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    if (getenv("http_proxy") == 0)
 315    {
 316       string DefProxy = _config->Find("Acquire::http::Proxy");
 317       string SpecificProxy = _config->Find("Acquire::http::Proxy::" + ServerName.Host);
 318       if (SpecificProxy.empty() == false)
 319       {
 320          if (SpecificProxy == "DIRECT")
 321             Proxy = "";
 322          else
 323             Proxy = SpecificProxy;
 324       }
 325       else
 326          Proxy = DefProxy;
 327    }
 328    else
 329       Proxy = getenv("http_proxy");
 330
 331    // Parse no_proxy, a , separated list of domains
 332    if (getenv("no_proxy") != 0)
 333    {
 334       if (CheckDomainList(ServerName.Host,getenv("no_proxy")) == true)
 335          Proxy = "";
 336    }
 337
 338    // Determine what host and port to use based on the proxy settings
 339    int Port = 0;
 340    string Host;
 341    if (Proxy.empty() == true || Proxy.Host.empty() == true)
 342    {
 343       if (ServerName.Port != 0)
 344          Port = ServerName.Port;
 345       Host = ServerName.Host;
 346    }
 347    else
 348    {
 349       if (Proxy.Port != 0)
 350          Port = Proxy.Port;
 351       Host = Proxy.Host;
 352    }
 353
 354    // Connect to the remote server
 355    if (Connect(Host,Port,"http",80,ServerFd,TimeOut,Owner) == false)
 356       return false;
 357
 358    return true;
 359 }
 360                                                                         /*}}}*/
 361 // ServerState::Close - Close a connection to the server                /*{{{*/
 362 // ---------------------------------------------------------------------
 363 /* */
 364 bool ServerState::Close()
 365 {
 366    close(ServerFd);
 367    ServerFd = -1;
 368    return true;
 369 }
 370                                                                         /*}}}*/
 371 // ServerState::RunHeaders - Get the headers before the data            /*{{{*/
 372 // ---------------------------------------------------------------------
 373 /* Returns 0 if things are OK, 1 if an IO error occurred and 2 if a header
 374    parse error occurred */
 375 int ServerState::RunHeaders()
 376 {
 377    State = Header;
 378
 379    Owner->Status(_("Waiting for headers"));
 380
 381    Major = 0;
 382    Minor = 0;
 383    Result = 0;
 384    Size = 0;
 385    StartPos = 0;
 386    Encoding = Closes;
 387    HaveContent = false;
 388    time(&Date);
 389
 390    do
 391    {
 392       string Data;
 393       if (In.WriteTillEl(Data) == false)
 394          continue;
 395
 396       if (Debug == true)
 397          clog << Data;
 398
 399       for (string::const_iterator I = Data.begin(); I < Data.end(); I++)
 400       {
 401          string::const_iterator J = I;
 402          for (; J != Data.end() && *J != '\n' && *J != '\r';J++);
 403          if (HeaderLine(string(I,J)) == false)
 404             return 2;
 405          I = J;
 406       }
 407
 408       // 100 Continue is a Nop...
 409       if (Result == 100)
 410          continue;
 411
 412       // Tidy up the connection persistance state.
 413       if (Encoding == Closes && HaveContent == true)
 414          Persistent = false;
 415
 416       return 0;
 417    }
 418    while (Owner->Go(false,this) == true);
 419
 420    return 1;
 421 }
 422                                                                         /*}}}*/
 423 // ServerState::RunData - Transfer the data from the socket             /*{{{*/
 424 // ---------------------------------------------------------------------
 425 /* */
 426 bool ServerState::RunData()
 427 {
 428    State = Data;
 429
 430    // Chunked transfer encoding is fun..
