gecko-dev/image/decoders/nsICODecoder.cpp

752 строки
25 KiB
C++
Исходник Обычный вид История

/* vim:set tw=80 expandtab softtabstop=2 ts=2 sw=2: */
2012-05-21 15:12:37 +04:00
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
/* This is a Cross-Platform ICO Decoder, which should work everywhere, including
* Big-Endian machines like the PowerPC. */
#include <stdlib.h>
#include "mozilla/Endian.h"
#include "mozilla/Move.h"
#include "nsICODecoder.h"
#include "RasterImage.h"
using namespace mozilla::gfx;
namespace mozilla {
namespace image {
// Constants.
static const uint32_t ICOHEADERSIZE = 6;
static const uint32_t BITMAPINFOSIZE = 40;
2001-11-03 12:50:31 +03:00
// ----------------------------------------
// Actual Data Processing
// ----------------------------------------
uint32_t
nsICODecoder::CalcAlphaRowSize()
{
// Calculate rowsize in DWORD's and then return in # of bytes
uint32_t rowSize = (GetRealWidth() + 31) / 32; // + 31 to round up
return rowSize * 4; // Return rowSize in bytes
}
// Obtains the number of colors from the bits per pixel
uint16_t
nsICODecoder::GetNumColors()
{
uint16_t numColors = 0;
if (mBPP <= 8) {
switch (mBPP) {
case 1:
numColors = 2;
break;
case 4:
numColors = 16;
break;
case 8:
numColors = 256;
break;
default:
numColors = (uint16_t)-1;
}
}
return numColors;
}
nsICODecoder::nsICODecoder(RasterImage* aImage)
: Decoder(aImage)
, mLexer(Transition::To(ICOState::HEADER, ICOHEADERSIZE))
, mBiggestResourceColorDepth(0)
, mBestResourceDelta(INT_MIN)
, mBestResourceColorDepth(0)
, mNumIcons(0)
, mCurrIcon(0)
, mBPP(0)
, mMaskRowSize(0)
, mCurrMaskLine(0)
, mIsCursor(false)
, mHasMaskAlpha(false)
{ }
void
nsICODecoder::FinishInternal()
{
// We shouldn't be called in error cases
MOZ_ASSERT(!HasError(), "Shouldn't call FinishInternal after error!");
GetFinalStateFromContainedDecoder();
}
void
nsICODecoder::FinishWithErrorInternal()
{
GetFinalStateFromContainedDecoder();
}
void
nsICODecoder::GetFinalStateFromContainedDecoder()
{
if (!mContainedDecoder) {
return;
}
// Finish the internally used decoder.
mContainedDecoder->CompleteDecode();
mDecodeDone = mContainedDecoder->GetDecodeDone();
mDataError = mDataError || mContainedDecoder->HasDataError();
mFailCode = NS_SUCCEEDED(mFailCode) ? mContainedDecoder->GetDecoderError()
: mFailCode;
mDecodeAborted = mContainedDecoder->WasAborted();
mProgress |= mContainedDecoder->TakeProgress();
mInvalidRect.UnionRect(mInvalidRect, mContainedDecoder->TakeInvalidRect());
mCurrentFrame = mContainedDecoder->GetCurrentFrameRef();
MOZ_ASSERT(HasError() || !mCurrentFrame || mCurrentFrame->IsImageComplete());
}
// Returns a buffer filled with the bitmap file header in little endian:
// Signature 2 bytes 'BM'
// FileSize 4 bytes File size in bytes
// reserved 4 bytes unused (=0)
// DataOffset 4 bytes File offset to Raster Data
// Returns true if successful
bool
nsICODecoder::FillBitmapFileHeaderBuffer(int8_t* bfh)
{
memset(bfh, 0, 14);
bfh[0] = 'B';
bfh[1] = 'M';
int32_t dataOffset = 0;
int32_t fileSize = 0;
Bug 1204394 (part 1) - Using StreamingLexer in the BMP decoder. r=seth. This patch is a major overhaul of nsBMPDecoder. The patch improves the code in the following ways. - It converts nsBMPDecoder to use StreamingLexer, which makes it much easier to read. - It adds a detailed comment about the BMP format at the top of nsBMPDecoder.cpp. - It fixes lots of inconsistent indenting. - It moves |bihsize| from |mBFH| to |mBIH| to match the file format and common sense. The avoids the need for the confusing LENGTH/INTERNAL_LENGTH distinction. - It renames most of the types in BMPFileHeader.h, so they have better names, in StudlyCaps form, and within the new |bmp| namespace. - It removes the BMP_HEADER_LENGTH struct and inlines its values directly into the two places they were used. - It removes the MOZ_LOG logging done on some of the failure cases. (Most failure cases lacked logging so why bother with some?) - It removes over 200 lines of code, despite the addition of the big format comment. The patch changes the way BMPs are decoded as follows. - It adds stricter testing of the InfoHeader length, rejecting files with bad values. - It moves all header sanity checking that can lead to file rejection into the metadata decode phase. (Previously, bpp/compression consistency checking did not occur during a metadata decode.) - It removes BMPINFOHEADER::ALPHABITFIELDS, which was (a) a weird WinCE-only thing, and (b) we didn't actually allow it, and (c) we used the value 4 instead of 6(!). - It rejects the previously-accepted compression==RLE4 && bpp=1 combination because it doesn't make sense. - It removes a fudge in RLE absolute mode handling that permitted one pixel too many in a row but only if the row's width was odd(!) - It now rejects a file with a negative gap between the color table and the pixel data. The patch leaves the following problems unaddressed. - If bpp==32 we totally ignore compression==BITFIELDS and treat it like compression=RGB. - Transparency as specified in WinBMPv{4,5} isn't handled at all. These will be fixed in follow-ups. All these changes affect (for the better) the results of the following tests that will be added in part 2: - g/pal8v4.bmp - g/pal8v5.bmp - q/pal8os2sp.bmp - q/pal8os2v2.bmp - q/pal8os2v2-16.bmp - b/badheadersize.bmp - b/badpalettesize.bmp - b/badrle.bmp --HG-- extra : rebase_source : 8ddc2f5fccce6998348097ff9f0a1072d273cdf4
