1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879
|
// =================================================================================================
// ADOBE SYSTEMS INCORPORATED
// Copyright 2009 Adobe Systems Incorporated
// All Rights Reserved
//
// NOTICE: Adobe permits you to use, modify, and distribute this file in accordance with the terms
// of the Adobe license agreement accompanying it.
// =================================================================================================
// must have access to handler class fields...
#include "RIFF.hpp"
#include "RIFF_Support.hpp"
#include "RIFF_Handler.hpp"
using namespace RIFF;
namespace RIFF {
// GENERAL STATIC FUNCTIONS ////////////////////////////////////////
Chunk* getChunk ( ContainerChunk* parent, RIFF_MetaHandler* handler )
{
LFA_FileRef file = handler->parent->fileRef;
XMP_Uns8 level = handler->level;
XMP_Uns32 peek = LFA_PeekUns32_LE( file );
if ( level == 0 )
{
XMP_Validate( peek == kChunk_RIFF, "expected RIFF chunk not found", kXMPErr_BadFileFormat );
XMP_Enforce( parent == NULL );
}
else
{
XMP_Validate( peek != kChunk_RIFF, "unexpected RIFF chunk below top-level", kXMPErr_BadFileFormat );
XMP_Enforce( parent != NULL );
}
switch( peek )
{
case kChunk_RIFF:
return new ContainerChunk( parent, handler );
case kChunk_LIST:
{
if ( level != 1 ) break; // only care on this level
// look further (beyond 4+4 = beyond id+size) to check on relevance
LFA_Seek( file, 8, SEEK_CUR );
XMP_Uns32 containerType = LFA_PeekUns32_LE( file );
LFA_Seek( file, -8, SEEK_CUR );
bool isRelevantList = ( containerType== kType_INFO || containerType == kType_Tdat );
if ( !isRelevantList ) break;
return new ContainerChunk( parent, handler );
}
case kChunk_XMP:
if ( level != 1 ) break; // ignore on inappropriate levels (might be compound metadata?)
return new XMPChunk( parent, handler );
case kChunk_DISP:
{
if ( level != 1 ) break; // only care on this level
// peek even further to see if type is 0x001 and size is reasonable
LFA_Seek( file , 4, SEEK_CUR ); // jump DISP
XMP_Uns32 dispSize = LFA_ReadUns32_LE( file );
XMP_Uns32 dispType = LFA_ReadUns32_LE( file );
LFA_Seek( file , -12, SEEK_CUR); // rewind, be in front of chunkID again
// only take as a relevant disp if both criteria met,
// otherwise treat as generic chunk!
if ( (dispType == 0x0001) && ( dispSize < 256 * 1024 ) )
{
ValueChunk* r = new ValueChunk( parent, handler );
handler->dispChunk = r;
return r;
}
break; // treat as irrelevant (non-0x1) DISP chunks as generic chunk
}
case kChunk_bext:
{
if ( level != 1 ) break; // only care on this level
// store for now in a value chunk
ValueChunk* r = new ValueChunk( parent, handler );
handler->bextChunk = r;
return r;
}
case kChunk_PrmL:
{
if ( level != 1 ) break; // only care on this level
ValueChunk* r = new ValueChunk( parent, handler );
handler->prmlChunk = r;
return r;
}
case kChunk_Cr8r:
{
if ( level != 1 ) break; // only care on this level
ValueChunk* r = new ValueChunk( parent, handler );
handler->cr8rChunk = r;
return r;
}
case kChunk_JUNQ:
case kChunk_JUNK:
{
JunkChunk* r = new JunkChunk( parent, handler );
return r;
}
}
// this "default:" section must be ouside switch bracket, to be
// reachable by all those break statements above:
// digest 'valuable' container chunks: LIST:INFO, LIST:Tdat
bool insideRelevantList = ( level==2 && parent->id == kChunk_LIST
&& ( parent->containerType== kType_INFO || parent->containerType == kType_Tdat ));
if ( insideRelevantList )
{
ValueChunk* r = new ValueChunk( parent, handler );
return r;
}
// general chunk of no interest, treat as unknown blob
