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 <iostream>
  42 #include <map>
  43 #include <apti18n.h>
  44
  45 // Internet stuff
  46 #include <netdb.h>
  47
  48 #include "config.h"
  49 #include "connect.h"
  50 #include "rfc2553emu.h"
  51 #include "http.h"
  52
  53                                                                         /*}}}*/
  54 using namespace std;
  55
  56 string HttpMethod::FailFile;
  57 int HttpMethod::FailFd = -1;
  58 time_t HttpMethod::FailTime = 0;
  59 unsigned long PipelineDepth = 10;
  60 unsigned long TimeOut = 120;
  61 bool AllowRedirect = false;
  62 bool Debug = false;
  63 URI Proxy;
  64
  65 unsigned long CircleBuf::BwReadLimit=0;
  66 unsigned long CircleBuf::BwTickReadData=0;
  67 struct timeval CircleBuf::BwReadTick={0,0};
  68 const unsigned int CircleBuf::BW_HZ=10;
  69
  70 // CircleBuf::CircleBuf - Circular input buffer                         /*{{{*/
  71 // ---------------------------------------------------------------------
  72 /* */
  73 CircleBuf::CircleBuf(unsigned long Size) : Size(Size), Hash(0)
  74 {
  75    Buf = new unsigned char[Size];
  76    Reset();
  77
  78    CircleBuf::BwReadLimit = _config->FindI("Acquire::http::Dl-Limit",0)*1024;
  79 }
  80                                                                         /*}}}*/
  81 // CircleBuf::Reset - Reset to the default state                        /*{{{*/
  82 // ---------------------------------------------------------------------
  83 /* */
  84 void CircleBuf::Reset()
  85 {
  86    InP = 0;
  87    OutP = 0;
  88    StrPos = 0;
  89    MaxGet = (unsigned int)-1;
  90    OutQueue = string();
  91    if (Hash != 0)
  92    {
  93       delete Hash;
  94       Hash = new Hashes;
  95    }
  96 };
  97                                                                         /*}}}*/
  98 // CircleBuf::Read - Read from a FD into the circular buffer            /*{{{*/
  99 // ---------------------------------------------------------------------
 100 /* This fills up the buffer with as much data as is in the FD, assuming it
 101    is non-blocking.. */
 102 bool CircleBuf::Read(int Fd)
 103 {
 104    unsigned long BwReadMax;
 105
 106    while (1)
 107    {
 108       // Woops, buffer is full
 109       if (InP - OutP == Size)
 110          return true;
 111
 112       // what's left to read in this tick
 113       BwReadMax = CircleBuf::BwReadLimit/BW_HZ;
 114
 115       if(CircleBuf::BwReadLimit) {
 116          struct timeval now;
 117          gettimeofday(&now,0);
 118
 119          unsigned long d = (now.tv_sec-CircleBuf::BwReadTick.tv_sec)*1000000 +
 120             now.tv_usec-CircleBuf::BwReadTick.tv_usec;
 121          if(d > 1000000/BW_HZ) {
 122             CircleBuf::BwReadTick = now;
 123             CircleBuf::BwTickReadData = 0;
 124          }
 125
 126          if(CircleBuf::BwTickReadData >= BwReadMax) {
 127             usleep(1000000/BW_HZ);
 128             return true;
 129          }
 130       }
 131
 132       // Write the buffer segment
 133       int Res;
 134       if(CircleBuf::BwReadLimit) {
 135          Res = read(Fd,Buf + (InP%Size),
 136                     BwReadMax > LeftRead() ? LeftRead() : BwReadMax);
 137       } else
 138          Res = read(Fd,Buf + (InP%Size),LeftRead());
 139
 140       if(Res > 0 && BwReadLimit > 0)
 141          CircleBuf::BwTickReadData += Res;
 142
 143       if (Res == 0)
 144          return false;
 145       if (Res < 0)
 146       {
 147          if (errno == EAGAIN)
 148             return true;
 149          return false;
 150       }
 151
 152       if (InP == 0)
 153          gettimeofday(&Start,0);
 154       InP += Res;
 155    }
 156 }
 157                                                                         /*}}}*/
 158 // CircleBuf::Read - Put the string into the buffer                     /*{{{*/
 159 // ---------------------------------------------------------------------
 160 /* This will hold the string in and fill the buffer with it as it empties */
 161 bool CircleBuf::Read(string Data)
 162 {
 163    OutQueue += Data;
 164    FillOut();
 165    return true;
 166 }
 167                                                                         /*}}}*/
 168 // CircleBuf::FillOut - Fill the buffer from the output queue           /*{{{*/
 169 // ---------------------------------------------------------------------
