apt-inst/contrib/extracttar.cc

   1 // -*- mode: cpp; mode: fold -*-
   2 // Description                                                          /*{{{*/
   3 // $Id: extracttar.cc,v 1.9 2004/01/07 20:39:37 mdz Exp $
   4 /* ######################################################################
   5
   6    Extract a Tar - Tar Extractor
   7
   8    Some performance measurements showed that zlib performed quite poorly
   9    in comparision to a forked gzip process. This tar extractor makes use
  10    of the fact that dup'd file descriptors have the same seek pointer
  11    and that gzip will not read past the end of a compressed stream,
  12    even if there is more data. We use the dup property to track extraction
  13    progress and the gzip feature to just feed gzip a fd in the middle
  14    of an AR file.
  15
  16    ##################################################################### */
  17                                                                         /*}}}*/
  18 // Include Files                                                        /*{{{*/
  19 #ifdef __GNUG__
  20 #pragma implementation "apt-pkg/extracttar.h"
  21 #endif
  22 #include <apt-pkg/extracttar.h>
  23
  24 #include <apt-pkg/error.h>
  25 #include <apt-pkg/strutl.h>
  26 #include <apt-pkg/configuration.h>
  27 #include <system.h>
  28
  29 #include <stdlib.h>
  30 #include <unistd.h>
  31 #include <signal.h>
  32 #include <fcntl.h>
  33 #include <iostream>
  34 #include <apti18n.h>
  35                                                                         /*}}}*/
  36
  37 using namespace std;
  38
  39 // The on disk header for a tar file.
  40 struct ExtractTar::TarHeader
  41 {
  42    char Name[100];
  43    char Mode[8];
  44    char UserID[8];
  45    char GroupID[8];
  46    char Size[12];
  47    char MTime[12];
  48    char Checksum[8];
  49    char LinkFlag;
  50    char LinkName[100];
  51    char MagicNumber[8];
  52    char UserName[32];
  53    char GroupName[32];
  54    char Major[8];
  55    char Minor[8];
  56 };
  57
  58 // ExtractTar::ExtractTar - Constructor                                 /*{{{*/
  59 // ---------------------------------------------------------------------
  60 /* */
  61 ExtractTar::ExtractTar(FileFd &Fd,unsigned long Max,string DecompressionProgram) : File(Fd),
  62                          MaxInSize(Max), DecompressProg(DecompressionProgram)
  63
  64 {
  65    GZPid = -1;
  66    InFd = -1;
  67    Eof = false;
  68 }
  69                                                                         /*}}}*/
  70 // ExtractTar::ExtractTar - Destructor                                  /*{{{*/
  71 // ---------------------------------------------------------------------
  72 /* */
  73 ExtractTar::~ExtractTar()
  74 {
  75    // Error close
  76    Done(true);
  77 }
  78                                                                         /*}}}*/
  79 // ExtractTar::Done - Reap the gzip sub process                         /*{{{*/
  80 // ---------------------------------------------------------------------
  81 /* If the force flag is given then error messages are suppressed - this
  82    means we hit the end of the tar file but there was still gzip data. */
  83 bool ExtractTar::Done(bool Force)
  84 {
  85    InFd.Close();
  86    if (GZPid <= 0)
  87       return true;
  88
  89    /* If there is a pending error then we are cleaning up gzip and are
  90       not interested in it's failures */
  91    if (_error->PendingError() == true)
  92       Force = true;
  93
  94    // Make sure we clean it up!
  95    kill(GZPid,SIGINT);
  96    string confvar = string("dir::bin::") + DecompressProg;
  97    if (ExecWait(GZPid,_config->Find(confvar.c_str(),DecompressProg.c_str()).c_str(),
  98                 Force) == false)
  99    {
 100       GZPid = -1;
 101       return Force;
 102    }
 103
 104    GZPid = -1;
 105    return true;
 106 }
 107                                                                         /*}}}*/
 108 // ExtractTar::StartGzip - Startup gzip                                 /*{{{*/
 109 // ---------------------------------------------------------------------
 110 /* This creates a gzip sub process that has its input as the file itself.
