зеркало из https://github.com/mozilla/gecko-dev.git
1525 строки
52 KiB
C++
1525 строки
52 KiB
C++
/* -*- Mode: C++; tab-width: 20; indent-tabs-mode: nil; c-basic-offset: 2 -*-
|
|
* 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/. */
|
|
|
|
#include "gfxUtils.h"
|
|
|
|
#include "cairo.h"
|
|
#include "gfxContext.h"
|
|
#include "gfxEnv.h"
|
|
#include "gfxImageSurface.h"
|
|
#include "gfxPlatform.h"
|
|
#include "gfxDrawable.h"
|
|
#include "gfxQuad.h"
|
|
#include "imgIEncoder.h"
|
|
#include "mozilla/Base64.h"
|
|
#include "mozilla/dom/ImageEncoder.h"
|
|
#include "mozilla/dom/WorkerPrivate.h"
|
|
#include "mozilla/dom/WorkerRunnable.h"
|
|
#include "mozilla/ipc/CrossProcessSemaphore.h"
|
|
#include "mozilla/gfx/2D.h"
|
|
#include "mozilla/gfx/DataSurfaceHelpers.h"
|
|
#include "mozilla/gfx/Logging.h"
|
|
#include "mozilla/gfx/PathHelpers.h"
|
|
#include "mozilla/gfx/Swizzle.h"
|
|
#include "mozilla/gfx/gfxVars.h"
|
|
#include "mozilla/image/nsBMPEncoder.h"
|
|
#include "mozilla/image/nsICOEncoder.h"
|
|
#include "mozilla/image/nsJPEGEncoder.h"
|
|
#include "mozilla/image/nsPNGEncoder.h"
|
|
#include "mozilla/layers/SynchronousTask.h"
|
|
#include "mozilla/Maybe.h"
|
|
#include "mozilla/RefPtr.h"
|
|
#include "mozilla/StaticPrefs_gfx.h"
|
|
#include "mozilla/StaticPrefs_layout.h"
|
|
#include "mozilla/UniquePtrExtensions.h"
|
|
#include "mozilla/Unused.h"
|
|
#include "mozilla/Vector.h"
|
|
#include "mozilla/webrender/webrender_ffi.h"
|
|
#include "nsAppRunner.h"
|
|
#include "nsComponentManagerUtils.h"
|
|
#include "nsIClipboardHelper.h"
|
|
#include "nsIFile.h"
|
|
#include "nsIGfxInfo.h"
|
|
#include "nsMimeTypes.h"
|
|
#include "nsPresContext.h"
|
|
#include "nsRegion.h"
|
|
#include "nsServiceManagerUtils.h"
|
|
#include "GeckoProfiler.h"
|
|
#include "ImageContainer.h"
|
|
#include "ImageRegion.h"
|
|
#include "gfx2DGlue.h"
|
|
|
|
#ifdef XP_WIN
|
|
# include "gfxWindowsPlatform.h"
|
|
#endif
|
|
|
|
using namespace mozilla;
|
|
using namespace mozilla::image;
|
|
using namespace mozilla::layers;
|
|
using namespace mozilla::gfx;
|
|
|
|
#undef compress
|
|
#include "mozilla/Compression.h"
|
|
|
|
using namespace mozilla::Compression;
|
|
extern "C" {
|
|
|
|
/**
|
|
* Dump a raw image to the default log. This function is exported
|
|
* from libxul, so it can be called from any library in addition to
|
|
* (of course) from a debugger.
|
|
*
|
|
* Note: this helper currently assumes that all 2-bytepp images are
|
|
* r5g6b5, and that all 4-bytepp images are r8g8b8a8.
|
|
*/
|
|
NS_EXPORT
|
|
void mozilla_dump_image(void* bytes, int width, int height, int bytepp,
|
|
int strideBytes) {
|
|
if (0 == strideBytes) {
|
|
strideBytes = width * bytepp;
|
|
}
|
|
SurfaceFormat format;
|
|
// TODO more flexible; parse string?
|
|
switch (bytepp) {
|
|
case 2:
|
|
format = SurfaceFormat::R5G6B5_UINT16;
|
|
break;
|
|
case 4:
|
|
default:
|
|
format = SurfaceFormat::R8G8B8A8;
|
|
break;
|
|
}
|
|
|
|
RefPtr<DataSourceSurface> surf = Factory::CreateWrappingDataSourceSurface(
|
|
(uint8_t*)bytes, strideBytes, IntSize(width, height), format);
|
|
gfxUtils::DumpAsDataURI(surf);
|
|
}
|
|
}
|
|
|
|
static bool MapSrcDest(DataSourceSurface* srcSurf, DataSourceSurface* destSurf,
|
|
DataSourceSurface::MappedSurface* out_srcMap,
|
|
DataSourceSurface::MappedSurface* out_destMap) {
|
|
MOZ_ASSERT(srcSurf && destSurf);
|
|
MOZ_ASSERT(out_srcMap && out_destMap);
|
|
|
|
if (srcSurf->GetSize() != destSurf->GetSize()) {
|
|
MOZ_ASSERT(false, "Width and height must match.");
|
|
return false;
|
|
}
|
|
|
|
if (srcSurf == destSurf) {
|
|
DataSourceSurface::MappedSurface map;
|
|
if (!srcSurf->Map(DataSourceSurface::MapType::READ_WRITE, &map)) {
|
|
NS_WARNING("Couldn't Map srcSurf/destSurf.");
|
|
return false;
|
|
}
|
|
|
|
*out_srcMap = map;
|
|
*out_destMap = map;
|
|
return true;
|
|
}
|
|
|
|
// Map src for reading.
|
|
DataSourceSurface::MappedSurface srcMap;
|
|
if (!srcSurf->Map(DataSourceSurface::MapType::READ, &srcMap)) {
|
|
NS_WARNING("Couldn't Map srcSurf.");
|
|
return false;
|
|
}
|
|
|
|
// Map dest for writing.
|
|
DataSourceSurface::MappedSurface destMap;
|
|
if (!destSurf->Map(DataSourceSurface::MapType::WRITE, &destMap)) {
|
|
NS_WARNING("Couldn't Map aDest.");
|
|
srcSurf->Unmap();
|
|
return false;
|
|
}
|
|
|
|
*out_srcMap = srcMap;
|
|
*out_destMap = destMap;
|
|
return true;
|
|
}
|
|
|
|
static void UnmapSrcDest(DataSourceSurface* srcSurf,
|
|
DataSourceSurface* destSurf) {
|
|
if (srcSurf == destSurf) {
|
|
srcSurf->Unmap();
|
|
} else {
|
|
srcSurf->Unmap();
|
|
destSurf->Unmap();
|
|
}
|
|
}
|
|
|
|
bool gfxUtils::PremultiplyDataSurface(DataSourceSurface* srcSurf,
|
|
DataSourceSurface* destSurf) {
|
|
MOZ_ASSERT(srcSurf && destSurf);
|
|
|
|
DataSourceSurface::MappedSurface srcMap;
|
|
DataSourceSurface::MappedSurface destMap;
|
|
if (!MapSrcDest(srcSurf, destSurf, &srcMap, &destMap)) return false;
|
|
|
|
PremultiplyData(srcMap.mData, srcMap.mStride, srcSurf->GetFormat(),
|
|
destMap.mData, destMap.mStride, destSurf->GetFormat(),
|
|
srcSurf->GetSize());
|
|
|
|
UnmapSrcDest(srcSurf, destSurf);
|
|
return true;
|
|
}
|
|
|
|
bool gfxUtils::UnpremultiplyDataSurface(DataSourceSurface* srcSurf,
|
|
DataSourceSurface* destSurf) {
|
|
MOZ_ASSERT(srcSurf && destSurf);
|
|
|
|
DataSourceSurface::MappedSurface srcMap;
|
|
DataSourceSurface::MappedSurface destMap;
|
|
if (!MapSrcDest(srcSurf, destSurf, &srcMap, &destMap)) return false;
|
|
|
|
UnpremultiplyData(srcMap.mData, srcMap.mStride, srcSurf->GetFormat(),
|
|
destMap.mData, destMap.mStride, destSurf->GetFormat(),
|
|
srcSurf->GetSize());
|
|
|
|
UnmapSrcDest(srcSurf, destSurf);
|
|
return true;
|
|
}
|
|
|
|
static bool MapSrcAndCreateMappedDest(
|
|
DataSourceSurface* srcSurf, RefPtr<DataSourceSurface>* out_destSurf,
|
|
DataSourceSurface::MappedSurface* out_srcMap,
|
|
DataSourceSurface::MappedSurface* out_destMap) {
|
|
MOZ_ASSERT(srcSurf);
|
|
MOZ_ASSERT(out_destSurf && out_srcMap && out_destMap);
|
|
|
|
// Ok, map source for reading.
|
|
DataSourceSurface::MappedSurface srcMap;
|
|
if (!srcSurf->Map(DataSourceSurface::MapType::READ, &srcMap)) {
|
|
MOZ_ASSERT(false, "Couldn't Map srcSurf.");
|
|
return false;
|
|
}
|
|
|
|
// Make our dest surface based on the src.