 170 /* */
 171 void CircleBuf::FillOut()
 172 {
 173    if (OutQueue.empty() == true)
 174       return;
 175    while (1)
 176    {
 177       // Woops, buffer is full
 178       if (InP - OutP == Size)
 179          return;
 180
 181       // Write the buffer segment
 182       unsigned long Sz = LeftRead();
 183       if (OutQueue.length() - StrPos < Sz)
 184          Sz = OutQueue.length() - StrPos;
 185       memcpy(Buf + (InP%Size),OutQueue.c_str() + StrPos,Sz);
 186
 187       // Advance
 188       StrPos += Sz;
 189       InP += Sz;
 190       if (OutQueue.length() == StrPos)
 191       {
 192          StrPos = 0;
 193          OutQueue = "";
 194          return;
 195       }
 196    }
 197 }
 198                                                                         /*}}}*/
 199 // CircleBuf::Write - Write from the buffer into a FD                   /*{{{*/
 200 // ---------------------------------------------------------------------
 201 /* This empties the buffer into the FD. */
 202 bool CircleBuf::Write(int Fd)
 203 {
 204    while (1)
 205    {
 206       FillOut();
 207
 208       // Woops, buffer is empty
 209       if (OutP == InP)
 210          return true;
 211
 212       if (OutP == MaxGet)
 213          return true;
 214
 215       // Write the buffer segment
 216       int Res;
 217       Res = write(Fd,Buf + (OutP%Size),LeftWrite());
 218
 219       if (Res == 0)
 220          return false;
 221       if (Res < 0)
 222       {
 223          if (errno == EAGAIN)
 224             return true;
 225
 226          return false;
 227       }
 228
 229       if (Hash != 0)
 230          Hash->Add(Buf + (OutP%Size),Res);
 231
 232       OutP += Res;
 233    }
 234 }
 235                                                                         /*}}}*/
 236 // CircleBuf::WriteTillEl - Write from the buffer to a string           /*{{{*/
 237 // ---------------------------------------------------------------------
 238 /* This copies till the first empty line */
 239 bool CircleBuf::WriteTillEl(string &Data,bool Single)
 240 {
 241    // We cheat and assume it is unneeded to have more than one buffer load
 242    for (unsigned long I = OutP; I < InP; I++)
 243    {
 244       if (Buf[I%Size] != '\n')
 245          continue;
 246       ++I;
 247
 248       if (Single == false)
 249       {
 250          if (I < InP  && Buf[I%Size] == '\r')
 251             ++I;
 252          if (I >= InP || Buf[I%Size] != '\n')
 253             continue;
 254          ++I;
 255       }
 256
 257       Data = "";
 258       while (OutP < I)
 259       {
 260          unsigned long Sz = LeftWrite();
 261          if (Sz == 0)
 262             return false;
 263          if (I - OutP < Sz)
 264             Sz = I - OutP;
 265          Data += string((char *)(Buf + (OutP%Size)),Sz);
 266          OutP += Sz;
 267       }
 268       return true;
 269    }
 270    return false;
 271 }
 272                                                                         /*}}}*/
 273 // CircleBuf::Stats - Print out stats information                       /*{{{*/
 274 // ---------------------------------------------------------------------
 275 /* */
 276 void CircleBuf::Stats()
 277 {
 278    if (InP == 0)
 279       return;
 280
 281    struct timeval Stop;
 282    gettimeofday(&Stop,0);
 283 /*   float Diff = Stop.tv_sec - Start.tv_sec +
 284              (float)(Stop.tv_usec - Start.tv_usec)/1000000;
 285    clog << "Got " << InP << " in " << Diff << " at " << InP/Diff << endl;*/
 286 }
 287                                                                         /*}}}*/
 288
 289 // ServerState::ServerState - Constructor                               /*{{{*/
 290 // ---------------------------------------------------------------------
 291 /* */
 292 ServerState::ServerState(URI Srv,HttpMethod *Owner) : Owner(Owner),
 293                         In(64*1024), Out(4*1024),
 294                         ServerName(Srv)
 295 {
 296    Reset();
 297 }
 298                                                                         /*}}}*/
 299 // ServerState::Open - Open a connection to the server                  /*{{{*/
 300 // ---------------------------------------------------------------------
 301 /* This opens a connection to the server. */
 302 bool ServerState::Open()
 303 {
 304    // Use the already open connection if possible.