 431    if (Encoding == Chunked)
 432    {
 433       while (1)
 434       {
 435          // Grab the block size
 436          bool Last = true;
 437          string Data;
 438          In.Limit(-1);
 439          do
 440          {
 441             if (In.WriteTillEl(Data,true) == true)
 442                break;
 443          }
 444          while ((Last = Owner->Go(false,this)) == true);
 445
 446          if (Last == false)
 447             return false;
 448
 449          // See if we are done
 450          unsigned long Len = strtol(Data.c_str(),0,16);
 451          if (Len == 0)
 452          {
 453             In.Limit(-1);
 454
 455             // We have to remove the entity trailer
 456             Last = true;
 457             do
 458             {
 459                if (In.WriteTillEl(Data,true) == true && Data.length() <= 2)
 460                   break;
 461             }
 462             while ((Last = Owner->Go(false,this)) == true);
 463             if (Last == false)
 464                return false;
 465             return !_error->PendingError();
 466          }
 467
 468          // Transfer the block
 469          In.Limit(Len);
 470          while (Owner->Go(true,this) == true)
 471             if (In.IsLimit() == true)
 472                break;
 473
 474          // Error
 475          if (In.IsLimit() == false)
 476             return false;
 477
 478          // The server sends an extra new line before the next block specifier..
 479          In.Limit(-1);
 480          Last = true;
 481          do
 482          {
 483             if (In.WriteTillEl(Data,true) == true)
 484                break;
 485          }
 486          while ((Last = Owner->Go(false,this)) == true);
 487          if (Last == false)
 488             return false;
 489       }
 490    }
 491    else
 492    {
 493       /* Closes encoding is used when the server did not specify a size, the
 494          loss of the connection means we are done */
 495       if (Encoding == Closes)
 496          In.Limit(-1);
 497       else
 498          In.Limit(Size - StartPos);
 499
 500       // Just transfer the whole block.
 501       do
 502       {
 503          if (In.IsLimit() == false)
 504             continue;
 505
 506          In.Limit(-1);
 507          return !_error->PendingError();
 508       }
 509       while (Owner->Go(true,this) == true);
 510    }
 511
 512    return Owner->Flush(this) && !_error->PendingError();
 513 }
 514                                                                         /*}}}*/
 515 // ServerState::HeaderLine - Process a header line                      /*{{{*/
 516 // ---------------------------------------------------------------------
 517 /* */
 518 bool ServerState::HeaderLine(string Line)
 519 {
 520    if (Line.empty() == true)
 521       return true;
 522
 523    // The http server might be trying to do something evil.
 524    if (Line.length() >= MAXLEN)
 525       return _error->Error(_("Got a single header line over %u chars"),MAXLEN);
 526
 527    string::size_type Pos = Line.find(' ');
 528    if (Pos == string::npos || Pos+1 > Line.length())
 529    {
 530       // Blah, some servers use "connection:closes", evil.
 531       Pos = Line.find(':');
 532       if (Pos == string::npos || Pos + 2 > Line.length())
 533          return _error->Error(_("Bad header line"));
 534       Pos++;
 535    }
 536
 537    // Parse off any trailing spaces between the : and the next word.
 538    string::size_type Pos2 = Pos;
 539    while (Pos2 < Line.length() && isspace(Line[Pos2]) != 0)
 540       Pos2++;
 541
 542    string Tag = string(Line,0,Pos);
 543    string Val = string(Line,Pos2);
 544
 545    if (stringcasecmp(Tag.c_str(),Tag.c_str()+4,"HTTP") == 0)
 546    {
 547       // Evil servers return no version
 548       if (Line[4] == '/')
 549       {
 550          if (sscanf(Line.c_str(),"HTTP/%u.%u %u %[^\n]",&Major,&Minor,
 551                     &Result,Code) != 4)
 552             return _error->Error(_("The HTTP server sent an invalid reply header"));
 553       }
 554       else
 555       {
 556          Major = 0;
 557          Minor = 9;
 558          if (sscanf(Line.c_str(),"HTTP %u %[^\n]",&Result,Code) != 2)
 559             return _error->Error(_("The HTTP server sent an invalid reply header"));
 560       }