2015-10-09 08:47:56 +03:00
dataOffset = bmp::FileHeader::LENGTH + BITMAPINFOSIZE;
// The color table is present only if BPP is <= 8
if (mDirEntry.mBitCount <= 8) {
uint16_t numColors = GetNumColors();
if (numColors == (uint16_t)-1) {
return false;
}
dataOffset += 4 * numColors;
fileSize = dataOffset + GetRealWidth() * GetRealHeight();
} else {
fileSize = dataOffset + (mDirEntry.mBitCount * GetRealWidth() *
GetRealHeight()) / 8;
}
NativeEndian::swapToLittleEndianInPlace(&fileSize, 1);
memcpy(bfh + 2, &fileSize, sizeof(fileSize));
NativeEndian::swapToLittleEndianInPlace(&dataOffset, 1);
memcpy(bfh + 10, &dataOffset, sizeof(dataOffset));
return true;
}
// A BMP inside of an ICO has *2 height because of the AND mask
// that follows the actual bitmap. The BMP shouldn't know about
// this difference though.
bool
nsICODecoder::FixBitmapHeight(int8_t* bih)
{
// Get the height from the BMP file information header
int32_t height;
memcpy(&height, bih + 8, sizeof(height));
NativeEndian::swapFromLittleEndianInPlace(&height, 1);
2012-07-21 21:57:35 +04:00
// BMPs can be stored inverted by having a negative height
height = abs(height);
// The bitmap height is by definition * 2 what it should be to account for
// the 'AND mask'. It is * 2 even if the `AND mask` is not present.
height /= 2;
if (height > 256) {
return false;
}
// We should always trust the height from the bitmap itself instead of
// the ICO height. So fix the ICO height.
if (height == 256) {
mDirEntry.mHeight = 0;
} else {
mDirEntry.mHeight = (int8_t)height;
}
// Fix the BMP height in the BIH so that the BMP decoder can work properly
NativeEndian::swapToLittleEndianInPlace(&height, 1);
memcpy(bih + 8, &height, sizeof(height));
return true;
}
// We should always trust the contained resource for the width
// information over our own information.
bool
nsICODecoder::FixBitmapWidth(int8_t* bih)
{
// Get the width from the BMP file information header
int32_t width;
memcpy(&width, bih + 4, sizeof(width));
NativeEndian::swapFromLittleEndianInPlace(&width, 1);
if (width > 256) {
return false;
}
// We should always trust the width from the bitmap itself instead of
// the ICO width.
if (width == 256) {
mDirEntry.mWidth = 0;
} else {
mDirEntry.mWidth = (int8_t)width;
}
return true;
}
// The BMP information header's bits per pixel should be trusted
// more than what we have. Usually the ICO's BPP is set to 0.
int32_t
nsICODecoder::ReadBPP(const char* aBIH)
{
const int8_t* bih = reinterpret_cast<const int8_t*>(aBIH);
int32_t bitsPerPixel;
memcpy(&bitsPerPixel, bih + 14, sizeof(bitsPerPixel));
NativeEndian::swapFromLittleEndianInPlace(&bitsPerPixel, 1);
return bitsPerPixel;
}
int32_t
nsICODecoder::ReadBIHSize(const char* aBIH)
{
const int8_t* bih = reinterpret_cast<const int8_t*>(aBIH);
int32_t headerSize;
memcpy(&headerSize, bih, sizeof(headerSize));
NativeEndian::swapFromLittleEndianInPlace(&headerSize, 1);
return headerSize;
}
LexerTransition<ICOState>
nsICODecoder::ReadHeader(const char* aData)
{
// If the third byte is 1, this is an icon. If 2, a cursor.
if ((aData[2] != 1) && (aData[2] != 2)) {
return Transition::Terminate(ICOState::FAILURE);
}
mIsCursor = (aData[2] == 2);
// The fifth and sixth bytes specify the number of resources in the file.
mNumIcons =
LittleEndian::readUint16(reinterpret_cast<const uint16_t*>(aData + 4));
if (mNumIcons == 0) {
return Transition::Terminate(ICOState::SUCCESS); // Nothing to do.
}
// Downscale-during-decode can end up decoding different resources in the ICO
// file depending on the target size. Since the resources are not necessarily
// scaled versions of the same image, some may be transparent and some may not
// be. We could be precise about transparency if we decoded the metadata of
// every resource, but for now we don't and it's safest to assume that
// transparency could be present.
PostHasTransparency();
return Transition::To(ICOState::DIR_ENTRY, ICODIRENTRYSIZE);
}
size_t
nsICODecoder::FirstResourceOffset() const
{
MOZ_ASSERT(mNumIcons > 0,
"Calling FirstResourceOffset before processing header");
// The first resource starts right after the directory, which starts right
// after the ICO header.
return ICOHEADERSIZE + mNumIcons * ICODIRENTRYSIZE;
}
LexerTransition<ICOState>
nsICODecoder::ReadDirEntry(const char* aData)
{
mCurrIcon++;
// Read the directory entry.