return new Chunk( parent, handler, true, chunk_GENERAL );
}
// BASE CLASS CHUNK ///////////////////////////////////////////////
// ad hoc creation
Chunk::Chunk( ContainerChunk* parent, ChunkType c, XMP_Uns32 id )
{
this->chunkType = c; // base class assumption
this->parent = parent;
this->id = id;
this->oldSize = 0;
this->newSize = 8;
this->oldPos = 0; // inevitable for ad-hoc
this->needSizeFix = false;
// good parenting for latter destruction
if ( this->parent != NULL )
{
this->parent->children.push_back( this );
if( this->chunkType == chunk_VALUE )
this->parent->childmap.insert( make_pair( this->id, (ValueChunk*) this ) );
}
}
// parsing creation
Chunk::Chunk( ContainerChunk* parent, RIFF_MetaHandler* handler, bool skip, ChunkType c )
{
chunkType = c; // base class assumption
this->parent = parent;
this->oldSize = 0;
this->hasChange = false; // [2414649] valid assumption at creation time
LFA_FileRef file = handler->parent->fileRef;
this->oldPos = LFA_Tell( file );
this->id = LFA_ReadUns32_LE( file );
this->oldSize = LFA_ReadUns32_LE( file ) + 8;
// Make sure the size is within expected bounds.
XMP_Int64 chunkEnd = this->oldPos + this->oldSize;
XMP_Int64 chunkLimit = handler->oldFileSize;
if ( parent != 0 ) chunkLimit = parent->oldPos + parent->oldSize;
if ( chunkEnd > chunkLimit ) {
bool isUpdate = XMP_OptionIsSet ( handler->parent->openFlags, kXMPFiles_OpenForUpdate );
bool repairFile = XMP_OptionIsSet ( handler->parent->openFlags, kXMPFiles_OpenRepairFile );
if ( (! isUpdate) || (repairFile && (parent == 0)) ) {
this->oldSize = chunkLimit - this->oldPos;
} else {
XMP_Throw ( "Bad RIFF chunk size", kXMPErr_BadFileFormat );
}
}
this->newSize = this->oldSize;
this->needSizeFix = false;
if ( skip )
{
bool ok;
LFA_Seek( file, this->oldSize - 8 , SEEK_CUR, &ok );
XMP_Validate( ok , "skipped beyond end of file (truncated file?)", kXMPErr_BadFileFormat );
}
// "good parenting", essential for latter destruction.
if ( this->parent != NULL )
{
this->parent->children.push_back( this );
if( this->chunkType == chunk_VALUE )
this->parent->childmap.insert( make_pair( this->id, (ValueChunk*) this ) );
}
}
void Chunk::changesAndSize( RIFF_MetaHandler* handler )
{
// only unknown chunks should reach this method,
// all others must reach overloads, hence little to do here:
hasChange = false; // unknown chunk ==> no change, naturally
this->newSize = this->oldSize;
}
std::string Chunk::toString(XMP_Uns8 level )
{
char buffer[256];
snprintf( buffer, 255, "%.4s -- "
"oldSize: 0x%.8llX, "
"newSize: 0x%.8llX, "
"oldPos: 0x%.8llX\n",
(char*)(&this->id), this->oldSize, this->newSize, this->oldPos );
return std::string(buffer);
}
void Chunk::write( RIFF_MetaHandler* handler, LFA_FileRef file , bool isMainChunk )
{
throw new XMP_Error(kXMPErr_InternalFailure, "Chunk::write never to be called for unknown chunks.");
}
Chunk::~Chunk()
{
//nothing
}
// CONTAINER CHUNK /////////////////////////////////////////////////
// a) creation
// [2376832] expectedSize - minimum padding "parking size" to use, if not available append to end
ContainerChunk::ContainerChunk( ContainerChunk* parent, XMP_Uns32 id, XMP_Uns32 containerType ) : Chunk( NULL /* !! */, chunk_CONTAINER, id )
{
// accept no unparented ConatinerChunks
XMP_Enforce( parent != NULL );
this->containerType = containerType;
this->newSize = 12;
this->parent = parent;
chunkVect* siblings = &parent->children;
// add at end. ( oldSize==0 will flag optimization later in the process)
siblings->push_back( this );
}
// b) parsing
ContainerChunk::ContainerChunk( ContainerChunk* parent, RIFF_MetaHandler* handler ) : Chunk( parent, handler, false, chunk_CONTAINER )
{