|
|
RefPtr<DataSourceSurface> destSurf =
|
|
Factory::CreateDataSourceSurfaceWithStride(
|
|
srcSurf->GetSize(), srcSurf->GetFormat(), srcMap.mStride);
|
|
if (NS_WARN_IF(!destSurf)) {
|
|
return false;
|
|
}
|
|
|
|
DataSourceSurface::MappedSurface destMap;
|
|
if (!destSurf->Map(DataSourceSurface::MapType::WRITE, &destMap)) {
|
|
MOZ_ASSERT(false, "Couldn't Map destSurf.");
|
|
srcSurf->Unmap();
|
|
return false;
|
|
}
|
|
|
|
*out_destSurf = destSurf;
|
|
*out_srcMap = srcMap;
|
|
*out_destMap = destMap;
|
|
return true;
|
|
}
|
|
|
|
already_AddRefed<DataSourceSurface> gfxUtils::CreatePremultipliedDataSurface(
|
|
DataSourceSurface* srcSurf) {
|
|
RefPtr<DataSourceSurface> destSurf;
|
|
DataSourceSurface::MappedSurface srcMap;
|
|
DataSourceSurface::MappedSurface destMap;
|
|
if (!MapSrcAndCreateMappedDest(srcSurf, &destSurf, &srcMap, &destMap)) {
|
|
MOZ_ASSERT(false, "MapSrcAndCreateMappedDest failed.");
|
|
RefPtr<DataSourceSurface> surface(srcSurf);
|
|
return surface.forget();
|
|
}
|
|
|
|
PremultiplyData(srcMap.mData, srcMap.mStride, srcSurf->GetFormat(),
|
|
destMap.mData, destMap.mStride, destSurf->GetFormat(),
|
|
srcSurf->GetSize());
|
|
|
|
UnmapSrcDest(srcSurf, destSurf);
|
|
return destSurf.forget();
|
|
}
|
|
|
|
already_AddRefed<DataSourceSurface> gfxUtils::CreateUnpremultipliedDataSurface(
|
|
DataSourceSurface* srcSurf) {
|
|
RefPtr<DataSourceSurface> destSurf;
|
|
DataSourceSurface::MappedSurface srcMap;
|
|
DataSourceSurface::MappedSurface destMap;
|
|
if (!MapSrcAndCreateMappedDest(srcSurf, &destSurf, &srcMap, &destMap)) {
|
|
MOZ_ASSERT(false, "MapSrcAndCreateMappedDest failed.");
|
|
RefPtr<DataSourceSurface> surface(srcSurf);
|
|
return surface.forget();
|
|
}
|
|
|
|
UnpremultiplyData(srcMap.mData, srcMap.mStride, srcSurf->GetFormat(),
|
|
destMap.mData, destMap.mStride, destSurf->GetFormat(),
|
|
srcSurf->GetSize());
|
|
|
|
UnmapSrcDest(srcSurf, destSurf);
|
|
return destSurf.forget();
|
|
}
|
|
|
|
void gfxUtils::ConvertBGRAtoRGBA(uint8_t* aData, uint32_t aLength) {
|
|
MOZ_ASSERT((aLength % 4) == 0, "Loop below will pass srcEnd!");
|
|
SwizzleData(aData, aLength, SurfaceFormat::B8G8R8A8, aData, aLength,
|
|
SurfaceFormat::R8G8B8A8, IntSize(aLength / 4, 1));
|
|
}
|
|
|
|
#if !defined(MOZ_GFX_OPTIMIZE_MOBILE)
|
|
/**
|
|
* This returns the fastest operator to use for solid surfaces which have no
|
|
* alpha channel or their alpha channel is uniformly opaque.
|
|
* This differs per render mode.
|
|
*/
|
|
static CompositionOp OptimalFillOp() {
|
|
# ifdef XP_WIN
|
|
if (gfxWindowsPlatform::GetPlatform()->IsDirect2DBackend()) {
|
|
// D2D -really- hates operator source.
|
|
return CompositionOp::OP_OVER;
|
|
}
|
|
# endif
|
|
return CompositionOp::OP_SOURCE;
|
|
}
|
|
|
|
// EXTEND_PAD won't help us here; we have to create a temporary surface to hold
|
|
// the subimage of pixels we're allowed to sample.
|
|
static already_AddRefed<gfxDrawable> CreateSamplingRestrictedDrawable(
|
|
gfxDrawable* aDrawable, gfxContext* aContext, const ImageRegion& aRegion,
|
|
const SurfaceFormat aFormat, bool aUseOptimalFillOp) {
|
|
AUTO_PROFILER_LABEL("CreateSamplingRestrictedDrawable", GRAPHICS);
|
|
|
|
DrawTarget* destDrawTarget = aContext->GetDrawTarget();
|
|
if (destDrawTarget->GetBackendType() == BackendType::DIRECT2D1_1) {
|
|
return nullptr;
|
|
}
|
|
|
|
gfxRect clipExtents = aContext->GetClipExtents();
|
|
|
|
// Inflate by one pixel because bilinear filtering will sample at most
|
|
// one pixel beyond the computed image pixel coordinate.
|
|
clipExtents.Inflate(1.0);
|
|
|
|
gfxRect needed = aRegion.IntersectAndRestrict(clipExtents);
|
|
needed.RoundOut();
|
|
|
|
// if 'needed' is empty, nothing will be drawn since aFill
|
|
// must be entirely outside the clip region, so it doesn't
|
|
// matter what we do here, but we should avoid trying to
|
|
// create a zero-size surface.
|
|
if (needed.IsEmpty()) return nullptr;
|
|
|
|
IntSize size(int32_t(needed.Width()), int32_t(needed.Height()));
|
|
|
|
RefPtr<DrawTarget> target =
|
|
gfxPlatform::GetPlatform()->CreateOffscreenContentDrawTarget(size,
|
|
aFormat);
|
|
if (!target || !target->IsValid()) {
|
|
return nullptr;
|
|
}
|
|
|
|
RefPtr<gfxContext> tmpCtx = gfxContext::CreateOrNull(target);
|
|
MOZ_ASSERT(tmpCtx); // already checked the target above
|
|
|
|
if (aUseOptimalFillOp) {
|
|
tmpCtx->SetOp(OptimalFillOp());
|
|
}
|
|
aDrawable->Draw(tmpCtx, needed - needed.TopLeft(), ExtendMode::REPEAT,
|
|
SamplingFilter::LINEAR, 1.0,
|
|
gfxMatrix::Translation(needed.TopLeft()));
|
|
RefPtr<SourceSurface> surface = target->Snapshot();
|
|
|
|
RefPtr<gfxDrawable> drawable = new gfxSurfaceDrawable(
|
|
surface, size, gfxMatrix::Translation(-needed.TopLeft()));
|
|
return drawable.forget();
|
|
}
|
|
#endif // !MOZ_GFX_OPTIMIZE_MOBILE
|
|
|
|
/* These heuristics are based on
|
|
* Source/WebCore/platform/graphics/skia/ImageSkia.cpp:computeResamplingMode()
|
|
*/
|
|
#ifdef MOZ_GFX_OPTIMIZE_MOBILE
|
|
static SamplingFilter ReduceResamplingFilter(SamplingFilter aSamplingFilter,
|
|
int aImgWidth, int aImgHeight,
|
|
float aSourceWidth,
|
|
float aSourceHeight) {
|
|
// Images smaller than this in either direction are considered "small" and
|
|
// are not resampled ever (see below).
|
|
const int kSmallImageSizeThreshold = 8;
|
|
|
|
// The amount an image can be stretched in a single direction before we
|
|
// say that it is being stretched so much that it must be a line or
|
|
// background that doesn't need resampling.
|
|
const float kLargeStretch = 3.0f;
|
|
|
|
if (aImgWidth <= kSmallImageSizeThreshold ||
|
|
aImgHeight <= kSmallImageSizeThreshold) {
|
|
// Never resample small images. These are often used for borders and
|
|
// rules (think 1x1 images used to make lines).
|
|
return SamplingFilter::POINT;
|
|
}
|
|
|
|
if (aImgHeight * kLargeStretch <= aSourceHeight ||
|
|
aImgWidth * kLargeStretch <= aSourceWidth) {
|
|
// Large image tiling detected.
|
|
|
|
// Don't resample if it is being tiled a lot in only one direction.
|
|
// This is trying to catch cases where somebody has created a border
|
|
// (which might be large) and then is stretching it to fill some part
|
|
// of the page.
|
|
if (fabs(aSourceWidth - aImgWidth) / aImgWidth < 0.5 ||
|
|
fabs(aSourceHeight - aImgHeight) / aImgHeight < 0.5)
|
|
return SamplingFilter::POINT;
|
|
|
|
// The image is growing a lot and in more than one direction. Resampling
|
|
// is slow and doesn't give us very much when growing a lot.
|
|
return aSamplingFilter;
|
|
}
|
|
|
|
/* Some notes on other heuristics:
|
|
The Skia backend also uses nearest for backgrounds that are stretched by
|
|
a large amount. I'm not sure this is common enough for us to worry about
|
|
now. It also uses nearest for backgrounds/avoids high quality for images
|
|
that are very slightly scaled. I'm also not sure that very slightly
|
|
scaled backgrounds are common enough us to worry about.