IconDirEntry e;
memset(&e, 0, sizeof(e));
memcpy(&e.mWidth, aData, sizeof(e.mWidth));
memcpy(&e.mHeight, aData + 1, sizeof(e.mHeight));
memcpy(&e.mColorCount, aData + 2, sizeof(e.mColorCount));
memcpy(&e.mReserved, aData + 3, sizeof(e.mReserved));
memcpy(&e.mPlanes, aData + 4, sizeof(e.mPlanes));
e.mPlanes = LittleEndian::readUint16(&e.mPlanes);
memcpy(&e.mBitCount, aData + 6, sizeof(e.mBitCount));
e.mBitCount = LittleEndian::readUint16(&e.mBitCount);
memcpy(&e.mBytesInRes, aData + 8, sizeof(e.mBytesInRes));
e.mBytesInRes = LittleEndian::readUint32(&e.mBytesInRes);
memcpy(&e.mImageOffset, aData + 12, sizeof(e.mImageOffset));
e.mImageOffset = LittleEndian::readUint32(&e.mImageOffset);
// Determine if this is the biggest resource we've seen so far. We always use
// the biggest resource for the intrinsic size, and if we're not downscaling,
// we select it as the best resource as well.
IntSize entrySize(GetRealWidth(e), GetRealHeight(e));
if (e.mBitCount >= mBiggestResourceColorDepth &&
entrySize.width * entrySize.height >=
mBiggestResourceSize.width * mBiggestResourceSize.height) {
mBiggestResourceSize = entrySize;
mBiggestResourceColorDepth = e.mBitCount;
mBiggestResourceHotSpot = IntSize(e.mXHotspot, e.mYHotspot);
if (!mDownscaler) {
mDirEntry = e;
}
}
if (mDownscaler) {
// Calculate the delta between this resource's size and the desired size, so
// we can see if it is better than our current-best option. In the case of
// several equally-good resources, we use the last one. "Better" in this
// case is determined by |delta|, a measure of the difference in size
// between the entry we've found and the downscaler's target size. We will
// choose the smallest resource that is >= the target size (i.e. we assume
// it's better to downscale a larger icon than to upscale a smaller one).
IntSize desiredSize = mDownscaler->TargetSize();
int32_t delta = entrySize.width - desiredSize.width +
entrySize.height - desiredSize.height;
if (e.mBitCount >= mBestResourceColorDepth &&
((mBestResourceDelta < 0 && delta >= mBestResourceDelta) ||
(delta >= 0 && delta <= mBestResourceDelta))) {
mBestResourceDelta = delta;
mBestResourceColorDepth = e.mBitCount;
mDirEntry = e;
}
}
if (mCurrIcon == mNumIcons) {
// Ensure the resource we selected has an offset past the ICO headers.
if (mDirEntry.mImageOffset < FirstResourceOffset()) {
return Transition::Terminate(ICOState::FAILURE);
}
// If this is a cursor, set the hotspot. We use the hotspot from the biggest
// resource since we also use that resource for the intrinsic size.
if (mIsCursor) {
mImageMetadata.SetHotspot(mBiggestResourceHotSpot.width,
mBiggestResourceHotSpot.height);
}
// We always report the biggest resource's size as the intrinsic size; this
// is necessary for downscale-during-decode to work since we won't even
// attempt to *upscale* while decoding.
PostSize(mBiggestResourceSize.width, mBiggestResourceSize.height);
if (IsMetadataDecode()) {
return Transition::Terminate(ICOState::SUCCESS);
}
// If the resource we selected matches the downscaler's target size
// perfectly, we don't need to do any downscaling.
if (mDownscaler && GetRealSize() == mDownscaler->TargetSize()) {
mDownscaler.reset();
}
size_t offsetToResource = mDirEntry.mImageOffset - FirstResourceOffset();
return Transition::ToUnbuffered(ICOState::FOUND_RESOURCE,
ICOState::SKIP_TO_RESOURCE,
offsetToResource);
}
return Transition::To(ICOState::DIR_ENTRY, ICODIRENTRYSIZE);
}
LexerTransition<ICOState>
nsICODecoder::SniffResource(const char* aData)
{
// We use the first PNGSIGNATURESIZE bytes to determine whether this resource
// is a PNG or a BMP.
bool isPNG = !memcmp(aData, nsPNGDecoder::pngSignatureBytes,
PNGSIGNATURESIZE);
if (isPNG) {
// Create a PNG decoder which will do the rest of the work for us.
mContainedDecoder = new nsPNGDecoder(mImage);
mContainedDecoder->SetMetadataDecode(IsMetadataDecode());
mContainedDecoder->SetDecoderFlags(GetDecoderFlags());
mContainedDecoder->SetSurfaceFlags(GetSurfaceFlags());
if (mDownscaler) {
mContainedDecoder->SetTargetSize(mDownscaler->TargetSize());
}
mContainedDecoder->Init();
if (!WriteToContainedDecoder(aData, PNGSIGNATURESIZE)) {
return Transition::Terminate(ICOState::FAILURE);
}
if (mDirEntry.mBytesInRes <= PNGSIGNATURESIZE) {
return Transition::Terminate(ICOState::FAILURE);
}
// Read in the rest of the PNG unbuffered.
