| 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. It also |
| 197 | indexes the tags in the section. This very simple hash function for the |
| 198 | last 8 letters gives very good performance on the debian package files */ |
| 199 | inline static unsigned long AlphaHash(const char *Text, const char *End = 0) |
| 200 | { |
| 201 | unsigned long Res = 0; |
| 202 | for (; Text != End && *Text != ':' && *Text != 0; Text++) |
| 203 | Res = ((unsigned long)(*Text) & 0xDF) ^ (Res << 1); |
| 204 | return Res & 0xFF; |
| 205 | } |
| 206 | |
| 207 | bool pkgTagSection::Scan(const char *Start,unsigned long MaxLength) |
| 208 | { |
| 209 | const char *End = Start + MaxLength; |
| 210 | Stop = Section = Start; |
| 211 | memset(AlphaIndexes,0,sizeof(AlphaIndexes)); |
| 212 | |
| 213 | if (Stop == 0) |
| 214 | return false; |
| 215 | |
| 216 | TagCount = 0; |
| 217 | while (TagCount+1 < sizeof(Indexes)/sizeof(Indexes[0]) && Stop < End) |
| 218 | { |
| 219 | // Start a new index and add it to the hash |
| 220 | if (isspace(Stop[0]) == 0) |
| 221 | { |
| 222 | Indexes[TagCount++] = Stop - Section; |
| 223 | AlphaIndexes[AlphaHash(Stop,End)] = TagCount; |
| 224 | } |
| 225 | |
| 226 | Stop = (const char *)memchr(Stop,'\n',End - Stop); |
| 227 | |
| 228 | if (Stop == 0) |
| 229 | return false; |
| 230 | |
| 231 | for (; Stop+1 < End && Stop[1] == '\r'; Stop++); |
| 232 | |
| 233 | // Double newline marks the end of the record |
| 234 | if (Stop+1 < End && Stop[1] == '\n') |
| 235 | { |
| 236 | Indexes[TagCount] = Stop - Section; |
| 237 | for (; Stop < End && (Stop[0] == '\n' || Stop[0] == '\r'); Stop++); |
| 238 | return true; |
| 239 | } |
| 240 | |
| 241 | Stop++; |
| 242 | } |
| 243 | |
| 244 | return false; |
| 245 | } |
| 246 | /*}}}*/ |
| 247 | // TagSection::Trim - Trim off any trailing garbage /*{{{*/ |
| 248 | // --------------------------------------------------------------------- |
| 249 | /* There should be exactly 1 newline at the end of the buffer, no more. */ |
| 250 | void pkgTagSection::Trim() |
| 251 | { |
| 252 | for (; Stop > Section + 2 && (Stop[-2] == '\n' || Stop[-2] == '\r'); Stop--); |
| 253 | } |
| 254 | /*}}}*/ |
| 255 | // TagSection::Find - Locate a tag /*{{{*/ |
| 256 | // --------------------------------------------------------------------- |
| 257 | /* This searches the section for a tag that matches the given string. */ |
| 258 | bool pkgTagSection::Find(const char *Tag,unsigned &Pos) const |
| 259 | { |
| 260 | unsigned int Length = strlen(Tag); |
| 261 | unsigned int I = AlphaIndexes[AlphaHash(Tag)]; |
| 262 | if (I == 0) |
| 263 | return false; |
| 264 | I--; |
| 265 | |
| 266 | for (unsigned int Counter = 0; Counter != TagCount; Counter++, |
| 267 | I = (I+1)%TagCount) |
| 268 | { |
| 269 | const char *St; |
| 270 | St = Section + Indexes[I]; |
| 271 | if (strncasecmp(Tag,St,Length) != 0) |
| 272 | continue; |
| 273 | |
| 274 | // Make sure the colon is in the right place |
| 275 | const char *C = St + Length; |
| 276 | for (; isspace(*C) != 0; C++); |
| 277 | if (*C != ':') |
| 278 | continue; |
| 279 | Pos = I; |
| 280 | return true; |
| 281 | } |
| 282 | |
| 283 | Pos = 0; |
| 284 | return false; |
| 285 | } |
| 286 | /*}}}*/ |