bool repairMode = ( 0 != ( handler->parent->openFlags & kXMPFiles_OpenRepairFile ));
try
{
LFA_FileRef file = handler->parent->fileRef;
XMP_Uns8 level = handler->level;
// get type of container chunk
this->containerType = LFA_ReadUns32_LE( file );
// ensure legality of top-level chunks
if ( level == 0 && handler->riffChunks.size() > 0 )
{
XMP_Validate( handler->parent->format == kXMP_AVIFile, "only AVI may have multiple top-level chunks", kXMPErr_BadFileFormat );
XMP_Validate( this->containerType == kType_AVIX, "all chunks beyond main chunk must be type AVIX", kXMPErr_BadFileFormat );
}
// has *relevant* subChunks? (there might be e.g. non-INFO LIST chunks we don't care about)
bool hasSubChunks = ( ( this->id == kChunk_RIFF ) ||
( this->id == kChunk_LIST && this->containerType == kType_INFO ) ||
( this->id == kChunk_LIST && this->containerType == kType_Tdat )
);
XMP_Int64 endOfChunk = this->oldPos + this->oldSize;
// this statement catches beyond-EoF-offsets on any level
// exception: level 0, tolerate if in repairMode
if ( (level == 0) && repairMode && (endOfChunk > handler->oldFileSize) )
{
endOfChunk = handler->oldFileSize; // assign actual file size
this->oldSize = endOfChunk - this->oldPos; //reversely calculate correct oldSize
}
XMP_Validate( endOfChunk <= handler->oldFileSize, "offset beyond EoF", kXMPErr_BadFileFormat );
Chunk* curChild = 0;
if ( hasSubChunks )
{
handler->level++;
while ( LFA_Tell( file ) < endOfChunk )
{
curChild = RIFF::getChunk( this, handler );
// digest pad byte - no value validation (0), since some 3rd party files have non-0-padding.
if ( LFA_Tell(file) % 2 == 1 )
{
// [1521093] tolerate missing pad byte at very end of file:
XMP_Uns8 pad;
LFA_Read ( file, &pad, 1 ); // Read the pad, tolerate being at EOF.
}
// within relevant LISTs, relentlesly delete junk chunks (create a single one
// at end as part of updateAndChanges()
if ( (containerType== kType_INFO || containerType == kType_Tdat)
&& ( curChild->chunkType == chunk_JUNK ) )
{
this->children.pop_back();
delete curChild;
} // for other chunks: join neighouring Junk chunks into one
else if ( (curChild->chunkType == chunk_JUNK) && ( this->children.size() >= 2 ) )
{
// nb: if there are e.g 2 chunks, then last one is at(1), prev one at(0) ==> '-2'
Chunk* prevChunk = this->children.at( this->children.size() - 2 );
if ( prevChunk->chunkType == chunk_JUNK )
{
// stack up size to prior chunk
prevChunk->oldSize += curChild->oldSize;
prevChunk->newSize += curChild->newSize;
XMP_Enforce( prevChunk->oldSize == prevChunk->newSize );
// destroy current chunk
this->children.pop_back();
delete curChild;
}
}
}
handler->level--;
XMP_Validate( LFA_Tell( file ) == endOfChunk, "subchunks exceed outer chunk size", kXMPErr_BadFileFormat );
// pointers for later legacy processing
if ( level==1 && this->id==kChunk_LIST && this->containerType == kType_INFO )
handler->listInfoChunk = this;
if ( level==1 && this->id==kChunk_LIST && this->containerType == kType_Tdat )
handler->listTdatChunk = this;
}
else // skip non-interest container chunk
{
bool ok;
LFA_Seek( file, this->oldSize - 8 - 4, SEEK_CUR, &ok );
XMP_Validate( ok , "skipped beyond end of file 2 (truncated file?)", kXMPErr_BadFileFormat );
} // if - else
} // try
catch (XMP_Error& e) {
this->release(); // free resources
if ( this->parent != 0)
this->parent->children.pop_back(); // hereby taken care of, so removing myself...
throw e; // re-throw
}
}
void ContainerChunk::changesAndSize( RIFF_MetaHandler* handler )
{
// Walk the container subtree adjusting the children that have size changes. The only containers
// are RIFF and LIST chunks, they are treated differently.