size_t toRead = mDirEntry.mBytesInRes - PNGSIGNATURESIZE;
return Transition::ToUnbuffered(ICOState::FINISHED_RESOURCE,
ICOState::READ_PNG,
toRead);
} else {
// Create a BMP decoder which will do most of the work for us; the exception
// is the AND mask, which isn't present in standalone BMPs.
nsBMPDecoder* bmpDecoder = new nsBMPDecoder(mImage);
mContainedDecoder = bmpDecoder;
Bug 1214072 (part 2) - Implement transparency properly for BMP images. r=seth. Currently we don't implement transparency at all in BMP images except for an odd-duck case of BMPs within ICOs. This patch does the following. - It implements transparency properly for 16bpp and 32bpp images via bitfield masking. (For 32bpp images this also requires handling colors via bitfield masking.) The patch maintains the existing BMP-within-ICO transparency handling. - It also reworks BitFields::Value::Set(). * It now works correctly if the run of 1s goes all the way to bit 31 (the old code didn't set mBitWidth). * If the mask is 0, will give an mRightShift of 0 (old code gave 32, and right-shifting by 32 is dodgy). * It's now easier to read. - It renames transparent.bmp as transparent-if-within-ico.bmp. Ironically enough this file currently uses BITFIELDS compression and is WinBMPv5 format, which means it contains well-specified alpha data. In order to use it to test the hacky BMP-within-ICO transparency scheme the patch changes it to be WinBMPv3 format with RGB compression (i.e. no compression). I left the now-excess bytes (including the bitfields) in the info header in place because that's allowed -- thanks to the start of the pixel data being specified by the |dataoffset| field -- they'll just be ignored. - It tweaks the naming of the relevant gtests and some of their finer details to work with the new way of doing things. This fixes all four remaining failures in bmpsuite. --HG-- rename : image/test/gtest/transparent.bmp => image/test/gtest/transparent-if-within-ico.bmp extra : rebase_source : 2f4838d04bbae4fac00cc69e8d75469105a5de3b
2015-10-14 07:20:10 +03:00
bmpDecoder->SetIsWithinICO();
mContainedDecoder->SetMetadataDecode(IsMetadataDecode());
mContainedDecoder->SetDecoderFlags(GetDecoderFlags());
mContainedDecoder->SetSurfaceFlags(GetSurfaceFlags());
if (mDownscaler) {
mContainedDecoder->SetTargetSize(mDownscaler->TargetSize());
}
mContainedDecoder->Init();
// Make sure we have a sane size for the bitmap information header.
int32_t bihSize = ReadBIHSize(aData);
if (bihSize != static_cast<int32_t>(BITMAPINFOSIZE)) {
return Transition::Terminate(ICOState::FAILURE);
}
// Buffer the first part of the bitmap information header.
memcpy(mBIHraw, aData, PNGSIGNATURESIZE);
// Read in the rest of the bitmap information header.
return Transition::To(ICOState::READ_BIH,
BITMAPINFOSIZE - PNGSIGNATURESIZE);
}
}
LexerTransition<ICOState>
nsICODecoder::ReadPNG(const char* aData, uint32_t aLen)
{
if (!WriteToContainedDecoder(aData, aLen)) {
return Transition::Terminate(ICOState::FAILURE);
}
// Raymond Chen says that 32bpp only are valid PNG ICOs
// http://blogs.msdn.com/b/oldnewthing/archive/2010/10/22/10079192.aspx
if (!static_cast<nsPNGDecoder*>(mContainedDecoder.get())->IsValidICO()) {
return Transition::Terminate(ICOState::FAILURE);
}
return Transition::ContinueUnbuffered(ICOState::READ_PNG);
}
LexerTransition<ICOState>
nsICODecoder::ReadBIH(const char* aData)
{
// Buffer the rest of the bitmap information header.
memcpy(mBIHraw + PNGSIGNATURESIZE, aData, BITMAPINFOSIZE - PNGSIGNATURESIZE);
// Extracting the BPP from the BIH header; it should be trusted over the one
// we have from the ICO header.
mBPP = ReadBPP(mBIHraw);
// The ICO format when containing a BMP does not include the 14 byte
// bitmap file header. To use the code of the BMP decoder we need to
// generate this header ourselves and feed it to the BMP decoder.
int8_t bfhBuffer[BMPFILEHEADERSIZE];
if (!FillBitmapFileHeaderBuffer(bfhBuffer)) {
return Transition::Terminate(ICOState::FAILURE);
}
if (!WriteToContainedDecoder(reinterpret_cast<const char*>(bfhBuffer),
sizeof(bfhBuffer))) {
return Transition::Terminate(ICOState::FAILURE);
}
// Fix the ICO height from the BIH. It needs to be halved so our BMP decoder
// will understand, because the BMP decoder doesn't expect the alpha mask that
// follows the BMP data in an ICO.
if (!FixBitmapHeight(reinterpret_cast<int8_t*>(mBIHraw))) {
return Transition::Terminate(ICOState::FAILURE);
}
// Fix the ICO width from the BIH.
if (!FixBitmapWidth(reinterpret_cast<int8_t*>(mBIHraw))) {
return Transition::Terminate(ICOState::FAILURE);
}
// Write out the BMP's bitmap info header.
if (!WriteToContainedDecoder(mBIHraw, sizeof(mBIHraw))) {
return Transition::Terminate(ICOState::FAILURE);
}
// Sometimes the ICO BPP header field is not filled out so we should trust the
// contained resource over our own information.