 305    if (ServerFd != -1)
 306       return true;
 307
 308    Close();
 309    In.Reset();
 310    Out.Reset();
 311    Persistent = true;
 312
 313    // Determine the proxy setting
 314    string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
 315    if (!SpecificProxy.empty())
 316    {
 317            if (SpecificProxy == "DIRECT")
 318                    Proxy = "";
 319            else
 320                    Proxy = SpecificProxy;
 321    }
 322    else
 323    {
 324            string DefProxy = _config->Find("Acquire::http::Proxy");
 325            if (!DefProxy.empty())
 326            {
 327                    Proxy = DefProxy;
 328            }
 329            else
 330            {
 331                    char* result = getenv("http_proxy");
 332                    Proxy = result ? result : "";
 333            }
 334    }
 335
 336    // Parse no_proxy, a , separated list of domains
 337    if (getenv("no_proxy") != 0)
 338    {
 339       if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 340          Proxy = "";
 341    }
 342
 343    // Determine what host and port to use based on the proxy settings
 344    int Port = 0;
 345    string Host;
 346    if (Proxy.empty() == true || Proxy.Host.empty() == true)
 347    {
 348       if (ServerName.Port != 0)
 349          Port = ServerName.Port;
 350       Host = ServerName.Host;
 351    }
 352    else
 353    {
 354       if (Proxy.Port != 0)
 355          Port = Proxy.Port;
 356       Host = Proxy.Host;
 357    }
 358
 359    // Connect to the remote server
 360    if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
 361       return false;
 362
 363    return true;
 364 }
 365                                                                         /*}}}*/
 366 // ServerState::Close - Close a connection to the server                /*{{{*/
 367 // ---------------------------------------------------------------------
 368 /* */
 369 bool ServerState::Close()
 370 {
 371    close(ServerFd);
 372    ServerFd = -1;
 373    return true;
 374 }
 375                                                                         /*}}}*/
 376 // ServerState::RunHeaders - Get the headers before the data            /*{{{*/
 377 // ---------------------------------------------------------------------
 378 /* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
 379    parse error occurred */
 380 int ServerState::RunHeaders()
 381 {
 382    State = Header;
 383
 384    Owner->Status(_("Waiting for headers"));
 385
 386    Major = 0;
 387    Minor = 0;
 388    Result = 0;
 389    Size = 0;
 390    StartPos = 0;
 391    Encoding = Closes;
 392    HaveContent = false;
 393    time(&Date);
 394
 395    do
 396    {
 397       string Data;
 398       if (In.WriteTillEl(Data) == false)
 399          continue;
 400
 401       if (Debug == true)
 402          clog << Data;
 403
 404       for (string::const_iterator I = Data.begin(); I < Data.end(); I++)
 405       {
 406          string::const_iterator J = I;
 407          for (; J != Data.end() && *J != '\n' && *J != '\r';J++);
 408          if (HeaderLine(string(I,J)) == false)
 409             return 2;
 410          I = J;
 411       }
 412
 413       // 100 Continue is a Nop...
 414       if (Result == 100)
 415          continue;
 416
 417       // Tidy up the connection persistance state.
 418       if (Encoding == Closes && HaveContent == true)
 419          Persistent = false;
 420
 421       return 0;
 422    }
 423    while (Owner->Go(false,this) == true);
 424
 425    return 1;
 426 }
 427                                                                         /*}}}*/
 428 // ServerState::RunData - Transfer the data from the socket             /*{{{*/
 429 // ---------------------------------------------------------------------
 430 /* */
 431 bool ServerState::RunData()
 432 {
 433    State = Data;
 434
 435    // Chunked transfer encoding is fun..
 436    if (Encoding == Chunked)
 437    {
 438       while (1)
 439       {
 440          // Grab the block size
 441          bool Last = true;
 442          string Data;
 443          In.Limit(-1);
 444          do
 445          {
 446             if (In.WriteTillEl(Data,true) == true)
 447                break;
 448          }
 449          while ((Last = Owner->Go(false,this)) == true);
 450
 451          if (Last == false)
 452             return false;
 453
 454          // See if we are done
 455          unsigned long Len = strtol(Data.c_str(),0,16);
 456          if (Len == 0)
 457          {
 458             In.Limit(-1);
 459
 460             // We have to remove the entity trailer
 461             Last = true;
 462             do
 463             {
 464                if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
 465                   break;
 466             }
 467             while ((Last = Owner->Go(false,this)) == true);
 468             if (Last == false)
 469                return false;
 470             return !_error->PendingError();
 471          }
 472
 473          // Transfer the block
 474          In.Limit(Len);
 475          while (Owner->Go(true,this) == true)
 476             if (In.IsLimit() == true)
 477                break;
 478
 479          // Error
 480          if (In.IsLimit() == false)
 481             return false;
 482
 483          // The server sends an extra new line before the next block specifier..
