkg/netrc.h>
  36
  37 #include <sys/stat.h>
  38 #include <sys/time.h>
  39 #include <utime.h>
  40 #include <unistd.h>
  41 #include <signal.h>
  42 #include <stdio.h>
  43 #include <errno.h>
  44 #include <string.h>
  45 #include <iostream>
  46 #include <map>
  47
  48 // Internet stuff
  49 #include <netdb.h>
  50
  51 #include "config.h"
  52 #include "connect.h"
  53 #include "rfc2553emu.h"
  54 #include "http.h"
  55
  56 #include <apti18n.h>
  57                                                                         /*}}}*/
  58 using namespace std;
  59
  60 string HttpMethod::FailFile;
  61 int HttpMethod::FailFd = -1;
  62 time_t HttpMethod::FailTime = 0;
  63 unsigned long PipelineDepth = 10;
  64 unsigned long TimeOut = 120;
  65 bool AllowRedirect = false;
  66 bool Debug = false;
  67 URI Proxy;
  68
  69 unsigned long long CircleBuf::BwReadLimit=0;
  70 unsigned long long CircleBuf::BwTickReadData=0;
  71 struct timeval CircleBuf::BwReadTick={0,0};
  72 const unsigned int CircleBuf::BW_HZ=10;
  73
  74 // CircleBuf::CircleBuf - Circular input buffer                         /*{{{*/
  75 // ---------------------------------------------------------------------
  76 /* */
  77 CircleBuf::CircleBuf(unsigned long long Size) : Size(Size), Hash(0)
  78 {
  79    Buf = new unsigned char[Size];
  80    Reset();
  81
  82    CircleBuf::BwReadLimit = _config->FindI("Acquire::http::Dl-Limit",0)*1024;
  83 }
  84                                                                         /*}}}*/
  85 // CircleBuf::Reset - Reset to the default state                        /*{{{*/
  86 // ---------------------------------------------------------------------
  87 /* */
  88 void CircleBuf::Reset()
  89 {
  90    InP = 0;
  91    OutP = 0;
  92    StrPos = 0;
  93    MaxGet = (unsigned long long)-1;
  94    OutQueue = string();
  95    if (Hash != 0)
  96    {
  97       delete Hash;
  98       Hash = new Hashes;
  99    }
 100 };
 101                                                                         /*}}}*/
 102 // CircleBuf::Read - Read from a FD into the circular buffer            /*{{{*/
 103 // ---------------------------------------------------------------------
 104 /* This fills up the buffer with as much data as is in the FD, assuming it
 105    is non-blocking.. */
 106 bool CircleBuf::Read(int Fd)
 107 {
 108    unsigned long long BwReadMax;
 109
 110    while (1)
 111    {
 112       // Woops, buffer is full
 113       if (InP - OutP == Size)
 114          return true;
 115
 116       // what's left to read in this tick
 117       BwReadMax = CircleBuf::BwReadLimit/BW_HZ;
 118
 119       if(CircleBuf::BwReadLimit) {
 120          struct timeval now;
 121          gettimeofday(&now,0);
 122
 123          unsigned long long d = (now.tv_sec-CircleBuf::BwReadTick.tv_sec)*1000000 +
 124             now.tv_usec-CircleBuf::BwReadTick.tv_usec;
 125          if(d > 1000000/BW_HZ) {
 126             CircleBuf::BwReadTick = now;
 127             CircleBuf::BwTickReadData = 0;
 128          }
 129
 130          if(CircleBuf::BwTickReadData >= BwReadMax) {
 131             usleep(1000000/BW_HZ);
 132             return true;
 133          }
 134       }
 135
 136       // Write the buffer segment
 137       ssize_t Res;
 138       if(CircleBuf::BwReadLimit) {
 139          Res = read(Fd,Buf + (InP%Size),
 140                     BwReadMax > LeftRead() ? LeftRead() : BwReadMax);
 141       } else
 142          Res = read(Fd,Buf + (InP%Size),LeftRead());
 143
 144       if(Res > 0 && BwReadLimit > 0)
 145          CircleBuf::BwTickReadData += Res;
 146
 147       if (Res == 0)
 148          return false;
 149       if (Res < 0)
 150       {
 151          if (errno == EAGAIN)
 152             return true;
 153          return false;
 154       }
 155
 156       if (InP == 0)
 157          gettimeofday(&Start,0);
 158       InP += Res;
 159    }
 160 }
 161                                                                         /*}}}*/
 162 // CircleBuf::Read - Put the string into the buffer                     /*{{{*/
 163 // ---------------------------------------------------------------------
 164 /* This will hold the string in and fill the buffer with it as it empties */
 165 bool CircleBuf::Read(string Data)
 166 {
 167    OutQueue += Data;
 168    FillOut();
 169    return true;
 170 }
 171                                                                         /*}}}*/
 172 // CircleBuf::FillOut - Fill the buffer from the output queue           /*{{{*/
