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 #include <apt-pkg/netrc.h>
  33
  34 #include <sys/stat.h>
  35 #include <sys/time.h>
  36 #include <utime.h>
  37 #include <unistd.h>
  38 #include <signal.h>
  39 #include <stdio.h>
  40 #include <errno.h>
  41 #include <string.h>
  42 #include <iostream>
  43 #include <map>
  44 #include <apti18n.h>
  45
  46
  47 // Internet stuff
  48 #include <netdb.h>
  49
  50 #include "config.h"
  51 #include "connect.h"
  52 #include "rfc2553emu.h"
  53 #include "http.h"
  54                                                                         /*}}}*/
  55 using namespace std;
  56
  57 string HttpMethod::FailFile;
  58 int HttpMethod::FailFd = -1;
  59 time_t HttpMethod::FailTime = 0;
  60 unsigned long PipelineDepth = 10;
  61 unsigned long TimeOut = 120;
  62 bool AllowRedirect = false;
  63 bool Debug = false;
  64 URI Proxy;
  65
  66 unsigned long CircleBuf::BwReadLimit=0;
  67 unsigned long CircleBuf::BwTickReadData=0;
  68 struct timeval CircleBuf::BwReadTick={0,0};
  69 const unsigned int CircleBuf::BW_HZ=10;
  70
  71 // CircleBuf::CircleBuf - Circular input buffer                         /*{{{*/
  72 // ---------------------------------------------------------------------
  73 /* */
  74 CircleBuf::CircleBuf(unsigned long Size) : Size(Size), Hash(0)
  75 {
  76    Buf = new unsigned char[Size];
  77    Reset();
  78
  79    CircleBuf::BwReadLimit = _config->FindI("Acquire::http::Dl-Limit",0)*1024;
  80 }
  81                                                                         /*}}}*/
  82 // CircleBuf::Reset - Reset to the default state                        /*{{{*/
  83 // ---------------------------------------------------------------------
  84 /* */
  85 void CircleBuf::Reset()
  86 {
  87    InP = 0;
  88    OutP = 0;
  89    StrPos = 0;
  90    MaxGet = (unsigned int)-1;
  91    OutQueue = string();
  92    if (Hash != 0)
  93    {
  94       delete Hash;
  95       Hash = new Hashes;
  96    }
  97 };
  98                                                                         /*}}}*/
  99 // CircleBuf::Read - Read from a FD into the circular buffer            /*{{{*/
 100 // ---------------------------------------------------------------------
 101 /* This fills up the buffer with as much data as is in the FD, assuming it
 102    is non-blocking.. */
 103 bool CircleBuf::Read(int Fd)
 104 {
 105    unsigned long BwReadMax;
 106
 107    while (1)
 108    {
 109       // Woops, buffer is full
 110       if (InP - OutP == Size)
 111          return true;
 112
 113       // what's left to read in this tick
 114       BwReadMax = CircleBuf::BwReadLimit/BW_HZ;
 115
 116       if(CircleBuf::BwReadLimit) {
 117          struct timeval now;
 118          gettimeofday(&now,0);
 119
 120          unsigned long d = (now.tv_sec-CircleBuf::BwReadTick.tv_sec)*1000000 +
 121             now.tv_usec-CircleBuf::BwReadTick.tv_usec;
 122          if(d > 1000000/BW_HZ) {
 123             CircleBuf::BwReadTick = now;
 124             CircleBuf::BwTickReadData = 0;
 125          }
 126
 127          if(CircleBuf::BwTickReadData >= BwReadMax) {
 128             usleep(1000000/BW_HZ);
 129             return true;
 130          }
 131       }
 132
 133       // Write the buffer segment
 134       int Res;
 135       if(CircleBuf::BwReadLimit) {
 136          Res = read(Fd,Buf + (InP%Size),
 137                     BwReadMax > LeftRead() ? LeftRead() : BwReadMax);
 138       } else
 139          Res = read(Fd,Buf + (InP%Size),LeftRead());
 140
 141       if(Res > 0 && BwReadLimit > 0)
 142          CircleBuf::BwTickReadData += Res;
 143
 144       if (Res == 0)
 145          return false;
 146       if (Res < 0)
 147       {
 148          if (errno == EAGAIN)
 149             return true;
 150          return false;
 151       }
 152
 153       if (InP == 0)
 154          gettimeofday(&Start,0);
 155       InP += Res;
 156    }
 157 }
 158                                                                         /*}}}*/
 159 // CircleBuf::Read - Put the string into the buffer                     /*{{{*/
 160 // ---------------------------------------------------------------------
 161 /* This will hold the string in and fill the buffer with it as it empties */
 162 bool CircleBuf::Read(string Data)
 163 {
 164    OutQueue += Data;
 165    FillOut();
 166    return true;
 167 }
 168                                                                         /*}}}*/
