Add UYVYToY function

This function reads 2 byte values and writes the 2nd byte to the destination.
It turns out this is useful for P010ToNV12 as well, so adding the planar function allows a high level to call this.
And adds UYVY support for something YUY2 already had.  Which is writing the 1st byte.

Bug: b/233233302, b/233634772
Change-Id: I10a9454cb4f5b2c4ac5532fa86feddf78284d8b8
Reviewed-on: https://chromium-review.googlesource.com/c/libyuv/libyuv/+/3659055
Commit-Queue: Frank Barchard <fbarchard@chromium.org>
Reviewed-by: richard winterton <rrwinterton@gmail.com>
Reviewed-by: Frank Barchard <fbarchard@chromium.org>

7 files changed, 164 insertions, 7 deletions

diff --git a/BUILD.gn b/BUILD.gn
index a72ff065..2c9c3e6e 100644
--- a/BUILD.gn
+++ b/BUILD.gn
@@ -27,10 +27,6 @@ config("libyuv_config") {
   if (is_android && current_cpu != "arm64") {
     ldflags = [ "-Wl,--dynamic-linker,/system/bin/linker" ]
   }
-
-  if (!libyuv_use_neon) {
-    defines = [ "LIBYUV_DISABLE_NEON" ]
-  }
 }
 
 # This target is built when no specific target is specified on the command line.
diff --git a/OWNERS b/OWNERS
index a96669f9..66bb995f 100644
--- a/OWNERS
+++ b/OWNERS
@@ -1,7 +1,6 @@
 mbonadei@chromium.org
 fbarchard@chromium.org
 magjed@chromium.org
-pbos@chromium.org
 
 per-file *.gn=mbonadei@chromium.org
 per-file .gitignore=*
diff --git a/README.chromium b/README.chromium
index 76b54121..b6d85b73 100644
--- a/README.chromium
+++ b/README.chromium
@@ -1,6 +1,6 @@
 Name: libyuv
 URL: http://code.google.com/p/libyuv/
-Version: 1823
+Version: 1824
 License: BSD
 License File: LICENSE
 
diff --git a/include/libyuv/planar_functions.h b/include/libyuv/planar_functions.h
index 972ca9e3..15c7d457 100644
--- a/include/libyuv/planar_functions.h
+++ b/include/libyuv/planar_functions.h
@@ -461,6 +461,14 @@ int YUY2ToY(const uint8_t* src_yuy2,
             int width,
             int height);
 
+LIBYUV_API
+int UYVYToY(const uint8_t* src_uyvy,
+            int src_stride_uyvy,
+            uint8_t* dst_y,
+            int dst_stride_y,
+            int width,
+            int height);
+
 // Convert I420 to I400. (calls CopyPlane ignoring u/v).
 LIBYUV_API
 int I420ToI400(const uint8_t* src_y,
diff --git a/include/libyuv/version.h b/include/libyuv/version.h
index d0b94fee..8afef0ed 100644
--- a/include/libyuv/version.h
+++ b/include/libyuv/version.h
@@ -11,6 +11,6 @@
 #ifndef INCLUDE_LIBYUV_VERSION_H_
 #define INCLUDE_LIBYUV_VERSION_H_
 
-#define LIBYUV_VERSION 1823
+#define LIBYUV_VERSION 1824
 
 #endif  // INCLUDE_LIBYUV_VERSION_H_
diff --git a/source/planar_functions.cc b/source/planar_functions.cc
index c698b01a..53e74679 100644
--- a/source/planar_functions.cc
+++ b/source/planar_functions.cc
@@ -2053,6 +2053,73 @@ int YUY2ToY(const uint8_t* src_yuy2,
   return 0;
 }
 