| 287 | // TagSection::Find - Locate a tag /*{{{*/ |
| 288 | // --------------------------------------------------------------------- |
| 289 | /* This searches the section for a tag that matches the given string. */ |
| 290 | bool pkgTagSection::Find(const char *Tag,const char *&Start, |
| 291 | const char *&End) const |
| 292 | { |
| 293 | unsigned int Length = strlen(Tag); |
| 294 | unsigned int I = AlphaIndexes[AlphaHash(Tag)]; |
| 295 | if (I == 0) |
| 296 | return false; |
| 297 | I--; |
| 298 | |
| 299 | for (unsigned int Counter = 0; Counter != TagCount; Counter++, |
| 300 | I = (I+1)%TagCount) |
| 301 | { |
| 302 | const char *St; |
| 303 | St = Section + Indexes[I]; |
| 304 | if (strncasecmp(Tag,St,Length) != 0) |
| 305 | continue; |
| 306 | |
| 307 | // Make sure the colon is in the right place |
| 308 | const char *C = St + Length; |
| 309 | for (; isspace(*C) != 0; C++); |
| 310 | if (*C != ':') |
| 311 | continue; |
| 312 | |
| 313 | // Strip off the gunk from the start end |
| 314 | Start = C; |
| 315 | End = Section + Indexes[I+1]; |
| 316 | if (Start >= End) |
| 317 | return _error->Error("Internal parsing error"); |
| 318 | |
| 319 | for (; (isspace(*Start) != 0 || *Start == ':') && Start < End; Start++); |
| 320 | for (; isspace(End[-1]) != 0 && End > Start; End--); |
| 321 | |
| 322 | return true; |
| 323 | } |
| 324 | |
| 325 | Start = End = 0; |
| 326 | return false; |
| 327 | } |
| 328 | /*}}}*/ |
| 329 | // TagSection::FindS - Find a string /*{{{*/ |
| 330 | // --------------------------------------------------------------------- |
| 331 | /* */ |
| 332 | string pkgTagSection::FindS(const char *Tag) const |
| 333 | { |
| 334 | const char *Start; |
| 335 | const char *End; |
| 336 | if (Find(Tag,Start,End) == false) |
| 337 | return string(); |
| 338 | return string(Start,End); |
| 339 | } |
| 340 | /*}}}*/ |
| 341 | // TagSection::FindI - Find an integer /*{{{*/ |
| 342 | // --------------------------------------------------------------------- |
| 343 | /* */ |
| 344 | signed int pkgTagSection::FindI(const char *Tag,signed long Default) const |
| 345 | { |
| 346 | const char *Start; |
| 347 | const char *Stop; |
| 348 | if (Find(Tag,Start,Stop) == false) |
| 349 | return Default; |
| 350 | |
| 351 | // Copy it into a temp buffer so we can use strtol |
| 352 | char S[300]; |
| 353 | if ((unsigned)(Stop - Start) >= sizeof(S)) |
| 354 | return Default; |
| 355 | strncpy(S,Start,Stop-Start); |
| 356 | S[Stop - Start] = 0; |
| 357 | |
| 358 | char *End; |
| 359 | signed long Result = strtol(S,&End,10); |
| 360 | if (S == End) |
| 361 | return Default; |
| 362 | return Result; |
| 363 | } |
| 364 | /*}}}*/ |
| 365 | // TagSection::FindFlag - Locate a yes/no type flag /*{{{*/ |
| 366 | // --------------------------------------------------------------------- |
| 367 | /* The bits marked in Flag are masked on/off in Flags */ |
| 368 | bool pkgTagSection::FindFlag(const char *Tag,unsigned long &Flags, |
| 369 | unsigned long Flag) const |
| 370 | { |
| 371 | const char *Start; |
| 372 | const char *Stop; |
| 373 | if (Find(Tag,Start,Stop) == false) |
| 374 | return true; |
| 375 | |
| 376 | switch (StringToBool(string(Start,Stop))) |
| 377 | { |
| 378 | case 0: |
| 379 | Flags &= ~Flag; |