 169 // CircleBuf::FillOut - Fill the buffer from the output queue           /*{{{*/
 170 // ---------------------------------------------------------------------
 171 /* */
 172 void CircleBuf::FillOut()
 173 {
 174    if (OutQueue.empty() == true)
 175       return;
 176    while (1)
 177    {
 178       // Woops, buffer is full
 179       if (InP - OutP == Size)
 180          return;
 181
 182       // Write the buffer segment
 183       unsigned long Sz = LeftRead();
 184       if (OutQueue.length() - StrPos < Sz)
 185          Sz = OutQueue.length() - StrPos;
 186       memcpy(Buf + (InP%Size),OutQueue.c_str() + StrPos,Sz);
 187
 188       // Advance
 189       StrPos += Sz;
 190       InP += Sz;
 191       if (OutQueue.length() == StrPos)
 192       {
 193          StrPos = 0;
 194          OutQueue = "";
 195          return;
 196       }
 197    }
 198 }
 199                                                                         /*}}}*/
 200 // CircleBuf::Write - Write from the buffer into a FD                   /*{{{*/
 201 // ---------------------------------------------------------------------
 202 /* This empties the buffer into the FD. */
 203 bool CircleBuf::Write(int Fd)
 204 {
 205    while (1)
 206    {
 207       FillOut();
 208
 209       // Woops, buffer is empty
 210       if (OutP == InP)
 211          return true;
 212
 213       if (OutP == MaxGet)
 214          return true;
 215
 216       // Write the buffer segment
 217       int Res;
 218       Res = write(Fd,Buf + (OutP%Size),LeftWrite());
 219
 220       if (Res == 0)
 221          return false;
 222       if (Res < 0)
 223       {
 224          if (errno == EAGAIN)
 225             return true;
 226
 227          return false;
 228       }
 229
 230       if (Hash != 0)
 231          Hash->Add(Buf + (OutP%Size),Res);
 232
 233       OutP += Res;
 234    }
 235 }
 236                                                                         /*}}}*/
 237 // CircleBuf::WriteTillEl - Write from the buffer to a string           /*{{{*/
 238 // ---------------------------------------------------------------------
 239 /* This copies till the first empty line */
 240 bool CircleBuf::WriteTillEl(string &Data,bool Single)
 241 {
 242    // We cheat and assume it is unneeded to have more than one buffer load
 243    for (unsigned long I = OutP; I < InP; I++)
 244    {
 245       if (Buf[I%Size] != '\n')
 246          continue;
 247       ++I;
 248
 249       if (Single == false)
 250       {
 251          if (I < InP  && Buf[I%Size] == '\r')
 252             ++I;
 253          if (I >= InP || Buf[I%Size] != '\n')
 254             continue;
 255          ++I;
 256       }
 257
 258       Data = "";
 259       while (OutP < I)
 260       {
 261          unsigned long Sz = LeftWrite();
 262          if (Sz == 0)
 263             return false;
 264          if (I - OutP < Sz)
 265             Sz = I - OutP;
 266          Data += string((char *)(Buf + (OutP%Size)),Sz);
 267          OutP += Sz;
 268       }
 269       return true;
 270    }
 271    return false;
 272 }
 273                                                                         /*}}}*/
 274 // CircleBuf::Stats - Print out stats information                       /*{{{*/
 275 // ---------------------------------------------------------------------
 276 /* */
 277 void CircleBuf::Stats()
 278 {
 279    if (InP == 0)
 280       return;
 281
 282    struct timeval Stop;
 283    gettimeofday(&Stop,0);
 284 /*   float Diff = Stop.tv_sec - Start.tv_sec +
 285              (float)(Stop.tv_usec - Start.tv_usec)/1000000;
 286    clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
 287 }
 288                                                                         /*}}}*/
 289
 290 // ServerState::ServerState - Constructor                               /*{{{*/
 291 // ---------------------------------------------------------------------
 292 /* */
 293 ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner),
 294                         In(64*1024), Out(4*1024),
 295                         ServerName(Srv)
 296 {
 297    Reset();
 298 }
 299                                                                         /*}}}*/
 300 // ServerState::Open - Open a connection to the server                  /*{{{*/
 301 // ---------------------------------------------------------------------
 302 /* This opens a connection to the server. */
 303 bool ServerState::Open()
 304 {
 305    // Use the already open connection if possible.
