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 <apt-pkg/fileutl.h>
  29 #include <apt-pkg/acquire-method.h>
  30 #include <apt-pkg/error.h>
  31 #include <apt-pkg/hashes.h>
  32
  33 #include <sys/stat.h>
  34 #include <sys/time.h>
  35 #include <utime.h>
  36 #include <unistd.h>
  37 #include <signal.h>
  38 #include <stdio.h>
  39 #include <errno.h>
  40 #include <string.h>
  41 #include <map>
  42 #include <apti18n.h>
  43
  44 // Internet stuff
  45 #include <netdb.h>
  46
  47 #include "config.h"
  48 #include "connect.h"
  49 #include "rfc2553emu.h"
  50 #include "http.h"
  51
  52                                                                         /*}}}*/
  53 using namespace std;
  54
  55 string HttpMethod::FailFile;
  56 int HttpMethod::FailFd = -1;
  57 time_t HttpMethod::FailTime = 0;
  58 unsigned long PipelineDepth = 10;
  59 unsigned long TimeOut = 120;
  60 bool AllowRedirect = false;
  61 bool Debug = false;
  62 URI Proxy;
  63
  64 unsigned long CircleBuf::BwReadLimit=0;
  65 unsigned long CircleBuf::BwTickReadData=0;
  66 struct timeval CircleBuf::BwReadTick={0,0};
  67 const unsigned int CircleBuf::BW_HZ=10;
  68
  69 // CircleBuf::CircleBuf - Circular input buffer                         /*{{{*/
  70 // ---------------------------------------------------------------------
  71 /* */
  72 CircleBuf::CircleBuf(unsigned long Size) : Size(Size), Hash(0)
  73 {
  74    Buf = new unsigned char[Size];
  75    Reset();
  76
  77    CircleBuf::BwReadLimit = _config->FindI("Acquire::http::Dl-Limit",0)*1024;
  78 }
  79                                                                         /*}}}*/
  80 // CircleBuf::Reset - Reset to the default state                        /*{{{*/
  81 // ---------------------------------------------------------------------
  82 /* */
  83 void CircleBuf::Reset()
  84 {
  85    InP = 0;
  86    OutP = 0;
  87    StrPos = 0;
  88    MaxGet = (unsigned int)-1;
  89    OutQueue = string();
  90    if (Hash != 0)
  91    {
  92       delete Hash;
  93       Hash = new Hashes;
  94    }
  95 };
  96                                                                         /*}}}*/
  97 // CircleBuf::Read - Read from a FD into the circular buffer            /*{{{*/
  98 // ---------------------------------------------------------------------
  99 /* This fills up the buffer with as much data as is in the FD, assuming it
 100    is non-blocking.. */
 101 bool CircleBuf::Read(int Fd)
 102 {
 103    unsigned long BwReadMax;
 104
 105    while (1)
 106    {
 107       // Woops, buffer is full
 108       if (InP - OutP == Size)
 109          return true;
 110
 111       // what's left to read in this tick
 112       BwReadMax = CircleBuf::BwReadLimit/BW_HZ;
 113
 114       if(CircleBuf::BwReadLimit) {
 115          struct timeval now;
 116          gettimeofday(&now,0);
 117
 118          unsigned long d = (now.tv_sec-CircleBuf::BwReadTick.tv_sec)*1000000 +
 119             now.tv_usec-CircleBuf::BwReadTick.tv_usec;
 120          if(d > 1000000/BW_HZ) {
 121             CircleBuf::BwReadTick = now;
 122             CircleBuf::BwTickReadData = 0;
 123          }
 124
 125          if(CircleBuf::BwTickReadData >= BwReadMax) {
 126             usleep(1000000/BW_HZ);
 127             return true;
 128          }
 129       }
 130
 131       // Write the buffer segment
 132       int Res;
 133       if(CircleBuf::BwReadLimit) {
 134          Res = read(Fd,Buf + (InP%Size),
 135                     BwReadMax > LeftRead() ? LeftRead() : BwReadMax);
 136       } else
 137          Res = read(Fd,Buf + (InP%Size),LeftRead());
 138
 139       if(Res > 0 && BwReadLimit > 0)
 140          CircleBuf::BwTickReadData += Res;
 141
 142       if (Res == 0)
 143          return false;
 144       if (Res < 0)
 145       {
 146          if (errno == EAGAIN)
 147             return true;
 148          return false;
 149       }
 150
 151       if (InP == 0)
 152          gettimeofday(&Start,0);
 153       InP += Res;
 154    }
 155 }
 156                                                                         /*}}}*/
 157 // CircleBuf::Read - Put the string into the buffer                     /*{{{*/
 158 // ---------------------------------------------------------------------
 159 /* This will hold the string in and fill the buffer with it as it empties */
 160 bool CircleBuf::Read(string Data)
 161 {
 162    OutQueue += Data;
 163    FillOut();
 164    return true;
 165 }
 166                                                                         /*}}}*/
