ImageBuffer-less SVG masking and clipping.

Source/core/platform/graphics/ImageBuffer.cpp

Index: Source/core/platform/graphics/ImageBuffer.cpp
diff --git a/Source/core/platform/graphics/ImageBuffer.cpp b/Source/core/platform/graphics/ImageBuffer.cpp
index 5e87788df1bbdcc6f12c99e83879b7eab90bdc32..d945b5cf168d8fe53ef983e8aaf78a7255968520 100644
--- a/Source/core/platform/graphics/ImageBuffer.cpp
+++ b/Source/core/platform/graphics/ImageBuffer.cpp
@@ -50,8 +50,10 @@
 #include "public/platform/Platform.h"
 #include "skia/ext/platform_canvas.h"
 #include "third_party/skia/include/core/SkBitmapDevice.h"
+#include "third_party/skia/include/core/SkColorFilter.h"
 #include "third_party/skia/include/core/SkColorPriv.h"
 #include "third_party/skia/include/core/SkSurface.h"
+#include "third_party/skia/include/effects/SkTableColorFilter.h"
 #include "third_party/skia/include/gpu/GrContext.h"
 #include "third_party/skia/include/gpu/SkGpuDevice.h"
 #include "wtf/MathExtras.h"
@@ -265,6 +267,36 @@ void ImageBuffer::drawPattern(GraphicsContext* context, const FloatRect& srcRect
     image->drawPattern(context, srcRect, scale, phase, op, destRect, blendMode);
 }
 
+static const Vector<uint8_t>& getLinearRgbLUT()
+{
+    DEFINE_STATIC_LOCAL(Vector<uint8_t>, linearRgbLUT, ());

[pdr., 2013/08/28 03:32:36]

Since you know the final size already, can you initialize this Vector with the
correct size?

[f(malita), 2013/08/28 19:54:38]

Good idea.

Simply passing a constructor size argument doesn't work well though, as the
vector ends up initializing $size elements upfront. Not only does this mess up
the initialization detection logic, but it also causes an unnecessary iteration.

I believe what we want to use is the template inlineCapacity arg
(Vector<typename T, size_t inlineCapacity = 0>), but unfortunately that doesn't
seem to be usable for statically defined vectors.

So the next best thing is an explicit reserveCapacity(256) call before entering
the initialization loop - to avoid growing the capacity in smaller increments
(not sure that was ever the case but doesn't hurt to be explicit). So let's do
this.

+    if (linearRgbLUT.isEmpty()) {
+        for (unsigned i = 0; i < 256; i++) {
+            float color = i  / 255.0f;
+            color = (color <= 0.04045f ? color / 12.92f : pow((color + 0.055f) / 1.055f, 2.4f));
+            color = std::max(0.0f, color);
+            color = std::min(1.0f, color);
+            linearRgbLUT.append(static_cast<uint8_t>(round(color * 255)));
+        }
+    }
+    return linearRgbLUT;
+}
+
+static const Vector<uint8_t>& getDeviceRgbLUT()
+{
+    DEFINE_STATIC_LOCAL(Vector<uint8_t>, deviceRgbLUT, ());
+    if (deviceRgbLUT.isEmpty()) {
+        for (unsigned i = 0; i < 256; i++) {
+            float color = i / 255.0f;
+            color = (powf(color, 1.0f / 2.4f) * 1.055f) - 0.055f;
+            color = std::max(0.0f, color);
+            color = std::min(1.0f, color);
+            deviceRgbLUT.append(static_cast<uint8_t>(round(color * 255)));
+        }
+    }
+    return deviceRgbLUT;
+}
+
 void ImageBuffer::transformColorSpace(ColorSpace srcColorSpace, ColorSpace dstColorSpace)
 {
     if (srcColorSpace == dstColorSpace)
@@ -301,36 +333,24 @@ void ImageBuffer::transformColorSpace(ColorSpace srcColorSpace, ColorSpace dstCo
     }
 }
 
-const Vector<uint8_t>& ImageBuffer::getLinearRgbLUT()
+PassRefPtr<SkColorFilter> ImageBuffer::createColorSpaceFilter(ColorSpace srcColorSpace,

[Stephen White, 2013/08/28 14:11:10]

Maybe at some point we can use this in place of
ImageBuffer::transformColorSpace() (or at least, use it in its implementation).