| 380 | return true; |
| 381 | |
| 382 | case 1: |
| 383 | Flags |= Flag; |
| 384 | return true; |
| 385 | |
| 386 | default: |
| 387 | _error->Warning("Unknown flag value: %s",string(Start,Stop).c_str()); |
| 388 | return true; |
| 389 | } |
| 390 | return true; |
| 391 | } |
| 392 | /*}}}*/ |
| 393 | |
| 394 | // TFRewrite - Rewrite a control record /*{{{*/ |
| 395 | // --------------------------------------------------------------------- |
| 396 | /* This writes the control record to stdout rewriting it as necessary. The |
| 397 | override map item specificies the rewriting rules to follow. This also |
| 398 | takes the time to sort the feild list. */ |
| 399 | |
| 400 | /* The order of this list is taken from dpkg source lib/parse.c the fieldinfos |
| 401 | array. */ |
| 402 | static const char *iTFRewritePackageOrder[] = { |
| 403 | "Package", |
| 404 | "Essential", |
| 405 | "Status", |
| 406 | "Priority", |
| 407 | "Section", |
| 408 | "Installed-Size", |
| 409 | "Maintainer", |
| 410 | "Original-Maintainer", |
| 411 | "Architecture", |
| 412 | "Source", |
| 413 | "Version", |
| 414 | "Revision", // Obsolete |
| 415 | "Config-Version", // Obsolete |
| 416 | "Replaces", |
| 417 | "Provides", |
| 418 | "Depends", |
| 419 | "Pre-Depends", |
| 420 | "Recommends", |
| 421 | "Suggests", |
| 422 | "Conflicts", |
| 423 | "Breaks", |
| 424 | "Conffiles", |
| 425 | "Filename", |
| 426 | "Size", |
| 427 | "MD5Sum", |
| 428 | "SHA1", |
| 429 | "SHA256", |
| 430 | "MSDOS-Filename", // Obsolete |
| 431 | "Description", |
| 432 | 0}; |
| 433 | static const char *iTFRewriteSourceOrder[] = {"Package", |
| 434 | "Source", |
| 435 | "Binary", |
| 436 | "Version", |
| 437 | "Priority", |
| 438 | "Section", |
| 439 | "Maintainer", |
| 440 | "Original-Maintainer", |
| 441 | "Build-Depends", |
| 442 | "Build-Depends-Indep", |
| 443 | "Build-Conflicts", |
| 444 | "Build-Conflicts-Indep", |
| 445 | "Architecture", |
| 446 | "Standards-Version", |
| 447 | "Format", |
| 448 | "Directory", |
| 449 | "Files", |
| 450 | 0}; |
| 451 | |
| 452 | /* Two levels of initialization are used because gcc will set the symbol |
| 453 | size of an array to the length of the array, causing dynamic relinking |
| 454 | errors. Doing this makes the symbol size constant */ |
| 455 | const char **TFRewritePackageOrder = iTFRewritePackageOrder; |
| 456 | const char **TFRewriteSourceOrder = iTFRewriteSourceOrder; |
| 457 | |
| 458 | bool TFRewrite(FILE *Output,pkgTagSection const &Tags,const char *Order[], |
| 459 | TFRewriteData *Rewrite) |
| 460 | { |
| 461 | unsigned char Visited[256]; // Bit 1 is Order, Bit 2 is Rewrite |
| 462 | for (unsigned I = 0; I != 256; I++) |
| 463 | Visited[I] = 0; |
| 464 | |
| 465 | // Set new tag up as necessary. |
| 466 | for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) |
| 467 | { |
| 468 | if (Rewrite[J].NewTag == 0) |
| 469 | Rewrite[J].NewTag = Rewrite[J].Tag; |
| 470 | } |
| 471 | |
| 472 | // Write all all of the tags, in order. |
| 473 | for (unsigned int I = 0; Order[I] != 0; I++) |
| 474 | { |
| 475 | bool Rewritten = false; |
| 476 | |
| 477 | // See if this is a field that needs to be rewritten |
| 478 | for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) |