// XXX(seth): Is this ever different than the value we obtained from
// ReadBPP() above?
Bug 1207245 - part 6 - rename nsRefPtr<T> to RefPtr<T>; r=ehsan; a=Tomcat The bulk of this commit was generated with a script, executed at the top level of a typical source code checkout. The only non-machine-generated part was modifying MFBT's moz.build to reflect the new naming. CLOSED TREE makes big refactorings like this a piece of cake. # The main substitution. find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \ xargs perl -p -i -e ' s/nsRefPtr\.h/RefPtr\.h/g; # handle includes s/nsRefPtr ?</RefPtr</g; # handle declarations and variables ' # Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h. perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h # Handle nsRefPtr.h itself, a couple places that define constructors # from nsRefPtr, and code generators specially. We do this here, rather # than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename # things like nsRefPtrHashtable. perl -p -i -e 's/nsRefPtr/RefPtr/g' \ mfbt/nsRefPtr.h \ xpcom/glue/nsCOMPtr.h \ xpcom/base/OwningNonNull.h \ ipc/ipdl/ipdl/lower.py \ ipc/ipdl/ipdl/builtin.py \ dom/bindings/Codegen.py \ python/lldbutils/lldbutils/utils.py # In our indiscriminate substitution above, we renamed # nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up. find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \ xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g' if [ -d .git ]; then git mv mfbt/nsRefPtr.h mfbt/RefPtr.h else hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h fi --HG-- rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
2015-10-18 08:24:48 +03:00
RefPtr<nsBMPDecoder> bmpDecoder =
static_cast<nsBMPDecoder*>(mContainedDecoder.get());
mBPP = bmpDecoder->GetBitsPerPixel();
// Check to make sure we have valid color settings.
uint16_t numColors = GetNumColors();
if (numColors == uint16_t(-1)) {
return Transition::Terminate(ICOState::FAILURE);
}
// Do we have an AND mask on this BMP? If so, we need to read it after we read
// the BMP data itself.
uint32_t bmpDataLength = bmpDecoder->GetCompressedImageSize() + 4 * numColors;
bool hasANDMask = (BITMAPINFOSIZE + bmpDataLength) < mDirEntry.mBytesInRes;
ICOState afterBMPState = hasANDMask ? ICOState::PREPARE_FOR_MASK
: ICOState::FINISHED_RESOURCE;
// Read in the rest of the BMP unbuffered.
return Transition::ToUnbuffered(afterBMPState,
ICOState::READ_BMP,
bmpDataLength);
}
LexerTransition<ICOState>
nsICODecoder::ReadBMP(const char* aData, uint32_t aLen)
{
if (!WriteToContainedDecoder(aData, aLen)) {
return Transition::Terminate(ICOState::FAILURE);
}
return Transition::ContinueUnbuffered(ICOState::READ_BMP);
}
LexerTransition<ICOState>
nsICODecoder::PrepareForMask()
{
Bug 1207245 - part 6 - rename nsRefPtr<T> to RefPtr<T>; r=ehsan; a=Tomcat The bulk of this commit was generated with a script, executed at the top level of a typical source code checkout. The only non-machine-generated part was modifying MFBT's moz.build to reflect the new naming. CLOSED TREE makes big refactorings like this a piece of cake. # The main substitution. find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \ xargs perl -p -i -e ' s/nsRefPtr\.h/RefPtr\.h/g; # handle includes s/nsRefPtr ?</RefPtr</g; # handle declarations and variables ' # Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h. perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h # Handle nsRefPtr.h itself, a couple places that define constructors # from nsRefPtr, and code generators specially. We do this here, rather # than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename # things like nsRefPtrHashtable. perl -p -i -e 's/nsRefPtr/RefPtr/g' \ mfbt/nsRefPtr.h \ xpcom/glue/nsCOMPtr.h \ xpcom/base/OwningNonNull.h \ ipc/ipdl/ipdl/lower.py \ ipc/ipdl/ipdl/builtin.py \ dom/bindings/Codegen.py \ python/lldbutils/lldbutils/utils.py # In our indiscriminate substitution above, we renamed # nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up. find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \ xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g' if [ -d .git ]; then git mv mfbt/nsRefPtr.h mfbt/RefPtr.h else hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h fi --HG-- rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
2015-10-18 08:24:48 +03:00
RefPtr<nsBMPDecoder> bmpDecoder =
static_cast<nsBMPDecoder*>(mContainedDecoder.get());
uint16_t numColors = GetNumColors();
MOZ_ASSERT(numColors != uint16_t(-1));
// Determine the length of the AND mask.