//
// LISTs get recomposed as a whole. Existing JUNK children of a LIST are removed, existing real
// children are left in order with their new size, new children have already been appended. The
// LIST as a whole gets a new size that is the sum of the final children.
//
// Special rules apply to various children of a RIFF container. FIrst, adjacent JUNK children
// are combined, this simplifies maximal reuse. The children are recursively adjusted in order
// to get their final size.
//
// Try to determine the final placement of each RIFF child using general rules:
// - if the size is unchanged: leave at current location
// - if the chunk is at the end of the last RIFF chunk and grows: leave at current location
// - if there is enough following JUNK: add part of the JUNK, adjust remaining JUNK size
// - if it shrinks by 9 bytes or more: carve off trailing JUNK
// - try to find adequate JUNK in the current parent
//
// Use child-specific rules as a last resort:
// - if it is LIST:INFO: delete it, must be in first RIFF chunk
// - for others: move to end of last RIFF chunk, make old space JUNK
// ! Don't create any junk chunks of exactly 8 bytes, just a header and no content. That has a
// ! size field of zero, which hits a crashing bug in some versions of Windows Media Player.
bool isRIFFContainer = (this->id == kChunk_RIFF);
bool isLISTContainer = (this->id == kChunk_LIST);
XMP_Enforce ( isRIFFContainer | isLISTContainer );
XMP_Index childIndex; // Could be local to the loops, this simplifies debuging. Need a signed type!
Chunk * currChild;
if ( this->children.empty() ) {
if ( isRIFFContainer) {
this->newSize = 12; // Keep a minimal size container.
} else {
this->newSize = 0; // Will get removed from parent in outer call.
}
this->hasChange = true;
return; // Nothing more to do without children.
}
// Collapse adjacent RIFF junk children, remove all LIST junk children. Work back to front to
// simplify the effect of .erase() on the loop. Purposely ignore the first chunk.
for ( childIndex = (XMP_Index)this->children.size() - 1; childIndex > 0; --childIndex ) {
currChild = this->children[childIndex];
if ( currChild->chunkType != chunk_JUNK ) continue;
if ( isRIFFContainer ) {
Chunk * prevChild = this->children[childIndex-1];
if ( prevChild->chunkType != chunk_JUNK ) continue;
prevChild->oldSize += currChild->oldSize;
prevChild->newSize += currChild->newSize;
prevChild->hasChange = true;
}
this->children.erase ( this->children.begin() + childIndex );
delete currChild;
this->hasChange = true;
}
// Process the children of RIFF and LIST containers to get their final size. Remove empty
// children. Work back to front to simplify the effect of .erase() on the loop. Do not ignore
// the first chunk.
for ( childIndex = (XMP_Index)this->children.size() - 1; childIndex >= 0; --childIndex ) {
currChild = this->children[childIndex];
++handler->level;
currChild->changesAndSize ( handler );
--handler->level;
if ( (currChild->newSize == 8) || (currChild->newSize == 0) ) { // ! The newSIze is supposed to include the header.
this->children.erase ( this->children.begin() + childIndex );
delete currChild;
this->hasChange = true;
} else {
this->hasChange |= currChild->hasChange;
currChild->needSizeFix = (currChild->newSize != currChild->oldSize);
if ( currChild->needSizeFix && (currChild->newSize > currChild->oldSize) &&
(this == handler->lastChunk) && (childIndex+1 == (XMP_Index)this->children.size()) ) {
// Let an existing last-in-file chunk grow in-place. Shrinking is conceptually OK,
// but complicates later sanity check that the main AVI chunk is not OK to append
// other chunks later. Ignore new chunks, they might reuse junk space.
if ( currChild->oldSize != 0 ) currChild->needSizeFix = false;
}
}
}
// Go through the children of a RIFF container, adjusting the placement as necessary. In brief,
// things can only grow at the end of the last RIFF chunk, and non-junk chunks can't be shifted.
if ( isRIFFContainer ) {
for ( childIndex = 0; childIndex < (XMP_Index)this->children.size(); ++childIndex ) {
currChild = this->children[childIndex];
if ( ! currChild->needSizeFix ) continue;
currChild->needSizeFix = false;
XMP_Int64 sizeDiff = currChild->newSize - currChild->oldSize; // Positive for growth.