 561
 562       /* Check the HTTP response header to get the default persistance
 563          state. */
 564       if (Major < 1)
 565          Persistent = false;
 566       else
 567       {
 568          if (Major == 1 && Minor <= 0)
 569             Persistent = false;
 570          else
 571             Persistent = true;
 572       }
 573
 574       return true;
 575    }
 576
 577    if (stringcasecmp(Tag,"Content-Length:") == 0)
 578    {
 579       if (Encoding == Closes)
 580          Encoding = Stream;
 581       HaveContent = true;
 582
 583       // The length is already set from the Content-Range header
 584       if (StartPos != 0)
 585          return true;
 586
 587       if (sscanf(Val.c_str(),"%lu",&Size) != 1)
 588          return _error->Error(_("The HTTP server sent an invalid Content-Length header"));
 589       return true;
 590    }
 591
 592    if (stringcasecmp(Tag,"Content-Type:") == 0)
 593    {
 594       HaveContent = true;
 595       return true;
 596    }
 597
 598    if (stringcasecmp(Tag,"Content-Range:") == 0)
 599    {
 600       HaveContent = true;
 601
 602       if (sscanf(Val.c_str(),"bytes %lu-%*u/%lu",&StartPos,&Size) != 2)
 603          return _error->Error(_("The HTTP server sent an invalid Content-Range header"));
 604       if ((unsigned)StartPos > Size)
 605          return _error->Error(_("This HTTP server has broken range support"));
 606       return true;
 607    }
 608
 609    if (stringcasecmp(Tag,"Transfer-Encoding:") == 0)
 610    {
 611       HaveContent = true;
 612       if (stringcasecmp(Val,"chunked") == 0)
 613          Encoding = Chunked;
 614       return true;
 615    }
 616
 617    if (stringcasecmp(Tag,"Connection:") == 0)
 618    {
 619       if (stringcasecmp(Val,"close") == 0)
 620          Persistent = false;
 621       if (stringcasecmp(Val,"keep-alive") == 0)
 622          Persistent = true;
 623       return true;
 624    }
 625
 626    if (stringcasecmp(Tag,"Last-Modified:") == 0)
 627    {
 628       if (StrToTime(Val,Date) == false)
 629          return _error->Error(_("Unknown date format"));
 630       return true;
 631    }
 632
 633    if (stringcasecmp(Tag,"Location:") == 0)
 634    {
 635       Location = Val;
 636       return true;
 637    }
 638
 639    return true;
 640 }
 641                                                                         /*}}}*/
 642
 643 // HttpMethod::SendReq - Send the HTTP request                          /*{{{*/
 644 // ---------------------------------------------------------------------
 645 /* This places the http request in the outbound buffer */
 646 void HttpMethod::SendReq(FetchItem *Itm,CircleBuf &Out)
 647 {
 648    URI Uri = Itm->Uri;
 649
 650    // The HTTP server expects a hostname with a trailing :port
 651    char Buf[1000];
 652    string ProperHost = Uri.Host;
 653    if (Uri.Port != 0)
 654    {
 655       sprintf(Buf,":%u",Uri.Port);
 656       ProperHost += Buf;
 657    }
 658
 659    // Just in case.
 660    if (Itm->Uri.length() >= sizeof(Buf))
 661        abort();
 662
 663    /* Build the request. We include a keep-alive header only for non-proxy
 664       requests. This is to tweak old http/1.0 servers that do support keep-alive
 665       but not HTTP/1.1 automatic keep-alive. Doing this with a proxy server
 666       will glitch HTTP/1.0 proxies because they do not filter it out and
 667       pass it on, HTTP/1.1 says the connection should default to keep alive
 668       and we expect the proxy to do this */
 669    if (Proxy.empty() == true || Proxy.Host.empty())
 670       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\nConnection: keep-alive\r\n",
 671               QuoteString(Uri.Path,"~").c_str(),ProperHost.c_str());
 672    else
 673    {
 674       /* Generate a cache control header if necessary. We place a max
 675          cache age on index files, optionally set a no-cache directive
 676          and a no-store directive for archives. */
 677       sprintf(Buf,"GET %s HTTP/1.1\r\nHost: %s\r\n",
 678               Itm->Uri.c_str(),ProperHost.c_str());
 679       // only generate a cache control header if we actually want to
 680       // use a cache
 681       if (_config->FindB("Acquire::http::No-Cache",false) == false)
 682       {
 683          if (Itm->IndexFile == true)