 173 // ---------------------------------------------------------------------
 174 /* */
 175 void CircleBuf::FillOut()
 176 {
 177    if (OutQueue.empty() == true)
 178       return;
 179    while (1)
 180    {
 181       // Woops, buffer is full
 182       if (InP - OutP == Size)
 183          return;
 184
 185       // Write the buffer segment
 186       unsigned long long Sz = LeftRead();
 187       if (OutQueue.length() - StrPos < Sz)
 188          Sz = OutQueue.length() - StrPos;
 189       memcpy(Buf + (InP%Size),OutQueue.c_str() + StrPos,Sz);
 190
 191       // Advance
 192       StrPos += Sz;
 193       InP += Sz;
 194       if (OutQueue.length() == StrPos)
 195       {
 196          StrPos = 0;
 197          OutQueue = "";
 198          return;
 199       }
 200    }
 201 }
 202                                                                         /*}}}*/
 203 // CircleBuf::Write - Write from the buffer into a FD                   /*{{{*/
 204 // ---------------------------------------------------------------------
 205 /* This empties the buffer into the FD. */
 206 bool CircleBuf::Write(int Fd)
 207 {
 208    while (1)
 209    {
 210       FillOut();
 211
 212       // Woops, buffer is empty
 213       if (OutP == InP)
 214          return true;
 215
 216       if (OutP == MaxGet)
 217          return true;
 218
 219       // Write the buffer segment
 220       ssize_t Res;
 221       Res = write(Fd,Buf + (OutP%Size),LeftWrite());
 222
 223       if (Res == 0)
 224          return false;
 225       if (Res < 0)
 226       {
 227          if (errno == EAGAIN)
 228             return true;
 229
 230          return false;
 231       }
 232
 233       if (Hash != 0)
 234          Hash->Add(Buf + (OutP%Size),Res);
 235
 236       OutP += Res;
 237    }
 238 }
 239                                                                         /*}}}*/
 240 // CircleBuf::WriteTillEl - Write from the buffer to a string           /*{{{*/
 241 // ---------------------------------------------------------------------
 242 /* This copies till the first empty line */
 243 bool CircleBuf::WriteTillEl(string &Data,bool Single)
 244 {
 245    // We cheat and assume it is unneeded to have more than one buffer load
 246    for (unsigned long long I = OutP; I < InP; I++)
 247    {
 248       if (Buf[I%Size] != '\n')
 249          continue;
 250       ++I;
 251
 252       if (Single == false)
 253       {
 254          if (I < InP  && Buf[I%Size] == '\r')
 255             ++I;
 256          if (I >= InP || Buf[I%Size] != '\n')
 257             continue;
 258          ++I;
 259       }
 260
 261       Data = "";
 262       while (OutP < I)
 263       {
 264          unsigned long long Sz = LeftWrite();
 265          if (Sz == 0)
 266             return false;
 267          if (I - OutP < Sz)
 268             Sz = I - OutP;
 269          Data += string((char *)(Buf + (OutP%Size)),Sz);
 270          OutP += Sz;
 271       }
 272       return true;
 273    }
 274    return false;
 275 }
 276                                                                         /*}}}*/
 277 // CircleBuf::Stats - Print out stats information                       /*{{{*/
 278 // ---------------------------------------------------------------------
 279 /* */
 280 void CircleBuf::Stats()
 281 {
 282    if (InP == 0)
 283       return;
 284
 285    struct timeval Stop;
 286    gettimeofday(&Stop,0);
 287 /*   float Diff = Stop.tv_sec - Start.tv_sec +
 288              (float)(Stop.tv_usec - Start.tv_usec)/1000000;
 289    clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
 290 }
 291                                                                         /*}}}*/
 292 CircleBuf::~CircleBuf()
 293 {
 294    delete [] Buf;
 295    delete Hash;
 296 }
 297
 298 // ServerState::ServerState - Constructor                               /*{{{*/
 299 // ---------------------------------------------------------------------
 300 /* */
 301 ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner),
 302                         In(64*1024), Out(4*1024),
 303                         ServerName(Srv)
 304 {
 305    Reset();
 306 }
 307                                                                         /*}}}*/
 308 // ServerState::Open - Open a connection to the server                  /*{{{*/
 309 // ---------------------------------------------------------------------
 310 /* This opens a connection to the server. */
 311 bool ServerState::Open()
 312 {
 313    // Use the already open connection if possible.