 306    if (ServerFd != -1)
 307       return true;
 308
 309    Close();
 310    In.Reset();
 311    Out.Reset();
 312    Persistent = true;
 313
 314    // Determine the proxy setting
 315    string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
 316    if (!SpecificProxy.empty())
 317    {
 318            if (SpecificProxy == "DIRECT")
 319                    Proxy = "";
 320            else
 321                    Proxy = SpecificProxy;
 322    }
 323    else
 324    {
 325            string DefProxy = _config->Find("Acquire::http::Proxy");
 326            if (!DefProxy.empty())
 327            {
 328                    Proxy = DefProxy;
 329            }
 330            else
 331            {
 332                    char* result = getenv("http_proxy");
 333                    Proxy = result ? result : "";
 334            }
 335    }
 336
 337    // Parse no_proxy, a , separated list of domains
 338    if (getenv("no_proxy") != 0)
 339    {
 340       if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 341          Proxy = "";
 342    }
 343
 344    // Determine what host and port to use based on the proxy settings
 345    int Port = 0;
 346    string Host;
 347    if (Proxy.empty() == true || Proxy.Host.empty() == true)
 348    {
 349       if (ServerName.Port != 0)
 350          Port = ServerName.Port;
 351       Host = ServerName.Host;
 352    }
 353    else
 354    {
 355       if (Proxy.Port != 0)
 356          Port = Proxy.Port;
 357       Host = Proxy.Host;
 358    }
 359
 360    // Connect to the remote server
 361    if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
 362       return false;
 363
 364    return true;
 365 }
 366                                                                         /*}}}*/
 367 // ServerState::Close - Close a connection to the server                /*{{{*/
 368 // ---------------------------------------------------------------------
 369 /* */
 370 bool ServerState::Close()
 371 {
 372    close(ServerFd);
 373    ServerFd = -1;
 374    return true;
 375 }
 376                                                                         /*}}}*/
 377 // ServerState::RunHeaders - Get the headers before the data            /*{{{*/
 378 // ---------------------------------------------------------------------
 379 /* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
 380    parse error occurred */
 381 ServerState::RunHeadersResult ServerState::RunHeaders()
 382 {
 383    State = Header;
 384
 385    Owner->Status(_("Waiting for headers"));
 386
 387    Major = 0;
 388    Minor = 0;
 389    Result = 0;
 390    Size = 0;
 391    StartPos = 0;
 392    Encoding = Closes;
 393    HaveContent = false;
 394    time(&Date);
 395
 396    do
 397    {
 398       string Data;
 399       if (In.WriteTillEl(Data) == false)
 400          continue;
 401
 402       if (Debug == true)
 403          clog << Data;
 404
 405       for (string::const_iterator I = Data.begin(); I < Data.end(); I++)
 406       {
 407          string::const_iterator J = I;
 408          for (; J != Data.end() && *J != '\n' && *J != '\r';J++);
 409          if (HeaderLine(string(I,J)) == false)
 410             return RUN_HEADERS_PARSE_ERROR;
 411          I = J;
 412       }
 413
 414       // 100 Continue is a Nop...
 415       if (Result == 100)
 416          continue;
 417
 418       // Tidy up the connection persistance state.
 419       if (Encoding == Closes && HaveContent == true)
 420          Persistent = false;
 421
 422       return RUN_HEADERS_OK;
 423    }
 424    while (Owner->Go(false,this) == true);
 425
 426    return RUN_HEADERS_IO_ERROR;
 427 }
 428                                                                         /*}}}*/
 429 // ServerState::RunData - Transfer the data from the socket             /*{{{*/
 430 // ---------------------------------------------------------------------
 431 /* */
 432 bool ServerState::RunData()
 433 {
 434    State = Data;
 435
 436    // Chunked transfer encoding is fun..
 437    if (Encoding == Chunked)
 438    {
 439       while (1)
 440       {
 441          // Grab the block size
 442          bool Last = true;
 443          string Data;
 444          In.Limit(-1);
 445          do
 446          {
 447             if (In.WriteTillEl(Data,true) == true)
 448                break;
 449          }
 450          while ((Last = Owner->Go(false,this)) == true);
 451
 452          if (Last == false)
 453             return false;
 454
 455          // See if we are done
 456          unsigned long Len = strtol(Data.c_str(),0,16);
 457          if (Len == 0)
 458          {
 459             In.Limit(-1);
 460
 461             // We have to remove the entity trailer
 462             Last = true;
 463             do
 464             {
 465                if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
 466                   break;
 467             }
 468             while ((Last = Owner->Go(false,this)) == true);
 469             if (Last == false)
 470                return false;
 471             return !_error->PendingError();
 472          }
 473
 474          // Transfer the block
 475          In.Limit(Len);
 476          while (Owner->Go(true,this) == true)
 477             if (In.IsLimit() == true)
 478                break;
 479
 480          // Error
 481          if (In.IsLimit() == false)
 482             return false;
 483
 484          // The server sends an extra new line before the next block specifier..
 485          In.Limit(-1);
 486          Last = true;
 487          do
 488          {
 489             if (In.WriteTillEl(Data,true) == true)
 490                break;
 491          }
 492          while ((Last = Owner->Go(false,this)) == true);
 493          if (Last == false)
 494             return false;
 495       }
 496    }
 497    else
 498    {
 499       /* Closes encoding is used when the server did not specify a size, the
 500          loss of the connection means we are done */
 501       if (Encoding == Closes)
 502          In.Limit(-1);
 503       else
 504          In.Limit(Size - StartPos);
 505
 506       // Just transfer the whole block.
 507       do
 508       {
 509          if (In.IsLimit() == false)
 510             continue;
 511
 512          In.Limit(-1);
 513          return !_error->PendingError();
 514       }
 515       while (Owner->Go(true,this) == true);
 516    }
 517
 518    return Owner->Flush(this) && !_error->PendingError();
 519 }
 520                                                                         /*}}}*/
 521 // ServerState::HeaderLine - Process a header line                      /*{{{*/
 522 // ---------------------------------------------------------------------
 523 /* */
 524 bool ServerState::HeaderLine(string Line)
 525 {
 526    if (Line.empty() == true)
 527       return true;
 528
 529    // The http server might be trying to do something evil.