|
|
|
|
We don't currently have much support for doing high quality interpolation.
|
|
The only place this currently happens is on Quartz and we don't have as
|
|
much control over it as would be needed. Webkit avoids using high quality
|
|
resampling during load. It also avoids high quality if the transformation
|
|
is not just a scale and translation
|
|
|
|
WebKit bug #40045 added code to avoid resampling different parts
|
|
of an image with different methods by using a resampling hint size.
|
|
It currently looks unused in WebKit but it's something to watch out for.
|
|
*/
|
|
|
|
return aSamplingFilter;
|
|
}
|
|
#else
|
|
static SamplingFilter ReduceResamplingFilter(SamplingFilter aSamplingFilter,
|
|
int aImgWidth, int aImgHeight,
|
|
int aSourceWidth,
|
|
int aSourceHeight) {
|
|
// Just pass the filter through unchanged
|
|
return aSamplingFilter;
|
|
}
|
|
#endif
|
|
|
|
#ifdef MOZ_WIDGET_COCOA
|
|
// Only prescale a temporary surface if we're going to repeat it often.
|
|
// Scaling is expensive on OS X and without prescaling, we'd scale
|
|
// every tile of the repeated rect. However, using a temp surface also
|
|
// potentially uses more memory if the scaled image is large. So only prescale
|
|
// on a temp surface if we know we're going to repeat the image in either the X
|
|
// or Y axis multiple times.
|
|
static bool ShouldUseTempSurface(Rect aImageRect, Rect aNeededRect) {
|
|
int repeatX = aNeededRect.width / aImageRect.width;
|
|
int repeatY = aNeededRect.height / aImageRect.height;
|
|
return (repeatX >= 5) || (repeatY >= 5);
|
|
}
|
|
|
|
static bool PrescaleAndTileDrawable(gfxDrawable* aDrawable,
|
|
gfxContext* aContext,
|
|
const ImageRegion& aRegion, Rect aImageRect,
|
|
const SamplingFilter aSamplingFilter,
|
|
const SurfaceFormat aFormat,
|
|
gfxFloat aOpacity, ExtendMode aExtendMode) {
|
|
Size scaleFactor = aContext->CurrentMatrix().ScaleFactors(true);
|
|
Matrix scaleMatrix = Matrix::Scaling(scaleFactor.width, scaleFactor.height);
|
|
const float fuzzFactor = 0.01;
|
|
|
|
// If we aren't scaling or translating, don't go down this path
|
|
if ((FuzzyEqual(scaleFactor.width, 1.0, fuzzFactor) &&
|
|
FuzzyEqual(scaleFactor.width, 1.0, fuzzFactor)) ||
|
|
aContext->CurrentMatrix().HasNonAxisAlignedTransform()) {
|
|
return false;
|
|
}
|
|
|
|
gfxRect clipExtents = aContext->GetClipExtents();
|
|
|
|
// Inflate by one pixel because bilinear filtering will sample at most
|
|
// one pixel beyond the computed image pixel coordinate.
|
|
clipExtents.Inflate(1.0);
|
|
|
|
gfxRect needed = aRegion.IntersectAndRestrict(clipExtents);
|
|
Rect scaledNeededRect = scaleMatrix.TransformBounds(ToRect(needed));
|
|
scaledNeededRect.RoundOut();
|
|
if (scaledNeededRect.IsEmpty()) {
|
|
return false;
|
|
}
|
|
|
|
Rect scaledImageRect = scaleMatrix.TransformBounds(aImageRect);
|
|
if (!ShouldUseTempSurface(scaledImageRect, scaledNeededRect)) {
|
|
return false;
|
|
}
|
|
|
|
IntSize scaledImageSize((int32_t)scaledImageRect.width,
|
|
(int32_t)scaledImageRect.height);
|
|
if (scaledImageSize.width != scaledImageRect.width ||
|
|
scaledImageSize.height != scaledImageRect.height) {
|
|
// If the scaled image isn't pixel aligned, we'll get artifacts
|
|
// so we have to take the slow path.
|
|
return false;
|
|
}
|
|
|
|
RefPtr<DrawTarget> scaledDT =
|
|
gfxPlatform::GetPlatform()->CreateOffscreenContentDrawTarget(
|
|
scaledImageSize, aFormat);
|
|
if (!scaledDT || !scaledDT->IsValid()) {
|
|
return false;
|
|
}
|
|
|
|
RefPtr<gfxContext> tmpCtx = gfxContext::CreateOrNull(scaledDT);
|
|
MOZ_ASSERT(tmpCtx); // already checked the target above
|
|
|
|
scaledDT->SetTransform(scaleMatrix);
|
|
gfxRect gfxImageRect(aImageRect.x, aImageRect.y, aImageRect.width,
|
|
aImageRect.height);
|
|
|
|
// Since this is just the scaled image, we don't want to repeat anything yet.
|
|
aDrawable->Draw(tmpCtx, gfxImageRect, ExtendMode::CLAMP, aSamplingFilter, 1.0,
|
|
gfxMatrix());
|
|
|
|
RefPtr<SourceSurface> scaledImage = scaledDT->Snapshot();
|
|
|
|
{
|
|
gfxContextMatrixAutoSaveRestore autoSR(aContext);
|
|
Matrix withoutScale = aContext->CurrentMatrix();
|
|
DrawTarget* destDrawTarget = aContext->GetDrawTarget();
|
|
|
|
// The translation still is in scaled units
|
|
withoutScale.PreScale(1.0 / scaleFactor.width, 1.0 / scaleFactor.height);
|
|
aContext->SetMatrix(withoutScale);
|
|
|
|
DrawOptions drawOptions(aOpacity, aContext->CurrentOp(),
|
|
aContext->CurrentAntialiasMode());
|
|
|
|
SurfacePattern scaledImagePattern(scaledImage, aExtendMode, Matrix(),
|
|
aSamplingFilter);
|
|
destDrawTarget->FillRect(scaledNeededRect, scaledImagePattern, drawOptions);
|
|
}
|
|
return true;
|
|
}
|
|
#endif // MOZ_WIDGET_COCOA
|
|
|
|
/* static */
|
|
void gfxUtils::DrawPixelSnapped(gfxContext* aContext, gfxDrawable* aDrawable,
|
|
const gfxSize& aImageSize,
|
|
const ImageRegion& aRegion,
|
|
const SurfaceFormat aFormat,
|
|
SamplingFilter aSamplingFilter,
|
|
uint32_t aImageFlags, gfxFloat aOpacity,
|
|
bool aUseOptimalFillOp) {
|
|
AUTO_PROFILER_LABEL("gfxUtils::DrawPixelSnapped", GRAPHICS);
|
|
|
|
gfxRect imageRect(gfxPoint(0, 0), aImageSize);
|
|
gfxRect region(aRegion.Rect());
|
|
ExtendMode extendMode = aRegion.GetExtendMode();
|
|
|
|
RefPtr<gfxDrawable> drawable = aDrawable;
|
|
|
|
aSamplingFilter = ReduceResamplingFilter(aSamplingFilter, imageRect.Width(),
|
|
imageRect.Height(), region.Width(),
|
|
region.Height());
|
|
|
|
// OK now, the hard part left is to account for the subimage sampling
|
|
// restriction. If all the transforms involved are just integer
|
|
// translations, then we assume no resampling will occur so there's
|
|
// nothing to do.
|
|
// XXX if only we had source-clipping in cairo!
|
|
|
|
if (aContext->CurrentMatrix().HasNonIntegerTranslation()) {
|
|
if ((extendMode != ExtendMode::CLAMP) ||
|
|
!aRegion.RestrictionContains(imageRect)) {
|
|
if (drawable->DrawWithSamplingRect(
|
|
aContext->GetDrawTarget(), aContext->CurrentOp(),
|
|
aContext->CurrentAntialiasMode(), aRegion.Rect(),
|
|
aRegion.Restriction(), extendMode, aSamplingFilter, aOpacity)) {
|
|
return;
|
|
}
|
|
|
|
#ifdef MOZ_WIDGET_COCOA
|
|
if (PrescaleAndTileDrawable(aDrawable, aContext, aRegion,
|
|
ToRect(imageRect), aSamplingFilter, aFormat,
|
|
aOpacity, extendMode)) {
|
|
return;
|
|
}
|
|
#endif
|
|
|
|
// On Mobile, we don't ever want to do this; it has the potential for
|
|
// allocating very large temporary surfaces, especially since we'll
|
|
// do full-page snapshots often (see bug 749426).
|
|
#if !defined(MOZ_GFX_OPTIMIZE_MOBILE)
|
|
RefPtr<gfxDrawable> restrictedDrawable = CreateSamplingRestrictedDrawable(
|
|
aDrawable, aContext, aRegion, aFormat, aUseOptimalFillOp);
|
|
if (restrictedDrawable) {
|
|
drawable.swap(restrictedDrawable);
|
|
|
|
// We no longer need to tile: Either we never needed to, or we already
|
|
// filled a surface with the tiled pattern; this surface can now be
|
|
// drawn without tiling.
