apt-pkg/tagfile.cc

   1 // -*- mode: cpp; mode: fold -*-
   2 // Description                                                          /*{{{*/
   3 // $Id: tagfile.cc,v 1.37.2.2 2003/12/31 16:02:30 mdz Exp $
   4 /* ######################################################################
   5
   6    Fast scanner for RFC-822 type header information
   7
   8    This uses a rotating buffer to load the package information into.
   9    The scanner runs over it and isolates and indexes a single section.
  10
  11    ##################################################################### */
  12                                                                         /*}}}*/
  13 // Include Files                                                        /*{{{*/
  14 #include <apt-pkg/tagfile.h>
  15 #include <apt-pkg/error.h>
  16 #include <apt-pkg/strutl.h>
  17
  18 #include <apti18n.h>
  19
  20 #include <string>
  21 #include <stdio.h>
  22 #include <ctype.h>
  23                                                                         /*}}}*/
  24
  25 using std::string;
  26
  27 // TagFile::pkgTagFile - Constructor                                    /*{{{*/
  28 // ---------------------------------------------------------------------
  29 /* */
  30 pkgTagFile::pkgTagFile(FileFd *pFd,unsigned long Size) :
  31      Fd(*pFd),
  32      Size(Size)
  33 {
  34    if (Fd.IsOpen() == false)
  35    {
  36       Buffer = 0;
  37       Start = End = Buffer = 0;
  38       Done = true;
  39       iOffset = 0;
  40       return;
  41    }
  42
  43    Buffer = new char[Size];
  44    Start = End = Buffer;
  45    Done = false;
  46    iOffset = 0;
  47    Fill();
  48 }
  49                                                                         /*}}}*/
  50 // TagFile::~pkgTagFile - Destructor                                    /*{{{*/
  51 // ---------------------------------------------------------------------
  52 /* */
  53 pkgTagFile::~pkgTagFile()
  54 {
  55    delete [] Buffer;
  56 }
  57                                                                         /*}}}*/
  58 // TagFile::Resize - Resize the internal buffer                         /*{{{*/
  59 // ---------------------------------------------------------------------
  60 /* Resize the internal buffer (double it in size). Fail if a maximum size
  61  * size is reached.
  62  */
  63 bool pkgTagFile::Resize()
  64 {
  65    char *tmp;
  66    unsigned long EndSize = End - Start;
  67
  68    // fail is the buffer grows too big
  69    if(Size > 1024*1024+1)
  70       return false;
  71
  72    // get new buffer and use it
  73    tmp = new char[2*Size];
  74    memcpy(tmp, Buffer, Size);
  75    Size = Size*2;
  76    delete [] Buffer;
  77    Buffer = tmp;
  78
  79    // update the start/end pointers to the new buffer
  80    Start = Buffer;
  81    End = Start + EndSize;
  82    return true;
  83 }
  84                                                                         /*}}}*/
  85 // TagFile::Step - Advance to the next section                          /*{{{*/
  86 // ---------------------------------------------------------------------
  87 /* If the Section Scanner fails we refill the buffer and try again.
  88  * If that fails too, double the buffer size and try again until a
  89  * maximum buffer is reached.
