Implementation of SkBitmapRegionDecoder using SkAndroidCodec

tools/SkBitmapRegionCodec.cpp

Index: tools/SkBitmapRegionCodec.cpp
diff --git a/tools/SkBitmapRegionCodec.cpp b/tools/SkBitmapRegionCodec.cpp
new file mode 100644
index 0000000000000000000000000000000000000000..053b965f6f8964db9b088d775f35b77fcff9446a
--- /dev/null
+++ b/tools/SkBitmapRegionCodec.cpp
@@ -0,0 +1,142 @@
+/*
+ * Copyright 2015 Google Inc.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE file.
+ */
+
+#include "SkBitmapRegionCodec.h"
+#include "SkAndroidCodec.h"
+#include "SkCodecPriv.h"
+#include "SkCodecTools.h"
+
+SkBitmapRegionCodec::SkBitmapRegionCodec(SkAndroidCodec* codec)
+    : INHERITED(codec->getInfo().width(), codec->getInfo().height())
+    , fCodec(codec)
+{}
+
+/*
+ * Three differences from the Android version:
+ *     Returns a skia bitmap instead of an Android bitmap.
+ *     Android version attempts to reuse a recycled bitmap.
+ *     Removed the options object and used parameters for color type and sample size.
+ */
+// FIXME: Should this function should take in SkIRect?
+SkBitmap* SkBitmapRegionCodec::decodeRegion(int inputX, int inputY, int inputWidth, int inputHeight,
+        int sampleSize, SkColorType dstColorType) {
+
+    // Fix the input sampleSize if necessary.
+    if (sampleSize < 1) {
+        sampleSize = 1;
+    }
+
+    // The size of the output bitmap is determined by the size of the
+    // requested region, not by the size of the intersection of the region

[scroggo, 2015/10/21 20:56:20]

nit: region -> subset (although, to be fair, this function is called
decodeRegion, to match BRD...)

[msarett, 2015/10/21 22:00:00]

Done.

+    // and the image dimensions.
+    // If inputX is negative, we will need to place decoded pixels into the
+    // output bitmap starting at a left offset.  Call this outX.
+    // If outX is non-zero, subsetX must be zero.
+    // If inputY is negative, we will need to place decoded pixels into the
+    // output bitmap starting at a top offset.  Call this outY.
+    // If outY is non-zero, subsetY must be zero.
+    int outX;
+    int outY;
+    SkIRect subset = SkIRect::MakeXYWH(inputX, inputY, inputWidth, inputHeight);
+    SubsetType type = adjust_subset_rect(fCodec->getInfo().dimensions(), &subset, &outX, &outY);
+    if (SubsetType::kInvalid_SubsetType == type) {
+        return NULL;
+    }
+
+    // Ask the codec for a scaled subset
+    if (!fCodec->getSupportedSubset(&subset)) {
+        SkCodecPrintf("Error: Could not get subset.\n");
+        return nullptr;
+    }
+    SkISize scaledSize = fCodec->getSampledSubsetDimensions(sampleSize, subset);
+
+    // Create the image info for the decode
+    SkAlphaType dstAlphaType = fCodec->getInfo().alphaType();
+    if (kUnpremul_SkAlphaType == dstAlphaType) {
+        dstAlphaType = kPremul_SkAlphaType;
+    }
+    SkImageInfo decodeInfo = SkImageInfo::Make(scaledSize.width(), scaledSize.height(),
+            dstColorType, dstAlphaType);
+
+    // Construct a color table for the decode if necessary
+    SkAutoTUnref<SkColorTable> colorTable(nullptr);
+    SkPMColor* colorPtr = nullptr;
+    int* colorCountPtr = nullptr;
+    int maxColors = 256;
+    SkPMColor colors[maxColors];
+    if (kIndex_8_SkColorType == dstColorType) {
+        // TODO (msarett): This performs a copy that is unnecessary since
+        //                 we have not yet initialized the color table.
+        //                 And then we need to use a const cast to get
+        //                 a pointer to the color table that we can
+        //                 modify during the decode.  Is there a better
+        //                 way to do this?  Can we allocate memory to
+        //                 decode into and then create the bitmap after

[scroggo, 2015/10/21 20:56:20]

We certainly can. Using an SkBitmap early on is just a convenience. Instead, we
would need to call SkImageInfo::getSize(SkImageInfo::minRowBytes()) to determine
how much memory to allocate.