|
|
extendMode = ExtendMode::CLAMP;
|
|
}
|
|
#endif
|
|
}
|
|
}
|
|
|
|
drawable->Draw(aContext, aRegion.Rect(), extendMode, aSamplingFilter,
|
|
aOpacity, gfxMatrix());
|
|
}
|
|
|
|
/* static */
|
|
int gfxUtils::ImageFormatToDepth(gfxImageFormat aFormat) {
|
|
switch (aFormat) {
|
|
case SurfaceFormat::A8R8G8B8_UINT32:
|
|
return 32;
|
|
case SurfaceFormat::X8R8G8B8_UINT32:
|
|
return 24;
|
|
case SurfaceFormat::R5G6B5_UINT16:
|
|
return 16;
|
|
default:
|
|
break;
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
/*static*/
|
|
void gfxUtils::ClipToRegion(gfxContext* aContext, const nsIntRegion& aRegion) {
|
|
aContext->NewPath();
|
|
for (auto iter = aRegion.RectIter(); !iter.Done(); iter.Next()) {
|
|
const IntRect& r = iter.Get();
|
|
aContext->Rectangle(gfxRect(r.X(), r.Y(), r.Width(), r.Height()));
|
|
}
|
|
aContext->Clip();
|
|
}
|
|
|
|
/*static*/
|
|
void gfxUtils::ClipToRegion(DrawTarget* aTarget, const nsIntRegion& aRegion) {
|
|
uint32_t numRects = aRegion.GetNumRects();
|
|
// If there is only one rect, then the region bounds are equivalent to the
|
|
// contents. So just use push a single clip rect with the bounds.
|
|
if (numRects == 1) {
|
|
aTarget->PushClipRect(Rect(aRegion.GetBounds()));
|
|
return;
|
|
}
|
|
|
|
// Check if the target's transform will preserve axis-alignment and
|
|
// pixel-alignment for each rect. For now, just handle the common case
|
|
// of integer translations.
|
|
Matrix transform = aTarget->GetTransform();
|
|
if (transform.IsIntegerTranslation()) {
|
|
IntPoint translation = RoundedToInt(transform.GetTranslation());
|
|
AutoTArray<IntRect, 16> rects;
|
|
rects.SetLength(numRects);
|
|
uint32_t i = 0;
|
|
// Build the list of transformed rects by adding in the translation.
|
|
for (auto iter = aRegion.RectIter(); !iter.Done(); iter.Next()) {
|
|
IntRect rect = iter.Get();
|
|
rect.MoveBy(translation);
|
|
rects[i++] = rect;
|
|
}
|
|
aTarget->PushDeviceSpaceClipRects(rects.Elements(), rects.Length());
|
|
} else {
|
|
// The transform does not produce axis-aligned rects or a rect was not
|
|
// pixel-aligned. So just build a path with all the rects and clip to it
|
|
// instead.
|
|
RefPtr<PathBuilder> pathBuilder = aTarget->CreatePathBuilder();
|
|
for (auto iter = aRegion.RectIter(); !iter.Done(); iter.Next()) {
|
|
AppendRectToPath(pathBuilder, Rect(iter.Get()));
|
|
}
|
|
RefPtr<Path> path = pathBuilder->Finish();
|
|
aTarget->PushClip(path);
|
|
}
|
|
}
|
|
|
|
/*static*/
|
|
float gfxUtils::ClampToScaleFactor(float aVal, bool aRoundDown) {
|
|
// Arbitary scale factor limitation. We can increase this
|
|
// for better scaling performance at the cost of worse
|
|
// quality.
|
|
static const float kScaleResolution = 2;
|
|
|
|
// Negative scaling is just a flip and irrelevant to
|
|
// our resolution calculation.
|
|
if (aVal < 0.0) {
|
|
aVal = -aVal;
|
|
}
|
|
|
|
bool inverse = false;
|
|
if (aVal < 1.0) {
|
|
inverse = true;
|
|
aVal = 1 / aVal;
|
|
}
|
|
|
|
float power = logf(aVal) / logf(kScaleResolution);
|
|
|
|
// If power is within 1e-5 of an integer, round to nearest to
|
|
// prevent floating point errors, otherwise round up to the
|
|
// next integer value.
|
|
if (fabs(power - NS_round(power)) < 1e-5) {
|
|
power = NS_round(power);
|
|
// Use floor when we are either inverted or rounding down, but
|
|
// not both.
|
|
} else if (inverse != aRoundDown) {
|
|
power = floor(power);
|
|
// Otherwise, ceil when we are not inverted and not rounding
|
|
// down, or we are inverted and rounding down.
|
|
} else {
|
|
power = ceil(power);
|
|
}
|
|
|
|
float scale = powf(kScaleResolution, power);
|
|
|
|
if (inverse) {
|
|
scale = 1 / scale;
|
|
}
|
|
|
|
return scale;
|
|
}
|
|
|
|
gfxMatrix gfxUtils::TransformRectToRect(const gfxRect& aFrom,
|
|
const gfxPoint& aToTopLeft,
|
|
const gfxPoint& aToTopRight,
|
|
const gfxPoint& aToBottomRight) {
|
|
gfxMatrix m;
|
|
if (aToTopRight.y == aToTopLeft.y && aToTopRight.x == aToBottomRight.x) {
|
|
// Not a rotation, so xy and yx are zero
|
|
m._21 = m._12 = 0.0;
|
|
m._11 = (aToBottomRight.x - aToTopLeft.x) / aFrom.Width();
|
|
m._22 = (aToBottomRight.y - aToTopLeft.y) / aFrom.Height();
|
|
m._31 = aToTopLeft.x - m._11 * aFrom.X();
|
|
m._32 = aToTopLeft.y - m._22 * aFrom.Y();
|
|
} else {
|
|
NS_ASSERTION(
|
|
aToTopRight.y == aToBottomRight.y && aToTopRight.x == aToTopLeft.x,
|
|
"Destination rectangle not axis-aligned");
|
|
m._11 = m._22 = 0.0;
|
|
m._21 = (aToBottomRight.x - aToTopLeft.x) / aFrom.Height();
|
|
m._12 = (aToBottomRight.y - aToTopLeft.y) / aFrom.Width();
|
|
m._31 = aToTopLeft.x - m._21 * aFrom.Y();
|
|
m._32 = aToTopLeft.y - m._12 * aFrom.X();
|
|
}
|
|
return m;
|
|
}
|
|
|
|
Matrix gfxUtils::TransformRectToRect(const gfxRect& aFrom,
|
|
const IntPoint& aToTopLeft,
|
|
const IntPoint& aToTopRight,
|
|
const IntPoint& aToBottomRight) {
|
|
Matrix m;
|
|
if (aToTopRight.y == aToTopLeft.y && aToTopRight.x == aToBottomRight.x) {
|
|
// Not a rotation, so xy and yx are zero
|
|
m._12 = m._21 = 0.0;
|
|
m._11 = (aToBottomRight.x - aToTopLeft.x) / aFrom.Width();
|
|
m._22 = (aToBottomRight.y - aToTopLeft.y) / aFrom.Height();
|
|
m._31 = aToTopLeft.x - m._11 * aFrom.X();
|
|
m._32 = aToTopLeft.y - m._22 * aFrom.Y();
|
|
} else {
|
|
NS_ASSERTION(
|
|
aToTopRight.y == aToBottomRight.y && aToTopRight.x == aToTopLeft.x,
|
|
"Destination rectangle not axis-aligned");
|
|
m._11 = m._22 = 0.0;
|
|
m._21 = (aToBottomRight.x - aToTopLeft.x) / aFrom.Height();
|
|
m._12 = (aToBottomRight.y - aToTopLeft.y) / aFrom.Width();
|
|
m._31 = aToTopLeft.x - m._21 * aFrom.Y();
|
|
m._32 = aToTopLeft.y - m._12 * aFrom.X();
|
|
}
|
|
return m;
|
|
}
|
|
|
|
/* This function is sort of shitty. We truncate doubles
|
|
* to ints then convert those ints back to doubles to make sure that
|
|
* they equal the doubles that we got in. */
|
|
bool gfxUtils::GfxRectToIntRect(const gfxRect& aIn, IntRect* aOut) {
|
|
*aOut = IntRect(int32_t(aIn.X()), int32_t(aIn.Y()), int32_t(aIn.Width()),
|
|
int32_t(aIn.Height()));
|
|
return gfxRect(aOut->X(), aOut->Y(), aOut->Width(), aOut->Height())
|
|
.IsEqualEdges(aIn);
|
|
}
|
|
|
|
/* Clamp r to CAIRO_COORD_MIN .. CAIRO_COORD_MAX
|
|
* these are to be device coordinates.