XMP_Uns8 padSize = (currChild->newSize & 1); // Need a pad for odd size.
// See if the following chunk is junk that can be utilized.
Chunk * nextChild = 0;
if ( childIndex+1 < (XMP_Index)this->children.size() ) nextChild = this->children[childIndex+1];
if ( (nextChild != 0) && (nextChild->chunkType == chunk_JUNK) ) {
if ( nextChild->newSize >= (9 + sizeDiff + padSize) ) {
// Incorporate part of the trailing junk, or make the trailing junk grow.
nextChild->newSize -= sizeDiff;
nextChild->newSize -= padSize;
nextChild->hasChange = true;
continue;
} else if ( nextChild->newSize == (sizeDiff + padSize) ) {
// Incorporate all of the trailing junk.
this->children.erase ( this->children.begin() + childIndex + 1 );
delete nextChild;
continue;
}
}
// See if the chunk shrinks enough to turn the leftover space into junk.
if ( (sizeDiff + padSize) <= -9 ) {
this->children.insert ( (this->children.begin() + childIndex + 1), new JunkChunk ( NULL, ((-sizeDiff) - padSize) ) );
continue;
}
// Look through the parent for a usable span of junk.
XMP_Index junkIndex;
Chunk * junkChunk = 0;
for ( junkIndex = 0; junkIndex < (XMP_Index)this->children.size(); ++junkIndex ) {
junkChunk = this->children[junkIndex];
if ( junkChunk->chunkType != chunk_JUNK ) continue;
if ( (junkChunk->newSize >= (9 + currChild->newSize + padSize)) ||
(junkChunk->newSize == (currChild->newSize + padSize)) ) break;
}
if ( junkIndex < (XMP_Index)this->children.size() ) {
// Use part or all of the junk for the relocated chunk, replace the old space with junk.
if ( junkChunk->newSize == (currChild->newSize + padSize) ) {
// The found junk is an exact fit.
this->children[junkIndex] = currChild;
delete junkChunk;
} else {
// The found junk has excess space. Insert the moving chunk and shrink the junk.
XMP_Assert ( junkChunk->newSize >= (9 + currChild->newSize + padSize) );
junkChunk->newSize -= (currChild->newSize + padSize);
junkChunk->hasChange = true;
this->children.insert ( (this->children.begin() + junkIndex), currChild );
if ( junkIndex < childIndex ) ++childIndex; // The insertion moved the current child.
}
if ( currChild->oldSize != 0 ) {
this->children[childIndex] = new JunkChunk ( 0, currChild->oldSize ); // Replace the old space with junk.
} else {
this->children.erase ( this->children.begin() + childIndex ); // Remove the newly created chunk's old location.
--childIndex; // Make the next loop iteration not skip a chunk.
}
continue;
}
// If this is a LIST:INFO chunk not in the last of multiple RIFF chunks, then give up
// and replace it with oldSize junk. Preserve the first RIFF chunk's original size.
bool isListInfo = (currChild->id == kChunk_LIST) && (currChild->chunkType == chunk_CONTAINER) &&
(((ContainerChunk*)currChild)->containerType == kType_INFO);
if ( isListInfo && (handler->riffChunks.size() > 1) &&
(this->id == kChunk_RIFF) && (this != handler->lastChunk) ) {
if ( currChild->oldSize != 0 ) {
this->children[childIndex] = new JunkChunk ( 0, currChild->oldSize );
} else {
this->children.erase ( this->children.begin() + childIndex );
--childIndex; // Make the next loop iteration not skip a chunk.