 484          In.Limit(-1);
 485          Last = true;
 486          do
 487          {
 488             if (In.WriteTillEl(Data,true) == true)
 489                break;
 490          }
 491          while ((Last = Owner->Go(false,this)) == true);
 492          if (Last == false)
 493             return false;
 494       }
 495    }
 496    else
 497    {
 498       /* Closes encoding is used when the server did not specify a size, the
 499          loss of the connection means we are done */
 500       if (Encoding == Closes)
 501          In.Limit(-1);
 502       else
 503          In.Limit(Size - StartPos);
 504
 505       // Just transfer the whole block.
 506       do
 507       {
 508          if (In.IsLimit() == false)
 509             continue;
 510
 511          In.Limit(-1);
 512          return !_error->PendingError();
 513       }
 514       while (Owner->Go(true,this) == true);
 515    }
 516
 517    return Owner->Flush(this) && !_error->PendingError();
 518 }
 519                                                                         /*}}}*/
 520 // ServerState::HeaderLine - Process a header line                      /*{{{*/
 521 // ---------------------------------------------------------------------
 522 /* */
 523 bool ServerState::HeaderLine(string Line)
 524 {
 525    if (Line.empty() == true)
 526       return true;
 527
 528    // The http server might be trying to do something evil.
 529    if (Line.length() >= MAXLEN)
 530       return _error->Error(_("Got a single header line over %u chars"),MAXLEN);
 531
 532    string::size_type Pos = Line.find(' ');
 533    if (Pos == string::npos || Pos+1 > Line.length())
 534    {
 535       // Blah, some servers use "connection:closes", evil.
 536       Pos = Line.find(':');
 537       if (Pos == string::npos || Pos + 2 > Line.length())
 538          return _error->Error(_("Bad header line"));
 539       Pos++;
 540    }
 541
 542    // Parse off any trailing spaces between the : and the next word.
 543    string::size_type Pos2 = Pos;
 544    while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
 545       Pos2++;
 546
 547    string Tag = string(Line,0,Pos);
 548    string Val = string(Line,Pos2);
 549
 550    if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
 551    {
 552       // Evil servers return no version
 553       if (Line[4] == '/')
 554       {
 555          if (sscanf(Line.c_str(),"HTTP/%u.%u %u%[^\n]",&Major,&Minor,
 556                     &Result,Code) != 4)
 557             return _error->Error(_("The HTTP server sent an invalid reply header"));
 558       }
 559       else
 560       {
 561          Major = 0;
 562          Minor = 9;
 563          if (sscanf(Line.c_str(),"HTTP %u%[^\n]",&Result,Code) != 2)
 564             return _error->Error(_("The HTTP server sent an invalid reply header"));
 565       }
 566
 567       /* Check the HTTP response header to get the default persistance
 568          state. */
 569       if (Major < 1)
 570          Persistent = false;
 571       else
 572       {
 573          if (Major == 1 && Minor <= 0)
 574             Persistent = false;
 575          else
 576             Persistent = true;
 577       }
 578
 579       return true;
 580    }
 581
 582    if (stringcasecmp(Tag,"Content-Length:") == 0)
 583    {
 584       if (Encoding == Closes)
 585          Encoding = Stream;
 586       HaveContent = true;
 587
 588       // The length is already set from the Content-Range header
 589       if (StartPos != 0)
 590          return true;
 591
 592       if (sscanf(Val.c_str(),"%lu",&Size) != 1)
 593          return _error->Error(_("The HTTP server sent an invalid Content-Length header"));
 594       return true;
 595    }
 596
 597    if (stringcasecmp(Tag,"Content-Type:") == 0)
 598    {
 599       HaveContent = true;
 600       return true;
 601    }
 602
 603    if (stringcasecmp(Tag,"Content-Range:") == 0)
 604    {
 605       HaveContent = true;
 606
 607       if (sscanf(Val.c_str(),"bytes %lu-%*u/%lu",&StartPos,&Size) != 2)
 608          return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
 609       if ((unsigned)StartPos > Size)
 610          return _error->Error(_("This HTTP server has broken range support"));
 611       return true;
 612    }
 613
 614    if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
 615    {
 616       HaveContent = true;
 617       if (stringcasecmp(Val,"chunked") == 0)
 618          Encoding = Chunked;
 619       return true;
 620    }
 621
 622    if (stringcasecmp(Tag,"Connection:") == 0)
 623    {
 624       if (stringcasecmp(Val,"close") == 0)
 625          Persistent = false;
 626       if (stringcasecmp(Val,"keep-alive") == 0)
 627          Persistent = true;
 628       return true;
 629    }
 630
 631    if (stringcasecmp(Tag,"Last-Modified:") == 0)
 632    {
 633       if (StrToTime(Val,Date) == false)
 634          return _error->Error(_("Unknown date format"));
 635       return true;
 636    }
 637
 638    if (stringcasecmp(Tag,"Location:") == 0)
 639    {
 640       Location = Val;
 641       return true;
 642    }
 643
 644    return true;
 645 }
 646                                                                         /*}}}*/
 647
 648 // HttpMethod::SendReq - Send the HTTP request                          /*{{{*/
 649 // ---------------------------------------------------------------------
 650 /* This places the http request in the outbound buffer */
 651 void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
 652 {
 653    URI Uri = Itm->Uri;
 654
 655    // The HTTP server expects a hostname with a trailing :port
 656    char Buf[1000];
 657    string ProperHost = Uri.Host;
 658    if (Uri.Port != 0)
 659    {
 660       sprintf(Buf,":%u",Uri.Port);
 661       ProperHost += Buf;
 662    }
 663
 664    // Just in case.