|
|
*
|
|
* Cairo is currently using 24.8 fixed point,
|
|
* so -2^24 .. 2^24-1 is our valid
|
|
*/
|
|
/*static*/
|
|
void gfxUtils::ConditionRect(gfxRect& aRect) {
|
|
#define CAIRO_COORD_MAX (16777215.0)
|
|
#define CAIRO_COORD_MIN (-16777216.0)
|
|
// if either x or y is way out of bounds;
|
|
// note that we don't handle negative w/h here
|
|
if (aRect.X() > CAIRO_COORD_MAX) {
|
|
aRect.SetRectX(CAIRO_COORD_MAX, 0.0);
|
|
}
|
|
|
|
if (aRect.Y() > CAIRO_COORD_MAX) {
|
|
aRect.SetRectY(CAIRO_COORD_MAX, 0.0);
|
|
}
|
|
|
|
if (aRect.X() < CAIRO_COORD_MIN) {
|
|
aRect.SetWidth(aRect.XMost() - CAIRO_COORD_MIN);
|
|
if (aRect.Width() < 0.0) {
|
|
aRect.SetWidth(0.0);
|
|
}
|
|
aRect.MoveToX(CAIRO_COORD_MIN);
|
|
}
|
|
|
|
if (aRect.Y() < CAIRO_COORD_MIN) {
|
|
aRect.SetHeight(aRect.YMost() - CAIRO_COORD_MIN);
|
|
if (aRect.Height() < 0.0) {
|
|
aRect.SetHeight(0.0);
|
|
}
|
|
aRect.MoveToY(CAIRO_COORD_MIN);
|
|
}
|
|
|
|
if (aRect.XMost() > CAIRO_COORD_MAX) {
|
|
aRect.SetRightEdge(CAIRO_COORD_MAX);
|
|
}
|
|
|
|
if (aRect.YMost() > CAIRO_COORD_MAX) {
|
|
aRect.SetBottomEdge(CAIRO_COORD_MAX);
|
|
}
|
|
#undef CAIRO_COORD_MAX
|
|
#undef CAIRO_COORD_MIN
|
|
}
|
|
|
|
/*static*/
|
|
gfxQuad gfxUtils::TransformToQuad(const gfxRect& aRect,
|
|
const mozilla::gfx::Matrix4x4& aMatrix) {
|
|
gfxPoint points[4];
|
|
|
|
points[0] = aMatrix.TransformPoint(aRect.TopLeft());
|
|
points[1] = aMatrix.TransformPoint(aRect.TopRight());
|
|
points[2] = aMatrix.TransformPoint(aRect.BottomRight());
|
|
points[3] = aMatrix.TransformPoint(aRect.BottomLeft());
|
|
|
|
// Could this ever result in lines that intersect? I don't think so.
|
|
return gfxQuad(points[0], points[1], points[2], points[3]);
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::ClearThebesSurface(gfxASurface* aSurface) {
|
|
if (aSurface->CairoStatus()) {
|
|
return;
|
|
}
|
|
cairo_surface_t* surf = aSurface->CairoSurface();
|
|
if (cairo_surface_status(surf)) {
|
|
return;
|
|
}
|
|
cairo_t* ctx = cairo_create(surf);
|
|
cairo_set_source_rgba(ctx, 0.0, 0.0, 0.0, 0.0);
|
|
cairo_set_operator(ctx, CAIRO_OPERATOR_SOURCE);
|
|
IntRect bounds(nsIntPoint(0, 0), aSurface->GetSize());
|
|
cairo_rectangle(ctx, bounds.X(), bounds.Y(), bounds.Width(), bounds.Height());
|
|
cairo_fill(ctx);
|
|
cairo_destroy(ctx);
|
|
}
|
|
|
|
/* static */
|
|
already_AddRefed<DataSourceSurface>
|
|
gfxUtils::CopySurfaceToDataSourceSurfaceWithFormat(SourceSurface* aSurface,
|
|
SurfaceFormat aFormat) {
|
|
MOZ_ASSERT(aFormat != aSurface->GetFormat(),
|
|
"Unnecessary - and very expersive - surface format conversion");
|
|
|
|
Rect bounds(0, 0, aSurface->GetSize().width, aSurface->GetSize().height);
|
|
|
|
if (!aSurface->IsDataSourceSurface()) {
|
|
// If the surface is NOT of type DATA then its data is not mapped into main
|
|
// memory. Format conversion is probably faster on the GPU, and by doing it
|
|
// there we can avoid any expensive uploads/readbacks except for (possibly)
|
|
// a single readback due to the unavoidable GetDataSurface() call. Using
|
|
// CreateOffscreenContentDrawTarget ensures the conversion happens on the
|
|
// GPU.
|
|
RefPtr<DrawTarget> dt =
|
|
gfxPlatform::GetPlatform()->CreateOffscreenContentDrawTarget(
|
|
aSurface->GetSize(), aFormat);
|
|
if (!dt) {
|
|
gfxWarning() << "gfxUtils::CopySurfaceToDataSourceSurfaceWithFormat "
|
|
"failed in CreateOffscreenContentDrawTarget";
|
|
return nullptr;
|
|
}
|
|
|
|
// Using DrawSurface() here rather than CopySurface() because CopySurface
|
|
// is optimized for memcpy and therefore isn't good for format conversion.
|
|
// Using OP_OVER since in our case it's equivalent to OP_SOURCE and
|
|
// generally more optimized.
|
|
dt->DrawSurface(aSurface, bounds, bounds, DrawSurfaceOptions(),
|
|
DrawOptions(1.0f, CompositionOp::OP_OVER));
|
|
RefPtr<SourceSurface> surface = dt->Snapshot();
|
|
return surface->GetDataSurface();
|
|
}
|
|
|
|
// If the surface IS of type DATA then it may or may not be in main memory
|
|
// depending on whether or not it has been mapped yet. We have no way of
|
|
// knowing, so we can't be sure if it's best to create a data wrapping
|
|
// DrawTarget for the conversion or an offscreen content DrawTarget. We could
|
|
// guess it's not mapped and create an offscreen content DrawTarget, but if
|
|
// it is then we'll end up uploading the surface data, and most likely the
|
|
// caller is going to be accessing the resulting surface data, resulting in a
|
|
// readback (both very expensive operations). Alternatively we could guess
|
|
// the data is mapped and create a data wrapping DrawTarget and, if the
|
|
// surface is not in main memory, then we will incure a readback. The latter
|
|
// of these two "wrong choices" is the least costly (a readback, vs an
|
|
// upload and a readback), and more than likely the DATA surface that we've
|
|
// been passed actually IS in main memory anyway. For these reasons it's most
|
|
// likely best to create a data wrapping DrawTarget here to do the format
|
|
// conversion.
|
|
RefPtr<DataSourceSurface> dataSurface =
|
|
Factory::CreateDataSourceSurface(aSurface->GetSize(), aFormat);
|
|
DataSourceSurface::MappedSurface map;
|
|
if (!dataSurface ||
|
|
!dataSurface->Map(DataSourceSurface::MapType::READ_WRITE, &map)) {
|
|
return nullptr;
|
|
}
|
|
RefPtr<DrawTarget> dt = Factory::CreateDrawTargetForData(
|
|
BackendType::CAIRO, map.mData, dataSurface->GetSize(), map.mStride,
|
|
aFormat);
|
|
if (!dt) {
|
|
dataSurface->Unmap();
|
|
return nullptr;
|
|
}
|
|
// Using DrawSurface() here rather than CopySurface() because CopySurface
|
|
// is optimized for memcpy and therefore isn't good for format conversion.
|
|
// Using OP_OVER since in our case it's equivalent to OP_SOURCE and
|
|
// generally more optimized.
|
|
dt->DrawSurface(aSurface, bounds, bounds, DrawSurfaceOptions(),
|
|
DrawOptions(1.0f, CompositionOp::OP_OVER));
|
|
dataSurface->Unmap();
|
|
return dataSurface.forget();
|
|
}
|
|
|
|
const uint32_t gfxUtils::sNumFrameColors = 8;
|
|
|
|
/* static */
|
|
const gfx::DeviceColor& gfxUtils::GetColorForFrameNumber(
|
|
uint64_t aFrameNumber) {
|
|
static bool initialized = false;
|
|
static gfx::DeviceColor colors[sNumFrameColors];
|
|
|
|
if (!initialized) {
|
|
// This isn't truly device color, but it is just for debug.