 684             sprintf(Buf+strlen(Buf),"Cache-Control: max-age=%u\r\n",
 685                     _config->FindI("Acquire::http::Max-Age",0));
 686          else
 687          {
 688             if (_config->FindB("Acquire::http::No-Store",false) == true)
 689                strcat(Buf,"Cache-Control: no-store\r\n");
 690          }
 691       }
 692    }
 693    // generate a no-cache header if needed
 694    if (_config->FindB("Acquire::http::No-Cache",false) == true)
 695       strcat(Buf,"Cache-Control: no-cache\r\nPragma: no-cache\r\n");
 696
 697
 698    string Req = Buf;
 699
 700    // Check for a partial file
 701    struct stat SBuf;
 702    if (stat(Itm->DestFile.c_str(),&SBuf) >= 0 && SBuf.st_size > 0)
 703    {
 704       // In this case we send an if-range query with a range header
 705       sprintf(Buf,"Range: bytes=%li-\r\nIf-Range: %s\r\n",(long)SBuf.st_size - 1,
 706               TimeRFC1123(SBuf.st_mtime).c_str());
 707       Req += Buf;
 708    }
 709    else
 710    {
 711       if (Itm->LastModified != 0)
 712       {
 713          sprintf(Buf,"If-Modified-Since: %s\r\n",TimeRFC1123(Itm->LastModified).c_str());
 714          Req += Buf;
 715       }
 716    }
 717
 718    if (Proxy.User.empty() == false || Proxy.Password.empty() == false)
 719       Req += string("Proxy-Authorization: Basic ") +
 720           Base64Encode(Proxy.User + ":" + Proxy.Password) + "\r\n";
 721
 722    if (Uri.User.empty() == false || Uri.Password.empty() == false)
 723       Req += string("Authorization: Basic ") +
 724           Base64Encode(Uri.User + ":" + Uri.Password) + "\r\n";
 725
 726    Req += "User-Agent: Debian APT-HTTP/1.3 ("VERSION")\r\n\r\n";
 727
 728    if (Debug == true)
 729       cerr << Req << endl;
 730
 731    Out.Read(Req);
 732 }
 733                                                                         /*}}}*/
 734 // HttpMethod::Go - Run a single loop                                   /*{{{*/
 735 // ---------------------------------------------------------------------
 736 /* This runs the select loop over the server FDs, Output file FDs and
 737    stdin. */
 738 bool HttpMethod::Go(bool ToFile,ServerState *Srv)
 739 {
 740    // Server has closed the connection
 741    if (Srv->ServerFd == -1 && (Srv->In.WriteSpace() == false ||
 742                                ToFile == false))
 743       return false;
 744
 745    fd_set rfds,wfds;
 746    FD_ZERO(&rfds);
 747    FD_ZERO(&wfds);
 748
 749    /* Add the server. We only send more requests if the connection will
 750       be persisting */
 751    if (Srv->Out.WriteSpace() == true && Srv->ServerFd != -1
 752        && Srv->Persistent == true)
 753       FD_SET(Srv->ServerFd,&wfds);
 754    if (Srv->In.ReadSpace() == true && Srv->ServerFd != -1)
 755       FD_SET(Srv->ServerFd,&rfds);
 756
 757    // Add the file
 758    int FileFD = -1;
 759    if (File != 0)
 760       FileFD = File->Fd();
 761
 762    if (Srv->In.WriteSpace() == true && ToFile == true && FileFD != -1)
 763       FD_SET(FileFD,&wfds);
 764
 765    // Add stdin
 766    FD_SET(STDIN_FILENO,&rfds);
 767
 768    // Figure out the max fd
 769    int MaxFd = FileFD;
 770    if (MaxFd < Srv->ServerFd)
 771       MaxFd = Srv->ServerFd;
 772
 773    // Select
 774    struct timeval tv;
 775    tv.tv_sec = TimeOut;
 776    tv.tv_usec = 0;
 777    int Res = 0;
 778    if ((Res = select(MaxFd+1,&rfds,&wfds,0,&tv)) < 0)
 779    {
 780       if (errno == EINTR)
 781          return true;
 782       return _error->Errno("select",_("Select failed"));
 783    }
 784
 785    if (Res == 0)
 786    {
 787       _error->Error(_("Connection timed out"));
 788       return ServerDie(Srv);
 789    }
 790
 791    // Handle server IO
 792    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&rfds))
 793    {
 794       errno = 0;
 795       if (Srv->In.Read(Srv->ServerFd) == false)
 796          return ServerDie(Srv);
 797    }
 798
 799    if (Srv->ServerFd != -1 && FD_ISSET(Srv->ServerFd,&wfds))
 800    {
 801       errno = 0;
 802       if (Srv->Out.Write(Srv->ServerFd) == false)
 803          return ServerDie(Srv);
 804    }
 805
 806    // Send data to the file
 807    if (FileFD != -1 && FD_ISSET(FileFD,&wfds))
 808    {
 809       if (Srv->In.Write(FileFD) == false)
 810          return _error->Errno("write",_("Error writing to output file"));