 314    if (ServerFd != -1)
 315       return true;
 316
 317    Close();
 318    In.Reset();
 319    Out.Reset();
 320    Persistent = true;
 321
 322    // Determine the proxy setting
 323    string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
 324    if (!SpecificProxy.empty())
 325    {
 326            if (SpecificProxy == "DIRECT")
 327                    Proxy = "";
 328            else
 329                    Proxy = SpecificProxy;
 330    }
 331    else
 332    {
 333            string DefProxy = _config->Find("Acquire::http::Proxy");
 334            if (!DefProxy.empty())
 335            {
 336                    Proxy = DefProxy;
 337            }
 338            else
 339            {
 340                    char* result = getenv("http_proxy");
 341                    Proxy = result ? result : "";
 342            }
 343    }
 344
 345    // Parse no_proxy, a , separated list of domains
 346    if (getenv("no_proxy") != 0)
 347    {
 348       if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 349          Proxy = "";
 350    }
 351
 352    // Determine what host and port to use based on the proxy settings
 353    int Port = 0;
 354    string Host;
 355    if (Proxy.empty() == true || Proxy.Host.empty() == true)
 356    {
 357       if (ServerName.Port != 0)
 358          Port = ServerName.Port;
 359       Host = ServerName.Host;
 360    }
 361    else
 362    {
 363       if (Proxy.Port != 0)
 364          Port = Proxy.Port;
 365       Host = Proxy.Host;
 366    }
 367
 368    // Connect to the remote server
 369    if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
 370       return false;
 371
 372    return true;
 373 }
 374                                                                         /*}}}*/
 375 // ServerState::Close - Close a connection to the server                /*{{{*/
 376 // ---------------------------------------------------------------------
 377 /* */
 378 bool ServerState::Close()
 379 {
 380    close(ServerFd);
 381    ServerFd = -1;
 382    return true;
 383 }
 384                                                                         /*}}}*/
 385 // ServerState::RunHeaders - Get the headers before the data            /*{{{*/
 386 // ---------------------------------------------------------------------
 387 /* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
 388    parse error occurred */
 389 ServerState::RunHeadersResult ServerState::RunHeaders()
 390 {
 391    State = Header;
 392
 393    Owner->Status(_("Waiting for headers"));
 394
 395    Major = 0;
 396    Minor = 0;
 397    Result = 0;
 398    Size = 0;
 399    StartPos = 0;
 400    Encoding = Closes;
 401    HaveContent = false;
 402    time(&Date);
 403
 404    do
 405    {
 406       string Data;
 407       if (In.WriteTillEl(Data) == false)
 408          continue;
 409
 410       if (Debug == true)
 411          clog << Data;
 412
 413       for (string::const_iterator I = Data.begin(); I < Data.end(); ++I)
 414       {
 415          string::const_iterator J = I;
 416          for (; J != Data.end() && *J != '\n' && *J != '\r'; ++J);
 417          if (HeaderLine(string(I,J)) == false)
 418             return RUN_HEADERS_PARSE_ERROR;
 419          I = J;
 420       }
 421
 422       // 100 Continue is a Nop...
 423       if (Result == 100)
 424          continue;
 425
 426       // Tidy up the connection persistance state.
 427       if (Encoding == Closes && HaveContent == true)
 428          Persistent = false;
 429
 430       return RUN_HEADERS_OK;
 431    }
 432    while (Owner->Go(false,this) == true);
 433
 434    return RUN_HEADERS_IO_ERROR;
 435 }
 436                                                                         /*}}}*/
 437 // ServerState::RunData - Transfer the data from the socket             /*{{{*/
 438 // ---------------------------------------------------------------------
 439 /* */
 440 bool ServerState::RunData()
 441 {
 442    State = Data;
 443
 444    // Chunked transfer encoding is fun..
 445    if (Encoding == Chunked)
 446    {
 447       while (1)
 448       {
 449          // Grab the block size
 450          bool Last = true;
 451          string Data;
 452          In.Limit(-1);
 453          do
 454          {
 455             if (In.WriteTillEl(Data,true) == true)
 456                break;
 457          }
 458          while ((Last = Owner->Go(false,this)) == true);
 459
 460          if (Last == false)
 461             return false;
 462
 463          // See if we are done
 464          unsigned long long Len = strtoull(Data.c_str(),0,16);
 465          if (Len == 0)
 466          {
 467             In.Limit(-1);
 468
 469             // We have to remove the entity trailer
 470             Last = true;
 471             do
 472             {
 473                if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
 474                   break;
 475             }
 476             while ((Last = Owner->Go(false,this)) == true);
 477             if (Last == false)
 478                return false;
 479             return !_error->PendingError();
 480          }
 481
 482          // Transfer the block
 483          In.Limit(Len);
 484          while (Owner->Go(true,this) == true)
 485             if (In.IsLimit() == true)
 486                break;
 487
 488          // Error
 489          if (In.IsLimit() == false)
 490             return false;
 491
 492          // The server sends an extra new line before the next block specifier..
 493          In.Limit(-1);
 494          Last = true;
 495          do
 496          {
 497             if (In.WriteTillEl(Data,true) == true)
 498                break;
 499          }
 500          while ((Last = Owner->Go(false,this)) == true);
 501          if (Last == false)
 502             return false;
 503       }
 504    }
 505    else
 506    {
 507       /* Closes encoding is used when the server did not specify a size, the
 508          loss of the connection means we are done */
 509       if (Encoding == Closes)
 510          In.Limit(-1);
 511       else
 512          In.Limit(Size - StartPos);
 513
 514       // Just transfer the whole block.
 515       do
 516       {
 517          if (In.IsLimit() == false)
 518             continue;
 519
 520          In.Limit(-1);
 521          return !_error->PendingError();
 522       }
 523       while (Owner->Go(true,this) == true);
 524    }
 525
 526    return Owner->Flush(this) && !_error->PendingError();
 527 }
 528                                                                         /*}}}*/
 529 // ServerState::HeaderLine - Process a header line                      /*{{{*/
 530 // ---------------------------------------------------------------------
 531 /* */
 532 bool ServerState::HeaderLine(string Line)
 533 {
 534    if (Line.empty() == true)
 535       return true;
 536
 537    // The http server might be trying to do something evil.
 538    if (Line.length() >= MAXLEN)
 539       return _error->Error(_("Got a single header line over %u chars"),MAXLEN);
 540
 541    string::size_type Pos = Line.find(' ');
 542    if (Pos == string::npos || Pos+1 > Line.length())
 543    {
 544       // Blah, some servers use "connection:closes", evil.