  90  */
  91 bool pkgTagFile::Step(pkgTagSection &Tag)
  92 {
  93    while (Tag.Scan(Start,End - Start) == false)
  94    {
  95       if (Fill() == false)
  96          return false;
  97
  98       if(Tag.Scan(Start,End - Start))
  99          break;
 100
 101       if (Resize() == false)
 102          return _error->Error(_("Unable to parse package file %s (1)"),
 103                                  Fd.Name().c_str());
 104    }
 105    Start += Tag.size();
 106    iOffset += Tag.size();
 107
 108    Tag.Trim();
 109    return true;
 110 }
 111                                                                         /*}}}*/
 112 // TagFile::Fill - Top up the buffer                                    /*{{{*/
 113 // ---------------------------------------------------------------------
 114 /* This takes the bit at the end of the buffer and puts it at the start
 115    then fills the rest from the file */
 116 bool pkgTagFile::Fill()
 117 {
 118    unsigned long EndSize = End - Start;
 119    unsigned long Actual = 0;
 120
 121    memmove(Buffer,Start,EndSize);
 122    Start = Buffer;
 123    End = Buffer + EndSize;
 124
 125    if (Done == false)
 126    {
 127       // See if only a bit of the file is left
 128       if (Fd.Read(End,Size - (End - Buffer),&Actual) == false)
 129          return false;
 130       if (Actual != Size - (End - Buffer))
 131          Done = true;
 132       End += Actual;
 133    }
 134
 135    if (Done == true)
 136    {
 137       if (EndSize <= 3 && Actual == 0)
 138          return false;
 139       if (Size - (End - Buffer) < 4)
 140          return true;
 141
 142       // Append a double new line if one does not exist
 143       unsigned int LineCount = 0;
 144       for (const char *E = End - 1; E - End < 6 && (*E == '\n' || *E == '\r'); E--)
 145          if (*E == '\n')
 146             LineCount++;
 147       for (; LineCount < 2; LineCount++)
 148          *End++ = '\n';
 149
 150       return true;
 151    }
 152
 153    return true;
 154 }
 155                                                                         /*}}}*/
 156 // TagFile::Jump - Jump to a pre-recorded location in the file          /*{{{*/
 157 // ---------------------------------------------------------------------
 158 /* This jumps to a pre-recorded file location and reads the record
 159    that is there */
 160 bool pkgTagFile::Jump(pkgTagSection &Tag,unsigned long Offset)
 161 {
 162    // We are within a buffer space of the next hit..
 163    if (Offset >= iOffset && iOffset + (End - Start) > Offset)
 164    {
 165       unsigned long Dist = Offset - iOffset;
 166       Start += Dist;
 167       iOffset += Dist;
 168       return Step(Tag);
 169    }
 170
 171    // Reposition and reload..
 172    iOffset = Offset;
 173    Done = false;
 174    if (Fd.Seek(Offset) == false)
 175       return false;
 176    End = Start = Buffer;
 177
 178    if (Fill() == false)
 179       return false;
 180
 181    if (Tag.Scan(Start,End - Start) == true)
 182       return true;
 183
 184    // This appends a double new line (for the real eof handling)
 185    if (Fill() == false)
 186       return false;
 187
 188    if (Tag.Scan(Start,End - Start) == false)
 189       return _error->Error(_("Unable to parse package file %s (2)"),Fd.Name().c_str());
 190
 191    return true;
 192 }
 193                                                                         /*}}}*/
 194 // TagSection::Scan - Scan for the end of the header information        /*{{{*/
 195 // ---------------------------------------------------------------------
 196 /* This looks for the first double new line in the data stream.
 197    It also indexes the tags in the section. */
 198 bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength)
 199 {
 200    const char *End = Start + MaxLength;
 201    Stop = Section = Start;
 202    memset(AlphaIndexes,0,sizeof(AlphaIndexes));
 203
 204    if (Stop == 0)
 205       return false;
 206
 207    TagCount = 0;
 208    while (TagCount+1 < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End)
 209    {
 210        TrimRecord(true,End);
 211
 212       // Start a new index and add it to the hash
 213       if (isspace(Stop[0]) == 0)
 214       {
 215          Indexes[TagCount++] = Stop - Section;
 216          AlphaIndexes[AlphaHash(Stop,End)] = TagCount;
 217       }
 218
 219       Stop = (const char *)memchr(Stop,'\n',End - Stop);
 220
 221       if (Stop == 0)
 222          return false;
 223
 224       for (; Stop+1 < End && Stop[1] == '\r'; Stop++);
 225
 226       // Double newline marks the end of the record
 227       if (Stop+1 < End && Stop[1] == '\n')
 228       {
 229          Indexes[TagCount] = Stop - Section;
 230          TrimRecord(false,End);
 231          return true;
 232       }
 233
 234       Stop++;
 235    }
 236
 237    return false;
 238 }
 239                                                                         /*}}}*/
 240 // TagSection::TrimRecord - Trim off any garbage before/after a record  /*{{{*/
 241 // ---------------------------------------------------------------------