 811    }
 812
 813    // Handle commands from APT
 814    if (FD_ISSET(STDIN_FILENO,&rfds))
 815    {
 816       if (Run(true) != -1)
 817          exit(100);
 818    }
 819
 820    return true;
 821 }
 822                                                                         /*}}}*/
 823 // HttpMethod::Flush - Dump the buffer into the file                    /*{{{*/
 824 // ---------------------------------------------------------------------
 825 /* This takes the current input buffer from the Server FD and writes it
 826    into the file */
 827 bool HttpMethod::Flush(ServerState *Srv)
 828 {
 829    if (File != 0)
 830    {
 831       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 832       // can't be set
 833       if (File->Name() != "/dev/null")
 834          SetNonBlock(File->Fd(),false);
 835       if (Srv->In.WriteSpace() == false)
 836          return true;
 837
 838       while (Srv->In.WriteSpace() == true)
 839       {
 840          if (Srv->In.Write(File->Fd()) == false)
 841             return _error->Errno("write",_("Error writing to file"));
 842          if (Srv->In.IsLimit() == true)
 843             return true;
 844       }
 845
 846       if (Srv->In.IsLimit() == true || Srv->Encoding == ServerState::Closes)
 847          return true;
 848    }
 849    return false;
 850 }
 851                                                                         /*}}}*/
 852 // HttpMethod::ServerDie - The server has closed the connection.        /*{{{*/
 853 // ---------------------------------------------------------------------
 854 /* */
 855 bool HttpMethod::ServerDie(ServerState *Srv)
 856 {
 857    unsigned int LErrno = errno;
 858
 859    // Dump the buffer to the file
 860    if (Srv->State == ServerState::Data)
 861    {
 862       // on GNU/kFreeBSD, apt dies on /dev/null because non-blocking
 863       // can't be set
 864       if (File->Name() != "/dev/null")
 865          SetNonBlock(File->Fd(),false);
 866       while (Srv->In.WriteSpace() == true)
 867       {
 868          if (Srv->In.Write(File->Fd()) == false)
 869             return _error->Errno("write",_("Error writing to the file"));
 870
 871          // Done
 872          if (Srv->In.IsLimit() == true)
 873             return true;
 874       }
 875    }
 876
 877    // See if this is because the server finished the data stream
 878    if (Srv->In.IsLimit() == false && Srv->State != ServerState::Header &&
 879        Srv->Encoding != ServerState::Closes)
 880    {
 881       Srv->Close();
 882       if (LErrno == 0)
 883          return _error->Error(_("Error reading from server. Remote end closed connection"));
 884       errno = LErrno;
 885       return _error->Errno("read",_("Error reading from server"));
 886    }
 887    else
 888    {
 889       Srv->In.Limit(-1);
 890
 891       // Nothing left in the buffer
 892       if (Srv->In.WriteSpace() == false)
 893          return false;
 894
 895       // We may have got multiple responses back in one packet..
 896       Srv->Close();
 897       return true;
 898    }
 899
 900    return false;
 901 }
 902                                                                         /*}}}*/
 903 // HttpMethod::DealWithHeaders - Handle the retrieved header data       /*{{{*/
 904 // ---------------------------------------------------------------------
 905 /* We look at the header data we got back from the server and decide what
 906    to do. Returns
 907      0 - File is open,
 908      1 - IMS hit
 909      3 - Unrecoverable error
 910      4 - Error with error content page
 911      5 - Unrecoverable non-server error (close the connection)
 912      6 - Try again with a new or changed URI
 913  */
 914 int HttpMethod::DealWithHeaders(FetchResult &Res,ServerState *Srv)
 915 {
 916    // Not Modified
 917    if (Srv->Result == 304)
 918    {
 919       unlink(Queue->DestFile.c_str());
 920       Res.IMSHit = true;
 921       Res.LastModified = Queue->LastModified;
 922       return 1;
 923    }
 924
 925    /* Redirect
 926     *
 927     * Note that it is only OK for us to treat all redirection the same
 928     * because we *always* use GET, not other HTTP methods.  There are
 929     * three redirection codes for which it is not appropriate that we
 930     * redirect.  Pass on those codes so the error handling kicks in.