 545       Pos = Line.find(':');
 546       if (Pos == string::npos || Pos + 2 > Line.length())
 547          return _error->Error(_("Bad header line"));
 548       Pos++;
 549    }
 550
 551    // Parse off any trailing spaces between the : and the next word.
 552    string::size_type Pos2 = Pos;
 553    while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
 554       Pos2++;
 555
 556    string Tag = string(Line,0,Pos);
 557    string Val = string(Line,Pos2);
 558
 559    if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
 560    {
 561       // Evil servers return no version
 562       if (Line[4] == '/')
 563       {
 564          int const elements = sscanf(Line.c_str(),"HTTP/%u.%u %u%[^\n]",&Major,&Minor,&Result,Code);
 565          if (elements == 3)
 566          {
 567             Code[0] = '\0';
 568             if (Debug == true)
 569                clog << "HTTP server doesn't give Reason-Phrase for " << Result << std::endl;
 570          }
 571          else if (elements != 4)
 572             return _error->Error(_("The HTTP server sent an invalid reply header"));
 573       }
 574       else
 575       {
 576          Major = 0;
 577          Minor = 9;
 578          if (sscanf(Line.c_str(),"HTTP %u%[^\n]",&Result,Code) != 2)
 579             return _error->Error(_("The HTTP server sent an invalid reply header"));
 580       }
 581
 582       /* Check the HTTP response header to get the default persistance
 583          state. */
 584       if (Major < 1)
 585          Persistent = false;
 586       else
 587       {
 588          if (Major == 1 && Minor <= 0)
 589             Persistent = false;
 590          else
 591             Persistent = true;
 592       }
 593
 594       return true;
 595    }
 596
 597    if (stringcasecmp(Tag,"Content-Length:") == 0)
 598    {
 599       if (Encoding == Closes)
 600          Encoding = Stream;
 601       HaveContent = true;
 602
 603       // The length is already set from the Content-Range header
 604       if (StartPos != 0)
 605          return true;
 606
 607       if (sscanf(Val.c_str(),"%llu",&Size) != 1)
 608          return _error->Error(_("The HTTP server sent an invalid Content-Length header"));
 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       if (sscanf(Val.c_str(),"bytes %llu-%*u/%llu",&StartPos,&Size) != 2)
 623          return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
 624       if ((unsigned long long)StartPos > Size)
 625          return _error->Error(_("This HTTP server has broken range support"));
 626       return true;
 627    }
 628
 629    if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
 630    {
 631       HaveContent = true;
 632       if (stringcasecmp(Val,"chunked") == 0)
 633          Encoding = Chunked;
 634       return true;
 635    }
 636
 637    if (stringcasecmp(Tag,"Connection:") == 0)
 638    {
 639       if (stringcasecmp(Val,"close") == 0)
 640          Persistent = false;
 641       if (stringcasecmp(Val,"keep-alive") == 0)
 642          Persistent = true;
 643       return true;
 644    }
 645
 646    if (stringcasecmp(Tag,"Last-Modified:") == 0)
 647    {
 648       if (RFC1123StrToTime(Val.c_str(), Date) == false)
 649          return _error->Error(_("Unknown date format"));
 650       return true;
 651    }
 652
 653    if (stringcasecmp(Tag,"Location:") == 0)
 654    {
 655       Location = Val;
 656       return true;
 657    }
 658
 659    return true;
 660 }
 661                                                                         /*}}}*/
 662
 663 // HttpMethod::SendReq - Send the HTTP request                          /*{{{*/
 664 // ---------------------------------------------------------------------
 665 /* This places the http request in the outbound buffer */
 666 void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
 667 {
 668    URI Uri = Itm->Uri;
 669
 670    // The HTTP server expects a hostname with a trailing :port
 671    char Buf[1000];
 672    string ProperHost = Uri.Host;
 673    if (Uri.Port != 0)
 674    {
 675       sprintf(Buf,":%u",Uri.Port);
 676       ProperHost += Buf;
 677    }
 678
 679    // Just in case.
 680    if (Itm->Uri.length() >= sizeof(Buf))
 681        abort();
 682
 683    /* Build the request. We include a keep-alive header only for non-proxy
 684       requests. This is to tweak old http/1.0 servers that do support keep-alive
 685       but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
 686       will glitch HTTP/1.0 proxies because they do not filter it out and
 687       pass it on, HTTP/1.1 says the connection should default to keep alive
 688       and we expect the proxy to do this */
 689    if (Proxy.empty() == true || Proxy.Host.empty())
 690       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
 691               QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
 692    else
 693    {
 694       /* Generate a cache control header if necessary. We place a max
 695          cache age on index files, optionally set a no-cache directive
 696          and a no-store directive for archives. */
 697       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 698               Itm->Uri.c_str(),ProperHost.c_str());
 699    }
 700    // generate a cache control header (if needed)
 701    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 702    {
 703       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 704    }
 705    else
 706    {
 707       if (Itm->IndexFile == true)
 708       {
 709          sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 710                  _config->FindI("Acquire::http::Max-Age",0));
 711       }
 712       else
 713       {
 714          if (_config->FindB("Acquire::http::No-Store",false) == true)
 715             strcat(Buf,"Cache-Control: no-store\r\n");
 716       }
 717    }
 718
 719    // If we ask for uncompressed files servers might respond with content-
 720    // negotation which lets us end up with compressed files we do not support,
 721    // see 657029, 657560 and co, so if we have no extension on the request
 722    // ask for text only. As a sidenote: If there is nothing to negotate servers
 723    // seem to be nice and ignore it.