|
|
uint32_t i = 0;
|
|
colors[i++] = gfx::DeviceColor::FromABGR(0xffff0000);
|
|
colors[i++] = gfx::DeviceColor::FromABGR(0xffcc00ff);
|
|
colors[i++] = gfx::DeviceColor::FromABGR(0xff0066cc);
|
|
colors[i++] = gfx::DeviceColor::FromABGR(0xff00ff00);
|
|
colors[i++] = gfx::DeviceColor::FromABGR(0xff33ffff);
|
|
colors[i++] = gfx::DeviceColor::FromABGR(0xffff0099);
|
|
colors[i++] = gfx::DeviceColor::FromABGR(0xff0000ff);
|
|
colors[i++] = gfx::DeviceColor::FromABGR(0xff999999);
|
|
MOZ_ASSERT(i == sNumFrameColors);
|
|
initialized = true;
|
|
}
|
|
|
|
return colors[aFrameNumber % sNumFrameColors];
|
|
}
|
|
|
|
/* static */
|
|
nsresult gfxUtils::EncodeSourceSurface(SourceSurface* aSurface,
|
|
const ImageType aImageType,
|
|
const nsAString& aOutputOptions,
|
|
BinaryOrData aBinaryOrData, FILE* aFile,
|
|
nsACString* aStrOut) {
|
|
MOZ_ASSERT(aBinaryOrData == gfxUtils::eDataURIEncode || aFile || aStrOut,
|
|
"Copying binary encoding to clipboard not currently supported");
|
|
|
|
const IntSize size = aSurface->GetSize();
|
|
if (size.IsEmpty()) {
|
|
return NS_ERROR_INVALID_ARG;
|
|
}
|
|
const Size floatSize(size.width, size.height);
|
|
|
|
RefPtr<DataSourceSurface> dataSurface;
|
|
if (aSurface->GetFormat() != SurfaceFormat::B8G8R8A8) {
|
|
// FIXME bug 995807 (B8G8R8X8), bug 831898 (R5G6B5)
|
|
dataSurface = gfxUtils::CopySurfaceToDataSourceSurfaceWithFormat(
|
|
aSurface, SurfaceFormat::B8G8R8A8);
|
|
} else {
|
|
dataSurface = aSurface->GetDataSurface();
|
|
}
|
|
if (!dataSurface) {
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
DataSourceSurface::MappedSurface map;
|
|
if (!dataSurface->Map(DataSourceSurface::MapType::READ, &map)) {
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
RefPtr<imgIEncoder> encoder = nullptr;
|
|
|
|
switch (aImageType) {
|
|
case ImageType::BMP:
|
|
encoder = MakeRefPtr<nsBMPEncoder>();
|
|
break;
|
|
|
|
case ImageType::ICO:
|
|
encoder = MakeRefPtr<nsICOEncoder>();
|
|
break;
|
|
|
|
case ImageType::JPEG:
|
|
encoder = MakeRefPtr<nsJPEGEncoder>();
|
|
break;
|
|
|
|
case ImageType::PNG:
|
|
encoder = MakeRefPtr<nsPNGEncoder>();
|
|
break;
|
|
|
|
default:
|
|
break;
|
|
}
|
|
|
|
MOZ_RELEASE_ASSERT(encoder != nullptr);
|
|
|
|
nsresult rv = encoder->InitFromData(
|
|
map.mData, BufferSizeFromStrideAndHeight(map.mStride, size.height),
|
|
size.width, size.height, map.mStride, imgIEncoder::INPUT_FORMAT_HOSTARGB,
|
|
aOutputOptions);
|
|
dataSurface->Unmap();
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
|
|
nsCOMPtr<nsIInputStream> imgStream(encoder);
|
|
if (!imgStream) {
|
|
return NS_ERROR_FAILURE;
|
|
}
|
|
|
|
uint64_t bufSize64;
|
|
rv = imgStream->Available(&bufSize64);
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
|
|
NS_ENSURE_TRUE(bufSize64 < UINT32_MAX - 16, NS_ERROR_FAILURE);
|
|
|
|
uint32_t bufSize = (uint32_t)bufSize64;
|
|
|
|
// ...leave a little extra room so we can call read again and make sure we
|
|
// got everything. 16 bytes for better padding (maybe)
|
|
bufSize += 16;
|
|
uint32_t imgSize = 0;
|
|
Vector<char> imgData;
|
|
if (!imgData.initCapacity(bufSize)) {
|
|
return NS_ERROR_OUT_OF_MEMORY;
|
|
}
|
|
uint32_t numReadThisTime = 0;
|
|
while ((rv = imgStream->Read(imgData.begin() + imgSize, bufSize - imgSize,
|
|
&numReadThisTime)) == NS_OK &&
|
|
numReadThisTime > 0) {
|
|
// Update the length of the vector without overwriting the new data.
|
|
if (!imgData.growByUninitialized(numReadThisTime)) {
|
|
return NS_ERROR_OUT_OF_MEMORY;
|
|
}
|
|
|
|
imgSize += numReadThisTime;
|
|
if (imgSize == bufSize) {
|
|
// need a bigger buffer, just double
|
|
bufSize *= 2;
|
|
if (!imgData.resizeUninitialized(bufSize)) {
|
|
return NS_ERROR_OUT_OF_MEMORY;
|
|
}
|
|
}
|
|
}
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
NS_ENSURE_TRUE(!imgData.empty(), NS_ERROR_FAILURE);
|
|
|
|
if (aBinaryOrData == gfxUtils::eBinaryEncode) {
|
|
if (aFile) {
|
|
Unused << fwrite(imgData.begin(), 1, imgSize, aFile);
|
|
}
|
|
return NS_OK;
|
|
}
|
|
|
|
// base 64, result will be null-terminated
|
|
nsCString encodedImg;
|
|
rv = Base64Encode(Substring(imgData.begin(), imgSize), encodedImg);
|
|
NS_ENSURE_SUCCESS(rv, rv);
|
|
|
|
nsCString stringBuf;
|
|
nsACString& dataURI = aStrOut ? *aStrOut : stringBuf;
|
|
dataURI.AppendLiteral("data:");
|
|
|
|
switch (aImageType) {
|
|
case ImageType::BMP:
|
|
dataURI.AppendLiteral(IMAGE_BMP);
|
|
break;
|
|
|
|
case ImageType::ICO:
|
|
dataURI.AppendLiteral(IMAGE_ICO_MS);
|
|
break;
|
|
case ImageType::JPEG:
|
|
dataURI.AppendLiteral(IMAGE_JPEG);
|
|
break;
|
|
|
|
case ImageType::PNG:
|
|
dataURI.AppendLiteral(IMAGE_PNG);
|
|
break;
|
|
|
|
default:
|
|
break;
|
|
}
|
|
|
|
dataURI.AppendLiteral(";base64,");
|
|
dataURI.Append(encodedImg);
|
|
|
|
if (aFile) {
|
|
#ifdef ANDROID
|
|
if (aFile == stdout || aFile == stderr) {
|
|
// ADB logcat cuts off long strings so we will break it down
|
|
const char* cStr = dataURI.BeginReading();
|
|
size_t len = strlen(cStr);
|
|
while (true) {
|
|
printf_stderr("IMG: %.140s\n", cStr);
|
|
if (len <= 140) break;
|
|
len -= 140;
|
|
cStr += 140;
|
|
}
|
|
}
|
|
#endif
|
|
fprintf(aFile, "%s", dataURI.BeginReading());
|
|
} else if (!aStrOut) {
|
|
nsCOMPtr<nsIClipboardHelper> clipboard(
|
|
do_GetService("@mozilla.org/widget/clipboardhelper;1", &rv));
|
|
if (clipboard) {
|
|
clipboard->CopyString(NS_ConvertASCIItoUTF16(dataURI));
|
|
}
|
|
}
|
|
return NS_OK;
|
|
}
|
|
|
|
static nsCString EncodeSourceSurfaceAsPNGURI(SourceSurface* aSurface) {
|
|
nsCString string;
|
|
gfxUtils::EncodeSourceSurface(aSurface, ImageType::PNG, EmptyString(),
|
|
gfxUtils::eDataURIEncode, nullptr, &string);
|
|
return string;
|
|
}
|
|
|
|
// https://jdashg.github.io/misc/colors/from-coeffs.html
|
|
const float kBT601NarrowYCbCrToRGB_RowMajor[16] = {
|
|
1.16438f, 0.00000f, 1.59603f, -0.87420f, 1.16438f, -0.39176f,
|
|
-0.81297f, 0.53167f, 1.16438f, 2.01723f, 0.00000f, -1.08563f,
|
|
0.00000f, 0.00000f, 0.00000f, 1.00000f};
|
|
const float kBT709NarrowYCbCrToRGB_RowMajor[16] = {
|
|
1.16438f, 0.00000f, 1.79274f, -0.97295f, 1.16438f, -0.21325f,
|
|
-0.53291f, 0.30148f, 1.16438f, 2.11240f, 0.00000f, -1.13340f,
|
|
0.00000f, 0.00000f, 0.00000f, 1.00000f};
|
|
const float kBT2020NarrowYCbCrToRGB_RowMajor[16] = {
|
|
1.16438f, 0.00000f, 1.67867f, -0.91569f, 1.16438f, -0.18733f,
|
|
-0.65042f, 0.34746f, 1.16438f, 2.14177f, 0.00000f, -1.14815f,
|
|
0.00000f, 0.00000f, 0.00000f, 1.00000f};
|
|
|
|
/* static */ const float* gfxUtils::YuvToRgbMatrix4x3RowMajor(
|
|
gfx::YUVColorSpace aYUVColorSpace) {
|
|
#define X(x) \
|
|
{ x[0], x[1], x[2], 0.0f, x[4], x[5], x[6], 0.0f, x[8], x[9], x[10], 0.0f }
|
|
|
|
static const float rec601[12] = X(kBT601NarrowYCbCrToRGB_RowMajor);
|
|
static const float rec709[12] = X(kBT709NarrowYCbCrToRGB_RowMajor);
|
|