 931     */
 932    if (AllowRedirect
 933        && (Srv->Result > 300 && Srv->Result < 400)
 934        && (Srv->Result != 300       // Multiple Choices
 935            && Srv->Result != 304    // Not Modified
 936            && Srv->Result != 306))  // (Not part of HTTP/1.1, reserved)
 937    {
 938       if (!Srv->Location.empty())
 939       {
 940          NextURI = Srv->Location;
 941          return 6;
 942       }
 943       /* else pass through for error message */
 944    }
 945
 946    /* We have a reply we dont handle. This should indicate a perm server
 947       failure */
 948    if (Srv->Result < 200 || Srv->Result >= 300)
 949    {
 950       _error->Error("%u %s",Srv->Result,Srv->Code);
 951       if (Srv->HaveContent == true)
 952          return 4;
 953       return 3;
 954    }
 955
 956    // This is some sort of 2xx 'data follows' reply
 957    Res.LastModified = Srv->Date;
 958    Res.Size = Srv->Size;
 959
 960    // Open the file
 961    delete File;
 962    File = new FileFd(Queue->DestFile,FileFd::WriteAny);
 963    if (_error->PendingError() == true)
 964       return 5;
 965
 966    FailFile = Queue->DestFile;
 967    FailFile.c_str();   // Make sure we dont do a malloc in the signal handler
 968    FailFd = File->Fd();
 969    FailTime = Srv->Date;
 970
 971    // Set the expected size
 972    if (Srv->StartPos >= 0)
 973    {
 974       Res.ResumePoint = Srv->StartPos;
 975       if (ftruncate(File->Fd(),Srv->StartPos) < 0)
 976          _error->Errno("ftruncate", _("Failed to truncate file"));
 977    }
 978
 979    // Set the start point
 980    lseek(File->Fd(),0,SEEK_END);
 981
 982    delete Srv->In.Hash;
 983    Srv->In.Hash = new Hashes;
 984
 985    // Fill the Hash if the file is non-empty (resume)
 986    if (Srv->StartPos > 0)
 987    {
 988       lseek(File->Fd(),0,SEEK_SET);
 989       if (Srv->In.Hash->AddFD(File->Fd(),Srv->StartPos) == false)
 990       {
 991          _error->Errno("read",_("Problem hashing file"));
 992          return 5;
 993       }
 994       lseek(File->Fd(),0,SEEK_END);
 995    }
 996
 997    SetNonBlock(File->Fd(),true);
 998    return 0;
 999 }
1000                                                                         /*}}}*/
1001 // HttpMethod::SigTerm - Handle a fatal signal                          /*{{{*/
1002 // ---------------------------------------------------------------------
1003 /* This closes and timestamps the open file. This is neccessary to get
1004    resume behavoir on user abort */
1005 void HttpMethod::SigTerm(int)
1006 {
1007    if (FailFd == -1)
1008       _exit(100);
1009    close(FailFd);
1010
1011    // Timestamp
1012    struct utimbuf UBuf;
1013    UBuf.actime = FailTime;
1014    UBuf.modtime = FailTime;
1015    utime(FailFile.c_str(),&UBuf);
1016
1017    _exit(100);
1018 }
1019                                                                         /*}}}*/
1020 // HttpMethod::Fetch - Fetch an item                                    /*{{{*/
1021 // ---------------------------------------------------------------------
1022 /* This adds an item to the pipeline. We keep the pipeline at a fixed
1023    depth. */
1024 bool HttpMethod::Fetch(FetchItem *)
1025 {
1026    if (Server == 0)
1027       return true;
1028
1029    // Queue the requests
1030    int Depth = -1;
1031    for (FetchItem *I = Queue; I != 0 && Depth < (signed)PipelineDepth;
1032         I = I->Next, Depth++)
1033    {
1034       // If pipelining is disabled, we only queue 1 request
1035       if (Server->Pipeline == false && Depth >= 0)
1036          break;
1037
1038       // Make sure we stick with the same server
1039       if (Server->Comp(I->Uri) == false)
1040          break;
1041       if (QueueBack == I)
1042       {
1043          QueueBack = I->Next;
1044          SendReq(I,Server->Out);
1045          continue;
1046       }
1047    }
1048
1049    return true;