 111    If this tar file is embedded into something like an ar file then
 112    gzip will efficiently ignore the extra bits. */
 113 bool ExtractTar::StartGzip()
 114 {
 115    int Pipes[2];
 116    if (pipe(Pipes) != 0)
 117       return _error->Errno("pipe",_("Failed to create pipes"));
 118
 119    // Fork off the process
 120    GZPid = ExecFork();
 121
 122    // Spawn the subprocess
 123    if (GZPid == 0)
 124    {
 125       // Setup the FDs
 126       dup2(Pipes[1],STDOUT_FILENO);
 127       dup2(File.Fd(),STDIN_FILENO);
 128       int Fd = open("/dev/null",O_RDWR);
 129       if (Fd == -1)
 130          _exit(101);
 131       dup2(Fd,STDERR_FILENO);
 132       close(Fd);
 133       SetCloseExec(STDOUT_FILENO,false);
 134       SetCloseExec(STDIN_FILENO,false);
 135       SetCloseExec(STDERR_FILENO,false);
 136
 137       const char *Args[3];
 138       string confvar = string("dir::bin::") + DecompressProg;
 139       Args[0] = _config->Find(confvar.c_str(),DecompressProg.c_str()).c_str();
 140       Args[1] = "-d";
 141       Args[2] = 0;
 142       execvp(Args[0],(char **)Args);
 143       cerr << _("Failed to exec gzip ") << Args[0] << endl;
 144       _exit(100);
 145    }
 146
 147    // Fix up our FDs
 148    InFd.Fd(Pipes[0]);
 149    close(Pipes[1]);
 150    return true;
 151 }
 152                                                                         /*}}}*/
 153 // ExtractTar::Go - Perform extraction                                  /*{{{*/
 154 // ---------------------------------------------------------------------
 155 /* This reads each 512 byte block from the archive and extracts the header
 156    information into the Item structure. Then it resolves the UID/GID and
 157    invokes the correct processing function. */
 158 bool ExtractTar::Go(pkgDirStream &Stream)
 159 {
 160    if (StartGzip() == false)
 161       return false;
 162
 163    // Loop over all blocks
 164    string LastLongLink;
 165    string LastLongName;
 166    while (1)
 167    {
 168       bool BadRecord = false;
 169       unsigned char Block[512];
 170       if (InFd.Read(Block,sizeof(Block),true) == false)
 171          return false;
 172
 173       if (InFd.Eof() == true)
 174          break;
 175
 176       // Get the checksum
 177       TarHeader *Tar = (TarHeader *)Block;
 178       unsigned long CheckSum;
 179       if (StrToNum(Tar->Checksum,CheckSum,sizeof(Tar->Checksum),8) == false)
 180          return _error->Error(_("Corrupted archive"));
 181
 182       /* Compute the checksum field. The actual checksum is blanked out
 183          with spaces so it is not included in the computation */
 184       unsigned long NewSum = 0;
 185       memset(Tar->Checksum,' ',sizeof(Tar->Checksum));
 186       for (int I = 0; I != sizeof(Block); I++)
 187          NewSum += Block[I];
 188
 189       /* Check for a block of nulls - in this case we kill gzip, GNU tar
 190          does this.. */
 191       if (NewSum == ' '*sizeof(Tar->Checksum))
 192          return Done(true);
 193
 194       if (NewSum != CheckSum)
 195          return _error->Error(_("Tar Checksum failed, archive corrupted"));
 196
 197       // Decode all of the fields
 198       pkgDirStream::Item Itm;
 199       if (StrToNum(Tar->Mode,Itm.Mode,sizeof(Tar->Mode),8) == false ||
 200           StrToNum(Tar->UserID,Itm.UID,sizeof(Tar->UserID),8) == false ||
 201           StrToNum(Tar->GroupID,Itm.GID,sizeof(Tar->GroupID),8) == false ||
 202           StrToNum(Tar->Size,Itm.Size,sizeof(Tar->Size),8) == false ||
 203           StrToNum(Tar->MTime,Itm.MTime,sizeof(Tar->MTime),8) == false ||
 204           StrToNum(Tar->Major,Itm.Major,sizeof(Tar->Major),8) == false ||
 205           StrToNum(Tar->Minor,Itm.Minor,sizeof(Tar->Minor),8) == false)
 206          return _error->Error(_("Corrupted archive"));
 207
 208       // Grab the filename
 209       if (LastLongName.empty() == false)
 210          Itm.Name = (char *)LastLongName.c_str();
 211       else
 212       {
 213          Tar->Name[sizeof(Tar->Name)] = 0;
 214          Itm.Name = Tar->Name;
 215       }
 216       if (Itm.Name[0] == '.' && Itm.Name[1] == '/' && Itm.Name[2] != 0)
 217          Itm.Name += 2;