| 479 | { |
| 480 | if (strcasecmp(Rewrite[J].Tag,Order[I]) == 0) |
| 481 | { |
| 482 | Visited[J] |= 2; |
| 483 | if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) |
| 484 | { |
| 485 | if (isspace(Rewrite[J].Rewrite[0])) |
| 486 | fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); |
| 487 | else |
| 488 | fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); |
| 489 | } |
| 490 | |
| 491 | Rewritten = true; |
| 492 | break; |
| 493 | } |
| 494 | } |
| 495 | |
| 496 | // See if it is in the fragment |
| 497 | unsigned Pos; |
| 498 | if (Tags.Find(Order[I],Pos) == false) |
| 499 | continue; |
| 500 | Visited[Pos] |= 1; |
| 501 | |
| 502 | if (Rewritten == true) |
| 503 | continue; |
| 504 | |
| 505 | /* Write out this element, taking a moment to rewrite the tag |
| 506 | in case of changes of case. */ |
| 507 | const char *Start; |
| 508 | const char *Stop; |
| 509 | Tags.Get(Start,Stop,Pos); |
| 510 | |
| 511 | if (fputs(Order[I],Output) < 0) |
| 512 | return _error->Errno("fputs","IO Error to output"); |
| 513 | Start += strlen(Order[I]); |
| 514 | if (fwrite(Start,Stop - Start,1,Output) != 1) |
| 515 | return _error->Errno("fwrite","IO Error to output"); |
| 516 | if (Stop[-1] != '\n') |
| 517 | fprintf(Output,"\n"); |
| 518 | } |
| 519 | |
| 520 | // Now write all the old tags that were missed. |
| 521 | for (unsigned int I = 0; I != Tags.Count(); I++) |
| 522 | { |
| 523 | if ((Visited[I] & 1) == 1) |
| 524 | continue; |
| 525 | |
| 526 | const char *Start; |
| 527 | const char *Stop; |
| 528 | Tags.Get(Start,Stop,I); |
| 529 | const char *End = Start; |
| 530 | for (; End < Stop && *End != ':'; End++); |
| 531 | |
| 532 | // See if this is a field that needs to be rewritten |
| 533 | bool Rewritten = false; |
| 534 | for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) |
| 535 | { |
| 536 | if (stringcasecmp(Start,End,Rewrite[J].Tag) == 0) |
| 537 | { |
| 538 | Visited[J] |= 2; |
| 539 | if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) |
| 540 | { |
| 541 | if (isspace(Rewrite[J].Rewrite[0])) |
| 542 | fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); |
| 543 | else |
| 544 | fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); |
| 545 | } |
| 546 | |
| 547 | Rewritten = true; |
| 548 | break; |
| 549 | } |
| 550 | } |
| 551 | |
| 552 | if (Rewritten == true) |
| 553 | continue; |
| 554 | |
| 555 | // Write out this element |
| 556 | if (fwrite(Start,Stop - Start,1,Output) != 1) |
| 557 | return _error->Errno("fwrite","IO Error to output"); |
| 558 | if (Stop[-1] != '\n') |
| 559 | fprintf(Output,"\n"); |
| 560 | } |
| 561 | |
| 562 | // Now write all the rewrites that were missed |
| 563 | for (unsigned int J = 0; Rewrite != 0 && Rewrite[J].Tag != 0; J++) |
| 564 | { |
| 565 | if ((Visited[J] & 2) == 2) |
| 566 | continue; |
| 567 | |
| 568 | if (Rewrite[J].Rewrite != 0 && Rewrite[J].Rewrite[0] != 0) |
| 569 | { |
| 570 | if (isspace(Rewrite[J].Rewrite[0])) |
| 571 | fprintf(Output,"%s:%s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); |
| 572 | else |
| 573 | fprintf(Output,"%s: %s\n",Rewrite[J].NewTag,Rewrite[J].Rewrite); |
| 574 | } |
| 575 | } |
| 576 | |
| 577 | return true; |
| 578 | } |
| 579 | /*}}}*/ |