 167 // CircleBuf::FillOut - Fill the buffer from the output queue           /*{{{*/
 168 // ---------------------------------------------------------------------
 169 /* */
 170 void CircleBuf::FillOut()
 171 {
 172    if (OutQueue.empty() == true)
 173       return;
 174    while (1)
 175    {
 176       // Woops, buffer is full
 177       if (InP - OutP == Size)
 178          return;
 179
 180       // Write the buffer segment
 181       unsigned long Sz = LeftRead();
 182       if (OutQueue.length() - StrPos < Sz)
 183          Sz = OutQueue.length() - StrPos;
 184       memcpy(Buf + (InP%Size),OutQueue.c_str() + StrPos,Sz);
 185
 186       // Advance
 187       StrPos += Sz;
 188       InP += Sz;
 189       if (OutQueue.length() == StrPos)
 190       {
 191          StrPos = 0;
 192          OutQueue = "";
 193          return;
 194       }
 195    }
 196 }
 197                                                                         /*}}}*/
 198 // CircleBuf::Write - Write from the buffer into a FD                   /*{{{*/
 199 // ---------------------------------------------------------------------
 200 /* This empties the buffer into the FD. */
 201 bool CircleBuf::Write(int Fd)
 202 {
 203    while (1)
 204    {
 205       FillOut();
 206
 207       // Woops, buffer is empty
 208       if (OutP == InP)
 209          return true;
 210
 211       if (OutP == MaxGet)
 212          return true;
 213
 214       // Write the buffer segment
 215       int Res;
 216       Res = write(Fd,Buf + (OutP%Size),LeftWrite());
 217
 218       if (Res == 0)
 219          return false;
 220       if (Res < 0)
 221       {
 222          if (errno == EAGAIN)
 223             return true;
 224
 225          return false;
 226       }
 227
 228       if (Hash != 0)
 229          Hash->Add(Buf + (OutP%Size),Res);
 230
 231       OutP += Res;
 232    }
 233 }
 234                                                                         /*}}}*/
 235 // CircleBuf::WriteTillEl - Write from the buffer to a string           /*{{{*/
 236 // ---------------------------------------------------------------------
 237 /* This copies till the first empty line */
 238 bool CircleBuf::WriteTillEl(string &Data,bool Single)
 239 {
 240    // We cheat and assume it is unneeded to have more than one buffer load
 241    for (unsigned long I = OutP; I < InP; I++)
 242    {
 243       if (Buf[I%Size] != '\n')
 244          continue;
 245       ++I;
 246
 247       if (Single == false)
 248       {
 249          if (I < InP  && Buf[I%Size] == '\r')
 250             ++I;
 251          if (I >= InP || Buf[I%Size] != '\n')
 252             continue;
 253          ++I;
 254       }
 255
 256       Data = "";
 257       while (OutP < I)
 258       {
 259          unsigned long Sz = LeftWrite();
 260          if (Sz == 0)
 261             return false;
 262          if (I - OutP < Sz)
 263             Sz = I - OutP;
 264          Data += string((char *)(Buf + (OutP%Size)),Sz);
 265          OutP += Sz;
 266       }
 267       return true;
 268    }
 269    return false;
 270 }
 271                                                                         /*}}}*/
 272 // CircleBuf::Stats - Print out stats information                       /*{{{*/
 273 // ---------------------------------------------------------------------
 274 /* */
 275 void CircleBuf::Stats()
 276 {
 277    if (InP == 0)
 278       return;
 279
 280    struct timeval Stop;
 281    gettimeofday(&Stop,0);
 282 /*   float Diff = Stop.tv_sec - Start.tv_sec +
 283              (float)(Stop.tv_usec - Start.tv_usec)/1000000;
 284    clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
 285 }
 286                                                                         /*}}}*/
 287
 288 // ServerState::ServerState - Constructor                               /*{{{*/
 289 // ---------------------------------------------------------------------
 290 /* */
 291 ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner),
 292                         In(64*1024), Out(4*1024),
 293                         ServerName(Srv)
 294 {
 295    Reset();
 296 }
 297                                                                         /*}}}*/
 298 // ServerState::Open - Open a connection to the server                  /*{{{*/
 299 // ---------------------------------------------------------------------
 300 /* This opens a connection to the server. */
 301 bool ServerState::Open()
 302 {
 303    // Use the already open connection if possible.
