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
|
/*
* Copyright (c) 2008, 2009, Google Inc. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are
* met:
*
* * Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following disclaimer
* in the documentation and/or other materials provided with the
* distribution.
* * Neither the name of Google Inc. nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "config.h"
#include "BMPImageReader.h"
namespace WebCore {
BMPImageReader::BMPImageReader(ImageDecoder* parent, size_t decodedAndHeaderOffset, size_t imgDataOffset, bool usesAndMask)
: m_parent(parent)
, m_buffer(0)
, m_decodedOffset(decodedAndHeaderOffset)
, m_headerOffset(decodedAndHeaderOffset)
, m_imgDataOffset(imgDataOffset)
, m_isOS21x(false)
, m_isOS22x(false)
, m_isTopDown(false)
, m_needToProcessBitmasks(false)
, m_needToProcessColorTable(false)
, m_tableSizeInBytes(0)
, m_seenNonZeroAlphaPixel(false)
, m_seenZeroAlphaPixel(false)
, m_andMaskState(usesAndMask ? NotYetDecoded : None)
{
// Clue-in decodeBMP() that we need to detect the correct info header size.
memset(&m_infoHeader, 0, sizeof(m_infoHeader));
}
bool BMPImageReader::decodeBMP(bool onlySize)
{
// Calculate size of info header.
if (!m_infoHeader.biSize && !readInfoHeaderSize())
return false;
// Read and process info header.
if ((m_decodedOffset < (m_headerOffset + m_infoHeader.biSize)) && !processInfoHeader())
return false;
// processInfoHeader() set the size, so if that's all we needed, we're done.
if (onlySize)
return true;
// Read and process the bitmasks, if needed.
if (m_needToProcessBitmasks && !processBitmasks())
return false;
// Read and process the color table, if needed.
if (m_needToProcessColorTable && !processColorTable())
return false;
// Initialize the framebuffer if needed.
ASSERT(m_buffer); // Parent should set this before asking us to decode!
if (m_buffer->status() == ImageFrame::FrameEmpty) {
if (!m_buffer->setSize(m_parent->size().width(), m_parent->size().height()))
return m_parent->setFailed(); // Unable to allocate.
m_buffer->setStatus(ImageFrame::FramePartial);
// setSize() calls eraseARGB(), which resets the alpha flag, so we force
// it back to false here. We'll set it true below in all cases where
// these 0s could actually show through.
m_buffer->setHasAlpha(false);
// For BMPs, the frame always fills the entire image.
m_buffer->setOriginalFrameRect(IntRect(IntPoint(), m_parent->size()));
if (!m_isTopDown)
m_coord.setY(m_parent->size().height() - 1);
}
// Decode the data.
if ((m_andMaskState != Decoding) && !pastEndOfImage(0)) {
if ((m_infoHeader.biCompression != RLE4) && (m_infoHeader.biCompression != RLE8) && (m_infoHeader.biCompression != RLE24)) {
const ProcessingResult result = processNonRLEData(false, 0);
if (result != Success)
return (result == Failure) ? m_parent->setFailed() : false;
} else if (!processRLEData())
return false;
}
// If the image has an AND mask and there was no alpha data, process the
// mask.
if ((m_andMaskState == NotYetDecoded) && !m_buffer->hasAlpha()) {
// Reset decoding coordinates to start of image.
m_coord.setX(0);
m_coord.setY(m_isTopDown ? 0 : (m_parent->size().height() - 1));
// The AND mask is stored as 1-bit data.
m_infoHeader.biBitCount = 1;
m_andMaskState = Decoding;
}
if (m_andMaskState == Decoding) {
const ProcessingResult result = processNonRLEData(false, 0);
if (result != Success)
return (result == Failure) ? m_parent->setFailed() : false;
}
// Done!
m_buffer->setStatus(ImageFrame::FrameComplete);
return true;
}
bool BMPImageReader::readInfoHeaderSize()
{
// Get size of info header.