 724    if (_config->FindB("Acquire::http::SendAccept", true) == true)
 725    {
 726       size_t const filepos = Itm->Uri.find_last_of('/');
 727       string const file = Itm->Uri.substr(filepos + 1);
 728       if (flExtension(file) == file)
 729          strcat(Buf,"Accept: text/*\r\n");
 730    }
 731
 732    string Req = Buf;
 733
 734    // Check for a partial file
 735    struct stat SBuf;
 736    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 737    {
 738       // In this case we send an if-range query with a range header
 739       sprintf(Buf,"Range: bytes=%lli-\r\nIf-Range: %s\r\n",(long long)SBuf.st_size - 1,
 740               TimeRFC1123(SBuf.st_mtime).c_str());
 741       Req += Buf;
 742    }
 743    else
 744    {
 745       if (Itm->LastModified != 0)
 746       {
 747          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 748          Req += Buf;
 749       }
 750    }
 751
 752    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 753       Req += string("Proxy-Authorization: Basic ") +
 754           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 755
 756    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 757    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 758    {
 759       Req += string("Authorization: Basic ") +
 760           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 761    }
 762    Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
 763                 "Debian APT-HTTP/1.3 ("VERSION")") + "\r\n\r\n";
 764
 765    if (Debug == true)
 766       cerr << Req << endl;
 767
 768    Out.Read(Req);
 769 }
 770                                                                         /*}}}*/
 771 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 772 // ---------------------------------------------------------------------
 773 /* This runs the select loop over the server FDs, Output file FDs and
 774    stdin. */
 775 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 776 {
 777    // Server has closed the connection
 778    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 779                                ToFile == false))
 780       return false;
 781
 782    fd_set rfds,wfds;
 783    FD_ZERO(&rfds);
 784    FD_ZERO(&wfds);
 785
 786    /* Add the server. We only send more requests if the connection will
 787       be persisting */
 788    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 789        && Srv->Persistent == true)
 790       FD_SET(Srv->ServerFd,&wfds);
 791    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 792       FD_SET(Srv->ServerFd,&rfds);
 793
 794    // Add the file
 795    int FileFD = -1;
 796    if (File != 0)
 797       FileFD = File->Fd();
 798
 799    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 800       FD_SET(FileFD,&wfds);
 801
 802    // Add stdin
 803    if (_config->FindB("Acquire::http::DependOnSTDIN", true) == true)
 804       FD_SET(STDIN_FILENO,&rfds);
 805
 806    // Figure out the max fd
 807    int MaxFd = FileFD;
 808    if (MaxFd < Srv->ServerFd)
 809       MaxFd = Srv->ServerFd;
 810
 811    // Select
 812    struct timeval tv;
 813    tv.tv_sec = TimeOut;
 814    tv.tv_usec = 0;
 815    int Res = 0;
 816    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 817    {
 818       if (errno == EINTR)
 819          return true;
 820       return _error->Errno("select",_("Select failed"));
 821    }
 822
 823    if (Res == 0)
 824    {
 825       _error->Error(_("Connection timed out"));
 826       return ServerDie(Srv);
 827    }
 828
 829    // Handle server IO
 830    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 831    {
 832       errno = 0;
 833       if (Srv->In.Read(Srv->ServerFd) == false)
 834          return ServerDie(Srv);
 835    }
 836
 837    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 838    {
 839       errno = 0;
 840       if (Srv->Out.Write(Srv->ServerFd) == false)
 841          return ServerDie(Srv);
 842    }
 843
 844    // Send data to the file
 845    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 846    {
 847       if (Srv->In.Write(FileFD) == false)
 848          return _error->Errno("write",_("Error writing to output file"));
 849    }
 850
 851    // Handle commands from APT
 852    if (FD_ISSET(STDIN_FILENO,&rfds))
 853    {
 854       if (Run(true) != -1)
 855          exit(100);
 856    }
 857
 858    return true;
 859 }
 860                                                                         /*}}}*/
 861 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 862 // ---------------------------------------------------------------------
 863 /* This takes the current input buffer from the Server FD and writes it
 864    into the file */
 865 bool HttpMethod::Flush(ServerState *Srv)
 866 {
 867    if (File != 0)
 868    {
 869       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 870       // can't be set
 871       if (File->Name() != "/dev/null")
 872          SetNonBlock(File->Fd(),false);
 873       if (Srv->In.WriteSpace() == false)
 874          return true;
 875
 876       while (Srv->In.WriteSpace() == true)
 877       {
 878          if (Srv->In.Write(File->Fd()) == false)
 879             return _error->Errno("write",_("Error writing to file"));
 880          if (Srv->In.IsLimit() == true)
 881             return true;
 882       }
 883
 884       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 885          return true;
 886    }
 887    return false;
 888 }
 889                                                                         /*}}}*/
 890 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 891 // ---------------------------------------------------------------------
 892 /* */
 893 bool HttpMethod::ServerDie(ServerState *Srv)
 894 {
 895    unsigned int LErrno = errno;
 896
 897    // Dump the buffer to the file
 898    if (Srv->State == ServerState::Data)
 899    {
 900       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 901       // can't be set
 902       if (File->Name() != "/dev/null")
 903          SetNonBlock(File->Fd(),false);
 904       while (Srv->In.WriteSpace() == true)
 905       {
 906          if (Srv->In.Write(File->Fd()) == false)
 907             return _error->Errno("write",_("Error writing to the file"));
 908
 909          // Done
 910          if (Srv->In.IsLimit() == true)
 911             return true;
 912       }
 913    }
 914
 915    // See if this is because the server finished the data stream
 916    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 917        Srv->Encoding != ServerState::Closes)
 918    {
 919       Srv->Close();
 920       if (LErrno == 0)
 921          return _error->Error(_("Error reading from server. Remote end closed connection"));
 922       errno = LErrno;
 923       return _error->Errno("read",_("Error reading from server"));
 924    }
 925    else
 926    {
 927       Srv->In.Limit(-1);
 928
 929       // Nothing left in the buffer
 930       if (Srv->In.WriteSpace() == false)
 931          return false;
 932
 933       // We may have got multiple responses back in one packet..