 242 /* There should be exactly 2 newline at the end of the record, no more. */
 243 void pkgTagSection::TrimRecord(bool BeforeRecord, const char*& End)
 244 {
 245    if (BeforeRecord == true)
 246       return;
 247    for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++);
 248 }
 249                                                                         /*}}}*/
 250 // TagSection::Trim - Trim off any trailing garbage                     /*{{{*/
 251 // ---------------------------------------------------------------------
 252 /* There should be exactly 1 newline at the end of the buffer, no more. */
 253 void pkgTagSection::Trim()
 254 {
 255    for (; Stop > Section + 2 && (Stop[-2] == '\n' || Stop[-2] == '\r'); Stop--);
 256 }
 257                                                                         /*}}}*/
 258 // TagSection::Find - Locate a tag                                      /*{{{*/
 259 // ---------------------------------------------------------------------
 260 /* This searches the section for a tag that matches the given string. */
 261 bool pkgTagSection::Find(const char *Tag,unsigned &Pos) const
 262 {
 263    unsigned int Length = strlen(Tag);
 264    unsigned int I = AlphaIndexes[AlphaHash(Tag)];
 265    if (I == 0)
 266       return false;
 267    I--;
 268
 269    for (unsigned int Counter = 0; Counter != TagCount; Counter++,
 270         I = (I+1)%TagCount)
 271    {
 272       const char *St;
 273       St = Section + Indexes[I];
 274       if (strncasecmp(Tag,St,Length) != 0)
 275          continue;
 276
 277       // Make sure the colon is in the right place
 278       const char *C = St + Length;
 279       for (; isspace(*C) != 0; C++);
 280       if (*C != ':')
 281          continue;
 282       Pos = I;
 283       return true;
 284    }
 285
 286    Pos = 0;
 287    return false;
 288 }
 289                                                                         /*}}}*/
 290 // TagSection::Find - Locate a tag                                      /*{{{*/
 291 // ---------------------------------------------------------------------
 292 /* This searches the section for a tag that matches the given string. */
 293 bool pkgTagSection::Find(const char *Tag,const char *&Start,
 294                          const char *&End) const
 295 {
 296    unsigned int Length = strlen(Tag);
 297    unsigned int I = AlphaIndexes[AlphaHash(Tag)];
 298    if (I == 0)
 299       return false;
 300    I--;
 301
 302    for (unsigned int Counter = 0; Counter != TagCount; Counter++,
 303         I = (I+1)%TagCount)
 304    {
 305       const char *St;
 306       St = Section + Indexes[I];
 307       if (strncasecmp(Tag,St,Length) != 0)
 308          continue;
 309
 310       // Make sure the colon is in the right place
 311       const char *C = St + Length;
 312       for (; isspace(*C) != 0; C++);
 313       if (*C != ':')
 314          continue;
 315
 316       // Strip off the gunk from the start end
 317       Start = C;
 318       End = Section + Indexes[I+1];
 319       if (Start >= End)
 320          return _error->Error("Internal parsing error");
 321
 322       for (; (isspace(*Start) != 0 || *Start == ':') && Start < End; Start++);
 323       for (; isspace(End[-1]) != 0 && End > Start; End--);
 324
 325       return true;
 326    }
 327
 328    Start = End = 0;
 329    return false;
 330 }
 331                                                                         /*}}}*/
 332 // TagSection::FindS - Find a string                                    /*{{{*/
 333 // ---------------------------------------------------------------------
 334 /* */
 335 string pkgTagSection::FindS(const char *Tag) const
 336 {
 337    const char *Start;
 338    const char *End;
 339    if (Find(Tag,Start,End) == false)
 340       return string();
 341    return string(Start,End);
 342 }
 343                                                                         /*}}}*/
 344 // TagSection::FindI - Find an integer                                  /*{{{*/
 345 // ---------------------------------------------------------------------
 346 /* */
 347 signed int pkgTagSection::FindI(const char *Tag,signed long Default) const
 348 {
 349    const char *Start;
 350    const char *Stop;
 351    if (Find(Tag,Start,Stop) == false)
 352       return Default;
 353
 354    // Copy it into a temp buffer so we can use strtol
 355    char S[300];
 356    if ((unsigned)(Stop - Start) >= sizeof(S))
 357       return Default;
 358    strncpy(S,Start,Stop-Start);
 359    S[Stop - Start] = 0;
 360
 361    char *End;
 362    signed long Result = strtol(S,&End,10);
 363    if (S == End)
 364       return Default;
 365    return Result;
 366 }
 367                                                                         /*}}}*/
 368 // TagSection::FindULL - Find an unsigned long long integer             /*{{{*/
 369 // ---------------------------------------------------------------------
 370 /* */
 371 unsigned long long pkgTagSection::FindULL(const char *Tag, unsigned long long const &Default) const
 372 {
 373    const char *Start;