 665    if (Itm->Uri.length() >= sizeof(Buf))
 666        abort();
 667
 668    /* Build the request. We include a keep-alive header only for non-proxy
 669       requests. This is to tweak old http/1.0 servers that do support keep-alive
 670       but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
 671       will glitch HTTP/1.0 proxies because they do not filter it out and
 672       pass it on, HTTP/1.1 says the connection should default to keep alive
 673       and we expect the proxy to do this */
 674    if (Proxy.empty() == true || Proxy.Host.empty())
 675       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
 676               QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
 677    else
 678    {
 679       /* Generate a cache control header if necessary. We place a max
 680          cache age on index files, optionally set a no-cache directive
 681          and a no-store directive for archives. */
 682       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 683               Itm->Uri.c_str(),ProperHost.c_str());
 684       // only generate a cache control header if we actually want to
 685       // use a cache
 686       if (_config->FindB("Acquire::http::No-Cache",false) == false)
 687       {
 688          if (Itm->IndexFile == true)
 689             sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 690                     _config->FindI("Acquire::http::Max-Age",0));
 691          else
 692          {
 693             if (_config->FindB("Acquire::http::No-Store",false) == true)
 694                strcat(Buf,"Cache-Control: no-store\r\n");
 695          }
 696       }
 697    }
 698    // generate a no-cache header if needed
 699    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 700       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 701
 702
 703    string Req = Buf;
 704
 705    // Check for a partial file
 706    struct stat SBuf;
 707    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 708    {
 709       // In this case we send an if-range query with a range header
 710       sprintf(Buf,"Range: bytes=%li-\r\nIf-Range: %s\r\n",(long)SBuf.st_size - 1,
 711               TimeRFC1123(SBuf.st_mtime).c_str());
 712       Req += Buf;
 713    }
 714    else
 715    {
 716       if (Itm->LastModified != 0)
 717       {
 718          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 719          Req += Buf;
 720       }
 721    }
 722
 723    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 724       Req += string("Proxy-Authorization: Basic ") +
 725           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 726
 727    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 728       Req += string("Authorization: Basic ") +
 729           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 730
 731    Req += "User-Agent: " + _config->Find("Acquire::http::User-Agent",
 732                 "Debian APT-HTTP/1.3 ("VERSION")") + "\r\n\r\n";
 733
 734    if (Debug == true)
 735       cerr << Req << endl;
 736
 737    Out.Read(Req);
 738 }
 739                                                                         /*}}}*/
 740 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 741 // ---------------------------------------------------------------------
 742 /* This runs the select loop over the server FDs, Output file FDs and
 743    stdin. */
 744 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 745 {
 746    // Server has closed the connection
 747    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 748                                ToFile == false))
 749       return false;
 750
 751    fd_set rfds,wfds;
 752    FD_ZERO(&rfds);
 753    FD_ZERO(&wfds);
 754
 755    /* Add the server. We only send more requests if the connection will
 756       be persisting */
 757    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 758        && Srv->Persistent == true)
 759       FD_SET(Srv->ServerFd,&wfds);
 760    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 761       FD_SET(Srv->ServerFd,&rfds);
 762
 763    // Add the file
 764    int FileFD = -1;
 765    if (File != 0)
 766       FileFD = File->Fd();
 767
 768    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 769       FD_SET(FileFD,&wfds);
 770
 771    // Add stdin
 772    FD_SET(STDIN_FILENO,&rfds);
 773
 774    // Figure out the max fd
 775    int MaxFd = FileFD;
 776    if (MaxFd < Srv->ServerFd)
 777       MaxFd = Srv->ServerFd;
 778
 779    // Select
 780    struct timeval tv;
 781    tv.tv_sec = TimeOut;
 782    tv.tv_usec = 0;
 783    int Res = 0;
 784    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 785    {
 786       if (errno == EINTR)
 787          return true;
 788       return _error->Errno("select",_("Select failed"));
 789    }
 790
 791    if (Res == 0)
 792    {
 793       _error->Error(_("Connection timed out"));
 794       return ServerDie(Srv);
 795    }
 796
 797    // Handle server IO
 798    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 799    {
 800       errno = 0;
 801       if (Srv->In.Read(Srv->ServerFd) == false)
 802          return ServerDie(Srv);
 803    }
 804
 805    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 806    {
 807       errno = 0;