1050 };
1051                                                                         /*}}}*/
1052 // HttpMethod::Configuration - Handle a configuration message           /*{{{*/
1053 // ---------------------------------------------------------------------
1054 /* We stash the desired pipeline depth */
1055 bool HttpMethod::Configuration(string Message)
1056 {
1057    if (pkgAcqMethod::Configuration(Message) == false)
1058       return false;
1059
1060    AllowRedirect = _config->FindB("Acquire::http::AllowRedirect",true);
1061    TimeOut = _config->FindI("Acquire::http::Timeout",TimeOut);
1062    PipelineDepth = _config->FindI("Acquire::http::Pipeline-Depth",
1063                                   PipelineDepth);
1064    Debug = _config->FindB("Debug::Acquire::http",false);
1065
1066    return true;
1067 }
1068                                                                         /*}}}*/
1069 // HttpMethod::Loop - Main loop                                         /*{{{*/
1070 // ---------------------------------------------------------------------
1071 /* */
1072 int HttpMethod::Loop()
1073 {
1074    typedef vector<string> StringVector;
1075    typedef vector<string>::iterator StringVectorIterator;
1076    map<string, StringVector> Redirected;
1077
1078    signal(SIGTERM,SigTerm);
1079    signal(SIGINT,SigTerm);
1080
1081    Server = 0;
1082
1083    int FailCounter = 0;
1084    while (1)
1085    {
1086       // We have no commands, wait for some to arrive
1087       if (Queue == 0)
1088       {
1089          if (WaitFd(STDIN_FILENO) == false)
1090             return 0;
1091       }
1092
1093       /* Run messages, we can accept 0 (no message) if we didn't
1094          do a WaitFd above.. Otherwise the FD is closed. */
1095       int Result = Run(true);
1096       if (Result != -1 && (Result != 0 || Queue == 0))
1097          return 100;
1098
1099       if (Queue == 0)
1100          continue;
1101
1102       // Connect to the server
1103       if (Server == 0 || Server->Comp(Queue->Uri) == false)
1104       {
1105          delete Server;
1106          Server = new ServerState(Queue->Uri,this);
1107       }
1108       /* If the server has explicitly said this is the last connection
1109          then we pre-emptively shut down the pipeline and tear down
1110          the connection. This will speed up HTTP/1.0 servers a tad
1111          since we don't have to wait for the close sequence to
1112          complete */
1113       if (Server->Persistent == false)
1114          Server->Close();
1115
1116       // Reset the pipeline
1117       if (Server->ServerFd == -1)
1118          QueueBack = Queue;
1119
1120       // Connnect to the host
1121       if (Server->Open() == false)
1122       {
1123          Fail(true);
1124          delete Server;
1125          Server = 0;
1126          continue;
1127       }
1128
1129       // Fill the pipeline.
1130       Fetch(0);
1131
1132       // Fetch the next URL header data from the server.
1133       switch (Server->RunHeaders())
1134       {
1135          case 0:
1136          break;
1137
1138          // The header data is bad
1139          case 2:
1140          {
1141             _error->Error(_("Bad header data"));
1142             Fail(true);
1143             RotateDNS();
1144             continue;
1145          }
1146
1147          // The server closed a connection during the header get..
1148          default:
1149          case 1:
1150          {
1151             FailCounter++;
1152             _error->Discard();
1153             Server->Close();
1154             Server->Pipeline = false;
1155
1156             if (FailCounter >= 2)
1157             {
1158                Fail(_("Connection failed"),true);
1159                FailCounter = 0;
1160             }
1161
1162             RotateDNS();
1163             continue;
1164          }
1165       };
1166
1167       // Decide what to do.