ASSERT(m_decodedOffset == m_headerOffset);
if ((m_decodedOffset > m_data->size()) || ((m_data->size() - m_decodedOffset) < 4))
return false;
m_infoHeader.biSize = readUint32(0);
// Don't increment m_decodedOffset here, it just makes the code in
// processInfoHeader() more confusing.
// Don't allow the header to overflow (which would be harmless here, but
// problematic or at least confusing in other places), or to overrun the
// image data.
if (((m_headerOffset + m_infoHeader.biSize) < m_headerOffset) || (m_imgDataOffset && (m_imgDataOffset < (m_headerOffset + m_infoHeader.biSize))))
return m_parent->setFailed();
// See if this is a header size we understand:
// OS/2 1.x: 12
if (m_infoHeader.biSize == 12)
m_isOS21x = true;
// Windows V3: 40
else if ((m_infoHeader.biSize == 40) || isWindowsV4Plus())
;
// OS/2 2.x: any multiple of 4 between 16 and 64, inclusive, or 42 or 46
else if ((m_infoHeader.biSize >= 16) && (m_infoHeader.biSize <= 64) && (!(m_infoHeader.biSize & 3) || (m_infoHeader.biSize == 42) || (m_infoHeader.biSize == 46)))
m_isOS22x = true;
else
return m_parent->setFailed();
return true;
}
bool BMPImageReader::processInfoHeader()
{
// Read info header.
ASSERT(m_decodedOffset == m_headerOffset);
if ((m_decodedOffset > m_data->size()) || ((m_data->size() - m_decodedOffset) < m_infoHeader.biSize) || !readInfoHeader())
return false;
m_decodedOffset += m_infoHeader.biSize;
// Sanity-check header values.
if (!isInfoHeaderValid())
return m_parent->setFailed();
// Set our size.
if (!m_parent->setSize(m_infoHeader.biWidth, m_infoHeader.biHeight))
return false;
// For paletted images, bitmaps can set biClrUsed to 0 to mean "all
// colors", so set it to the maximum number of colors for this bit depth.
// Also do this for bitmaps that put too large a value here.
if (m_infoHeader.biBitCount < 16) {
const uint32_t maxColors = static_cast<uint32_t>(1) << m_infoHeader.biBitCount;
if (!m_infoHeader.biClrUsed || (m_infoHeader.biClrUsed > maxColors))
m_infoHeader.biClrUsed = maxColors;
}
// For any bitmaps that set their BitCount to the wrong value, reset the
// counts now that we've calculated the number of necessary colors, since
// other code relies on this value being correct.
if (m_infoHeader.biCompression == RLE8)
m_infoHeader.biBitCount = 8;
else if (m_infoHeader.biCompression == RLE4)
m_infoHeader.biBitCount = 4;
// Tell caller what still needs to be processed.
if (m_infoHeader.biBitCount >= 16)
m_needToProcessBitmasks = true;
else if (m_infoHeader.biBitCount)
m_needToProcessColorTable = true;
return true;
}
bool BMPImageReader::readInfoHeader()
{
// Pre-initialize some fields that not all headers set.
m_infoHeader.biCompression = RGB;
m_infoHeader.biClrUsed = 0;
if (m_isOS21x) {
m_infoHeader.biWidth = readUint16(4);
m_infoHeader.biHeight = readUint16(6);
ASSERT(m_andMaskState == None); // ICO is a Windows format, not OS/2!
m_infoHeader.biBitCount = readUint16(10);
return true;
}
m_infoHeader.biWidth = readUint32(4);
m_infoHeader.biHeight = readUint32(8);
if (m_andMaskState != None)
m_infoHeader.biHeight /= 2;
m_infoHeader.biBitCount = readUint16(14);
// Read compression type, if present.
if (m_infoHeader.biSize >= 20) {
uint32_t biCompression = readUint32(16);
// Detect OS/2 2.x-specific compression types.
if ((biCompression == 3) && (m_infoHeader.biBitCount == 1)) {
m_infoHeader.biCompression = HUFFMAN1D;
m_isOS22x = true;
} else if ((biCompression == 4) && (m_infoHeader.biBitCount == 24)) {
m_infoHeader.biCompression = RLE24;
m_isOS22x = true;
} else if (biCompression > 5)
return m_parent->setFailed(); // Some type we don't understand.