[msarett, 2015/10/21 22:00:00]

Acknowledged.

+        //                 the decode?  Can we create an SkColorTable
+        //                 without copying the colors?
+        colorTable.reset(new SkColorTable(colors, maxColors));
+        colorPtr = const_cast<SkPMColor*>(colorTable->readColors());
+        colorCountPtr = &maxColors;
+    }
+
+    // Initialize the destination bitmap
+    SkAutoTDelete<SkBitmap> bitmap(new SkBitmap());
+    int scaledOutX = 0;
+    int scaledOutY = 0;
+    int scaledOutWidth = scaledSize.width();
+    int scaledOutHeight = scaledSize.height();
+    if (SubsetType::kPartiallyInside_SubsetType == type) {
+        scaledOutX = outX / sampleSize;
+        scaledOutY = outY / sampleSize;
+        // We need to be safe here because getSupportedSubset() may have modified the subset.
+        const int extraX = SkTMax(0, inputWidth - outX - subset.width());
+        const int extraY = SkTMax(0, inputHeight - outY - subset.height());
+        const int scaledExtraX = extraX / sampleSize;
+        const int scaledExtraY = extraY / sampleSize;
+        scaledOutWidth += scaledOutX + scaledExtraX;
+        scaledOutHeight += scaledOutY + scaledExtraY;
+    }
+    SkImageInfo outInfo = decodeInfo.makeWH(scaledOutWidth, scaledOutHeight);
+    if (!bitmap->tryAllocPixels(outInfo, nullptr, colorTable.get())) {
+        SkCodecPrintf("Error: Could not allocate pixels.\n");
+        return nullptr;
+    }
+
+    // Zero the bitmap if the region is not completely within the image.
+    // TODO (msarett): Can we make this faster by implementing it to only
+    //                 zero parts of the image that we won't overwrite with
+    //                 pixels?
+    // TODO (msarett): This could be skipped if memory is zero initialized.
+    //                 This would matter if this code is moved to Android and
+    //                 uses Android bitmaps.
+    if (SubsetType::kPartiallyInside_SubsetType == type) {
+        bitmap->eraseColor(0);

[scroggo, 2015/10/21 20:56:20]

Won't this not work for kIndex_8?

[msarett, 2015/10/21 22:00:00]

Ah yes...
Will fix.

+    }
+
+    // Decode into the destination bitmap
+    SkAndroidCodec::AndroidOptions options;
+    options.fSampleSize = sampleSize;
+    options.fSubset = ⊂
+    options.fColorPtr = colorPtr;
+    options.fColorCount = colorCountPtr;
+    void* dst = bitmap->getAddr(scaledOutX, scaledOutY);
+    size_t rowBytes = bitmap->rowBytes();
+    SkCodec::Result result = fCodec->getAndroidPixels(decodeInfo, dst, rowBytes, &options);
+    if (SkCodec::kSuccess != result && SkCodec::kIncompleteInput != result) {
+        SkCodecPrintf("Error: Could not get pixels.\n");
+        return nullptr;
+    }
+
+    return bitmap.detach();
+}
+
+bool SkBitmapRegionCodec::conversionSupported(SkColorType colorType) {
+    // FIXME: Call virtual function when it lands.
+    SkImageInfo info = SkImageInfo::Make(0, 0, colorType, fCodec->getInfo().alphaType(),
+            fCodec->getInfo().profileType());
+    return conversion_possible(info, fCodec->getInfo());
+}