 530    if (Line.length() >= MAXLEN)
 531       return _error->Error(_("Got a single header line over %u chars"),MAXLEN);
 532
 533    string::size_type Pos = Line.find(' ');
 534    if (Pos == string::npos || Pos+1 > Line.length())
 535    {
 536       // Blah, some servers use "connection:closes", evil.
 537       Pos = Line.find(':');
 538       if (Pos == string::npos || Pos + 2 > Line.length())
 539          return _error->Error(_("Bad header line"));
 540       Pos++;
 541    }
 542
 543    // Parse off any trailing spaces between the : and the next word.
 544    string::size_type Pos2 = Pos;
 545    while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
 546       Pos2++;
 547
 548    string Tag = string(Line,0,Pos);
 549    string Val = string(Line,Pos2);
 550
 551    if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
 552    {
 553       // Evil servers return no version
 554       if (Line[4] == '/')
 555       {
 556          if (sscanf(Line.c_str(),"HTTP/%u.%u %u%[^\n]",&Major,&Minor,
 557                     &Result,Code) != 4)
 558             return _error->Error(_("The HTTP server sent an invalid reply header"));
 559       }
 560       else
 561       {
 562          Major = 0;
 563          Minor = 9;
 564          if (sscanf(Line.c_str(),"HTTP %u%[^\n]",&Result,Code) != 2)
 565             return _error->Error(_("The HTTP server sent an invalid reply header"));
 566       }
 567
 568       /* Check the HTTP response header to get the default persistance
 569          state. */
 570       if (Major < 1)
 571          Persistent = false;
 572       else
 573       {
 574          if (Major == 1 && Minor <= 0)
 575             Persistent = false;
 576          else
 577             Persistent = true;
 578       }
 579
 580       return true;
 581    }
 582
 583    if (stringcasecmp(Tag,"Content-Length:") == 0)
 584    {
 585       if (Encoding == Closes)
 586          Encoding = Stream;
 587       HaveContent = true;
 588
 589       // The length is already set from the Content-Range header
 590       if (StartPos != 0)
 591          return true;
 592
 593       if (sscanf(Val.c_str(),"%lu",&Size) != 1)
 594          return _error->Error(_("The HTTP server sent an invalid Content-Length header"));
 595       return true;
 596    }
 597
 598    if (stringcasecmp(Tag,"Content-Type:") == 0)
 599    {
 600       HaveContent = true;
 601       return true;
 602    }
 603
 604    if (stringcasecmp(Tag,"Content-Range:") == 0)
 605    {
 606       HaveContent = true;
 607
 608       if (sscanf(Val.c_str(),"bytes %lu-%*u/%lu",&StartPos,&Size) != 2)
 609          return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
 610       if ((unsigned)StartPos > Size)
 611          return _error->Error(_("This HTTP server has broken range support"));
 612       return true;
 613    }
 614
 615    if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
 616    {
 617       HaveContent = true;
 618       if (stringcasecmp(Val,"chunked") == 0)
 619          Encoding = Chunked;
 620       return true;
 621    }
 622
 623    if (stringcasecmp(Tag,"Connection:") == 0)
 624    {
 625       if (stringcasecmp(Val,"close") == 0)
 626          Persistent = false;
 627       if (stringcasecmp(Val,"keep-alive") == 0)
 628          Persistent = true;
 629       return true;
 630    }
 631
 632    if (stringcasecmp(Tag,"Last-Modified:") == 0)
 633    {
 634       if (StrToTime(Val,Date) == false)
 635          return _error->Error(_("Unknown date format"));
 636       return true;
 637    }
 638
 639    if (stringcasecmp(Tag,"Location:") == 0)
 640    {
 641       Location = Val;
 642       return true;
 643    }
 644
 645    return true;
 646 }
 647                                                                         /*}}}*/
 648
 649 // HttpMethod::SendReq - Send the HTTP request                          /*{{{*/
 650 // ---------------------------------------------------------------------
 651 /* This places the http request in the outbound buffer */
 652 void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
 653 {
 654    URI Uri = Itm->Uri;
 655
 656    // The HTTP server expects a hostname with a trailing :port
 657    char Buf[1000];
 658    string ProperHost = Uri.Host;
 659    if (Uri.Port != 0)
 660    {
 661       sprintf(Buf,":%u",Uri.Port);
 662       ProperHost += Buf;
 663    }
 664
 665    // Just in case.