else
m_infoHeader.biCompression = static_cast<CompressionType>(biCompression);
}
// Read colors used, if present.
if (m_infoHeader.biSize >= 36)
m_infoHeader.biClrUsed = readUint32(32);
// Windows V4+ can safely read the four bitmasks from 40-56 bytes in, so do
// that here. If the bit depth is less than 16, these values will be
// ignored by the image data decoders. If the bit depth is at least 16 but
// the compression format isn't BITFIELDS, these values will be ignored and
// overwritten* in processBitmasks().
// NOTE: We allow alpha here. Microsoft doesn't really document this well,
// but some BMPs appear to use it.
//
// For non-Windows V4+, m_bitMasks[] et. al will be initialized later
// during processBitmasks().
//
// *Except the alpha channel. Bizarrely, some RGB bitmaps expect decoders
// to pay attention to the alpha mask here, so there's a special case in
// processBitmasks() that doesn't always overwrite that value.
if (isWindowsV4Plus()) {
m_bitMasks[0] = readUint32(40);
m_bitMasks[1] = readUint32(44);
m_bitMasks[2] = readUint32(48);
m_bitMasks[3] = readUint32(52);
}
// Detect top-down BMPs.
if (m_infoHeader.biHeight < 0) {
m_isTopDown = true;
m_infoHeader.biHeight = -m_infoHeader.biHeight;
}
return true;
}
bool BMPImageReader::isInfoHeaderValid() const
{
// Non-positive widths/heights are invalid. (We've already flipped the
// sign of the height for top-down bitmaps.)
if ((m_infoHeader.biWidth <= 0) || !m_infoHeader.biHeight)
return false;
// Only Windows V3+ has top-down bitmaps.
if (m_isTopDown && (m_isOS21x || m_isOS22x))
return false;
// Only bit depths of 1, 4, 8, or 24 are universally supported.
if ((m_infoHeader.biBitCount != 1) && (m_infoHeader.biBitCount != 4) && (m_infoHeader.biBitCount != 8) && (m_infoHeader.biBitCount != 24)) {
// Windows V3+ additionally supports bit depths of 0 (for embedded
// JPEG/PNG images), 16, and 32.
if (m_isOS21x || m_isOS22x || (m_infoHeader.biBitCount && (m_infoHeader.biBitCount != 16) && (m_infoHeader.biBitCount != 32)))
return false;
}
// Each compression type is only valid with certain bit depths (except RGB,
// which can be used with any bit depth). Also, some formats do not
// some compression types.
switch (m_infoHeader.biCompression) {
case RGB:
if (!m_infoHeader.biBitCount)
return false;
break;
case RLE8:
// Supposedly there are undocumented formats like "BitCount = 1,
// Compression = RLE4" (which means "4 bit, but with a 2-color table"),
// so also allow the paletted RLE compression types to have too low a
// bit count; we'll correct this later.
if (!m_infoHeader.biBitCount || (m_infoHeader.biBitCount > 8))
return false;
break;
case RLE4:
// See comments in RLE8.
if (!m_infoHeader.biBitCount || (m_infoHeader.biBitCount > 4))
return false;
break;
case BITFIELDS:
// Only valid for Windows V3+.
if (m_isOS21x || m_isOS22x || ((m_infoHeader.biBitCount != 16) && (m_infoHeader.biBitCount != 32)))
return false;
break;
case JPEG:
case PNG:
// Only valid for Windows V3+.
if (m_isOS21x || m_isOS22x || m_infoHeader.biBitCount)
return false;
break;
case HUFFMAN1D:
// Only valid for OS/2 2.x.
if (!m_isOS22x || (m_infoHeader.biBitCount != 1))
return false;
break;
case RLE24:
// Only valid for OS/2 2.x.
if (!m_isOS22x || (m_infoHeader.biBitCount != 24))
return false;
break;
default:
// Some type we don't understand. This should have been caught in
// readInfoHeader().