 934       Srv->Close();
 935       return true;
 936    }
 937
 938    return false;
 939 }
 940                                                                         /*}}}*/
 941 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 942 // ---------------------------------------------------------------------
 943 /* We look at the header data we got back from the server and decide what
 944    to do. Returns DealWithHeadersResult (see http.h for details).
 945  */
 946 HttpMethod::DealWithHeadersResult
 947 HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 948 {
 949    // Not Modified
 950    if (Srv->Result == 304)
 951    {
 952       unlink(Queue->DestFile.c_str());
 953       Res.IMSHit = true;
 954       Res.LastModified = Queue->LastModified;
 955       return IMS_HIT;
 956    }
 957
 958    /* Redirect
 959     *
 960     * Note that it is only OK for us to treat all redirection the same
 961     * because we *always* use GET, not other HTTP methods.  There are
 962     * three redirection codes for which it is not appropriate that we
 963     * redirect.  Pass on those codes so the error handling kicks in.
 964     */
 965    if (AllowRedirect
 966        && (Srv->Result > 300 && Srv->Result < 400)
 967        && (Srv->Result != 300       // Multiple Choices
 968            && Srv->Result != 304    // Not Modified
 969            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 970    {
 971       if (Srv->Location.empty() == true);
 972       else if (Srv->Location[0] == '/' && Queue->Uri.empty() == false)
 973       {
 974          URI Uri = Queue->Uri;
 975          if (Uri.Host.empty() == false)
 976          {
 977             if (Uri.Port != 0)
 978                strprintf(NextURI, "http://%s:%u", Uri.Host.c_str(), Uri.Port);
 979             else
 980                NextURI = "http://" + Uri.Host;
 981          }
 982          else
 983             NextURI.clear();
 984          NextURI.append(DeQuoteString(Srv->Location));
 985          return TRY_AGAIN_OR_REDIRECT;
 986       }
 987       else
 988       {
 989          NextURI = DeQuoteString(Srv->Location);
 990          return TRY_AGAIN_OR_REDIRECT;
 991       }
 992       /* else pass through for error message */
 993    }
 994
 995    /* We have a reply we dont handle. This should indicate a perm server
 996       failure */
 997    if (Srv->Result < 200 || Srv->Result >= 300)
 998    {
 999       char err[255];
1000       snprintf(err,sizeof(err)-1,"HttpError%i",Srv->Result);
1001       SetFailReason(err);
1002       _error->Error("%u %s",Srv->Result,Srv->Code);
1003       if (Srv->HaveContent == true)
1004          return ERROR_WITH_CONTENT_PAGE;
1005       return ERROR_UNRECOVERABLE;
1006    }
1007
1008    // This is some sort of 2xx 'data follows' reply
1009    Res.LastModified = Srv->Date;
1010    Res.Size = Srv->Size;
1011
1012    // Open the file
1013    delete File;
1014    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
1015    if (_error->PendingError() == true)
1016       return ERROR_NOT_FROM_SERVER;
1017
1018    FailFile = Queue->DestFile;
1019    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
1020    FailFd = File->Fd();
1021    FailTime = Srv->Date;
1022
1023    delete Srv->In.Hash;
1024    Srv->In.Hash = new Hashes;
1025
1026    // Set the expected size and read file for the hashes
1027    if (Srv->StartPos >= 0)
1028    {
1029       Res.ResumePoint = Srv->StartPos;
1030       File->Truncate(Srv->StartPos);
1031
1032       if (Srv->In.Hash->AddFD(*File,Srv->StartPos) == false)
1033       {
1034          _error->Errno("read",_("Problem hashing file"));
1035          return ERROR_NOT_FROM_SERVER;
1036       }
1037    }
1038
1039    SetNonBlock(File->Fd(),true);
1040    return FILE_IS_OPEN;
1041 }
1042                                                                         /*}}}*/
1043 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1044 // ---------------------------------------------------------------------
1045 /* This closes and timestamps the open file. This is neccessary to get
1046    resume behavoir on user abort */
1047 void HttpMethod::SigTerm(int)
1048 {
1049    if (FailFd == -1)
1050       _exit(100);
1051    close(FailFd);
1052
1053    // Timestamp
1054    struct utimbuf UBuf;
1055    UBuf.actime = FailTime;
1056    UBuf.modtime = FailTime;
1057    utime(FailFile.c_str(),&UBuf);
1058
1059    _exit(100);
1060 }
1061                                                                         /*}}}*/
1062 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1063 // ---------------------------------------------------------------------
1064 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1065    depth. */
1066 bool HttpMethod::Fetch(FetchItem *)
1067 {
1068    if (Server == 0)
1069       return true;
1070
1071    // Queue the requests