 304    if (ServerFd != -1)
 305       return true;
 306
 307    Close();
 308    In.Reset();
 309    Out.Reset();
 310    Persistent = true;
 311
 312    // Determine the proxy setting
 313    if (getenv("http_proxy") == 0)
 314    {
 315       string DefProxy = _config->Find("Acquire::http::Proxy");
 316       string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
 317       if (SpecificProxy.empty() == false)
 318       {
 319          if (SpecificProxy == "DIRECT")
 320             Proxy = "";
 321          else
 322             Proxy = SpecificProxy;
 323       }
 324       else
 325          Proxy = DefProxy;
 326    }
 327    else
 328       Proxy = getenv("http_proxy");
 329
 330    // Parse no_proxy, a , separated list of domains
 331    if (getenv("no_proxy") != 0)
 332    {
 333       if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 334          Proxy = "";
 335    }
 336
 337    // Determine what host and port to use based on the proxy settings
 338    int Port = 0;
 339    string Host;
 340    if (Proxy.empty() == true || Proxy.Host.empty() == true)
 341    {
 342       if (ServerName.Port != 0)
 343          Port = ServerName.Port;
 344       Host = ServerName.Host;
 345    }
 346    else
 347    {
 348       if (Proxy.Port != 0)
 349          Port = Proxy.Port;
 350       Host = Proxy.Host;
 351    }
 352
 353    // Connect to the remote server
 354    if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
 355       return false;
 356
 357    return true;
 358 }
 359                                                                         /*}}}*/
 360 // ServerState::Close - Close a connection to the server                /*{{{*/
 361 // ---------------------------------------------------------------------
 362 /* */
 363 bool ServerState::Close()
 364 {
 365    close(ServerFd);
 366    ServerFd = -1;
 367    return true;
 368 }
 369                                                                         /*}}}*/
 370 // ServerState::RunHeaders - Get the headers before the data            /*{{{*/
 371 // ---------------------------------------------------------------------
 372 /* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
 373    parse error occurred */
 374 int ServerState::RunHeaders()
 375 {
 376    State = Header;
 377
 378    Owner->Status(_("Waiting for headers"));
 379
 380    Major = 0;
 381    Minor = 0;
 382    Result = 0;
 383    Size = 0;
 384    StartPos = 0;
 385    Encoding = Closes;
 386    HaveContent = false;
 387    time(&Date);
 388
 389    do
 390    {
 391       string Data;
 392       if (In.WriteTillEl(Data) == false)
 393          continue;
 394
 395       if (Debug == true)
 396          clog << Data;
 397
 398       for (string::const_iterator I = Data.begin(); I < Data.end(); I++)
 399       {
 400          string::const_iterator J = I;
 401          for (; J != Data.end() && *J != '\n' && *J != '\r';J++);
 402          if (HeaderLine(string(I,J)) == false)
 403             return 2;
 404          I = J;
 405       }
 406
 407       // 100 Continue is a Nop...
 408       if (Result == 100)
 409          continue;
 410
 411       // Tidy up the connection persistance state.
 412       if (Encoding == Closes && HaveContent == true)
 413          Persistent = false;
 414
 415       return 0;
 416    }
 417    while (Owner->Go(false,this) == true);
 418
 419    return 1;
 420 }
 421                                                                         /*}}}*/
 422 // ServerState::RunData - Transfer the data from the socket             /*{{{*/
 423 // ---------------------------------------------------------------------
 424 /* */
 425 bool ServerState::RunData()
 426 {
 427    State = Data;
 428
 429    // Chunked transfer encoding is fun..
 430    if (Encoding == Chunked)
 431    {
 432       while (1)
 433       {
 434          // Grab the block size
 435          bool Last = true;
 436          string Data;
 437          In.Limit(-1);
 438          do
 439          {
 440             if (In.WriteTillEl(Data,true) == true)
 441                break;
 442          }
 443          while ((Last = Owner->Go(false,this)) == true);
 444
 445          if (Last == false)
 446             return false;
 447
 448          // See if we are done
 449          unsigned long Len = strtol(Data.c_str(),0,16);
 450          if (Len == 0)
 451          {
 452             In.Limit(-1);
 453
 454             // We have to remove the entity trailer
 455             Last = true;
 456             do
 457             {
 458                if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
 459                   break;
 460             }
 461             while ((Last = Owner->Go(false,this)) == true);
 462             if (Last == false)
 463                return false;
 464             return !_error->PendingError();
 465          }
 466
 467          // Transfer the block
 468          In.Limit(Len);
 469          while (Owner->Go(true,this) == true)
 470             if (In.IsLimit() == true)
 471                break;
 472
 473          // Error
 474          if (In.IsLimit() == false)
 475             return false;
 476
 477          // The server sends an extra new line before the next block specifier..
 478          In.Limit(-1);
 479          Last = true;
 480          do
 481          {
 482             if (In.WriteTillEl(Data,true) == true)
 483                break;
 484          }
 485          while ((Last = Owner->Go(false,this)) == true);
 486          if (Last == false)
 487             return false;
 488       }
 489    }
 490    else
 491    {
 492       /* Closes encoding is used when the server did not specify a size, the
 493          loss of the connection means we are done */
 494       if (Encoding == Closes)
 495          In.Limit(-1);
 496       else
 497          In.Limit(Size - StartPos);
 498
 499       // Just transfer the whole block.