 666    if (Itm->Uri.length() >= sizeof(Buf))
 667        abort();
 668
 669    /* Build the request. We include a keep-alive header only for non-proxy
 670       requests. This is to tweak old http/1.0 servers that do support keep-alive
 671       but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
 672       will glitch HTTP/1.0 proxies because they do not filter it out and
 673       pass it on, HTTP/1.1 says the connection should default to keep alive
 674       and we expect the proxy to do this */
 675    if (Proxy.empty() == true || Proxy.Host.empty())
 676       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
 677               QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
 678    else
 679    {
 680       /* Generate a cache control header if necessary. We place a max
 681          cache age on index files, optionally set a no-cache directive
 682          and a no-store directive for archives. */
 683       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 684               Itm->Uri.c_str(),ProperHost.c_str());
 685    }
 686    // generate a cache control header (if needed)
 687    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 688    {
 689       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 690    }
 691    else
 692    {
 693       if (Itm->IndexFile == true)
 694       {
 695          sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 696                  _config->FindI("Acquire::http::Max-Age",0));
 697       }
 698       else
 699       {
 700          if (_config->FindB("Acquire::http::No-Store",false) == true)
 701             strcat(Buf,"Cache-Control: no-store\r\n");
 702       }
 703    }
 704
 705
 706    string Req = Buf;
 707
 708    // Check for a partial file
 709    struct stat SBuf;
 710    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 711    {
 712       // In this case we send an if-range query with a range header
 713       sprintf(Buf,"Range: bytes=%li-\r\nIf-Range: %s\r\n",(long)SBuf.st_size - 1,
 714               TimeRFC1123(SBuf.st_mtime).c_str());
 715       Req += Buf;
 716    }
 717    else
 718    {
 719       if (Itm->LastModified != 0)
 720       {
 721          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 722          Req += Buf;
 723       }
 724    }
 725
 726    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 727       Req += string("Proxy-Authorization: Basic ") +
 728           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 729
 730    maybe_add_auth (Uri, _config->FindFile("Dir::Etc::netrc"));
 731    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 732    {
 733       Req += string("Authorization: Basic ") +
 734           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 735    }
 736    Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
 737                 "Debian APT-HTTP/1.3 ("VERSION")") + "\r\n\r\n";
 738
 739    if (Debug == true)
 740       cerr << Req << endl;
 741
 742    Out.Read(Req);
 743 }
 744                                                                         /*}}}*/
 745 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 746 // ---------------------------------------------------------------------
 747 /* This runs the select loop over the server FDs, Output file FDs and
 748    stdin. */
 749 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 750 {
 751    // Server has closed the connection
 752    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 753                                ToFile == false))
 754       return false;
 755
 756    fd_set rfds,wfds;
 757    FD_ZERO(&rfds);
 758    FD_ZERO(&wfds);
 759
 760    /* Add the server. We only send more requests if the connection will
 761       be persisting */
 762    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 763        && Srv->Persistent == true)
 764       FD_SET(Srv->ServerFd,&wfds);
 765    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 766       FD_SET(Srv->ServerFd,&rfds);
 767
 768    // Add the file
 769    int FileFD = -1;
 770    if (File != 0)
 771       FileFD = File->Fd();
 772
 773    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 774       FD_SET(FileFD,&wfds);
 775
 776    // Add stdin
 777    FD_SET(STDIN_FILENO,&rfds);
 778
 779    // Figure out the max fd
 780    int MaxFd = FileFD;
 781    if (MaxFd < Srv->ServerFd)
 782       MaxFd = Srv->ServerFd;
 783
 784    // Select
 785    struct timeval tv;
 786    tv.tv_sec = TimeOut;
 787    tv.tv_usec = 0;
 788    int Res = 0;
 789    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 790    {
 791       if (errno == EINTR)
 792          return true;
 793       return _error->Errno("select",_("Select failed"));
 794    }
 795
 796    if (Res == 0)
 797    {
 798       _error->Error(_("Connection timed out"));
 799       return ServerDie(Srv);
 800    }
 801
 802    // Handle server IO
 803    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 804    {
 805       errno = 0;
 806       if (Srv->In.Read(Srv->ServerFd) == false)
 807          return ServerDie(Srv);
 808    }
 809
 810    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 811    {
 812       errno = 0;
 813       if (Srv->Out.Write(Srv->ServerFd) == false)
 814          return ServerDie(Srv);
 815    }
 816
 817    // Send data to the file
 818    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 819    {
 820       if (Srv->In.Write(FileFD) == false)
 821          return _error->Errno("write",_("Error writing to output file"));
 822    }
 823
 824    // Handle commands from APT
 825    if (FD_ISSET(STDIN_FILENO,&rfds))
 826    {
 827       if (Run(true) != -1)
 828          exit(100);
 829    }
 830
 831    return true;
 832 }
 833                                                                         /*}}}*/
 834 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 835 // ---------------------------------------------------------------------
 836 /* This takes the current input buffer from the Server FD and writes it
 837    into the file */
 838 bool HttpMethod::Flush(ServerState *Srv)
 839 {
 840    if (File != 0)
 841    {
 842       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 843       // can't be set
 844       if (File->Name() != "/dev/null")
 845          SetNonBlock(File->Fd(),false);
 846       if (Srv->In.WriteSpace() == false)
 847          return true;
 848
 849       while (Srv->In.WriteSpace() == true)
 850       {
 851          if (Srv->In.Write(File->Fd()) == false)
 852             return _error->Errno("write",_("Error writing to file"));
 853          if (Srv->In.IsLimit() == true)
 854             return true;
 855       }
 856
 857       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 858          return true;
 859    }
 860    return false;
 861 }
 862                                                                         /*}}}*/
 863 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 864 // ---------------------------------------------------------------------
 865 /* */
 866 bool HttpMethod::ServerDie(ServerState *Srv)
 867 {
 868    unsigned int LErrno = errno;
 869
 870    // Dump the buffer to the file
 871    if (Srv->State == ServerState::Data)
 872    {
 873       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 874       // can't be set
 875       if (File->Name() != "/dev/null")
 876          SetNonBlock(File->Fd(),false);
 877       while (Srv->In.WriteSpace() == true)
 878       {
 879          if (Srv->In.Write(File->Fd()) == false)
 880             return _error->Errno("write",_("Error writing to the file"));
 881
 882          // Done
 883          if (Srv->In.IsLimit() == true)
 884             return true;
 885       }
 886    }
 887
 888    // See if this is because the server finished the data stream
 889    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 890        Srv->Encoding != ServerState::Closes)
 891    {
 892       Srv->Close();
 893       if (LErrno == 0)
 894          return _error->Error(_("Error reading from server. Remote end closed connection"));
 895       errno = LErrno;
 896       return _error->Errno("read",_("Error reading from server"));
 897    }
 898    else
 899    {
 900       Srv->In.Limit(-1);
 901
 902       // Nothing left in the buffer
 903       if (Srv->In.WriteSpace() == false)
 904          return false;
 905
 906       // We may have got multiple responses back in one packet..