uint32_t bmpLengthWithHeader =
BITMAPINFOSIZE + bmpDecoder->GetCompressedImageSize() + 4 * numColors;
MOZ_ASSERT(bmpLengthWithHeader < mDirEntry.mBytesInRes);
uint32_t maskLength = mDirEntry.mBytesInRes - bmpLengthWithHeader;
Bug 1214072 (part 2) - Implement transparency properly for BMP images. r=seth. Currently we don't implement transparency at all in BMP images except for an odd-duck case of BMPs within ICOs. This patch does the following. - It implements transparency properly for 16bpp and 32bpp images via bitfield masking. (For 32bpp images this also requires handling colors via bitfield masking.) The patch maintains the existing BMP-within-ICO transparency handling. - It also reworks BitFields::Value::Set(). * It now works correctly if the run of 1s goes all the way to bit 31 (the old code didn't set mBitWidth). * If the mask is 0, will give an mRightShift of 0 (old code gave 32, and right-shifting by 32 is dodgy). * It's now easier to read. - It renames transparent.bmp as transparent-if-within-ico.bmp. Ironically enough this file currently uses BITFIELDS compression and is WinBMPv5 format, which means it contains well-specified alpha data. In order to use it to test the hacky BMP-within-ICO transparency scheme the patch changes it to be WinBMPv3 format with RGB compression (i.e. no compression). I left the now-excess bytes (including the bitfields) in the info header in place because that's allowed -- thanks to the start of the pixel data being specified by the |dataoffset| field -- they'll just be ignored. - It tweaks the naming of the relevant gtests and some of their finer details to work with the new way of doing things. This fixes all four remaining failures in bmpsuite. --HG-- rename : image/test/gtest/transparent.bmp => image/test/gtest/transparent-if-within-ico.bmp extra : rebase_source : 2f4838d04bbae4fac00cc69e8d75469105a5de3b
2015-10-14 07:20:10 +03:00
// If the BMP provides its own transparency, we ignore the AND mask. We can
// also obviously ignore it if the image has zero width or zero height.
Bug 1214072 (part 2) - Implement transparency properly for BMP images. r=seth. Currently we don't implement transparency at all in BMP images except for an odd-duck case of BMPs within ICOs. This patch does the following. - It implements transparency properly for 16bpp and 32bpp images via bitfield masking. (For 32bpp images this also requires handling colors via bitfield masking.) The patch maintains the existing BMP-within-ICO transparency handling. - It also reworks BitFields::Value::Set(). * It now works correctly if the run of 1s goes all the way to bit 31 (the old code didn't set mBitWidth). * If the mask is 0, will give an mRightShift of 0 (old code gave 32, and right-shifting by 32 is dodgy). * It's now easier to read. - It renames transparent.bmp as transparent-if-within-ico.bmp. Ironically enough this file currently uses BITFIELDS compression and is WinBMPv5 format, which means it contains well-specified alpha data. In order to use it to test the hacky BMP-within-ICO transparency scheme the patch changes it to be WinBMPv3 format with RGB compression (i.e. no compression). I left the now-excess bytes (including the bitfields) in the info header in place because that's allowed -- thanks to the start of the pixel data being specified by the |dataoffset| field -- they'll just be ignored. - It tweaks the naming of the relevant gtests and some of their finer details to work with the new way of doing things. This fixes all four remaining failures in bmpsuite. --HG-- rename : image/test/gtest/transparent.bmp => image/test/gtest/transparent-if-within-ico.bmp extra : rebase_source : 2f4838d04bbae4fac00cc69e8d75469105a5de3b
2015-10-14 07:20:10 +03:00
if (bmpDecoder->HasTransparency() ||
GetRealWidth() == 0 || GetRealHeight() == 0) {
return Transition::ToUnbuffered(ICOState::FINISHED_RESOURCE,
ICOState::SKIP_MASK,
maskLength);
}
// Compute the row size for the mask.
mMaskRowSize = ((GetRealWidth() + 31) / 32) * 4; // + 31 to round up
// If the expected size of the AND mask is larger than its actual size, then
// we must have a truncated (and therefore corrupt) AND mask.
uint32_t expectedLength = mMaskRowSize * GetRealHeight();
if (maskLength < expectedLength) {
return Transition::Terminate(ICOState::FAILURE);
}
// If we're downscaling, the mask is the wrong size for the surface we've
// produced, so we need to downscale the mask into a temporary buffer and then
// combine the mask's alpha values with the color values from the image.
if (mDownscaler) {
MOZ_ASSERT(bmpDecoder->GetImageDataLength() ==
mDownscaler->TargetSize().width *
mDownscaler->TargetSize().height *
sizeof(uint32_t));
mMaskBuffer = MakeUnique<uint8_t[]>(bmpDecoder->GetImageDataLength());
nsresult rv = mDownscaler->BeginFrame(GetRealSize(), Nothing(),
mMaskBuffer.get(),
/* aHasAlpha = */ true,
/* aFlipVertically = */ true);
if (NS_FAILED(rv)) {
return Transition::Terminate(ICOState::FAILURE);
}
}
mCurrMaskLine = GetRealHeight();
return Transition::To(ICOState::READ_MASK_ROW, mMaskRowSize);
}
LexerTransition<ICOState>
nsICODecoder::ReadMaskRow(const char* aData)
{
mCurrMaskLine--;
uint8_t sawTransparency = 0;
// Get the mask row we're reading.
const uint8_t* mask = reinterpret_cast<const uint8_t*>(aData);
const uint8_t* maskRowEnd = mask + mMaskRowSize;
// Get the corresponding row of the mask buffer (if we're downscaling) or the
// decoded image data (if we're not).
uint32_t* decoded = nullptr;
if (mDownscaler) {
// Initialize the row to all white and fully opaque.