}
delete currChild;
continue;
}
// Move the chunk to the end of the last RIFF chunk and make the old space junk.
if ( (this == handler->lastChunk) && (childIndex+1 == (XMP_Index)this->children.size()) ) continue; // Already last.
handler->lastChunk->children.push_back( currChild );
if ( currChild->oldSize != 0 ) {
this->children[childIndex] = new JunkChunk ( 0, currChild->oldSize ); // Replace the old space with junk.
} else {
this->children.erase ( this->children.begin() + childIndex ); // Remove the newly created chunk's old location.
--childIndex; // Make the next loop iteration not skip a chunk.
}
}
}
// Compute the finished container's new size (for both RIFF and LIST).
this->newSize = 12; // Start with standard container header.
for ( childIndex = 0; childIndex < (XMP_Index)this->children.size(); ++childIndex ) {
currChild = this->children[childIndex];
this->newSize += currChild->newSize;
this->newSize += (this->newSize & 1); // Round up if odd.
}
XMP_Validate ( (this->newSize <= 0xFFFFFFFFLL), "No single chunk may be above 4 GB", kXMPErr_Unimplemented );
}
std::string ContainerChunk::toString(XMP_Uns8 level )
{
XMP_Int64 offset= 12; // compute offsets, just for informational purposes
// (actually only correct for first chunk)
char buffer[256];
snprintf( buffer, 255, "%.4s:%.4s, "
"oldSize: 0x%8llX, "
"newSize: 0x%.8llX, "
"oldPos: 0x%.8llX\n",
(char*)(&this->id), (char*)(&this->containerType), this->oldSize, this->newSize, this->oldPos );
std::string r(buffer);
chunkVectIter iter;
for( iter = this->children.begin(); iter != this->children.end(); iter++ )
{
char buffer[256];
snprintf( buffer, 250, "offset 0x%.8llX", offset );
r += std::string ( level*4, ' ' ) + std::string( buffer ) + ":" + (*iter)->toString( level + 1 );
offset += (*iter)->newSize;
if ( offset % 2 == 1 )
offset++;
}
return std::string(r);
}
void ContainerChunk::write( RIFF_MetaHandler* handler, LFA_FileRef file, bool isMainChunk )
{
bool ok;
if ( isMainChunk )
LFA_Rewind( file );
// enforce even position
XMP_Int64 chunkStart = LFA_Tell(file);
XMP_Int64 chunkEnd = chunkStart + this->newSize;
XMP_Enforce( chunkStart % 2 == 0 );
chunkVect *rc = &this->children;
// [2473303] have to write back-to-front to avoid stomp-on-feet
XMP_Int64 childStart = chunkEnd;
for ( XMP_Int32 chunkNo = (XMP_Int32)(rc->size() -1); chunkNo >= 0; chunkNo-- )
{
Chunk* cur = rc->at(chunkNo);
// pad byte first
if ( cur->newSize % 2 == 1 )
{
childStart--;
LFA_Seek( file, childStart, SEEK_SET );
LFA_WriteUns8( file, 0 );
}
// then contents
childStart-= cur->newSize;
LFA_Seek( file, childStart, SEEK_SET );
switch ( cur->chunkType )
{
case chunk_GENERAL: //COULDDO enfore no change, since not write-out-able
if ( cur->oldPos != childStart )
LFA_Move( file, cur->oldPos, file, childStart, cur->oldSize, 0, 0 );
break;
default:
cur->write( handler, file, false );
break;
} // switch
} // for
XMP_Enforce ( chunkStart + 12 == childStart);
LFA_Seek( file, chunkStart, SEEK_SET, &ok );
LFA_WriteUns32_LE( file, this->id );
LFA_WriteUns32_LE( file, (XMP_Uns32) this->newSize - 8 ); // validated in changesAndSize() above
LFA_WriteUns32_LE( file, this->containerType );
}
void ContainerChunk::release()
{
// free subchunks
Chunk* curChunk;
while( ! this->children.empty() )
{
curChunk = this->children.back();