 374    const char *Stop;
 375    if (Find(Tag,Start,Stop) == false)
 376       return Default;
 377
 378    // Copy it into a temp buffer so we can use strtoull
 379    char S[100];
 380    if ((unsigned)(Stop - Start) >= sizeof(S))
 381       return Default;
 382    strncpy(S,Start,Stop-Start);
 383    S[Stop - Start] = 0;
 384
 385    char *End;
 386    unsigned long long Result = strtoull(S,&End,10);
 387    if (S == End)
 388       return Default;
 389    return Result;
 390 }
 391                                                                         /*}}}*/
 392 // TagSection::FindFlag - Locate a yes/no type flag                     /*{{{*/
 393 // ---------------------------------------------------------------------
 394 /* The bits marked in Flag are masked on/off in Flags */
 395 bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags,
 396                              unsigned long Flag) const
 397 {
 398    const char *Start;
 399    const char *Stop;
 400    if (Find(Tag,Start,Stop) == false)
 401       return true;
 402
 403    switch (StringToBool(string(Start,Stop)))
 404    {
 405       case 0:
 406       Flags &= ~Flag;
 407       return true;
 408
 409       case 1:
 410       Flags |= Flag;
 411       return true;
 412
 413       default:
 414       _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str());
 415       return true;
 416    }
 417    return true;
 418 }
 419                                                                         /*}}}*/
 420 // TFRewrite - Rewrite a control record                                 /*{{{*/
 421 // ---------------------------------------------------------------------
 422 /* This writes the control record to stdout rewriting it as necessary. The
 423    override map item specificies the rewriting rules to follow. This also
 424    takes the time to sort the feild list. */
 425
 426 /* The order of this list is taken from dpkg source lib/parse.c the fieldinfos
 427    array. */
 428 static const char *iTFRewritePackageOrder[] = {
 429                           "Package",
 430                           "Essential",
 431                           "Status",
 432                           "Priority",
 433                           "Section",
 434                           "Installed-Size",
 435                           "Maintainer",
 436                           "Original-Maintainer",
 437                           "Architecture",
 438                           "Source",
 439                           "Version",
 440                            "Revision",         // Obsolete
 441                            "Config-Version",   // Obsolete
 442                           "Replaces",
 443                           "Provides",
 444                           "Depends",
 445                           "Pre-Depends",
 446                           "Recommends",
 447                           "Suggests",
 448                           "Conflicts",
 449                           "Breaks",
 450                           "Conffiles",
 451                           "Filename",
 452                           "Size",
 453                           "MD5Sum",
 454                           "SHA1",
 455                           "SHA256",
 456                            "MSDOS-Filename",   // Obsolete
 457                           "Description",
 458                           0};
 459 static const char *iTFRewriteSourceOrder[] = {"Package",
 460                                       "Source",
 461                                       "Binary",
 462                                       "Version",
 463                                       "Priority",
 464                                       "Section",
 465                                       "Maintainer",
 466                                       "Original-Maintainer",
 467                                       "Build-Depends",
 468                                       "Build-Depends-Indep",
 469                                       "Build-Conflicts",
 470                                       "Build-Conflicts-Indep",
 471                                       "Architecture",
 472                                       "Standards-Version",
 473                                       "Format",
 474                                       "Directory",
 475                                       "Files",
 476                                       0};
 477
 478 /* Two levels of initialization are used because gcc will set the symbol
 479    size of an array to the length of the array, causing dynamic relinking
 480    errors. Doing this makes the symbol size constant */
 481 const char **TFRewritePackageOrder = iTFRewritePackageOrder;
 482 const char **TFRewriteSourceOrder = iTFRewriteSourceOrder;
 483
 484 bool TFRewrite(FILE *Output,pkgTagSection const &Tags,const char *Order[],
 485                TFRewriteData *Rewrite)
 486 {
 487    unsigned char Visited[256];   // Bit 1 is Order, Bit 2 is Rewrite
 488    for (unsigned I = 0; I != 256; I++)
 489       Visited[I] = 0;
 490
 491    // Set new tag up as necessary.