 500       do
 501       {
 502          if (In.IsLimit() == false)
 503             continue;
 504
 505          In.Limit(-1);
 506          return !_error->PendingError();
 507       }
 508       while (Owner->Go(true,this) == true);
 509    }
 510
 511    return Owner->Flush(this) && !_error->PendingError();
 512 }
 513                                                                         /*}}}*/
 514 // ServerState::HeaderLine - Process a header line                      /*{{{*/
 515 // ---------------------------------------------------------------------
 516 /* */
 517 bool ServerState::HeaderLine(string Line)
 518 {
 519    if (Line.empty() == true)
 520       return true;
 521
 522    // The http server might be trying to do something evil.
 523    if (Line.length() >= MAXLEN)
 524       return _error->Error(_("Got a single header line over %u chars"),MAXLEN);
 525
 526    string::size_type Pos = Line.find(' ');
 527    if (Pos == string::npos || Pos+1 > Line.length())
 528    {
 529       // Blah, some servers use "connection:closes", evil.
 530       Pos = Line.find(':');
 531       if (Pos == string::npos || Pos + 2 > Line.length())
 532          return _error->Error(_("Bad header line"));
 533       Pos++;
 534    }
 535
 536    // Parse off any trailing spaces between the : and the next word.
 537    string::size_type Pos2 = Pos;
 538    while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
 539       Pos2++;
 540
 541    string Tag = string(Line,0,Pos);
 542    string Val = string(Line,Pos2);
 543
 544    if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
 545    {
 546       // Evil servers return no version
 547       if (Line[4] == '/')
 548       {
 549          if (sscanf(Line.c_str(),"HTTP/%u.%u %u %[^\n]",&Major,&Minor,
 550                     &Result,Code) != 4)
 551             return _error->Error(_("The HTTP server sent an invalid reply header"));
 552       }
 553       else
 554       {
 555          Major = 0;
 556          Minor = 9;
 557          if (sscanf(Line.c_str(),"HTTP %u %[^\n]",&Result,Code) != 2)
 558             return _error->Error(_("The HTTP server sent an invalid reply header"));
 559       }
 560
 561       /* Check the HTTP response header to get the default persistance
 562          state. */
 563       if (Major < 1)
 564          Persistent = false;
 565       else
 566       {
 567          if (Major == 1 && Minor <= 0)
 568             Persistent = false;
 569          else
 570             Persistent = true;
 571       }
 572
 573       return true;
 574    }
 575
 576    if (stringcasecmp(Tag,"Content-Length:") == 0)
 577    {
 578       if (Encoding == Closes)
 579          Encoding = Stream;
 580       HaveContent = true;
 581
 582       // The length is already set from the Content-Range header
 583       if (StartPos != 0)
 584          return true;
 585
 586       if (sscanf(Val.c_str(),"%lu",&Size) != 1)
 587          return _error->Error(_("The HTTP server sent an invalid Content-Length header"));
 588       return true;
 589    }
 590
 591    if (stringcasecmp(Tag,"Content-Type:") == 0)
 592    {
 593       HaveContent = true;
 594       return true;
 595    }
 596
 597    if (stringcasecmp(Tag,"Content-Range:") == 0)
 598    {
 599       HaveContent = true;
 600
 601       if (sscanf(Val.c_str(),"bytes %lu-%*u/%lu",&StartPos,&Size) != 2)
 602          return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
 603       if ((unsigned)StartPos > Size)
 604          return _error->Error(_("This HTTP server has broken range support"));
 605       return true;
 606    }
 607
 608    if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
 609    {
 610       HaveContent = true;
 611       if (stringcasecmp(Val,"chunked") == 0)
 612          Encoding = Chunked;
 613       return true;
 614    }
 615
 616    if (stringcasecmp(Tag,"Connection:") == 0)
 617    {
 618       if (stringcasecmp(Val,"close") == 0)
 619          Persistent = false;
 620       if (stringcasecmp(Val,"keep-alive") == 0)
 621          Persistent = true;
 622       return true;
 623    }
 624
 625    if (stringcasecmp(Tag,"Last-Modified:") == 0)
 626    {
 627       if (StrToTime(Val,Date) == false)
 628          return _error->Error(_("Unknown date format"));
 629       return true;
 630    }
 631
 632    if (stringcasecmp(Tag,"Location:") == 0)
 633    {
 634       Location = Val;
 635       return true;
 636    }
 637
 638    return true;
 639 }
 640                                                                         /*}}}*/
 641
 642 // HttpMethod::SendReq - Send the HTTP request                          /*{{{*/
 643 // ---------------------------------------------------------------------
 644 /* This places the http request in the outbound buffer */
 645 void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
 646 {
 647    URI Uri = Itm->Uri;
 648
 649    // The HTTP server expects a hostname with a trailing :port
 650    char Buf[1000];
 651    string ProperHost = Uri.Host;
 652    if (Uri.Port != 0)
 653    {
 654       sprintf(Buf,":%u",Uri.Port);
 655       ProperHost += Buf;
 656    }
 657
 658    // Just in case.