delete curChunk;
this->children.pop_back();
}
}
ContainerChunk::~ContainerChunk()
{
this->release(); // free resources
}
// XMP CHUNK ///////////////////////////////////////////////
// a) create
// a) creation
XMPChunk::XMPChunk( ContainerChunk* parent ) : Chunk( parent, chunk_XMP , kChunk_XMP )
{
// nothing
}
// b) parse
XMPChunk::XMPChunk( ContainerChunk* parent, RIFF_MetaHandler* handler ) : Chunk( parent, handler, false, chunk_XMP )
{
chunkType = chunk_XMP;
LFA_FileRef file = handler->parent->fileRef;
XMP_Uns8 level = handler->level;
handler->packetInfo.offset = this->oldPos + 8;
handler->packetInfo.length = (XMP_Int32) this->oldSize - 8;
handler->xmpPacket.reserve ( handler->packetInfo.length );
handler->xmpPacket.assign ( handler->packetInfo.length, ' ' );
LFA_Read ( file, (void*)handler->xmpPacket.data(), handler->packetInfo.length, kLFA_RequireAll );
handler->containsXMP = true; // last, after all possible failure
// pointer for later processing
handler->xmpChunk = this;
}
void XMPChunk::changesAndSize( RIFF_MetaHandler* handler )
{
XMP_Enforce( &handler->xmpPacket != 0 );
XMP_Enforce( handler->xmpPacket.size() > 0 );
this->newSize = 8 + handler->xmpPacket.size();
XMP_Validate( this->newSize <= 0xFFFFFFFFLL, "no single chunk may be above 4 GB", kXMPErr_InternalFailure );
// a complete no-change would have been caught in XMPFiles common code anyway
this->hasChange = true;
}
void XMPChunk::write( RIFF_MetaHandler* handler, LFA_FileRef file, bool isMainChunk )
{
LFA_WriteUns32_LE( file, kChunk_XMP );
LFA_WriteUns32_LE( file, (XMP_Uns32) this->newSize - 8 ); // validated in changesAndSize() above
LFA_Write( file, handler->xmpPacket.data(), (XMP_Int32)handler->xmpPacket.size() );
}
// Value CHUNK ///////////////////////////////////////////////
// a) creation
ValueChunk::ValueChunk( ContainerChunk* parent, std::string value, XMP_Uns32 id ) : Chunk( parent, chunk_VALUE, id )
{
this->oldValue = std::string();
this->SetValue( value );
}
// b) parsing
ValueChunk::ValueChunk( ContainerChunk* parent, RIFF_MetaHandler* handler ) : Chunk( parent, handler, false, chunk_VALUE )
{
// set value: -----------------
LFA_FileRef file = handler->parent->fileRef;
XMP_Uns8 level = handler->level;
// unless changed through reconciliation, assume for now.
// IMPORTANT to stay true to the original (no \0 cleanup or similar)
// since unknown value chunks might not be fully understood,
// hence must be precisely preserved !!!
XMP_Int32 length = (XMP_Int32) this->oldSize - 8;
this->oldValue.reserve( length );
this->oldValue.assign( length + 1, '\0' );
LFA_Read ( file, (void*)this->oldValue.data(), length, kLFA_RequireAll );
this->newValue = this->oldValue;
this->newSize = this->oldSize;
}
void ValueChunk::SetValue( std::string value, bool optionalNUL /* = false */ )
{
this->newValue.assign( value );
if ( (! optionalNUL) || ((value.size() & 1) == 1) ) {
// ! The NUL should be optional in WAV to avoid a parsing bug in Audition 3 - can't handle implicit pad byte.
this->newValue.append( 1, '\0' ); // append zero termination as explicit part of string
}
this->newSize = this->newValue.size() + 8;
}
void ValueChunk::changesAndSize( RIFF_MetaHandler* handler )
{
// Don't simply assign to this->hasChange, it might already be true.