 218
 219       // Grab the link target
 220       Tar->Name[sizeof(Tar->LinkName)] = 0;
 221       Itm.LinkTarget = Tar->LinkName;
 222
 223       if (LastLongLink.empty() == false)
 224          Itm.LinkTarget = (char *)LastLongLink.c_str();
 225
 226       // Convert the type over
 227       switch (Tar->LinkFlag)
 228       {
 229          case NormalFile0:
 230          case NormalFile:
 231          Itm.Type = pkgDirStream::Item::File;
 232          break;
 233
 234          case HardLink:
 235          Itm.Type = pkgDirStream::Item::HardLink;
 236          break;
 237
 238          case SymbolicLink:
 239          Itm.Type = pkgDirStream::Item::SymbolicLink;
 240          break;
 241
 242          case CharacterDevice:
 243          Itm.Type = pkgDirStream::Item::CharDevice;
 244          break;
 245
 246          case BlockDevice:
 247          Itm.Type = pkgDirStream::Item::BlockDevice;
 248          break;
 249
 250          case Directory:
 251          Itm.Type = pkgDirStream::Item::Directory;
 252          break;
 253
 254          case FIFO:
 255          Itm.Type = pkgDirStream::Item::FIFO;
 256          break;
 257
 258          case GNU_LongLink:
 259          {
 260             unsigned long Length = Itm.Size;
 261             unsigned char Block[512];
 262             while (Length > 0)
 263             {
 264                if (InFd.Read(Block,sizeof(Block),true) == false)
 265                   return false;
 266                if (Length <= sizeof(Block))
 267                {
 268                   LastLongLink.append(Block,Block+sizeof(Block));
 269                   break;
 270                }
 271                LastLongLink.append(Block,Block+sizeof(Block));
 272                Length -= sizeof(Block);
 273             }
 274             continue;
 275          }
 276
 277          case GNU_LongName:
 278          {
 279             unsigned long Length = Itm.Size;
 280             unsigned char Block[512];
 281             while (Length > 0)
 282             {
 283                if (InFd.Read(Block,sizeof(Block),true) == false)
 284                   return false;
 285                if (Length < sizeof(Block))
 286                {
 287                   LastLongName.append(Block,Block+sizeof(Block));
 288                   break;
 289                }
 290                LastLongName.append(Block,Block+sizeof(Block));
 291                Length -= sizeof(Block);
 292             }
 293             continue;
 294          }
 295
 296          default:
 297          BadRecord = true;
 298          _error->Warning(_("Unknown TAR header type %u, member %s"),(unsigned)Tar->LinkFlag,Tar->Name);
 299          break;
 300       }
 301
 302       int Fd = -1;
 303       if (BadRecord == false)
 304          if (Stream.DoItem(Itm,Fd) == false)
 305             return false;
 306
 307       // Copy the file over the FD
 308       unsigned long Size = Itm.Size;
 309       while (Size != 0)
 310       {
 311          unsigned char Junk[32*1024];
 312          unsigned long Read = MIN(Size,sizeof(Junk));
 313          if (InFd.Read(Junk,((Read+511)/512)*512) == false)
 314             return false;
 315
 316          if (BadRecord == false)
 317          {
 318             if (Fd > 0)
 319             {
 320                if (write(Fd,Junk,Read) != (signed)Read)
 321                   return Stream.Fail(Itm,Fd);
 322             }
 323             else
 324             {
 325                /* An Fd of -2 means to send to a special processing
 326                   function */
 327                if (Fd == -2)
 328                   if (Stream.Process(Itm,Junk,Read,Itm.Size - Size) == false)
 329                      return Stream.Fail(Itm,Fd);
 330             }
 331          }
 332
 333          Size -= Read;
 334       }
 335
 336       // And finish up
 337       if (Itm.Size != 0 && BadRecord == false)
 338          if (Stream.FinishedFile(Itm,Fd) == false)
 339             return false;
 340
 341       LastLongName.erase();
 342       LastLongLink.erase();
 343    }
 344
 345    return Done(false);
 346 }
 347                                                                         /*}}}*/