Allowing YUV data to be retrieved from the JPEG Decoder.

Source/platform/image-decoders/ImageDecoder.h

Index: Source/platform/image-decoders/ImageDecoder.h
diff --git a/Source/platform/image-decoders/ImageDecoder.h b/Source/platform/image-decoders/ImageDecoder.h
index 559213011fae93fa14db64b85678ba0c704652f1..8bc902f86eab8818d214315a0bec74fc09cdabcd 100644
--- a/Source/platform/image-decoders/ImageDecoder.h
+++ b/Source/platform/image-decoders/ImageDecoder.h
@@ -49,6 +49,20 @@
 
 namespace blink {
 
+// DecodingBuffers can be used to decode color components into provided buffers instead of using an ImageFrame.
+class DecodingBuffers {

[Alpha Left Google, 2014/07/24 22:07:27]

This should be called ImagePlanes to be clear.

[sugoi1, 2014/07/24 23:16:35]

Acknowledged.

+public:
+    DecodingBuffers();
+
+    void set(void* planes[3], size_t rowBytes[3]);

[Alpha Left Google, 2014/07/24 22:07:27]

It will not always be 3 planes but it's good for now.

[sugoi1, 2014/07/24 23:16:35]

Acknowledged.

+    void* getPlane(int);

[Noel Gordon, 2014/07/25 15:55:01]

blink webcore convention: no "get" prefix in class getters.

[sugoi1, 2014/07/25 16:56:48]

Done.

+    size_t getRowBytes(int) const;
+
+private:
+    void* m_planes[3];
+    size_t m_rowBytes[3];
+};
+
 // ImageDecoder is a base for all format-specific decoders
 // (e.g. JPEGImageDecoder). This base manages the ImageFrame cache.
 //
@@ -100,7 +114,7 @@ public:
 
     // Decoders which downsample images should override this method to
     // return the actual decoded size.
-    virtual IntSize decodedSize() const { return size(); }
+    virtual IntSize decodedSize(int component = 0) const { return size(); }

[Noel Gordon, 2014/07/25 15:55:01]

I believe the Androids might downsample when decoding images.  Did you check if
yuv decoding interworks with that?

[sugoi1, 2014/07/25 16:56:48]

Looking at DecodingImageGenerator::onGetPixels() :
"Implementation doesn't support scaling yet so make sure we're not given a
different size."

For now, I could bail out if any scaling factor is given since it's not yet
supported. I can add that functionality as a follow up.

 
     // This will only differ from size() for ICO (where each frame is a
     // different icon) or other formats where different frames are different
@@ -252,6 +266,11 @@ public:
         m_frameBufferCache[0].setMemoryAllocator(allocator);
     }
 

[Alpha Left Google, 2014/07/24 22:07:26]

How do I know if the code that uses this decoder can draw the YUV subsampled
data? I think it's good to have method to query YUV subsampling.

[sugoi1, 2014/07/24 23:16:35]

I'm not sure I understand your question. The code would only call these methods
if it wants the YUV data, no? There shouldn't be any surprises for the code
calling these.

[Alpha Left Google, 2014/07/24 23:30:56]

I mean there's different types of subsampling. My question is if the caller
enables YUV decoding, it must handle all subsampling types. Is it true?

[sugoi1, 2014/07/25 02:55:13]

Well, the caller first receives the size of the Y, U and V planes and could
choose to support it or not based on the sizes received. It's fairly easy to
deduce the subsampling type from the sizes.

But, yes, currently the caller can handle all subsampling types. Note that the
JPEG decoder will only accept to set the output type to YUV if the subsampling
is one of : 4:4:4, 4:4:0, 4:2:2, 4:2:0, 4:1:1, 4:1:0, so only 6 subsampling
types need to be supported currently.

+    virtual void setAcceleratedYUVDecoding(bool) { }

[Alpha Left Google, 2014/07/24 22:07:26]

nit: drop "accelerated" from methods names. Decoding is still in software so
nothing is accelerated from the point of view of decoders.

[sugoi1, 2014/07/24 23:16:35]

Acknowledged.

+    virtual bool acceleratedYUVDecoding() const { return false; }
+    virtual bool doAcceleratedYUVDecoding() { return false; }
+    virtual void setDecodingBuffers(OwnPtr<DecodingBuffers>&) { }
+
 protected:
     // Calculates the most recent frame whose image data may be needed in
     // order to decode frame |frameIndex|, based on frame disposal methods