ASSERT_NOT_REACHED();
return false;
}
// Top-down bitmaps cannot be compressed; they must be RGB or BITFIELDS.
if (m_isTopDown && (m_infoHeader.biCompression != RGB) && (m_infoHeader.biCompression != BITFIELDS))
return false;
// Reject the following valid bitmap types that we don't currently bother
// decoding. Few other people decode these either, they're unlikely to be
// in much use.
// TODO(pkasting): Consider supporting these someday.
// * Bitmaps larger than 2^16 pixels in either dimension (Windows
// probably doesn't draw these well anyway, and the decoded data would
// take a lot of memory).
if ((m_infoHeader.biWidth >= (1 << 16)) || (m_infoHeader.biHeight >= (1 << 16)))
return false;
// * Windows V3+ JPEG-in-BMP and PNG-in-BMP bitmaps (supposedly not found
// in the wild, only used to send data to printers?).
if ((m_infoHeader.biCompression == JPEG) || (m_infoHeader.biCompression == PNG))
return false;
// * OS/2 2.x Huffman-encoded monochrome bitmaps (see
// http://www.fileformat.info/mirror/egff/ch09_05.htm , re: "G31D"
// algorithm).
if (m_infoHeader.biCompression == HUFFMAN1D)
return false;
return true;
}
bool BMPImageReader::processBitmasks()
{
// Create m_bitMasks[] values.
if (m_infoHeader.biCompression != BITFIELDS) {
// The format doesn't actually use bitmasks. To simplify the decode
// logic later, create bitmasks for the RGB data. For Windows V4+,
// this overwrites the masks we read from the header, which are
// supposed to be ignored in non-BITFIELDS cases.
// 16 bits: MSB <- xRRRRRGG GGGBBBBB -> LSB
// 24/32 bits: MSB <- [AAAAAAAA] RRRRRRRR GGGGGGGG BBBBBBBB -> LSB
const int numBits = (m_infoHeader.biBitCount == 16) ? 5 : 8;
for (int i = 0; i <= 2; ++i)
m_bitMasks[i] = ((static_cast<uint32_t>(1) << (numBits * (3 - i))) - 1) ^ ((static_cast<uint32_t>(1) << (numBits * (2 - i))) - 1);
// For Windows V4+ 32-bit RGB, don't overwrite the alpha mask from the
// header (see note in readInfoHeader()).
if (m_infoHeader.biBitCount < 32)
m_bitMasks[3] = 0;
else if (!isWindowsV4Plus())
m_bitMasks[3] = static_cast<uint32_t>(0xff000000);
} else if (!isWindowsV4Plus()) {
// For Windows V4+ BITFIELDS mode bitmaps, this was already done when
// we read the info header.
// Fail if we don't have enough file space for the bitmasks.
static const size_t SIZEOF_BITMASKS = 12;
if (((m_headerOffset + m_infoHeader.biSize + SIZEOF_BITMASKS) < (m_headerOffset + m_infoHeader.biSize)) || (m_imgDataOffset && (m_imgDataOffset < (m_headerOffset + m_infoHeader.biSize + SIZEOF_BITMASKS))))
return m_parent->setFailed();
// Read bitmasks.
if ((m_data->size() - m_decodedOffset) < SIZEOF_BITMASKS)
return false;
m_bitMasks[0] = readUint32(0);
m_bitMasks[1] = readUint32(4);
m_bitMasks[2] = readUint32(8);
// No alpha in anything other than Windows V4+.
m_bitMasks[3] = 0;
m_decodedOffset += SIZEOF_BITMASKS;
}
// We've now decoded all the non-image data we care about. Skip anything
// else before the actual raster data.
if (m_imgDataOffset)
m_decodedOffset = m_imgDataOffset;
m_needToProcessBitmasks = false;
// Check masks and set shift values.
for (int i = 0; i < 4; ++i) {
// Trim the mask to the allowed bit depth. Some Windows V4+ BMPs
// specify a bogus alpha channel in bits that don't exist in the pixel
// data (for example, bits 25-31 in a 24-bit RGB format).