 659    if (Itm->Uri.length() >= sizeof(Buf))
 660        abort();
 661
 662    /* Build the request. We include a keep-alive header only for non-proxy
 663       requests. This is to tweak old http/1.0 servers that do support keep-alive
 664       but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
 665       will glitch HTTP/1.0 proxies because they do not filter it out and
 666       pass it on, HTTP/1.1 says the connection should default to keep alive
 667       and we expect the proxy to do this */
 668    if (Proxy.empty() == true || Proxy.Host.empty())
 669       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
 670               QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
 671    else
 672    {
 673       /* Generate a cache control header if necessary. We place a max
 674          cache age on index files, optionally set a no-cache directive
 675          and a no-store directive for archives. */
 676       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 677               Itm->Uri.c_str(),ProperHost.c_str());
 678       // only generate a cache control header if we actually want to
 679       // use a cache
 680       if (_config->FindB("Acquire::http::No-Cache",false) == false)
 681       {
 682          if (Itm->IndexFile == true)
 683             sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 684                     _config->FindI("Acquire::http::Max-Age",0));
 685          else
 686          {
 687             if (_config->FindB("Acquire::http::No-Store",false) == true)
 688                strcat(Buf,"Cache-Control: no-store\r\n");
 689          }
 690       }
 691    }
 692    // generate a no-cache header if needed
 693    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 694       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 695
 696
 697    string Req = Buf;
 698
 699    // Check for a partial file
 700    struct stat SBuf;
 701    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 702    {
 703       // In this case we send an if-range query with a range header
 704       sprintf(Buf,"Range: bytes=%li-\r\nIf-Range: %s\r\n",(long)SBuf.st_size - 1,
 705               TimeRFC1123(SBuf.st_mtime).c_str());
 706       Req += Buf;
 707    }
 708    else
 709    {
 710       if (Itm->LastModified != 0)
 711       {
 712          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 713          Req += Buf;
 714       }
 715    }
 716
 717    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 718       Req += string("Proxy-Authorization: Basic ") +
 719           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 720
 721    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 722       Req += string("Authorization: Basic ") +
 723           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 724
 725    Req += "User-Agent: Ubuntu APT-HTTP/1.3 ("VERSION")\r\n\r\n";
 726
 727    if (Debug == true)
 728       cerr << Req << endl;
 729
 730    Out.Read(Req);
 731 }
 732                                                                         /*}}}*/
 733 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 734 // ---------------------------------------------------------------------
 735 /* This runs the select loop over the server FDs, Output file FDs and
 736    stdin. */
 737 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 738 {
 739    // Server has closed the connection
 740    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 741                                ToFile == false))
 742       return false;
 743
 744    fd_set rfds,wfds;
 745    FD_ZERO(&rfds);
 746    FD_ZERO(&wfds);
 747
 748    /* Add the server. We only send more requests if the connection will
 749       be persisting */
 750    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 751        && Srv->Persistent == true)
 752       FD_SET(Srv->ServerFd,&wfds);
 753    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 754       FD_SET(Srv->ServerFd,&rfds);
 755
 756    // Add the file
 757    int FileFD = -1;
 758    if (File != 0)
 759       FileFD = File->Fd();
 760
 761    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 762       FD_SET(FileFD,&wfds);
 763
 764    // Add stdin
 765    FD_SET(STDIN_FILENO,&rfds);
 766
 767    // Figure out the max fd
 768    int MaxFd = FileFD;
 769    if (MaxFd < Srv->ServerFd)
 770       MaxFd = Srv->ServerFd;
 771
 772    // Select
 773    struct timeval tv;
 774    tv.tv_sec = TimeOut;
 775    tv.tv_usec = 0;
 776    int Res = 0;
 777    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 778    {
 779       if (errno == EINTR)
 780          return true;
 781       return _error->Errno("select",_("Select failed"));
 782    }
 783
 784    if (Res == 0)
 785    {
 786       _error->Error(_("Connection timed out"));
 787       return ServerDie(Srv);
 788    }
 789