static const float rec2020[12] = X(kBT2020NarrowYCbCrToRGB_RowMajor);
|
|
|
|
#undef X
|
|
|
|
switch (aYUVColorSpace) {
|
|
case gfx::YUVColorSpace::BT601:
|
|
return rec601;
|
|
case gfx::YUVColorSpace::BT709:
|
|
return rec709;
|
|
case gfx::YUVColorSpace::BT2020:
|
|
return rec2020;
|
|
default: // YUVColorSpace::UNKNOWN
|
|
MOZ_ASSERT(false, "unknown aYUVColorSpace");
|
|
return rec601;
|
|
}
|
|
}
|
|
|
|
/* static */ const float* gfxUtils::YuvToRgbMatrix3x3ColumnMajor(
|
|
gfx::YUVColorSpace aYUVColorSpace) {
|
|
#define X(x) \
|
|
{ x[0], x[4], x[8], x[1], x[5], x[9], x[2], x[6], x[10] }
|
|
|
|
static const float rec601[9] = X(kBT601NarrowYCbCrToRGB_RowMajor);
|
|
static const float rec709[9] = X(kBT709NarrowYCbCrToRGB_RowMajor);
|
|
static const float rec2020[9] = X(kBT2020NarrowYCbCrToRGB_RowMajor);
|
|
|
|
#undef X
|
|
|
|
switch (aYUVColorSpace) {
|
|
case gfx::YUVColorSpace::BT601:
|
|
return rec601;
|
|
case YUVColorSpace::BT709:
|
|
return rec709;
|
|
case YUVColorSpace::BT2020:
|
|
return rec2020;
|
|
default: // YUVColorSpace::UNKNOWN
|
|
MOZ_ASSERT(false, "unknown aYUVColorSpace");
|
|
return rec601;
|
|
}
|
|
}
|
|
|
|
/* static */ const float* gfxUtils::YuvToRgbMatrix4x4ColumnMajor(
|
|
YUVColorSpace aYUVColorSpace) {
|
|
#define X(x) \
|
|
{ \
|
|
x[0], x[4], x[8], x[12], x[1], x[5], x[9], x[13], x[2], x[6], x[10], \
|
|
x[14], x[3], x[7], x[11], x[15] \
|
|
}
|
|
|
|
static const float rec601[16] = X(kBT601NarrowYCbCrToRGB_RowMajor);
|
|
static const float rec709[16] = X(kBT709NarrowYCbCrToRGB_RowMajor);
|
|
static const float rec2020[16] = X(kBT2020NarrowYCbCrToRGB_RowMajor);
|
|
|
|
#undef X
|
|
|
|
switch (aYUVColorSpace) {
|
|
case YUVColorSpace::BT601:
|
|
return rec601;
|
|
case YUVColorSpace::BT709:
|
|
return rec709;
|
|
case YUVColorSpace::BT2020:
|
|
return rec2020;
|
|
default: // YUVColorSpace::UNKNOWN
|
|
MOZ_ASSERT(false, "unknown aYUVColorSpace");
|
|
return rec601;
|
|
}
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::WriteAsPNG(SourceSurface* aSurface, const nsAString& aFile) {
|
|
WriteAsPNG(aSurface, NS_ConvertUTF16toUTF8(aFile).get());
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::WriteAsPNG(SourceSurface* aSurface, const char* aFile) {
|
|
FILE* file = fopen(aFile, "wb");
|
|
|
|
if (!file) {
|
|
// Maybe the directory doesn't exist; try creating it, then fopen again.
|
|
nsresult rv = NS_ERROR_FAILURE;
|
|
nsCOMPtr<nsIFile> comFile = do_CreateInstance("@mozilla.org/file/local;1");
|
|
if (comFile) {
|
|
NS_ConvertUTF8toUTF16 utf16path((nsDependentCString(aFile)));
|
|
rv = comFile->InitWithPath(utf16path);
|
|
if (NS_SUCCEEDED(rv)) {
|
|
nsCOMPtr<nsIFile> dirPath;
|
|
comFile->GetParent(getter_AddRefs(dirPath));
|
|
if (dirPath) {
|
|
rv = dirPath->Create(nsIFile::DIRECTORY_TYPE, 0777);
|
|
if (NS_SUCCEEDED(rv) || rv == NS_ERROR_FILE_ALREADY_EXISTS) {
|
|
file = fopen(aFile, "wb");
|
|
}
|
|
}
|
|
}
|
|
}
|
|
if (!file) {
|
|
NS_WARNING("Failed to open file to create PNG!");
|
|
return;
|
|
}
|
|
}
|
|
|
|
EncodeSourceSurface(aSurface, ImageType::PNG, EmptyString(), eBinaryEncode,
|
|
file);
|
|
fclose(file);
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::WriteAsPNG(DrawTarget* aDT, const nsAString& aFile) {
|
|
WriteAsPNG(aDT, NS_ConvertUTF16toUTF8(aFile).get());
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::WriteAsPNG(DrawTarget* aDT, const char* aFile) {
|
|
RefPtr<SourceSurface> surface = aDT->Snapshot();
|
|
if (surface) {
|
|
WriteAsPNG(surface, aFile);
|
|
} else {
|
|
NS_WARNING("Failed to get surface!");
|
|
}
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::DumpAsDataURI(SourceSurface* aSurface, FILE* aFile) {
|
|
EncodeSourceSurface(aSurface, ImageType::PNG, EmptyString(), eDataURIEncode,
|
|
aFile);
|
|
}
|
|
|
|
/* static */
|
|
nsCString gfxUtils::GetAsDataURI(SourceSurface* aSurface) {
|
|
return EncodeSourceSurfaceAsPNGURI(aSurface);
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::DumpAsDataURI(DrawTarget* aDT, FILE* aFile) {
|
|
RefPtr<SourceSurface> surface = aDT->Snapshot();
|
|
if (surface) {
|
|
DumpAsDataURI(surface, aFile);
|
|
} else {
|
|
NS_WARNING("Failed to get surface!");
|
|
}
|
|
}
|
|
|
|
/* static */
|
|
nsCString gfxUtils::GetAsLZ4Base64Str(DataSourceSurface* aSourceSurface) {
|
|
DataSourceSurface::ScopedMap map(aSourceSurface, DataSourceSurface::READ);
|
|
int32_t dataSize = aSourceSurface->GetSize().height * map.GetStride();
|
|
auto compressedData = MakeUnique<char[]>(LZ4::maxCompressedSize(dataSize));
|
|
if (compressedData) {
|
|
int nDataSize =
|
|
LZ4::compress((char*)map.GetData(), dataSize, compressedData.get());
|
|
if (nDataSize > 0) {
|
|
nsCString encodedImg;
|
|
nsresult rv =
|
|
Base64Encode(Substring(compressedData.get(), nDataSize), encodedImg);
|
|
if (rv == NS_OK) {
|
|
nsCString string("");
|
|
string.AppendPrintf("data:image/lz4bgra;base64,%i,%i,%i,",
|
|
aSourceSurface->GetSize().width, map.GetStride(),
|
|
aSourceSurface->GetSize().height);
|
|
string.Append(encodedImg);
|
|
return string;
|
|
}
|
|
}
|
|
}
|
|
return nsCString("");
|
|
}
|
|
|
|
/* static */
|
|
nsCString gfxUtils::GetAsDataURI(DrawTarget* aDT) {
|
|
RefPtr<SourceSurface> surface = aDT->Snapshot();
|
|
if (surface) {
|
|
return EncodeSourceSurfaceAsPNGURI(surface);
|
|
} else {
|
|
NS_WARNING("Failed to get surface!");
|
|
return nsCString("");
|
|
}
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::CopyAsDataURI(SourceSurface* aSurface) {
|
|
EncodeSourceSurface(aSurface, ImageType::PNG, EmptyString(), eDataURIEncode,
|
|
nullptr);
|
|
}
|
|
|
|
/* static */
|
|
void gfxUtils::CopyAsDataURI(DrawTarget* aDT) {
|
|
RefPtr<SourceSurface> surface = aDT->Snapshot();
|
|
if (surface) {
|
|
CopyAsDataURI(surface);
|
|
} else {
|
|
NS_WARNING("Failed to get surface!");
|
|
}
|
|
}
|
|
|
|
/* static */ UniquePtr<uint8_t[]> gfxUtils::GetImageBuffer(
|
|
gfx::DataSourceSurface* aSurface, bool aIsAlphaPremultiplied,
|
|
int32_t* outFormat) {
|
|
*outFormat = 0;
|
|
|
|
DataSourceSurface::MappedSurface map;
|
|
if (!aSurface->Map(DataSourceSurface::MapType::READ, &map)) return nullptr;
|
|
|
|
uint32_t bufferSize =
|
|
aSurface->GetSize().width * aSurface->GetSize().height * 4;
|
|
auto imageBuffer = MakeUniqueFallible<uint8_t[]>(bufferSize);
|
|
if (!imageBuffer) {
|
|
aSurface->Unmap();
|
|
return nullptr;
|
|
}
|
|
memcpy(imageBuffer.get(), map.mData, bufferSize);
|
|
|
|
aSurface->Unmap();
|
|
|
|
int32_t format = imgIEncoder::INPUT_FORMAT_HOSTARGB;
|
|
if (!aIsAlphaPremultiplied) {
|
|
// We need to convert to INPUT_FORMAT_RGBA, otherwise
|
|
// we are automatically considered premult, and unpremult'd.
|
|
// Yes, it is THAT silly.
|
|
// Except for different lossy conversions by color,
|
|
// we could probably just change the label, and not change the data.