 907       Srv->Close();
 908       return true;
 909    }
 910
 911    return false;
 912 }
 913                                                                         /*}}}*/
 914 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 915 // ---------------------------------------------------------------------
 916 /* We look at the header data we got back from the server and decide what
 917    to do. Returns DealWithHeadersResult (see http.h for details).
 918  */
 919 HttpMethod::DealWithHeadersResult
 920 HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 921 {
 922    // Not Modified
 923    if (Srv->Result == 304)
 924    {
 925       unlink(Queue->DestFile.c_str());
 926       Res.IMSHit = true;
 927       Res.LastModified = Queue->LastModified;
 928       return IMS_HIT;
 929    }
 930
 931    /* Redirect
 932     *
 933     * Note that it is only OK for us to treat all redirection the same
 934     * because we *always* use GET, not other HTTP methods.  There are
 935     * three redirection codes for which it is not appropriate that we
 936     * redirect.  Pass on those codes so the error handling kicks in.
 937     */
 938    if (AllowRedirect
 939        && (Srv->Result > 300 && Srv->Result < 400)
 940        && (Srv->Result != 300       // Multiple Choices
 941            && Srv->Result != 304    // Not Modified
 942            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 943    {
 944       if (!Srv->Location.empty())
 945       {
 946          NextURI = Srv->Location;
 947          return TRY_AGAIN_OR_REDIRECT;
 948       }
 949       /* else pass through for error message */
 950    }
 951
 952    /* We have a reply we dont handle. This should indicate a perm server
 953       failure */
 954    if (Srv->Result < 200 || Srv->Result >= 300)
 955    {
 956       char err[255];
 957       snprintf(err,sizeof(err)-1,"HttpError%i",Srv->Result);
 958       SetFailReason(err);
 959       _error->Error("%u %s",Srv->Result,Srv->Code);
 960       if (Srv->HaveContent == true)
 961          return ERROR_WITH_CONTENT_PAGE;
 962       return ERROR_UNRECOVERABLE;
 963    }
 964
 965    // This is some sort of 2xx 'data follows' reply
 966    Res.LastModified = Srv->Date;
 967    Res.Size = Srv->Size;
 968
 969    // Open the file
 970    delete File;
 971    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
 972    if (_error->PendingError() == true)
 973       return ERROR_NOT_FROM_SERVER;
 974
 975    FailFile = Queue->DestFile;
 976    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
 977    FailFd = File->Fd();
 978    FailTime = Srv->Date;
 979
 980    // Set the expected size
 981    if (Srv->StartPos >= 0)
 982    {
 983       Res.ResumePoint = Srv->StartPos;
 984       if (ftruncate(File->Fd(),Srv->StartPos) < 0)
 985          _error->Errno("ftruncate", _("Failed to truncate file"));
 986    }
 987
 988    // Set the start point
 989    lseek(File->Fd(),0,SEEK_END);
 990
 991    delete Srv->In.Hash;
 992    Srv->In.Hash = new Hashes;
 993
 994    // Fill the Hash if the file is non-empty (resume)
 995    if (Srv->StartPos > 0)
 996    {
 997       lseek(File->Fd(),0,SEEK_SET);
 998       if (Srv->In.Hash->AddFD(File->Fd(),Srv->StartPos) == false)
 999       {
1000          _error->Errno("read",_("Problem hashing file"));
1001          return ERROR_NOT_FROM_SERVER;
1002       }
1003       lseek(File->Fd(),0,SEEK_END);
1004    }
1005
1006    SetNonBlock(File->Fd(),true);
1007    return FILE_IS_OPEN;
1008 }
1009                                                                         /*}}}*/
1010 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1011 // ---------------------------------------------------------------------
1012 /* This closes and timestamps the open file. This is neccessary to get
1013    resume behavoir on user abort */
1014 void HttpMethod::SigTerm(int)
1015 {
1016    if (FailFd == -1)
1017       _exit(100);
1018    close(FailFd);
1019
1020    // Timestamp
1021    struct utimbuf UBuf;
1022    UBuf.actime = FailTime;
1023    UBuf.modtime = FailTime;
1024    utime(FailFile.c_str(),&UBuf);
1025
1026    _exit(100);
1027 }
1028                                                                         /*}}}*/
1029 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1030 // ---------------------------------------------------------------------
1031 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1032    depth. */
1033 bool HttpMethod::Fetch(FetchItem *)
1034 {
1035    if (Server == 0)
1036       return true;
1037
1038    // Queue the requests
1039    int Depth = -1;