 808       if (Srv->Out.Write(Srv->ServerFd) == false)
 809          return ServerDie(Srv);
 810    }
 811
 812    // Send data to the file
 813    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 814    {
 815       if (Srv->In.Write(FileFD) == false)
 816          return _error->Errno("write",_("Error writing to output file"));
 817    }
 818
 819    // Handle commands from APT
 820    if (FD_ISSET(STDIN_FILENO,&rfds))
 821    {
 822       if (Run(true) != -1)
 823          exit(100);
 824    }
 825
 826    return true;
 827 }
 828                                                                         /*}}}*/
 829 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 830 // ---------------------------------------------------------------------
 831 /* This takes the current input buffer from the Server FD and writes it
 832    into the file */
 833 bool HttpMethod::Flush(ServerState *Srv)
 834 {
 835    if (File != 0)
 836    {
 837       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 838       // can't be set
 839       if (File->Name() != "/dev/null")
 840          SetNonBlock(File->Fd(),false);
 841       if (Srv->In.WriteSpace() == false)
 842          return true;
 843
 844       while (Srv->In.WriteSpace() == true)
 845       {
 846          if (Srv->In.Write(File->Fd()) == false)
 847             return _error->Errno("write",_("Error writing to file"));
 848          if (Srv->In.IsLimit() == true)
 849             return true;
 850       }
 851
 852       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 853          return true;
 854    }
 855    return false;
 856 }
 857                                                                         /*}}}*/
 858 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 859 // ---------------------------------------------------------------------
 860 /* */
 861 bool HttpMethod::ServerDie(ServerState *Srv)
 862 {
 863    unsigned int LErrno = errno;
 864
 865    // Dump the buffer to the file
 866    if (Srv->State == ServerState::Data)
 867    {
 868       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 869       // can't be set
 870       if (File->Name() != "/dev/null")
 871          SetNonBlock(File->Fd(),false);
 872       while (Srv->In.WriteSpace() == true)
 873       {
 874          if (Srv->In.Write(File->Fd()) == false)
 875             return _error->Errno("write",_("Error writing to the file"));
 876
 877          // Done
 878          if (Srv->In.IsLimit() == true)
 879             return true;
 880       }
 881    }
 882
 883    // See if this is because the server finished the data stream
 884    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 885        Srv->Encoding != ServerState::Closes)
 886    {
 887       Srv->Close();
 888       if (LErrno == 0)
 889          return _error->Error(_("Error reading from server. Remote end closed connection"));
 890       errno = LErrno;
 891       return _error->Errno("read",_("Error reading from server"));
 892    }
 893    else
 894    {
 895       Srv->In.Limit(-1);
 896
 897       // Nothing left in the buffer
 898       if (Srv->In.WriteSpace() == false)
 899          return false;
 900
 901       // We may have got multiple responses back in one packet..
 902       Srv->Close();
 903       return true;
 904    }
 905
 906    return false;
 907 }
 908                                                                         /*}}}*/
 909 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 910 // ---------------------------------------------------------------------
 911 /* We look at the header data we got back from the server and decide what
 912    to do. Returns
 913      0 - File is open,
 914      1 - IMS hit
 915      3 - Unrecoverable error
 916      4 - Error with error content page
 917      5 - Unrecoverable non-server error (close the connection)
 918      6 - Try again with a new or changed URI
 919  */
 920 int 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 1;
 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 6;
 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       _error->Error("%u %s",Srv->Result,Srv->Code);
 957       if (Srv->HaveContent == true)
 958          return 4;
 959       return 3;
 960    }
 961
 962    // This is some sort of 2xx 'data follows' reply
 963    Res.LastModified = Srv->Date;
 964    Res.Size = Srv->Size;
 965
 966    // Open the file
 967    delete File;
 968    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
 969    if (_error->PendingError() == true)
 970       return 5;
 971
 972    FailFile = Queue->DestFile;
 973    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
 974    FailFd = File->Fd();
 975    FailTime = Srv->Date;
 976
 977    // Set the expected size
 978    if (Srv->StartPos >= 0)
 979    {
 980       Res.ResumePoint = Srv->StartPos;
 981       if (ftruncate(File->Fd(),Srv->StartPos) < 0)
 982          _error->Errno("ftruncate", _("Failed to truncate file"));
 983    }
 984
 985    // Set the start point
 986    lseek(File->Fd(),0,SEEK_END);
 987
 988    delete Srv->In.Hash;
 989    Srv->In.Hash = new Hashes;
 990
 991    // Fill the Hash if the file is non-empty (resume)
 992    if (Srv->StartPos > 0)
 993    {
 994       lseek(File->Fd(),0,SEEK_SET);