[f(malita), 2013/08/28 19:54:38]

Makes sense. The only transformColorSpace() user left appears to be
FilterEffect::transformResultColorSpace() so maybe it will just die off when
you're done Skia-fying the filter chain.

+    ColorSpace dstColorSpace)
 {
-    DEFINE_STATIC_LOCAL(Vector<uint8_t>, linearRgbLUT, ());
-    if (linearRgbLUT.isEmpty()) {
-        for (unsigned i = 0; i < 256; i++) {
-            float color = i  / 255.0f;
-            color = (color <= 0.04045f ? color / 12.92f : pow((color + 0.055f) / 1.055f, 2.4f));
-            color = std::max(0.0f, color);
-            color = std::min(1.0f, color);
-            linearRgbLUT.append(static_cast<uint8_t>(round(color * 255)));
-        }
-    }
-    return linearRgbLUT;
-}
+    if ((srcColorSpace == dstColorSpace)
+        || (srcColorSpace != ColorSpaceLinearRGB && srcColorSpace != ColorSpaceDeviceRGB)
+        || (dstColorSpace != ColorSpaceLinearRGB && dstColorSpace != ColorSpaceDeviceRGB))
+        return 0;
 
-const Vector<uint8_t>& ImageBuffer::getDeviceRgbLUT()
-{
-    DEFINE_STATIC_LOCAL(Vector<uint8_t>, deviceRgbLUT, ());
-    if (deviceRgbLUT.isEmpty()) {
-        for (unsigned i = 0; i < 256; i++) {
-            float color = i / 255.0f;
-            color = (powf(color, 1.0f / 2.4f) * 1.055f) - 0.055f;
-            color = std::max(0.0f, color);
-            color = std::min(1.0f, color);
-            deviceRgbLUT.append(static_cast<uint8_t>(round(color * 255)));
-        }
-    }
-    return deviceRgbLUT;
-}
+    const uint8_t* lut = 0;
+    if (dstColorSpace == ColorSpaceLinearRGB)
+        lut = &getLinearRgbLUT()[0];
+    else if (dstColorSpace == ColorSpaceDeviceRGB)
+        lut = &getDeviceRgbLUT()[0];
+    else
+        return 0;
 
+    return adoptRef(SkTableColorFilter::CreateARGB(0, lut, lut, lut));
+}
 
 template <Multiply multiplied>
 PassRefPtr<Uint8ClampedArray> getImageData(const IntRect& rect, GraphicsContext* context, const IntSize& size)
@@ -423,26 +443,6 @@ void ImageBuffer::putByteArray(Multiply multiplied, Uint8ClampedArray* source, c
     context()->writePixels(srcBitmap, destX, destY, config8888);
 }
 
-void ImageBuffer::convertToLuminanceMask()
-{
-    IntRect luminanceRect(IntPoint(), internalSize());
-    RefPtr<Uint8ClampedArray> srcPixelArray = getUnmultipliedImageData(luminanceRect);
-
-    unsigned pixelArrayLength = srcPixelArray->length();
-    for (unsigned pixelOffset = 0; pixelOffset < pixelArrayLength; pixelOffset += 4) {
-        unsigned char a = srcPixelArray->item(pixelOffset + 3);
-        if (!a)
-            continue;
-        unsigned char r = srcPixelArray->item(pixelOffset);
-        unsigned char g = srcPixelArray->item(pixelOffset + 1);
-        unsigned char b = srcPixelArray->item(pixelOffset + 2);
-
-        double luma = (r * 0.2125 + g * 0.7154 + b * 0.0721) * ((double)a / 255.0);
-        srcPixelArray->set(pixelOffset + 3, luma);
-    }
-    putByteArray(Unmultiplied, srcPixelArray.get(), luminanceRect.size(), luminanceRect, IntPoint());
-}
-
 template <typename T>
 static bool encodeImage(T& source, const String& mimeType, const double* quality, Vector<char>* output)
 {