if (m_infoHeader.biBitCount < 32)
m_bitMasks[i] &= ((static_cast<uint32_t>(1) << m_infoHeader.biBitCount) - 1);
// For empty masks (common on the alpha channel, especially after the
// trimming above), quickly clear the shifts and continue, to avoid an
// infinite loop in the counting code below.
uint32_t tempMask = m_bitMasks[i];
if (!tempMask) {
m_bitShiftsRight[i] = m_bitShiftsLeft[i] = 0;
continue;
}
// Make sure bitmask does not overlap any other bitmasks.
for (int j = 0; j < i; ++j) {
if (tempMask & m_bitMasks[j])
return m_parent->setFailed();
}
// Count offset into pixel data.
for (m_bitShiftsRight[i] = 0; !(tempMask & 1); tempMask >>= 1)
++m_bitShiftsRight[i];
// Count size of mask.
for (m_bitShiftsLeft[i] = 8; tempMask & 1; tempMask >>= 1)
--m_bitShiftsLeft[i];
// Make sure bitmask is contiguous.
if (tempMask)
return m_parent->setFailed();
// Since RGBABuffer tops out at 8 bits per channel, adjust the shift
// amounts to use the most significant 8 bits of the channel.
if (m_bitShiftsLeft[i] < 0) {
m_bitShiftsRight[i] -= m_bitShiftsLeft[i];
m_bitShiftsLeft[i] = 0;
}
}
return true;
}
bool BMPImageReader::processColorTable()
{
m_tableSizeInBytes = m_infoHeader.biClrUsed * (m_isOS21x ? 3 : 4);
// Fail if we don't have enough file space for the color table.
if (((m_headerOffset + m_infoHeader.biSize + m_tableSizeInBytes) < (m_headerOffset + m_infoHeader.biSize)) || (m_imgDataOffset && (m_imgDataOffset < (m_headerOffset + m_infoHeader.biSize + m_tableSizeInBytes))))
return m_parent->setFailed();
// Read color table.
if ((m_decodedOffset > m_data->size()) || ((m_data->size() - m_decodedOffset) < m_tableSizeInBytes))
return false;
m_colorTable.resize(m_infoHeader.biClrUsed);
for (size_t i = 0; i < m_infoHeader.biClrUsed; ++i) {
m_colorTable[i].rgbBlue = m_data->data()[m_decodedOffset++];
m_colorTable[i].rgbGreen = m_data->data()[m_decodedOffset++];
m_colorTable[i].rgbRed = m_data->data()[m_decodedOffset++];
// Skip padding byte (not present on OS/2 1.x).
if (!m_isOS21x)
++m_decodedOffset;
}
// We've now decoded all the non-image data we care about. Skip anything
// else before the actual raster data.
if (m_imgDataOffset)
m_decodedOffset = m_imgDataOffset;
m_needToProcessColorTable = false;
return true;
}
bool BMPImageReader::processRLEData()
{
if (m_decodedOffset > m_data->size())
return false;
// RLE decoding is poorly specified. Two main problems:
// (1) Are EOL markers necessary? What happens when we have too many
// pixels for one row?
// http://www.fileformat.info/format/bmp/egff.htm says extra pixels
// should wrap to the next line. Real BMPs I've encountered seem to
// instead expect extra pixels to be ignored until the EOL marker is
// seen, although this has only happened in a few cases and I suspect
// those BMPs may be invalid. So we only change lines on EOL (or Delta
// with dy > 0), and fail in most cases when pixels extend past the end
// of the line.
// (2) When Delta, EOL, or EOF are seen, what happens to the "skipped"
// pixels?
// http://www.daubnet.com/formats/BMP.html says these should be filled
// with color 0. However, the "do nothing" and "don't care" comments
// of other references suggest leaving these alone, i.e. letting them
// be transparent to the background behind the image. This seems to
// match how MSPAINT treats BMPs, so we do that. Note that when we
// actually skip pixels for a case like this, we need to note on the
// framebuffer that we have alpha.