1168       FetchResult Res;
1169       Res.Filename = Queue->DestFile;
1170       switch (DealWithHeaders(Res,Server))
1171       {
1172          // Ok, the file is Open
1173          case 0:
1174          {
1175             URIStart(Res);
1176
1177             // Run the data
1178             bool Result =  Server->RunData();
1179
1180             /* If the server is sending back sizeless responses then fill in
1181                the size now */
1182             if (Res.Size == 0)
1183                Res.Size = File->Size();
1184
1185             // Close the file, destroy the FD object and timestamp it
1186             FailFd = -1;
1187             delete File;
1188             File = 0;
1189
1190             // Timestamp
1191             struct utimbuf UBuf;
1192             time(&UBuf.actime);
1193             UBuf.actime = Server->Date;
1194             UBuf.modtime = Server->Date;
1195             utime(Queue->DestFile.c_str(),&UBuf);
1196
1197             // Send status to APT
1198             if (Result == true)
1199             {
1200                Res.TakeHashes(*Server->In.Hash);
1201                URIDone(Res);
1202             }
1203             else
1204             {
1205                if (Server->ServerFd == -1)
1206                {
1207                   FailCounter++;
1208                   _error->Discard();
1209                   Server->Close();
1210
1211                   if (FailCounter >= 2)
1212                   {
1213                      Fail(_("Connection failed"),true);
1214                      FailCounter = 0;
1215                   }
1216
1217                   QueueBack = Queue;
1218                }
1219                else
1220                   Fail(true);
1221             }
1222             break;
1223          }
1224
1225          // IMS hit
1226          case 1:
1227          {
1228             URIDone(Res);
1229             break;
1230          }
1231
1232          // Hard server error, not found or something
1233          case 3:
1234          {
1235             Fail();
1236             break;
1237          }
1238
1239          // Hard internal error, kill the connection and fail
1240          case 5:
1241          {
1242             delete File;
1243             File = 0;
1244
1245             Fail();
1246             RotateDNS();
1247             Server->Close();
1248             break;
1249          }
1250
1251          // We need to flush the data, the header is like a 404 w/ error text
1252          case 4:
1253          {
1254             Fail();
1255
1256             // Send to content to dev/null
1257             File = new FileFd("/dev/null",FileFd::WriteExists);
1258             Server->RunData();
1259             delete File;
1260             File = 0;
1261             break;
1262          }
1263
1264          // Try again with a new URL
1265          case 6:
1266          {
1267             // Clear rest of response if there is content
1268             if (Server->HaveContent)
1269             {
1270                File = new FileFd("/dev/null",FileFd::WriteExists);
1271                Server->RunData();
1272                delete File;
1273                File = 0;
1274             }
1275
1276             /* Detect redirect loops.  No more redirects are allowed
1277                after the same URI is seen twice in a queue item. */
1278             StringVector &R = Redirected[Queue->DestFile];
1279             bool StopRedirects = false;
1280             if (R.size() == 0)
1281                R.push_back(Queue->Uri);
1282             else if (R[0] == "STOP" || R.size() > 10)
1283                StopRedirects = true;
1284             else
1285             {
1286                for (StringVectorIterator I = R.begin(); I != R.end(); I++)
1287                   if (Queue->Uri == *I)
1288                   {
1289                      R[0] = "STOP";
1290                      break;
1291                   }
1292
1293                R.push_back(Queue->Uri);
1294             }
1295
1296             if (StopRedirects == false)
1297                Redirect(NextURI);
1298             else
1299                Fail();
1300
1301             break;
1302          }
1303
1304          default:
1305          Fail(_("Internal error"));
1306          break;
1307       }
1308
1309       FailCounter = 0;
1310    }
1311
1312    return 0;
1313 }
1314                                                                         /*}}}*/
1315
1316 int main()
1317 {
1318    setlocale(LC_ALL, "");
1319    // ignore SIGPIPE, this can happen on write() if the socket
1320    // closes the connection (this is dealt with via ServerDie())
1321    signal(SIGPIPE, SIG_IGN);
1322
1323    HttpMethod Mth;
1324    return Mth.Loop();
1325 }
1326
1327