+// Convert UYVY to Y.
+LIBYUV_API
+int UYVYToY(const uint8_t* src_uyvy,
+            int src_stride_uyvy,
+            uint8_t* dst_y,
+            int dst_stride_y,
+            int width,
+            int height) {
+  int y;
+  void (*UYVYToYRow)(const uint8_t* src_uyvy, uint8_t* dst_y, int width) =
+      UYVYToYRow_C;
+  if (!src_uyvy || !dst_y || width <= 0 || height == 0) {
+    return -1;
+  }
+  // Negative height means invert the image.
+  if (height < 0) {
+    height = -height;
+    src_uyvy = src_uyvy + (height - 1) * src_stride_uyvy;
+    src_stride_uyvy = -src_stride_uyvy;
+  }
+  // Coalesce rows.
+  if (src_stride_uyvy == width * 2 && dst_stride_y == width) {
+    width *= height;
+    height = 1;
+    src_stride_uyvy = dst_stride_y = 0;
+  }
+#if defined(HAS_UYVYTOYROW_SSE2)
+  if (TestCpuFlag(kCpuHasSSE2)) {
+    UYVYToYRow = UYVYToYRow_Any_SSE2;
+    if (IS_ALIGNED(width, 16)) {
+      UYVYToYRow = UYVYToYRow_SSE2;
+    }
+  }
+#endif
+#if defined(HAS_UYVYTOYROW_AVX2)
+  if (TestCpuFlag(kCpuHasAVX2)) {
+    UYVYToYRow = UYVYToYRow_Any_AVX2;
+    if (IS_ALIGNED(width, 32)) {
+      UYVYToYRow = UYVYToYRow_AVX2;
+    }
+  }
+#endif
+#if defined(HAS_UYVYTOYROW_NEON)
+  if (TestCpuFlag(kCpuHasNEON)) {
+    UYVYToYRow = UYVYToYRow_Any_NEON;
+    if (IS_ALIGNED(width, 16)) {
+      UYVYToYRow = UYVYToYRow_NEON;
+    }
+  }
+#endif
+#if defined(HAS_UYVYTOYROW_MSA)
+  if (TestCpuFlag(kCpuHasMSA)) {
+    UYVYToYRow = UYVYToYRow_Any_MSA;
+    if (IS_ALIGNED(width, 32)) {
+      UYVYToYRow = UYVYToYRow_MSA;
+    }
+  }
+#endif
+
+  for (y = 0; y < height; ++y) {
+    UYVYToYRow(src_uyvy, dst_y, width);
+    src_uyvy += src_stride_uyvy;
+    dst_y += dst_stride_y;
+  }
+  return 0;
+}
+
 // Mirror a plane of data.
 // See Also I400Mirror
 LIBYUV_API
diff --git a/unit_test/planar_test.cc b/unit_test/planar_test.cc
index 76fec3ec..833091a1 100644
--- a/unit_test/planar_test.cc
+++ b/unit_test/planar_test.cc
@@ -3463,6 +3463,64 @@ TEST_F(LibYUVPlanarTest, Convert16To8Plane) {
   free_aligned_buffer_page_end(dst_pixels_y_c);
 }
 