1072    int Depth = -1;
1073    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1074         I = I->Next, Depth++)
1075    {
1076       // If pipelining is disabled, we only queue 1 request
1077       if (Server->Pipeline == false && Depth >= 0)
1078          break;
1079
1080       // Make sure we stick with the same server
1081       if (Server->Comp(I->Uri) == false)
1082          break;
1083       if (QueueBack == I)
1084       {
1085          QueueBack = I->Next;
1086          SendReq(I,Server->Out);
1087          continue;
1088       }
1089    }
1090
1091    return true;
1092 };
1093                                                                         /*}}}*/
1094 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1095 // ---------------------------------------------------------------------
1096 /* We stash the desired pipeline depth */
1097 bool HttpMethod::Configuration(string Message)
1098 {
1099    if (pkgAcqMethod::Configuration(Message) == false)
1100       return false;
1101
1102    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1103    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1104    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1105                                   PipelineDepth);
1106    Debug = _config->FindB("Debug::Acquire::http",false);
1107    AutoDetectProxyCmd = _config->Find("Acquire::http::ProxyAutoDetect");
1108
1109    // Get the proxy to use
1110    AutoDetectProxy();
1111
1112    return true;
1113 }
1114                                                                         /*}}}*/
1115 // HttpMethod::Loop - Main loop                                         /*{{{*/
1116 // ---------------------------------------------------------------------
1117 /* */
1118 int HttpMethod::Loop()
1119 {
1120    typedef vector<string> StringVector;
1121    typedef vector<string>::iterator StringVectorIterator;
1122    map<string, StringVector> Redirected;
1123
1124    signal(SIGTERM,SigTerm);
1125    signal(SIGINT,SigTerm);
1126
1127    Server = 0;
1128
1129    int FailCounter = 0;
1130    while (1)
1131    {
1132       // We have no commands, wait for some to arrive
1133       if (Queue == 0)
1134       {
1135          if (WaitFd(STDIN_FILENO) == false)
1136             return 0;
1137       }
1138
1139       /* Run messages, we can accept 0 (no message) if we didn't
1140          do a WaitFd above.. Otherwise the FD is closed. */
1141       int Result = Run(true);
1142       if (Result != -1 && (Result != 0 || Queue == 0))
1143       {
1144          if(FailReason.empty() == false ||
1145             _config->FindB("Acquire::http::DependOnSTDIN", true) == true)
1146             return 100;
1147          else
1148             return 0;
1149       }
1150
1151       if (Queue == 0)
1152          continue;
1153
1154       // Connect to the server
1155       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1156       {
1157          delete Server;
1158          Server = new ServerState(Queue->Uri,this);
1159       }
1160       /* If the server has explicitly said this is the last connection
1161          then we pre-emptively shut down the pipeline and tear down
1162          the connection. This will speed up HTTP/1.0 servers a tad
1163          since we don't have to wait for the close sequence to
1164          complete */
1165       if (Server->Persistent == false)
1166          Server->Close();
1167
1168       // Reset the pipeline
1169       if (Server->ServerFd == -1)
1170          QueueBack = Queue;
1171
1172       // Connnect to the host
1173       if (Server->Open() == false)
1174       {
1175          Fail(true);
1176          delete Server;
1177          Server = 0;
1178          continue;
1179       }
1180
1181       // Fill the pipeline.
1182       Fetch(0);
1183
1184       // Fetch the next URL header data from the server.
1185       switch (Server->RunHeaders())
1186       {
1187          case ServerState::RUN_HEADERS_OK:
1188          break;
1189
1190          // The header data is bad
1191          case ServerState::RUN_HEADERS_PARSE_ERROR:
1192          {
1193             _error->Error(_("Bad header data"));
1194             Fail(true);
1195             RotateDNS();
1196             continue;
1197          }
1198
1199          // The server closed a connection during the header get..
1200          default:
1201          case ServerState::RUN_HEADERS_IO_ERROR:
1202          {
1203             FailCounter++;
1204             _error->Discard();
1205             Server->Close();
1206             Server->Pipeline = false;
1207
1208             if (FailCounter >= 2)
1209             {
1210                Fail(_("Connection failed"),true);
1211                FailCounter = 0;
1212             }
1213
1214             RotateDNS();
1215             continue;
1216          }
1217       };
1218
1219       // Decide what to do.
