Allowing YUV data to be retrieved from the JPEG Decoder.

Source/platform/image-decoders/jpeg/JPEGImageDecoder.h

Index: Source/platform/image-decoders/jpeg/JPEGImageDecoder.h
diff --git a/Source/platform/image-decoders/jpeg/JPEGImageDecoder.h b/Source/platform/image-decoders/jpeg/JPEGImageDecoder.h
index 9293a67e44a45766ff89348e833f5e203c0be4cf..a5d2fecc7c9583077e3b88e67c495958d7952f47 100644
--- a/Source/platform/image-decoders/jpeg/JPEGImageDecoder.h
+++ b/Source/platform/image-decoders/jpeg/JPEGImageDecoder.h
@@ -47,13 +47,17 @@ public:
     virtual String filenameExtension() const OVERRIDE { return "jpg"; }
     virtual bool isSizeAvailable() OVERRIDE;
     virtual bool hasColorProfile() const OVERRIDE { return m_hasColorProfile; }
-    virtual IntSize decodedSize() const OVERRIDE { return m_decodedSize; }
+    virtual IntSize decodedSize(int component = 0) const OVERRIDE;
     virtual bool setSize(unsigned width, unsigned height) OVERRIDE;
     virtual ImageFrame* frameBufferAtIndex(size_t) OVERRIDE;
     // CAUTION: setFailed() deletes |m_reader|.  Be careful to avoid
     // accessing deleted memory, especially when calling this from inside
     // JPEGImageReader!
     virtual bool setFailed() OVERRIDE;
+    virtual void setAcceleratedYUVDecoding(bool acceleratedYUVDecoding) OVERRIDE { m_acceleratedYUVDecoding = acceleratedYUVDecoding; }

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

This should be enableAcceleratedYUVDecoding() and be merged with
setDecodingBuffers().

+    virtual bool acceleratedYUVDecoding() const OVERRIDE { return m_acceleratedYUVDecoding; }
+    virtual bool doAcceleratedYUVDecoding() OVERRIDE;
+    virtual void setDecodingBuffers(OwnPtr<DecodingBuffers>&) OVERRIDE;
 
     bool outputScanlines();
     unsigned desiredScaleNumerator() const;
@@ -70,8 +74,10 @@ private:
     void decode(bool onlySize);
 
     OwnPtr<JPEGImageReader> m_reader;
+    OwnPtr<DecodingBuffers> m_decodingBuffers;
     IntSize m_decodedSize;
     bool m_hasColorProfile;
+    bool m_acceleratedYUVDecoding;

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

If there's DecodingBuffers there's no need to have this flag.

[sugoi1, 2014/07/24 23:16:38]

The current workflow does require a separate flag, let me explain why:

Step 1: Query the YUV size (by setting this flag)
Step 2: The caller allocates the memory
Step 3: The caller sends the allocated memory through setDecodingBuffers

I agree with you that, in step 3, the flag wouldn't be necessary, but the flag
is necessary for step 1, at a point where the buffers are not allocated yet.
Would you prefer that I send an empty DecodingBuffer object in step 1 instead of
the flag?

 };
 
 } // namespace blink