|
|
gfxUtils::ConvertBGRAtoRGBA(imageBuffer.get(), bufferSize);
|
|
format = imgIEncoder::INPUT_FORMAT_RGBA;
|
|
}
|
|
|
|
*outFormat = format;
|
|
return imageBuffer;
|
|
}
|
|
|
|
/* static */
|
|
nsresult gfxUtils::GetInputStream(gfx::DataSourceSurface* aSurface,
|
|
bool aIsAlphaPremultiplied,
|
|
const char* aMimeType,
|
|
const nsAString& aEncoderOptions,
|
|
nsIInputStream** outStream) {
|
|
nsCString enccid("@mozilla.org/image/encoder;2?type=");
|
|
enccid += aMimeType;
|
|
nsCOMPtr<imgIEncoder> encoder = do_CreateInstance(enccid.get());
|
|
if (!encoder) return NS_ERROR_FAILURE;
|
|
|
|
int32_t format = 0;
|
|
UniquePtr<uint8_t[]> imageBuffer =
|
|
GetImageBuffer(aSurface, aIsAlphaPremultiplied, &format);
|
|
if (!imageBuffer) return NS_ERROR_FAILURE;
|
|
|
|
return dom::ImageEncoder::GetInputStream(
|
|
aSurface->GetSize().width, aSurface->GetSize().height, imageBuffer.get(),
|
|
format, encoder, aEncoderOptions, outStream);
|
|
}
|
|
|
|
class GetFeatureStatusWorkerRunnable final
|
|
: public dom::WorkerMainThreadRunnable {
|
|
public:
|
|
GetFeatureStatusWorkerRunnable(dom::WorkerPrivate* workerPrivate,
|
|
const nsCOMPtr<nsIGfxInfo>& gfxInfo,
|
|
int32_t feature, nsACString& failureId,
|
|
int32_t* status)
|
|
: WorkerMainThreadRunnable(workerPrivate, "GFX :: GetFeatureStatus"_ns),
|
|
mGfxInfo(gfxInfo),
|
|
mFeature(feature),
|
|
mStatus(status),
|
|
mFailureId(failureId),
|
|
mNSResult(NS_OK) {}
|
|
|
|
bool MainThreadRun() override {
|
|
if (mGfxInfo) {
|
|
mNSResult = mGfxInfo->GetFeatureStatus(mFeature, mFailureId, mStatus);
|
|
}
|
|
return true;
|
|
}
|
|
|
|
nsresult GetNSResult() const { return mNSResult; }
|
|
|
|
protected:
|
|
~GetFeatureStatusWorkerRunnable() = default;
|
|
|
|
private:
|
|
nsCOMPtr<nsIGfxInfo> mGfxInfo;
|
|
int32_t mFeature;
|
|
int32_t* mStatus;
|
|
nsACString& mFailureId;
|
|
nsresult mNSResult;
|
|
};
|
|
|
|
/* static */
|
|
nsresult gfxUtils::ThreadSafeGetFeatureStatus(
|
|
const nsCOMPtr<nsIGfxInfo>& gfxInfo, int32_t feature, nsACString& failureId,
|
|
int32_t* status) {
|
|
if (NS_IsMainThread()) {
|
|
return gfxInfo->GetFeatureStatus(feature, failureId, status);
|
|
}
|
|
|
|
// In a content process, we must call this on the main thread.
|
|
// In a composition process (parent or GPU), this needs to be called on the
|
|
// compositor thread.
|
|
bool isCompositionProcess = XRE_IsGPUProcess() || XRE_IsParentProcess();
|
|
MOZ_ASSERT(!isCompositionProcess || NS_IsInCompositorThread());
|
|
|
|
// Content-process non-main-thread case:
|
|
if (!isCompositionProcess) {
|
|
dom::WorkerPrivate* workerPrivate = dom::GetCurrentThreadWorkerPrivate();
|
|
|
|
RefPtr<GetFeatureStatusWorkerRunnable> runnable =
|
|
new GetFeatureStatusWorkerRunnable(workerPrivate, gfxInfo, feature,
|
|
failureId, status);
|
|
|
|
ErrorResult rv;
|
|
runnable->Dispatch(dom::WorkerStatus::Canceling, rv);
|
|
if (rv.Failed()) {
|
|
// XXXbz This is totally broken, since we're supposed to just abort
|
|
// everything up the callstack but the callers basically eat the
|
|
// exception. Ah, well.
|
|
return rv.StealNSResult();
|
|
}
|
|
return runnable->GetNSResult();
|
|
}
|
|
|
|
nsresult rv;
|
|
SynchronousTask task("GetFeatureStatusSync");
|
|
NS_DispatchToMainThread(NS_NewRunnableFunction("GetFeatureStatusMain", [&]() {
|
|
AutoCompleteTask complete(&task);
|
|
rv = gfxInfo->GetFeatureStatus(feature, failureId, status);
|
|
}));
|
|
task.Wait();
|
|
return rv;
|
|
}
|
|
|
|
#define GFX_SHADER_CHECK_BUILD_VERSION_PREF "gfx-shader-check.build-version"
|
|
#define GFX_SHADER_CHECK_DEVICE_ID_PREF "gfx-shader-check.device-id"
|
|
#define GFX_SHADER_CHECK_DRIVER_VERSION_PREF "gfx-shader-check.driver-version"
|
|
|
|
/* static */
|
|
void gfxUtils::RemoveShaderCacheFromDiskIfNecessary() {
|
|
if (!gfxVars::UseWebRenderProgramBinaryDisk()) {
|
|
return;
|
|
}
|
|
|
|
nsCOMPtr<nsIGfxInfo> gfxInfo = services::GetGfxInfo();
|
|
|
|
// Get current values
|
|
nsCString buildID(mozilla::PlatformBuildID());
|
|
nsString deviceID, driverVersion;
|
|
gfxInfo->GetAdapterDeviceID(deviceID);
|
|
gfxInfo->GetAdapterDriverVersion(driverVersion);
|
|
|
|
// Get pref stored values
|
|
nsAutoCString buildIDChecked;
|
|
Preferences::GetCString(GFX_SHADER_CHECK_BUILD_VERSION_PREF, buildIDChecked);
|
|
nsAutoString deviceIDChecked, driverVersionChecked;
|
|
Preferences::GetString(GFX_SHADER_CHECK_DEVICE_ID_PREF, deviceIDChecked);
|
|
Preferences::GetString(GFX_SHADER_CHECK_DRIVER_VERSION_PREF,
|
|
driverVersionChecked);
|
|
|
|
if (buildID == buildIDChecked && deviceID == deviceIDChecked &&
|
|
driverVersion == driverVersionChecked) {
|
|
return;
|
|
}
|
|
|
|
nsAutoString path(gfx::gfxVars::ProfDirectory());
|
|
|
|
if (!wr::remove_program_binary_disk_cache(&path)) {
|
|
// Failed to remove program binary disk cache. The disk cache might have
|
|
// invalid data. Disable program binary disk cache usage.
|
|
gfxVars::SetUseWebRenderProgramBinaryDisk(false);
|
|
return;
|
|
}
|
|
|
|
Preferences::SetCString(GFX_SHADER_CHECK_BUILD_VERSION_PREF, buildID);
|
|
Preferences::SetString(GFX_SHADER_CHECK_DEVICE_ID_PREF, deviceID);
|
|
Preferences::SetString(GFX_SHADER_CHECK_DRIVER_VERSION_PREF, driverVersion);
|
|
}
|
|
|
|
/* static */
|
|
bool gfxUtils::DumpDisplayList() {
|
|
return StaticPrefs::layout_display_list_dump() ||
|
|
(StaticPrefs::layout_display_list_dump_parent() &&
|
|
XRE_IsParentProcess()) ||
|
|
(StaticPrefs::layout_display_list_dump_content() &&
|
|
XRE_IsContentProcess());
|
|
}
|
|
|
|
FILE* gfxUtils::sDumpPaintFile = stderr;
|
|
|
|
namespace mozilla {
|
|
namespace gfx {
|
|
|
|
DeviceColor ToDeviceColor(const sRGBColor& aColor) {
|
|
// aColor is pass-by-value since to get return value optimization goodness we
|
|
// need to return the same object from all return points in this function. We
|
|
// could declare a local Color variable and use that, but we might as well
|
|
// just use aColor.
|
|
if (gfxPlatform::GetCMSMode() == eCMSMode_All) {
|
|
qcms_transform* transform = gfxPlatform::GetCMSRGBTransform();
|
|
if (transform) {
|
|
return gfxPlatform::TransformPixel(aColor, transform);
|
|
// Use the original alpha to avoid unnecessary float->byte->float
|
|
// conversion errors
|
|
}
|
|
}
|
|
return DeviceColor(aColor.r, aColor.g, aColor.b, aColor.a);
|
|
}
|
|
|
|
DeviceColor ToDeviceColor(nscolor aColor) {
|
|
return ToDeviceColor(sRGBColor::FromABGR(aColor));
|
|
}
|
|
|
|
DeviceColor ToDeviceColor(const StyleRGBA& aColor) {
|
|
return ToDeviceColor(aColor.ToColor());
|
|
}
|
|
|
|
} // namespace gfx
|
|
} // namespace mozilla
|