// Impossible to decode row-at-a-time, so just do things as a stream of
// bytes.
while (true) {
// Every entry takes at least two bytes; bail if there isn't enough
// data.
if ((m_data->size() - m_decodedOffset) < 2)
return false;
// For every entry except EOF, we'd better not have reached the end of
// the image.
const uint8_t count = m_data->data()[m_decodedOffset];
const uint8_t code = m_data->data()[m_decodedOffset + 1];
if ((count || (code != 1)) && pastEndOfImage(0))
return m_parent->setFailed();
// Decode.
if (!count) {
switch (code) {
case 0: // Magic token: EOL
// Skip any remaining pixels in this row.
if (m_coord.x() < m_parent->size().width())
m_buffer->setHasAlpha(true);
moveBufferToNextRow();
m_decodedOffset += 2;
break;
case 1: // Magic token: EOF
// Skip any remaining pixels in the image.
if ((m_coord.x() < m_parent->size().width()) || (m_isTopDown ? (m_coord.y() < (m_parent->size().height() - 1)) : (m_coord.y() > 0)))
m_buffer->setHasAlpha(true);
return true;
case 2: { // Magic token: Delta
// The next two bytes specify dx and dy. Bail if there isn't
// enough data.
if ((m_data->size() - m_decodedOffset) < 4)
return false;
// Fail if this takes us past the end of the desired row or
// past the end of the image.
const uint8_t dx = m_data->data()[m_decodedOffset + 2];
const uint8_t dy = m_data->data()[m_decodedOffset + 3];
if (dx || dy)
m_buffer->setHasAlpha(true);
if (((m_coord.x() + dx) > m_parent->size().width()) || pastEndOfImage(dy))
return m_parent->setFailed();
// Skip intervening pixels.
m_coord.move(dx, m_isTopDown ? dy : -dy);
m_decodedOffset += 4;
break;
}
default: { // Absolute mode
// |code| pixels specified as in BI_RGB, zero-padded at the end
// to a multiple of 16 bits.
// Because processNonRLEData() expects m_decodedOffset to
// point to the beginning of the pixel data, bump it past
// the escape bytes and then reset if decoding failed.
m_decodedOffset += 2;
const ProcessingResult result = processNonRLEData(true, code);
if (result == Failure)
return m_parent->setFailed();
if (result == InsufficientData) {
m_decodedOffset -= 2;
return false;
}
break;
}
}
} else { // Encoded mode
// The following color data is repeated for |count| total pixels.
// Strangely, some BMPs seem to specify excessively large counts
// here; ignore pixels past the end of the row.
const int endX = std::min(m_coord.x() + count, m_parent->size().width());
if (m_infoHeader.biCompression == RLE24) {
// Bail if there isn't enough data.
if ((m_data->size() - m_decodedOffset) < 4)
return false;
// One BGR triple that we copy |count| times.
fillRGBA(endX, m_data->data()[m_decodedOffset + 3], m_data->data()[m_decodedOffset + 2], code, 0xff);
m_decodedOffset += 4;
} else {
// RLE8 has one color index that gets repeated; RLE4 has two
// color indexes in the upper and lower 4 bits of the byte,
// which are alternated.
size_t colorIndexes[2] = {code, code};
if (m_infoHeader.biCompression == RLE4) {
colorIndexes[0] = (colorIndexes[0] >> 4) & 0xf;
colorIndexes[1] &= 0xf;
}
if ((colorIndexes[0] >= m_infoHeader.biClrUsed) || (colorIndexes[1] >= m_infoHeader.biClrUsed))
return m_parent->setFailed();
for (int which = 0; m_coord.x() < endX; ) {
setI(colorIndexes[which]);
which = !which;
}
m_decodedOffset += 2;
}
}
}
}
BMPImageReader::ProcessingResult BMPImageReader::processNonRLEData(bool inRLE, int numPixels)
{
if (m_decodedOffset > m_data->size())
return InsufficientData;
if (!inRLE)
numPixels = m_parent->size().width();
// Fail if we're being asked to decode more pixels than remain in the row.