 790    // Handle server IO
 791    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 792    {
 793       errno = 0;
 794       if (Srv->In.Read(Srv->ServerFd) == false)
 795          return ServerDie(Srv);
 796    }
 797
 798    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 799    {
 800       errno = 0;
 801       if (Srv->Out.Write(Srv->ServerFd) == false)
 802          return ServerDie(Srv);
 803    }
 804
 805    // Send data to the file
 806    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 807    {
 808       if (Srv->In.Write(FileFD) == false)
 809          return _error->Errno("write",_("Error writing to output file"));
 810    }
 811
 812    // Handle commands from APT
 813    if (FD_ISSET(STDIN_FILENO,&rfds))
 814    {
 815       if (Run(true) != -1)
 816          exit(100);
 817    }
 818
 819    return true;
 820 }
 821                                                                         /*}}}*/
 822 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 823 // ---------------------------------------------------------------------
 824 /* This takes the current input buffer from the Server FD and writes it
 825    into the file */
 826 bool HttpMethod::Flush(ServerState *Srv)
 827 {
 828    if (File != 0)
 829    {
 830       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 831       // can't be set
 832       if (File->Name() != "/dev/null")
 833          SetNonBlock(File->Fd(),false);
 834       if (Srv->In.WriteSpace() == false)
 835          return true;
 836
 837       while (Srv->In.WriteSpace() == true)
 838       {
 839          if (Srv->In.Write(File->Fd()) == false)
 840             return _error->Errno("write",_("Error writing to file"));
 841          if (Srv->In.IsLimit() == true)
 842             return true;
 843       }
 844
 845       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 846          return true;
 847    }
 848    return false;
 849 }
 850                                                                         /*}}}*/
 851 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 852 // ---------------------------------------------------------------------
 853 /* */
 854 bool HttpMethod::ServerDie(ServerState *Srv)
 855 {
 856    unsigned int LErrno = errno;
 857
 858    // Dump the buffer to the file
 859    if (Srv->State == ServerState::Data)
 860    {
 861       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 862       // can't be set
 863       if (File->Name() != "/dev/null")
 864          SetNonBlock(File->Fd(),false);
 865       while (Srv->In.WriteSpace() == true)
 866       {
 867          if (Srv->In.Write(File->Fd()) == false)
 868             return _error->Errno("write",_("Error writing to the file"));
 869
 870          // Done
 871          if (Srv->In.IsLimit() == true)
 872             return true;
 873       }
 874    }
 875
 876    // See if this is because the server finished the data stream
 877    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 878        Srv->Encoding != ServerState::Closes)
 879    {
 880       Srv->Close();
 881       if (LErrno == 0)
 882          return _error->Error(_("Error reading from server. Remote end closed connection"));
 883       errno = LErrno;
 884       return _error->Errno("read",_("Error reading from server"));
 885    }
 886    else
 887    {
 888       Srv->In.Limit(-1);
 889
 890       // Nothing left in the buffer
 891       if (Srv->In.WriteSpace() == false)
 892          return false;
 893
 894       // We may have got multiple responses back in one packet..
 895       Srv->Close();
 896       return true;
 897    }
 898
 899    return false;
 900 }
 901                                                                         /*}}}*/
 902 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 903 // ---------------------------------------------------------------------
 904 /* We look at the header data we got back from the server and decide what
 905    to do. Returns
 906      0 - File is open,
 907      1 - IMS hit
 908      3 - Unrecoverable error
 909      4 - Error with error content page
 910      5 - Unrecoverable non-server error (close the connection)
 911      6 - Try again with a new or changed URI
 912  */
 913 int HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 914 {
 915    // Not Modified
 916    if (Srv->Result == 304)
 917    {
 918       unlink(Queue->DestFile.c_str());
 919       Res.IMSHit = true;
 920       Res.LastModified = Queue->LastModified;
 921       return 1;
 922    }
 923
 924    /* Redirect
 925     *
 926     * Note that it is only OK for us to treat all redirection the same
 927     * because we *always* use GET, not other HTTP methods.  There are
 928     * three redirection codes for which it is not appropriate that we
 929     * redirect.  Pass on those codes so the error handling kicks in.