memset(mDownscaler->RowBuffer(), 0xFF, GetRealWidth() * sizeof(uint32_t));
decoded = reinterpret_cast<uint32_t*>(mDownscaler->RowBuffer());
} else {
Bug 1207245 - part 6 - rename nsRefPtr<T> to RefPtr<T>; r=ehsan; a=Tomcat The bulk of this commit was generated with a script, executed at the top level of a typical source code checkout. The only non-machine-generated part was modifying MFBT's moz.build to reflect the new naming. CLOSED TREE makes big refactorings like this a piece of cake. # The main substitution. find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \ xargs perl -p -i -e ' s/nsRefPtr\.h/RefPtr\.h/g; # handle includes s/nsRefPtr ?</RefPtr</g; # handle declarations and variables ' # Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h. perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h # Handle nsRefPtr.h itself, a couple places that define constructors # from nsRefPtr, and code generators specially. We do this here, rather # than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename # things like nsRefPtrHashtable. perl -p -i -e 's/nsRefPtr/RefPtr/g' \ mfbt/nsRefPtr.h \ xpcom/glue/nsCOMPtr.h \ xpcom/base/OwningNonNull.h \ ipc/ipdl/ipdl/lower.py \ ipc/ipdl/ipdl/builtin.py \ dom/bindings/Codegen.py \ python/lldbutils/lldbutils/utils.py # In our indiscriminate substitution above, we renamed # nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up. find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \ xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g' if [ -d .git ]; then git mv mfbt/nsRefPtr.h mfbt/RefPtr.h else hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h fi --HG-- rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
2015-10-18 08:24:48 +03:00
RefPtr<nsBMPDecoder> bmpDecoder =
static_cast<nsBMPDecoder*>(mContainedDecoder.get());
uint32_t* imageData = bmpDecoder->GetImageData();
if (!imageData) {
return Transition::Terminate(ICOState::FAILURE);
}
decoded = imageData + mCurrMaskLine * GetRealWidth();
}
MOZ_ASSERT(decoded);
uint32_t* decodedRowEnd = decoded + GetRealWidth();
// Iterate simultaneously through the AND mask and the image data.
while (mask < maskRowEnd) {
uint8_t idx = *mask++;
sawTransparency |= idx;
for (uint8_t bit = 0x80; bit && decoded < decodedRowEnd; bit >>= 1) {
// Clear pixel completely for transparency.
if (idx & bit) {
*decoded = 0;
}
decoded++;
}
}
if (mDownscaler) {
mDownscaler->CommitRow();
}
// If any bits are set in sawTransparency, then we know at least one pixel was
// transparent.
if (sawTransparency) {
mHasMaskAlpha = true;
}
if (mCurrMaskLine == 0) {
return Transition::To(ICOState::FINISH_MASK, 0);
}
return Transition::To(ICOState::READ_MASK_ROW, mMaskRowSize);
}
LexerTransition<ICOState>
nsICODecoder::FinishMask()
{
// If we're downscaling, we now have the appropriate alpha values in
// mMaskBuffer. We just need to transfer them to the image.
if (mDownscaler) {
// Retrieve the image data.
Bug 1207245 - part 6 - rename nsRefPtr<T> to RefPtr<T>; r=ehsan; a=Tomcat The bulk of this commit was generated with a script, executed at the top level of a typical source code checkout. The only non-machine-generated part was modifying MFBT's moz.build to reflect the new naming. CLOSED TREE makes big refactorings like this a piece of cake. # The main substitution. find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \ xargs perl -p -i -e ' s/nsRefPtr\.h/RefPtr\.h/g; # handle includes s/nsRefPtr ?</RefPtr</g; # handle declarations and variables ' # Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h. perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h # Handle nsRefPtr.h itself, a couple places that define constructors # from nsRefPtr, and code generators specially. We do this here, rather # than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename # things like nsRefPtrHashtable. perl -p -i -e 's/nsRefPtr/RefPtr/g' \ mfbt/nsRefPtr.h \ xpcom/glue/nsCOMPtr.h \ xpcom/base/OwningNonNull.h \ ipc/ipdl/ipdl/lower.py \ ipc/ipdl/ipdl/builtin.py \ dom/bindings/Codegen.py \ python/lldbutils/lldbutils/utils.py # In our indiscriminate substitution above, we renamed # nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up. find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \ xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g' if [ -d .git ]; then git mv mfbt/nsRefPtr.h mfbt/RefPtr.h else hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h fi --HG-- rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
2015-10-18 08:24:48 +03:00
RefPtr<nsBMPDecoder> bmpDecoder =
static_cast<nsBMPDecoder*>(mContainedDecoder.get());
uint8_t* imageData = reinterpret_cast<uint8_t*>(bmpDecoder->GetImageData());
if (!imageData) {
return Transition::Terminate(ICOState::FAILURE);
}
// Iterate through the alpha values, copying from mask to image.
MOZ_ASSERT(mMaskBuffer);
MOZ_ASSERT(bmpDecoder->GetImageDataLength() > 0);
for (size_t i = 3 ; i < bmpDecoder->GetImageDataLength() ; i += 4) {
imageData[i] = mMaskBuffer[i];
}
}
// If the mask contained any transparent pixels, record that fact.