 492    for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
 493    {
 494       if (Rewrite[J].NewTag == 0)
 495          Rewrite[J].NewTag = Rewrite[J].Tag;
 496    }
 497
 498    // Write all all of the tags, in order.
 499    for (unsigned int I = 0; Order[I] != 0; I++)
 500    {
 501       bool Rewritten = false;
 502
 503       // See if this is a field that needs to be rewritten
 504       for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
 505       {
 506          if (strcasecmp(Rewrite[J].Tag,Order[I]) == 0)
 507          {
 508             Visited[J] |= 2;
 509             if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
 510             {
 511                if (isspace(Rewrite[J].Rewrite[0]))
 512                   fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
 513                else
 514                   fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
 515             }
 516
 517             Rewritten = true;
 518             break;
 519          }
 520       }
 521
 522       // See if it is in the fragment
 523       unsigned Pos;
 524       if (Tags.Find(Order[I],Pos) == false)
 525          continue;
 526       Visited[Pos] |= 1;
 527
 528       if (Rewritten == true)
 529          continue;
 530
 531       /* Write out this element, taking a moment to rewrite the tag
 532          in case of changes of case. */
 533       const char *Start;
 534       const char *Stop;
 535       Tags.Get(Start,Stop,Pos);
 536
 537       if (fputs(Order[I],Output) < 0)
 538          return _error->Errno("fputs","IO Error to output");
 539       Start += strlen(Order[I]);
 540       if (fwrite(Start,Stop - Start,1,Output) != 1)
 541          return _error->Errno("fwrite","IO Error to output");
 542       if (Stop[-1] != '\n')
 543          fprintf(Output,"\n");
 544    }
 545
 546    // Now write all the old tags that were missed.
 547    for (unsigned int I = 0; I != Tags.Count(); I++)
 548    {
 549       if ((Visited[I] & 1) == 1)
 550          continue;
 551
 552       const char *Start;
 553       const char *Stop;
 554       Tags.Get(Start,Stop,I);
 555       const char *End = Start;
 556       for (; End < Stop && *End != ':'; End++);
 557
 558       // See if this is a field that needs to be rewritten
 559       bool Rewritten = false;
 560       for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
 561       {
 562          if (stringcasecmp(Start,End,Rewrite[J].Tag) == 0)
 563          {
 564             Visited[J] |= 2;
 565             if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
 566             {
 567                if (isspace(Rewrite[J].Rewrite[0]))
 568                   fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
 569                else
 570                   fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
 571             }
 572
 573             Rewritten = true;
 574             break;
 575          }
 576       }
 577
 578       if (Rewritten == true)
 579          continue;
 580
 581       // Write out this element
 582       if (fwrite(Start,Stop - Start,1,Output) != 1)
 583          return _error->Errno("fwrite","IO Error to output");
 584       if (Stop[-1] != '\n')
 585          fprintf(Output,"\n");
 586    }
 587
 588    // Now write all the rewrites that were missed
 589    for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++)
 590    {
 591       if ((Visited[J] & 2) == 2)
 592          continue;
 593
 594       if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0)
 595       {
 596          if (isspace(Rewrite[J].Rewrite[0]))
 597             fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
 598          else
 599             fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite);
 600       }
 601    }
 602
 603    return true;
 604 }
 605                                                                         /*}}}*/