if ( this->newValue.size() != this->oldValue.size() ) {
this->hasChange = true;
} else if ( strncmp ( this->oldValue.c_str(), this->newValue.c_str(), this->newValue.size() ) != 0 ) {
this->hasChange = true;
}
}
void ValueChunk::write( RIFF_MetaHandler* handler, LFA_FileRef file, bool isMainChunk )
{
LFA_WriteUns32_LE( file, this->id );
LFA_WriteUns32_LE( file, (XMP_Uns32)this->newSize - 8 );
LFA_Write( file, this->newValue.data() , (XMP_Int32)this->newSize - 8 );
}
/* remove value chunk if existing.
return true if it was existing. */
bool ContainerChunk::removeValue( XMP_Uns32 id )
{
valueMap* cm = &this->childmap;
valueMapIter iter = cm->find( id );
if( iter == cm->end() )
return false; //not found
ValueChunk* propChunk = iter->second;
// remove from vector (difficult)
chunkVect* cv = &this->children;
chunkVectIter cvIter;
for (cvIter = cv->begin(); cvIter != cv->end(); ++cvIter )
{
if ( (*cvIter)->id == id )
break; // found!
}
XMP_Validate( cvIter != cv->end(), "property not found in children vector", kXMPErr_InternalFailure );
cv->erase( cvIter );
// remove from map (easy)
cm->erase( iter );
delete propChunk;
return true; // found and removed
}
/* returns iterator to (first) occurence of this chunk.
iterator to the end of the map if chunk pointer is not found */
chunkVectIter ContainerChunk::getChild( Chunk* needle )
{
chunkVectIter iter;
for( iter = this->children.begin(); iter != this->children.end(); iter++ )
{
Chunk* temp1 = *iter;
Chunk* temp2 = needle;
if ( (*iter) == needle ) return iter;
}
return this->children.end();
}
/* replaces a chunk by a JUNK chunk.
Also frees memory of prior chunk. */
void ContainerChunk::replaceChildWithJunk( Chunk* child, bool deleteChild )
{
chunkVectIter iter = getChild( child );
if ( iter == this->children.end() ) {
throw new XMP_Error(kXMPErr_InternalFailure, "replaceChildWithJunk: childChunk not found.");
}
*iter = new JunkChunk ( NULL, child->oldSize );
if ( deleteChild ) delete child;
this->hasChange = true;
}
// JunkChunk ///////////////////////////////////////////////////
// a) creation
JunkChunk::JunkChunk( ContainerChunk* parent, XMP_Int64 size ) : Chunk( parent, chunk_JUNK, kChunk_JUNK )
{
XMP_Assert( size >= 8 );
this->oldSize = size;
this->newSize = size;
this->hasChange = true;
}
// b) parsing
JunkChunk::JunkChunk( ContainerChunk* parent, RIFF_MetaHandler* handler ) : Chunk( parent, handler, true, chunk_JUNK )
{
chunkType = chunk_JUNK;
}
void JunkChunk::changesAndSize( RIFF_MetaHandler* handler )
{
this->newSize = this->oldSize; // optimization at a later stage
XMP_Validate( this->newSize <= 0xFFFFFFFFLL, "no single chunk may be above 4 GB", kXMPErr_InternalFailure );
if ( this->id == kChunk_JUNQ ) this->hasChange = true; // Force ID change to JUNK.
}
// zeroBuffer, etc to write out empty native padding
const static XMP_Uns32 kZeroBufferSize64K = 64 * 1024;
static XMP_Uns8 kZeroes64K [ kZeroBufferSize64K ]; // C semantics guarantee zero initialization.
void JunkChunk::write( RIFF_MetaHandler* handler, LFA_FileRef file, bool isMainChunk )
{
LFA_WriteUns32_LE( file, kChunk_JUNK ); // write JUNK, never JUNQ
XMP_Enforce( this->newSize < 0xFFFFFFFF );
XMP_Enforce( this->newSize >= 8 ); // minimum size of any chunk
XMP_Uns32 innerSize = (XMP_Uns32)this->newSize - 8;
LFA_WriteUns32_LE( file, innerSize );
// write out in 64K chunks
while ( innerSize > kZeroBufferSize64K )
{
LFA_Write( file, kZeroes64K , kZeroBufferSize64K );
innerSize -= kZeroBufferSize64K;
}
LFA_Write( file, kZeroes64K , innerSize );
}
} // namespace RIFF
|