1040    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1041         I = I->Next, Depth++)
1042    {
1043       // If pipelining is disabled, we only queue 1 request
1044       if (Server->Pipeline == false && Depth >= 0)
1045          break;
1046
1047       // Make sure we stick with the same server
1048       if (Server->Comp(I->Uri) == false)
1049          break;
1050       if (QueueBack == I)
1051       {
1052          QueueBack = I->Next;
1053          SendReq(I,Server->Out);
1054          continue;
1055       }
1056    }
1057
1058    return true;
1059 };
1060                                                                         /*}}}*/
1061 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1062 // ---------------------------------------------------------------------
1063 /* We stash the desired pipeline depth */
1064 bool HttpMethod::Configuration(string Message)
1065 {
1066    if (pkgAcqMethod::Configuration(Message) == false)
1067       return false;
1068
1069    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1070    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1071    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1072                                   PipelineDepth);
1073    Debug = _config->FindB("Debug::Acquire::http",false);
1074    AutoDetectProxyCmd = _config->Find("Acquire::http::ProxyAutoDetect");
1075
1076    // Get the proxy to use
1077    AutoDetectProxy();
1078
1079    return true;
1080 }
1081                                                                         /*}}}*/
1082 // HttpMethod::Loop - Main loop                                         /*{{{*/
1083 // ---------------------------------------------------------------------
1084 /* */
1085 int HttpMethod::Loop()
1086 {
1087    typedef vector<string> StringVector;
1088    typedef vector<string>::iterator StringVectorIterator;
1089    map<string, StringVector> Redirected;
1090
1091    signal(SIGTERM,SigTerm);
1092    signal(SIGINT,SigTerm);
1093
1094    Server = 0;
1095
1096    int FailCounter = 0;
1097    while (1)
1098    {
1099       // We have no commands, wait for some to arrive
1100       if (Queue == 0)
1101       {
1102          if (WaitFd(STDIN_FILENO) == false)
1103             return 0;
1104       }
1105
1106       /* Run messages, we can accept 0 (no message) if we didn't
1107          do a WaitFd above.. Otherwise the FD is closed. */
1108       int Result = Run(true);
1109       if (Result != -1 && (Result != 0 || Queue == 0))
1110          return 100;
1111
1112       if (Queue == 0)
1113          continue;
1114
1115       // Connect to the server
1116       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1117       {
1118          delete Server;
1119          Server = new ServerState(Queue->Uri,this);
1120       }
1121       /* If the server has explicitly said this is the last connection
1122          then we pre-emptively shut down the pipeline and tear down
1123          the connection. This will speed up HTTP/1.0 servers a tad
1124          since we don't have to wait for the close sequence to
1125          complete */
1126       if (Server->Persistent == false)
1127          Server->Close();
1128
1129       // Reset the pipeline
1130       if (Server->ServerFd == -1)
1131          QueueBack = Queue;
1132
1133       // Connnect to the host
1134       if (Server->Open() == false)
1135       {
1136          Fail(true);
1137          delete Server;
1138          Server = 0;
1139          continue;
1140       }
1141
1142       // Fill the pipeline.
1143       Fetch(0);
1144
1145       // Fetch the next URL header data from the server.
1146       switch (Server->RunHeaders())
1147       {
1148          case ServerState::RUN_HEADERS_OK:
1149          break;
1150
1151          // The header data is bad
1152          case ServerState::RUN_HEADERS_PARSE_ERROR:
1153          {
1154             _error->Error(_("Bad header data"));
1155             Fail(true);
1156             RotateDNS();
1157             continue;
1158          }
1159
1160          // The server closed a connection during the header get..
1161          default:
1162          case ServerState::RUN_HEADERS_IO_ERROR:
1163          {
1164             FailCounter++;
1165             _error->Discard();
1166             Server->Close();
1167             Server->Pipeline = false;
1168
1169             if (FailCounter >= 2)
1170             {
1171                Fail(_("Connection failed"),true);
1172                FailCounter = 0;
1173             }
1174
1175             RotateDNS();
1176             continue;
1177          }
1178       };
1179
1180       // Decide what to do.