 930     */
 931    if (AllowRedirect
 932        && (Srv->Result > 300 && Srv->Result < 400)
 933        && (Srv->Result != 300       // Multiple Choices
 934            && Srv->Result != 304    // Not Modified
 935            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 936    {
 937       if (!Srv->Location.empty())
 938       {
 939          NextURI = Srv->Location;
 940          return 6;
 941       }
 942       /* else pass through for error message */
 943    }
 944
 945    /* We have a reply we dont handle. This should indicate a perm server
 946       failure */
 947    if (Srv->Result < 200 || Srv->Result >= 300)
 948    {
 949       char err[255];
 950       snprintf(err,sizeof(err)-1,"HttpError%i",Srv->Result);
 951       SetFailReason(err);
 952       _error->Error("%u %s",Srv->Result,Srv->Code);
 953       if (Srv->HaveContent == true)
 954          return 4;
 955       return 3;
 956    }
 957
 958    // This is some sort of 2xx 'data follows' reply
 959    Res.LastModified = Srv->Date;
 960    Res.Size = Srv->Size;
 961
 962    // Open the file
 963    delete File;
 964    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
 965    if (_error->PendingError() == true)
 966       return 5;
 967
 968    FailFile = Queue->DestFile;
 969    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
 970    FailFd = File->Fd();
 971    FailTime = Srv->Date;
 972
 973    // Set the expected size
 974    if (Srv->StartPos >= 0)
 975    {
 976       Res.ResumePoint = Srv->StartPos;
 977       if (ftruncate(File->Fd(),Srv->StartPos) < 0)
 978          _error->Errno("ftruncate", _("Failed to truncate file"));
 979    }
 980
 981    // Set the start point
 982    lseek(File->Fd(),0,SEEK_END);
 983
 984    delete Srv->In.Hash;
 985    Srv->In.Hash = new Hashes;
 986
 987    // Fill the Hash if the file is non-empty (resume)
 988    if (Srv->StartPos > 0)
 989    {
 990       lseek(File->Fd(),0,SEEK_SET);
 991       if (Srv->In.Hash->AddFD(File->Fd(),Srv->StartPos) == false)
 992       {
 993          _error->Errno("read",_("Problem hashing file"));
 994          return 5;
 995       }
 996       lseek(File->Fd(),0,SEEK_END);
 997    }
 998
 999    SetNonBlock(File->Fd(),true);
1000    return 0;
1001 }
1002                                                                         /*}}}*/
1003 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1004 // ---------------------------------------------------------------------
1005 /* This closes and timestamps the open file. This is neccessary to get
1006    resume behavoir on user abort */
1007 void HttpMethod::SigTerm(int)
1008 {
1009    if (FailFd == -1)
1010       _exit(100);
1011    close(FailFd);
1012
1013    // Timestamp
1014    struct utimbuf UBuf;
1015    UBuf.actime = FailTime;
1016    UBuf.modtime = FailTime;
1017    utime(FailFile.c_str(),&UBuf);
1018
1019    _exit(100);
1020 }
1021                                                                         /*}}}*/
1022 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1023 // ---------------------------------------------------------------------
1024 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1025    depth. */
1026 bool HttpMethod::Fetch(FetchItem *)
1027 {
1028    if (Server == 0)
1029       return true;
1030
1031    // Queue the requests
1032    int Depth = -1;
1033    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1034         I = I->Next, Depth++)
1035    {
1036       // If pipelining is disabled, we only queue 1 request
1037       if (Server->Pipeline == false && Depth >= 0)
1038          break;
1039
1040       // Make sure we stick with the same server
1041       if (Server->Comp(I->Uri) == false)
1042          break;
1043       if (QueueBack == I)
1044       {
1045          QueueBack = I->Next;
1046          SendReq(I,Server->Out);
1047          continue;
1048       }
1049    }
1050
1051    return true;
1052 };
1053                                                                         /*}}}*/
1054 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1055 // ---------------------------------------------------------------------
1056 /* We stash the desired pipeline depth */
1057 bool HttpMethod::Configuration(string Message)
1058 {
1059    if (pkgAcqMethod::Configuration(Message) == false)
1060       return false;
1061
1062    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1063    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1064    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1065                                   PipelineDepth);
1066    Debug = _config->FindB("Debug::Acquire::http",false);
1067
1068    return true;
1069 }
1070                                                                         /*}}}*/
1071 // HttpMethod::Loop - Main loop                                         /*{{{*/
1072 // ---------------------------------------------------------------------
1073 /* */
1074 int HttpMethod::Loop()
1075 {
1076    typedef vector<string> StringVector;
1077    typedef vector<string>::iterator StringVectorIterator;
1078    map<string, StringVector> Redirected;
1079
1080    signal(SIGTERM,SigTerm);
1081    signal(SIGINT,SigTerm);
1082
1083    Server = 0;
1084
1085    int FailCounter = 0;
1086    while (1)
1087    {
1088       // We have no commands, wait for some to arrive
1089       if (Queue == 0)
1090       {
1091          if (WaitFd(STDIN_FILENO) == false)
1092             return 0;
1093       }
1094
1095       /* Run messages, we can accept 0 (no message) if we didn't
1096          do a WaitFd above.. Otherwise the FD is closed. */
1097       int Result = Run(true);
1098       if (Result != -1 && (Result != 0 || Queue == 0))
1099          return 100;
1100
1101       if (Queue == 0)
1102          continue;
1103
1104       // Connect to the server
1105       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1106       {
1107          delete Server;
1108          Server = new ServerState(Queue->Uri,this);
1109       }
1110       /* If the server has explicitly said this is the last connection
1111          then we pre-emptively shut down the pipeline and tear down
1112          the connection. This will speed up HTTP/1.0 servers a tad
1113          since we don't have to wait for the close sequence to
1114          complete */
1115       if (Server->Persistent == false)
1116          Server->Close();
1117
1118       // Reset the pipeline
1119       if (Server->ServerFd == -1)
1120          QueueBack = Queue;
1121
1122       // Connnect to the host
1123       if (Server->Open() == false)
1124       {
1125          Fail(true);
1126          delete Server;
1127          Server = 0;
1128          continue;
1129       }
1130
1131       // Fill the pipeline.