 995       if (Srv->In.Hash->AddFD(File->Fd(),Srv->StartPos) == false)
 996       {
 997          _error->Errno("read",_("Problem hashing file"));
 998          return 5;
 999       }
1000       lseek(File->Fd(),0,SEEK_END);
1001    }
1002
1003    SetNonBlock(File->Fd(),true);
1004    return 0;
1005 }
1006                                                                         /*}}}*/
1007 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1008 // ---------------------------------------------------------------------
1009 /* This closes and timestamps the open file. This is neccessary to get
1010    resume behavoir on user abort */
1011 void HttpMethod::SigTerm(int)
1012 {
1013    if (FailFd == -1)
1014       _exit(100);
1015    close(FailFd);
1016
1017    // Timestamp
1018    struct utimbuf UBuf;
1019    UBuf.actime = FailTime;
1020    UBuf.modtime = FailTime;
1021    utime(FailFile.c_str(),&UBuf);
1022
1023    _exit(100);
1024 }
1025                                                                         /*}}}*/
1026 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1027 // ---------------------------------------------------------------------
1028 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1029    depth. */
1030 bool HttpMethod::Fetch(FetchItem *)
1031 {
1032    if (Server == 0)
1033       return true;
1034
1035    // Queue the requests
1036    int Depth = -1;
1037    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1038         I = I->Next, Depth++)
1039    {
1040       // If pipelining is disabled, we only queue 1 request
1041       if (Server->Pipeline == false && Depth >= 0)
1042          break;
1043
1044       // Make sure we stick with the same server
1045       if (Server->Comp(I->Uri) == false)
1046          break;
1047       if (QueueBack == I)
1048       {
1049          QueueBack = I->Next;
1050          SendReq(I,Server->Out);
1051          continue;
1052       }
1053    }
1054
1055    return true;
1056 };
1057                                                                         /*}}}*/
1058 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1059 // ---------------------------------------------------------------------
1060 /* We stash the desired pipeline depth */
1061 bool HttpMethod::Configuration(string Message)
1062 {
1063    if (pkgAcqMethod::Configuration(Message) == false)
1064       return false;
1065
1066    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1067    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1068    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1069                                   PipelineDepth);
1070    Debug = _config->FindB("Debug::Acquire::http",false);
1071
1072    return true;
1073 }
1074                                                                         /*}}}*/
1075 // HttpMethod::Loop - Main loop                                         /*{{{*/
1076 // ---------------------------------------------------------------------
1077 /* */
1078 int HttpMethod::Loop()
1079 {
1080    typedef vector<string> StringVector;
1081    typedef vector<string>::iterator StringVectorIterator;
1082    map<string, StringVector> Redirected;
1083
1084    signal(SIGTERM,SigTerm);
1085    signal(SIGINT,SigTerm);
1086
1087    Server = 0;
1088
1089    int FailCounter = 0;
1090    while (1)
1091    {
1092       // We have no commands, wait for some to arrive
1093       if (Queue == 0)
1094       {
1095          if (WaitFd(STDIN_FILENO) == false)
1096             return 0;
1097       }
1098
1099       /* Run messages, we can accept 0 (no message) if we didn't
1100          do a WaitFd above.. Otherwise the FD is closed. */
1101       int Result = Run(true);
1102       if (Result != -1 && (Result != 0 || Queue == 0))
1103          return 100;
1104
1105       if (Queue == 0)
1106          continue;
1107
1108       // Connect to the server
1109       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1110       {
1111          delete Server;
1112          Server = new ServerState(Queue->Uri,this);
1113       }
1114       /* If the server has explicitly said this is the last connection
1115          then we pre-emptively shut down the pipeline and tear down
1116          the connection. This will speed up HTTP/1.0 servers a tad
1117          since we don't have to wait for the close sequence to
1118          complete */
1119       if (Server->Persistent == false)
1120          Server->Close();
1121
1122       // Reset the pipeline
1123       if (Server->ServerFd == -1)
1124          QueueBack = Queue;
1125
1126       // Connnect to the host
1127       if (Server->Open() == false)
1128       {
1129          Fail(true);
1130          delete Server;
1131          Server = 0;
1132          continue;
1133       }
1134
1135       // Fill the pipeline.
1136       Fetch(0);
1137
1138       // Fetch the next URL header data from the server.
1139       switch (Server->RunHeaders())
1140       {
1141          case 0:
1142          break;
1143
1144          // The header data is bad
1145          case 2:
1146          {
1147             _error->Error(_("Bad header data"));
1148             Fail(true);
1149             RotateDNS();
1150             continue;
1151          }
1152
1153          // The server closed a connection during the header get..