1181       FetchResult Res;
1182       Res.Filename = Queue->DestFile;
1183       switch (DealWithHeaders(Res,Server))
1184       {
1185          // Ok, the file is Open
1186          case FILE_IS_OPEN:
1187          {
1188             URIStart(Res);
1189
1190             // Run the data
1191             bool Result =  Server->RunData();
1192
1193             /* If the server is sending back sizeless responses then fill in
1194                the size now */
1195             if (Res.Size == 0)
1196                Res.Size = File->Size();
1197
1198             // Close the file, destroy the FD object and timestamp it
1199             FailFd = -1;
1200             delete File;
1201             File = 0;
1202
1203             // Timestamp
1204             struct utimbuf UBuf;
1205             time(&UBuf.actime);
1206             UBuf.actime = Server->Date;
1207             UBuf.modtime = Server->Date;
1208             utime(Queue->DestFile.c_str(),&UBuf);
1209
1210             // Send status to APT
1211             if (Result == true)
1212             {
1213                Res.TakeHashes(*Server->In.Hash);
1214                URIDone(Res);
1215             }
1216             else
1217             {
1218                if (Server->ServerFd == -1)
1219                {
1220                   FailCounter++;
1221                   _error->Discard();
1222                   Server->Close();
1223
1224                   if (FailCounter >= 2)
1225                   {
1226                      Fail(_("Connection failed"),true);
1227                      FailCounter = 0;
1228                   }
1229
1230                   QueueBack = Queue;
1231                }
1232                else
1233                   Fail(true);
1234             }
1235             break;
1236          }
1237
1238          // IMS hit
1239          case IMS_HIT:
1240          {
1241             URIDone(Res);
1242             break;
1243          }
1244
1245          // Hard server error, not found or something
1246          case ERROR_UNRECOVERABLE:
1247          {
1248             Fail();
1249             break;
1250          }
1251
1252          // Hard internal error, kill the connection and fail
1253          case ERROR_NOT_FROM_SERVER:
1254          {
1255             delete File;
1256             File = 0;
1257
1258             Fail();
1259             RotateDNS();
1260             Server->Close();
1261             break;
1262          }
1263
1264          // We need to flush the data, the header is like a 404 w/ error text
1265          case ERROR_WITH_CONTENT_PAGE:
1266          {
1267             Fail();
1268
1269             // Send to content to dev/null
1270             File = new FileFd("/dev/null",FileFd::WriteExists);
1271             Server->RunData();
1272             delete File;
1273             File = 0;
1274             break;
1275          }
1276
1277          // Try again with a new URL
1278          case TRY_AGAIN_OR_REDIRECT:
1279          {
1280             // Clear rest of response if there is content
1281             if (Server->HaveContent)
1282             {
1283                File = new FileFd("/dev/null",FileFd::WriteExists);
1284                Server->RunData();
1285                delete File;
1286                File = 0;
1287             }
1288
1289             /* Detect redirect loops.  No more redirects are allowed
1290                after the same URI is seen twice in a queue item. */
1291             StringVector &R = Redirected[Queue->DestFile];
1292             bool StopRedirects = false;
1293             if (R.size() == 0)
1294                R.push_back(Queue->Uri);
1295             else if (R[0] == "STOP" || R.size() > 10)
1296                StopRedirects = true;
1297             else
1298             {
1299                for (StringVectorIterator I = R.begin(); I != R.end(); I++)
1300                   if (Queue->Uri == *I)
1301                   {
1302                      R[0] = "STOP";
1303                      break;
1304                   }
1305
1306                R.push_back(Queue->Uri);
1307             }
1308
1309             if (StopRedirects == false)
1310                Redirect(NextURI);
1311             else
1312                Fail();
1313
1314             break;
1315          }
1316
1317          default:
1318          Fail(_("Internal error"));
1319          break;
1320       }
1321
1322       FailCounter = 0;
1323    }
1324
1325    return 0;
1326 }
1327                                                                         /*}}}*/
1328 // HttpMethod::AutoDetectProxy - auto detect proxy                      /*{{{*/
1329 // ---------------------------------------------------------------------
1330 /* */
1331 bool HttpMethod::AutoDetectProxy()
1332 {
1333    if (AutoDetectProxyCmd.empty())
1334       return true;
1335
1336    if (Debug)
1337       clog << "Using auto proxy detect command: " << AutoDetectProxyCmd << endl;
1338
1339    int Pipes[2] = {-1,-1};
1340    if (pipe(Pipes) != 0)
1341       return _error->Errno("pipe", "Failed to create Pipe");
1342
1343    pid_t Process = ExecFork();
1344    if (Process == 0)
1345    {
1346       dup2(Pipes[1],STDOUT_FILENO);
1347       SetCloseExec(STDOUT_FILENO,false);
1348
1349       const char *Args[2];
1350       Args[0] = AutoDetectProxyCmd.c_str();
1351       Args[1] = 0;
1352       execv(Args[0],(char **)Args);
1353       cerr << "Failed to exec method " << Args[0] << endl;
1354       _exit(100);
1355    }
1356    char buf[512];
1357    int InFd = Pipes[0];
1358    if (read(InFd, buf, sizeof(buf)) < 0)
1359       return _error->Errno("read", "Failed to read");
1360    ExecWait(Process, "ProxyAutoDetect");
1361
1362    if (Debug)
1363       clog << "auto detect command returned: '" << buf << "'" << endl;
1364
1365    if (strstr(buf, "http://") == buf)
1366       _config->Set("Acquire::http::proxy", _strstrip(buf));
1367
1368    return true;
1369 }
1370                                                                         /*}}}*/
1371
1372