1132       Fetch(0);
1133
1134       // Fetch the next URL header data from the server.
1135       switch (Server->RunHeaders())
1136       {
1137          case 0:
1138          break;
1139
1140          // The header data is bad
1141          case 2:
1142          {
1143             _error->Error(_("Bad header data"));
1144             Fail(true);
1145             RotateDNS();
1146             continue;
1147          }
1148
1149          // The server closed a connection during the header get..
1150          default:
1151          case 1:
1152          {
1153             FailCounter++;
1154             _error->Discard();
1155             Server->Close();
1156             Server->Pipeline = false;
1157
1158             if (FailCounter >= 2)
1159             {
1160                Fail(_("Connection failed"),true);
1161                FailCounter = 0;
1162             }
1163
1164             RotateDNS();
1165             continue;
1166          }
1167       };
1168
1169       // Decide what to do.
1170       FetchResult Res;
1171       Res.Filename = Queue->DestFile;
1172       switch (DealWithHeaders(Res,Server))
1173       {
1174          // Ok, the file is Open
1175          case 0:
1176          {
1177             URIStart(Res);
1178
1179             // Run the data
1180             bool Result =  Server->RunData();
1181
1182             /* If the server is sending back sizeless responses then fill in
1183                the size now */
1184             if (Res.Size == 0)
1185                Res.Size = File->Size();
1186
1187             // Close the file, destroy the FD object and timestamp it
1188             FailFd = -1;
1189             delete File;
1190             File = 0;
1191
1192             // Timestamp
1193             struct utimbuf UBuf;
1194             time(&UBuf.actime);
1195             UBuf.actime = Server->Date;
1196             UBuf.modtime = Server->Date;
1197             utime(Queue->DestFile.c_str(),&UBuf);
1198
1199             // Send status to APT
1200             if (Result == true)
1201             {
1202                Res.TakeHashes(*Server->In.Hash);
1203                URIDone(Res);
1204             }
1205             else
1206             {
1207                if (Server->ServerFd == -1)
1208                {
1209                   FailCounter++;
1210                   _error->Discard();
1211                   Server->Close();
1212
1213                   if (FailCounter >= 2)
1214                   {
1215                      Fail(_("Connection failed"),true);
1216                      FailCounter = 0;
1217                   }
1218
1219                   QueueBack = Queue;
1220                }
1221                else
1222                   Fail(true);
1223             }
1224             break;
1225          }
1226
1227          // IMS hit
1228          case 1:
1229          {
1230             URIDone(Res);
1231             break;
1232          }
1233
1234          // Hard server error, not found or something
1235          case 3:
1236          {
1237             Fail();
1238             break;
1239          }
1240
1241          // Hard internal error, kill the connection and fail
1242          case 5:
1243          {
1244             delete File;
1245             File = 0;
1246
1247             Fail();
1248             RotateDNS();
1249             Server->Close();
1250             break;
1251          }
1252
1253          // We need to flush the data, the header is like a 404 w/ error text
1254          case 4:
1255          {
1256             Fail();
1257
1258             // Send to content to dev/null
1259             File = new FileFd("/dev/null",FileFd::WriteExists);
1260             Server->RunData();
1261             delete File;
1262             File = 0;
1263             break;
1264          }
1265
1266          // Try again with a new URL
1267          case 6:
1268          {
1269             // Clear rest of response if there is content
1270             if (Server->HaveContent)
1271             {
1272                File = new FileFd("/dev/null",FileFd::WriteExists);
1273                Server->RunData();
1274                delete File;
1275                File = 0;
1276             }
1277
1278             /* Detect redirect loops.  No more redirects are allowed
1279                after the same URI is seen twice in a queue item. */
1280             StringVector &R = Redirected[Queue->DestFile];
1281             bool StopRedirects = false;
1282             if (R.size() == 0)
1283                R.push_back(Queue->Uri);
1284             else if (R[0] == "STOP" || R.size() > 10)
1285                StopRedirects = true;
1286             else
1287             {
1288                for (StringVectorIterator I = R.begin(); I != R.end(); I++)
1289                   if (Queue->Uri == *I)
1290                   {
1291                      R[0] = "STOP";
1292                      break;
1293                   }
1294
1295                R.push_back(Queue->Uri);
1296             }
1297
1298             if (StopRedirects == false)
1299                Redirect(NextURI);
1300             else
1301                Fail();
1302
1303             break;
1304          }
1305
1306          default:
1307          Fail(_("Internal error"));
1308          break;
1309       }
1310
1311       FailCounter = 0;
1312    }
1313
1314    return 0;
1315 }
1316                                                                         /*}}}*/
1317
1318
1319