const int endX = m_coord.x() + numPixels;
if (endX > m_parent->size().width())
return Failure;
// Determine how many bytes of data the requested number of pixels
// requires.
const size_t pixelsPerByte = 8 / m_infoHeader.biBitCount;
const size_t bytesPerPixel = m_infoHeader.biBitCount / 8;
const size_t unpaddedNumBytes = (m_infoHeader.biBitCount < 16) ? ((numPixels + pixelsPerByte - 1) / pixelsPerByte) : (numPixels * bytesPerPixel);
// RLE runs are zero-padded at the end to a multiple of 16 bits. Non-RLE
// data is in rows and is zero-padded to a multiple of 32 bits.
const size_t alignBits = inRLE ? 1 : 3;
const size_t paddedNumBytes = (unpaddedNumBytes + alignBits) & ~alignBits;
// Decode as many rows as we can. (For RLE, where we only want to decode
// one row, we've already checked that this condition is true.)
while (!pastEndOfImage(0)) {
// Bail if we don't have enough data for the desired number of pixels.
if ((m_data->size() - m_decodedOffset) < paddedNumBytes)
return InsufficientData;
if (m_infoHeader.biBitCount < 16) {
// Paletted data. Pixels are stored little-endian within bytes.
// Decode pixels one byte at a time, left to right (so, starting at
// the most significant bits in the byte).
const uint8_t mask = (1 << m_infoHeader.biBitCount) - 1;
for (size_t byte = 0; byte < unpaddedNumBytes; ++byte) {
uint8_t pixelData = m_data->data()[m_decodedOffset + byte];
for (size_t pixel = 0; (pixel < pixelsPerByte) && (m_coord.x() < endX); ++pixel) {
const size_t colorIndex = (pixelData >> (8 - m_infoHeader.biBitCount)) & mask;
if (m_andMaskState == Decoding) {
// There's no way to accurately represent an AND + XOR
// operation as an RGBA image, so where the AND values
// are 1, we simply set the framebuffer pixels to fully
// transparent, on the assumption that most ICOs on the
// web will not be doing a lot of inverting.
if (colorIndex) {
setRGBA(0, 0, 0, 0);
m_buffer->setHasAlpha(true);
} else
m_coord.move(1, 0);
} else {
if (colorIndex >= m_infoHeader.biClrUsed)
return Failure;
setI(colorIndex);
}
pixelData <<= m_infoHeader.biBitCount;
}
}
} else {
// RGB data. Decode pixels one at a time, left to right.
while (m_coord.x() < endX) {
const uint32_t pixel = readCurrentPixel(bytesPerPixel);
// Some BMPs specify an alpha channel but don't actually use it
// (it contains all 0s). To avoid displaying these images as
// fully-transparent, decode as if images are fully opaque
// until we actually see a non-zero alpha value; at that point,
// reset any previously-decoded pixels to fully transparent and
// continue decoding based on the real alpha channel values.
// As an optimization, avoid setting "hasAlpha" to true for
// images where all alpha values are 255; opaque images are
// faster to draw.
int alpha = getAlpha(pixel);
if (!m_seenNonZeroAlphaPixel && !alpha) {
m_seenZeroAlphaPixel = true;
alpha = 255;
} else {
m_seenNonZeroAlphaPixel = true;
if (m_seenZeroAlphaPixel) {
m_buffer->zeroFillPixelData();
m_seenZeroAlphaPixel = false;
} else if (alpha != 255)
m_buffer->setHasAlpha(true);
}
setRGBA(getComponent(pixel, 0), getComponent(pixel, 1),
getComponent(pixel, 2), alpha);
}
}
// Success, keep going.
m_decodedOffset += paddedNumBytes;
if (inRLE)
return Success;
moveBufferToNextRow();
}
// Finished decoding whole image.
return Success;
}
void BMPImageReader::moveBufferToNextRow()
{
m_coord.move(-m_coord.x(), m_isTopDown ? 1 : -1);
}
} // namespace WebCore
|