1154          default:
1155          case 1:
1156          {
1157             FailCounter++;
1158             _error->Discard();
1159             Server->Close();
1160             Server->Pipeline = false;
1161
1162             if (FailCounter >= 2)
1163             {
1164                Fail(_("Connection failed"),true);
1165                FailCounter = 0;
1166             }
1167
1168             RotateDNS();
1169             continue;
1170          }
1171       };
1172
1173       // Decide what to do.
1174       FetchResult Res;
1175       Res.Filename = Queue->DestFile;
1176       switch (DealWithHeaders(Res,Server))
1177       {
1178          // Ok, the file is Open
1179          case 0:
1180          {
1181             URIStart(Res);
1182
1183             // Run the data
1184             bool Result =  Server->RunData();
1185
1186             /* If the server is sending back sizeless responses then fill in
1187                the size now */
1188             if (Res.Size == 0)
1189                Res.Size = File->Size();
1190
1191             // Close the file, destroy the FD object and timestamp it
1192             FailFd = -1;
1193             delete File;
1194             File = 0;
1195
1196             // Timestamp
1197             struct utimbuf UBuf;
1198             time(&UBuf.actime);
1199             UBuf.actime = Server->Date;
1200             UBuf.modtime = Server->Date;
1201             utime(Queue->DestFile.c_str(),&UBuf);
1202
1203             // Send status to APT
1204             if (Result == true)
1205             {
1206                Res.TakeHashes(*Server->In.Hash);
1207                URIDone(Res);
1208             }
1209             else
1210             {
1211                if (Server->ServerFd == -1)
1212                {
1213                   FailCounter++;
1214                   _error->Discard();
1215                   Server->Close();
1216
1217                   if (FailCounter >= 2)
1218                   {
1219                      Fail(_("Connection failed"),true);
1220                      FailCounter = 0;
1221                   }
1222
1223                   QueueBack = Queue;
1224                }
1225                else
1226                   Fail(true);
1227             }
1228             break;
1229          }
1230
1231          // IMS hit
1232          case 1:
1233          {
1234             URIDone(Res);
1235             break;
1236          }
1237
1238          // Hard server error, not found or something
1239          case 3:
1240          {
1241             Fail();
1242             break;
1243          }
1244
1245          // Hard internal error, kill the connection and fail
1246          case 5:
1247          {
1248             delete File;
1249             File = 0;
1250
1251             Fail();
1252             RotateDNS();
1253             Server->Close();
1254             break;
1255          }
1256
1257          // We need to flush the data, the header is like a 404 w/ error text
1258          case 4:
1259          {
1260             Fail();
1261
1262             // Send to content to dev/null
1263             File = new FileFd("/dev/null",FileFd::WriteExists);
1264             Server->RunData();
1265             delete File;
1266             File = 0;
1267             break;
1268          }
1269
1270          // Try again with a new URL
1271          case 6:
1272          {
1273             // Clear rest of response if there is content
1274             if (Server->HaveContent)
1275             {
1276                File = new FileFd("/dev/null",FileFd::WriteExists);
1277                Server->RunData();
1278                delete File;
1279                File = 0;
1280             }
1281
1282             /* Detect redirect loops.  No more redirects are allowed
1283                after the same URI is seen twice in a queue item. */
1284             StringVector &R = Redirected[Queue->DestFile];
1285             bool StopRedirects = false;
1286             if (R.size() == 0)
1287                R.push_back(Queue->Uri);
1288             else if (R[0] == "STOP" || R.size() > 10)
1289                StopRedirects = true;
1290             else
1291             {
1292                for (StringVectorIterator I = R.begin(); I != R.end(); I++)
1293                   if (Queue->Uri == *I)
1294                   {
1295                      R[0] = "STOP";
1296                      break;
1297                   }
1298
1299                R.push_back(Queue->Uri);
1300             }
1301
1302             if (StopRedirects == false)
1303                Redirect(NextURI);
1304             else
1305                Fail();
1306
1307             break;
1308          }
1309
1310          default:
1311          Fail(_("Internal error"));
1312          break;
1313       }
1314
1315       FailCounter = 0;
1316    }
1317
1318    return 0;
1319 }
1320                                                                         /*}}}*/
1321
1322 int main()
1323 {
1324    setlocale(LC_ALL, "");
1325    // ignore SIGPIPE, this can happen on write() if the socket
1326    // closes the connection (this is dealt with via ServerDie())
1327    signal(SIGPIPE, SIG_IGN);
1328
1329    HttpMethod Mth;
1330    return Mth.Loop();
1331 }
1332
1333