if (mHasMaskAlpha) {
PostHasTransparency();
Bug 1207245 - part 6 - rename nsRefPtr<T> to RefPtr<T>; r=ehsan; a=Tomcat The bulk of this commit was generated with a script, executed at the top level of a typical source code checkout. The only non-machine-generated part was modifying MFBT's moz.build to reflect the new naming. CLOSED TREE makes big refactorings like this a piece of cake. # The main substitution. find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \ xargs perl -p -i -e ' s/nsRefPtr\.h/RefPtr\.h/g; # handle includes s/nsRefPtr ?</RefPtr</g; # handle declarations and variables ' # Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h. perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h # Handle nsRefPtr.h itself, a couple places that define constructors # from nsRefPtr, and code generators specially. We do this here, rather # than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename # things like nsRefPtrHashtable. perl -p -i -e 's/nsRefPtr/RefPtr/g' \ mfbt/nsRefPtr.h \ xpcom/glue/nsCOMPtr.h \ xpcom/base/OwningNonNull.h \ ipc/ipdl/ipdl/lower.py \ ipc/ipdl/ipdl/builtin.py \ dom/bindings/Codegen.py \ python/lldbutils/lldbutils/utils.py # In our indiscriminate substitution above, we renamed # nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up. find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \ xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g' if [ -d .git ]; then git mv mfbt/nsRefPtr.h mfbt/RefPtr.h else hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h fi --HG-- rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
2015-10-18 08:24:48 +03:00
RefPtr<nsBMPDecoder> bmpDecoder =
static_cast<nsBMPDecoder*>(mContainedDecoder.get());
Bug 1214072 (part 2) - Implement transparency properly for BMP images. r=seth. Currently we don't implement transparency at all in BMP images except for an odd-duck case of BMPs within ICOs. This patch does the following. - It implements transparency properly for 16bpp and 32bpp images via bitfield masking. (For 32bpp images this also requires handling colors via bitfield masking.) The patch maintains the existing BMP-within-ICO transparency handling. - It also reworks BitFields::Value::Set(). * It now works correctly if the run of 1s goes all the way to bit 31 (the old code didn't set mBitWidth). * If the mask is 0, will give an mRightShift of 0 (old code gave 32, and right-shifting by 32 is dodgy). * It's now easier to read. - It renames transparent.bmp as transparent-if-within-ico.bmp. Ironically enough this file currently uses BITFIELDS compression and is WinBMPv5 format, which means it contains well-specified alpha data. In order to use it to test the hacky BMP-within-ICO transparency scheme the patch changes it to be WinBMPv3 format with RGB compression (i.e. no compression). I left the now-excess bytes (including the bitfields) in the info header in place because that's allowed -- thanks to the start of the pixel data being specified by the |dataoffset| field -- they'll just be ignored. - It tweaks the naming of the relevant gtests and some of their finer details to work with the new way of doing things. This fixes all four remaining failures in bmpsuite. --HG-- rename : image/test/gtest/transparent.bmp => image/test/gtest/transparent-if-within-ico.bmp extra : rebase_source : 2f4838d04bbae4fac00cc69e8d75469105a5de3b
2015-10-14 07:20:10 +03:00
bmpDecoder->SetHasTransparency();
}
return Transition::To(ICOState::FINISHED_RESOURCE, 0);
}
LexerTransition<ICOState>
nsICODecoder::FinishResource()
{
// Make sure the actual size of the resource matches the size in the directory
// entry. If not, we consider the image corrupt.
if (mContainedDecoder->HasSize() &&
mContainedDecoder->GetSize() != GetRealSize()) {
return Transition::Terminate(ICOState::FAILURE);
}
return Transition::Terminate(ICOState::SUCCESS);
}
void
nsICODecoder::WriteInternal(const char* aBuffer, uint32_t aCount)
{
MOZ_ASSERT(!HasError(), "Shouldn't call WriteInternal after error!");
MOZ_ASSERT(aBuffer);
MOZ_ASSERT(aCount > 0);
Maybe<ICOState> terminalState =
mLexer.Lex(aBuffer, aCount,
[=](ICOState aState, const char* aData, size_t aLength) {
switch (aState) {
case ICOState::HEADER:
return ReadHeader(aData);
case ICOState::DIR_ENTRY:
return ReadDirEntry(aData);
case ICOState::SKIP_TO_RESOURCE:
return Transition::ContinueUnbuffered(ICOState::SKIP_TO_RESOURCE);
case ICOState::FOUND_RESOURCE:
return Transition::To(ICOState::SNIFF_RESOURCE, PNGSIGNATURESIZE);
case ICOState::SNIFF_RESOURCE:
return SniffResource(aData);
case ICOState::READ_PNG:
return ReadPNG(aData, aLength);
case ICOState::READ_BIH:
return ReadBIH(aData);
case ICOState::READ_BMP:
return ReadBMP(aData, aLength);
case ICOState::PREPARE_FOR_MASK:
return PrepareForMask();
case ICOState::READ_MASK_ROW:
return ReadMaskRow(aData);
case ICOState::FINISH_MASK:
return FinishMask();
case ICOState::SKIP_MASK:
return Transition::ContinueUnbuffered(ICOState::SKIP_MASK);
case ICOState::FINISHED_RESOURCE:
return FinishResource();
default:
MOZ_ASSERT_UNREACHABLE("Unknown ICOState");
return Transition::Terminate(ICOState::FAILURE);
}
});
if (!terminalState) {
return; // Need more data.
}
if (*terminalState == ICOState::FAILURE) {
PostDataError();
return;
}
MOZ_ASSERT(*terminalState == ICOState::SUCCESS);
}
bool
nsICODecoder::WriteToContainedDecoder(const char* aBuffer, uint32_t aCount)
{
mContainedDecoder->Write(aBuffer, aCount);
mProgress |= mContainedDecoder->TakeProgress();
mInvalidRect.UnionRect(mInvalidRect, mContainedDecoder->TakeInvalidRect());
if (mContainedDecoder->HasDataError()) {
PostDataError();
}
if (mContainedDecoder->HasDecoderError()) {
PostDecoderError(mContainedDecoder->GetDecoderError());
}
return !HasError();
}
} // namespace image
} // namespace mozilla