1220       FetchResult Res;
1221       Res.Filename = Queue->DestFile;
1222       switch (DealWithHeaders(Res,Server))
1223       {
1224          // Ok, the file is Open
1225          case FILE_IS_OPEN:
1226          {
1227             URIStart(Res);
1228
1229             // Run the data
1230             bool Result =  Server->RunData();
1231
1232             /* If the server is sending back sizeless responses then fill in
1233                the size now */
1234             if (Res.Size == 0)
1235                Res.Size = File->Size();
1236
1237             // Close the file, destroy the FD object and timestamp it
1238             FailFd = -1;
1239             delete File;
1240             File = 0;
1241
1242             // Timestamp
1243             struct utimbuf UBuf;
1244             time(&UBuf.actime);
1245             UBuf.actime = Server->Date;
1246             UBuf.modtime = Server->Date;
1247             utime(Queue->DestFile.c_str(),&UBuf);
1248
1249             // Send status to APT
1250             if (Result == true)
1251             {
1252                Res.TakeHashes(*Server->In.Hash);
1253                URIDone(Res);
1254             }
1255             else
1256             {
1257                if (Server->ServerFd == -1)
1258                {
1259                   FailCounter++;
1260                   _error->Discard();
1261                   Server->Close();
1262
1263                   if (FailCounter >= 2)
1264                   {
1265                      Fail(_("Connection failed"),true);
1266                      FailCounter = 0;
1267                   }
1268
1269                   QueueBack = Queue;
1270                }
1271                else
1272                   Fail(true);
1273             }
1274             break;
1275          }
1276
1277          // IMS hit
1278          case IMS_HIT:
1279          {
1280             URIDone(Res);
1281             break;
1282          }
1283
1284          // Hard server error, not found or something
1285          case ERROR_UNRECOVERABLE:
1286          {
1287             Fail();
1288             break;
1289          }
1290
1291          // Hard internal error, kill the connection and fail
1292          case ERROR_NOT_FROM_SERVER:
1293          {
1294             delete File;
1295             File = 0;
1296
1297             Fail();
1298             RotateDNS();
1299             Server->Close();
1300             break;
1301          }
1302
1303          // We need to flush the data, the header is like a 404 w/ error text
1304          case ERROR_WITH_CONTENT_PAGE:
1305          {
1306             Fail();
1307
1308             // Send to content to dev/null
1309             File = new FileFd("/dev/null",FileFd::WriteExists);
1310             Server->RunData();
1311             delete File;
1312             File = 0;
1313             break;
1314          }
1315
1316          // Try again with a new URL
1317          case TRY_AGAIN_OR_REDIRECT:
1318          {
1319             // Clear rest of response if there is content
1320             if (Server->HaveContent)
1321             {
1322                File = new FileFd("/dev/null",FileFd::WriteExists);
1323                Server->RunData();
1324                delete File;
1325                File = 0;
1326             }
1327
1328             /* Detect redirect loops.  No more redirects are allowed
1329                after the same URI is seen twice in a queue item. */
1330             StringVector &R = Redirected[Queue->DestFile];
1331             bool StopRedirects = false;
1332             if (R.size() == 0)
1333                R.push_back(Queue->Uri);
1334             else if (R[0] == "STOP" || R.size() > 10)
1335                StopRedirects = true;
1336             else
1337             {
1338                for (StringVectorIterator I = R.begin(); I != R.end(); ++I)
1339                   if (Queue->Uri == *I)
1340                   {
1341                      R[0] = "STOP";
1342                      break;
1343                   }
1344
1345                R.push_back(Queue->Uri);
1346             }
1347
1348             if (StopRedirects == false)
1349                Redirect(NextURI);
1350             else
1351                Fail();
1352
1353             break;
1354          }
1355
1356          default:
1357          Fail(_("Internal error"));
1358          break;
1359       }
1360
1361       FailCounter = 0;
1362    }
1363
1364    return 0;
1365 }
1366                                                                         /*}}}*/
1367 // HttpMethod::AutoDetectProxy - auto detect proxy                      /*{{{*/
1368 // ---------------------------------------------------------------------
1369 /* */
1370 bool HttpMethod::AutoDetectProxy()
1371 {
1372    if (AutoDetectProxyCmd.empty())
1373       return true;
1374
1375    if (Debug)
1376       clog << "Using auto proxy detect command: " << AutoDetectProxyCmd << endl;
1377
1378    int Pipes[2] = {-1,-1};
1379    if (pipe(Pipes) != 0)
1380       return _error->Errno("pipe", "Failed to create Pipe");
1381
1382    pid_t Process = ExecFork();
1383    if (Process == 0)
1384    {
1385       close(Pipes[0]);
1386       dup2(Pipes[1],STDOUT_FILENO);
1387       SetCloseExec(STDOUT_FILENO,false);
1388
1389       const char *Args[2];
1390       Args[0] = AutoDetectProxyCmd.c_str();
1391       Args[1] = 0;
1392       execv(Args[0],(char **)Args);
1393       cerr << "Failed to exec method " << Args[0] << endl;
1394       _exit(100);
1395    }
1396    char buf[512];
1397    int InFd = Pipes[0];
1398    close(Pipes[1]);
1399    int res = read(InFd, buf, sizeof(buf));
1400    ExecWait(Process, "ProxyAutoDetect", true);
1401
1402    if (res < 0)
1403       return _error->Errno("read", "Failed to read");
1404    if (res == 0)
1405       return _error->Warning("ProxyAutoDetect returned no data");
1406
1407    // add trailing \0
1408    buf[res] = 0;
1409
1410    if (Debug)
1411       clog << "auto detect command returned: '" << buf << "'" << endl;
1412
1413    if (strstr(buf, "http://") == buf)
1414       _config->Set("Acquire::http::proxy", _strstrip(buf));
1415
1416    return true;
1417 }
1418                                                                         /*}}}*/
1419
1420