+TEST_F(LibYUVPlanarTest, YUY2ToY) {
+  const int kPixels = benchmark_width_ * benchmark_height_;
+  align_buffer_page_end(src_pixels_y, kPixels * 2);
+  align_buffer_page_end(dst_pixels_y_opt, kPixels);
+  align_buffer_page_end(dst_pixels_y_c, kPixels);
+
+  MemRandomize(src_pixels_y, kPixels * 2);
+  memset(dst_pixels_y_opt, 0, kPixels);
+  memset(dst_pixels_y_c, 1, kPixels);
+
+  MaskCpuFlags(disable_cpu_flags_);
+  YUY2ToY(src_pixels_y, benchmark_width_ * 2, dst_pixels_y_c,
+          benchmark_width_, benchmark_width_, benchmark_height_);
+  MaskCpuFlags(benchmark_cpu_info_);
+
+  for (int i = 0; i < benchmark_iterations_; ++i) {
+    YUY2ToY(src_pixels_y, benchmark_width_ * 2, dst_pixels_y_opt,
+            benchmark_width_, benchmark_width_, benchmark_height_);
+  }
+
+  for (int i = 0; i < kPixels; ++i) {
+    EXPECT_EQ(dst_pixels_y_opt[i], dst_pixels_y_c[i]);
+  }
+
+  free_aligned_buffer_page_end(src_pixels_y);
+  free_aligned_buffer_page_end(dst_pixels_y_opt);
+  free_aligned_buffer_page_end(dst_pixels_y_c);
+}
+
+TEST_F(LibYUVPlanarTest, UYVYToY) {
+  const int kPixels = benchmark_width_ * benchmark_height_;
+  align_buffer_page_end(src_pixels_y, kPixels * 2);
+  align_buffer_page_end(dst_pixels_y_opt, kPixels);
+  align_buffer_page_end(dst_pixels_y_c, kPixels);
+
+  MemRandomize(src_pixels_y, kPixels * 2);
+  memset(dst_pixels_y_opt, 0, kPixels);
+  memset(dst_pixels_y_c, 1, kPixels);
+
+  MaskCpuFlags(disable_cpu_flags_);
+  UYVYToY(src_pixels_y, benchmark_width_ * 2, dst_pixels_y_c,
+          benchmark_width_, benchmark_width_, benchmark_height_);
+  MaskCpuFlags(benchmark_cpu_info_);
+
+  for (int i = 0; i < benchmark_iterations_; ++i) {
+    UYVYToY(src_pixels_y, benchmark_width_ * 2, dst_pixels_y_opt,
+            benchmark_width_, benchmark_width_, benchmark_height_);
+  }
+
+  for (int i = 0; i < kPixels; ++i) {
+    EXPECT_EQ(dst_pixels_y_opt[i], dst_pixels_y_c[i]);
+  }
+
+  free_aligned_buffer_page_end(src_pixels_y);
+  free_aligned_buffer_page_end(dst_pixels_y_opt);
+  free_aligned_buffer_page_end(dst_pixels_y_c);
+}
+
 #ifdef ENABLE_ROW_TESTS
 // TODO(fbarchard): Improve test for more platforms.
 #ifdef HAS_CONVERT16TO8ROW_AVX2
@@ -3509,6 +3567,35 @@ TEST_F(LibYUVPlanarTest, Convert16To8Row_Opt) {
   free_aligned_buffer_page_end(dst_pixels_y_c);
 }
 #endif  // HAS_CONVERT16TO8ROW_AVX2
+
+#ifdef HAS_UYVYTOYROW_NEON
+TEST_F(LibYUVPlanarTest, UYVYToYRow_Opt) {
+  // NEON does multiple of 16, so round count up
+  const int kPixels = (benchmark_width_ * benchmark_height_ + 15) & ~15;
+  align_buffer_page_end(src_pixels_y, kPixels * 2);
+  align_buffer_page_end(dst_pixels_y_opt, kPixels);
+  align_buffer_page_end(dst_pixels_y_c, kPixels);
+
+  MemRandomize(src_pixels_y, kPixels * 2);
+  memset(dst_pixels_y_opt, 0, kPixels);
+  memset(dst_pixels_y_c, 1, kPixels);
+
+  UYVYToYRow_C(src_pixels_y, dst_pixels_y_c, kPixels);
+
+  for (int i = 0; i < benchmark_iterations_; ++i) {
+    UYVYToYRow_NEON(src_pixels_y, dst_pixels_y_opt, kPixels);
+  }
+
+  for (int i = 0; i < kPixels; ++i) {
+    EXPECT_EQ(dst_pixels_y_opt[i], dst_pixels_y_c[i]);
+  }
+
+  free_aligned_buffer_page_end(src_pixels_y);
+  free_aligned_buffer_page_end(dst_pixels_y_opt);
+  free_aligned_buffer_page_end(dst_pixels_y_c);
+}
+#endif  // HAS_UYVYTOYROW_NEON
+
 #endif  // ENABLE_ROW_TESTS
 
 TEST_F(LibYUVPlanarTest, Convert8To16Plane) {