Shape Detection: Add two content browser tests for face detection

content/browser/shapedetection/shapedetection_browsertest.cc

Index: content/browser/shapedetection/shapedetection_browsertest.cc
diff --git a/content/browser/shapedetection/shapedetection_browsertest.cc b/content/browser/shapedetection/shapedetection_browsertest.cc
new file mode 100644
index 0000000000000000000000000000000000000000..14ab374a8c3794460397e1e7789fb36cb31ae45b
--- /dev/null
+++ b/content/browser/shapedetection/shapedetection_browsertest.cc
@@ -0,0 +1,68 @@
+// Copyright 2016 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "base/command_line.h"
+#include "content/public/common/content_switches.h"
+#include "content/public/test/browser_test_utils.h"
+#include "content/public/test/content_browser_test.h"
+#include "content/public/test/content_browser_test_utils.h"
+
+namespace content {
+
+namespace {
+
+const char kFaceDetectionTestHtml[] = "/media/face_detection_test.html";
+
+}  // namespace
+
+// This class contains content_browsertests for Shape Detection API, which
+// allows for generating bounding boxes of face for a still image.

[mcasas, 2016/10/18 23:15:13]

... for generating bounding boxes for faces on still images.
?

[xianglu, 2016/10/19 23:26:10]

Done.

+class ShapeDetectionBrowserTest : public ContentBrowserTest {
+ public:
+  void SetUpCommandLine(base::CommandLine* command_line) override {
+    // Specific flag to enable ShapeDetection and DOMRect API.
+    base::CommandLine::ForCurrentProcess()->AppendSwitchASCII(
+        switches::kEnableBlinkFeatures, "ShapeDetection, GeometryInterfaces");
+  }
+
+ protected:
+  void RunDetectFacesOnImageUrl(const std::string image_path,
+                                const std::string expected_results) {

[mcasas, 2016/10/18 23:15:13]

const std::string& (both parameters).

[xianglu, 2016/10/19 23:26:10]

Done.

+    ASSERT_TRUE(embedded_test_server()->Start());
+
+    GURL html_url(embedded_test_server()->GetURL(kFaceDetectionTestHtml));
+    GURL image_url(embedded_test_server()->GetURL(image_path));
+    NavigateToURL(shell(), html_url);
+    std::string js_command =

[mcasas, 2016/10/18 23:15:13]

const, here and in l. 34, 35 and 37.

[xianglu, 2016/10/19 23:26:10]

Done.

+        "detectFacesOnImageUrl('" + image_url.spec() + "')";
+    std::string results;
+    ASSERT_TRUE(ExecuteScriptAndExtractString(shell(), js_command, &results));
+    ASSERT_EQ(expected_results, results);
+  }
+};
+
+#if defined(OS_ANDROID)
+#define MAYBE_DetectFacesOnImageWithNoFaces DetectFacesOnImageWithNoFaces
+#else
+#define MAYBE_DetectFacesOnImageWithNoFaces DISABLED_DetectFacesOnImageWithNoFaces
+#endif

[mcasas, 2016/10/18 23:15:13]

Bundle these l.45-49 and l.56-60 around l.10
and add a reference to the bug we're using for
development in a comment, e.g.:
// TODO(xianglu): Enable other platforms with support: https://crbug.com/646803.

[xianglu, 2016/10/19 23:26:10]

Done.

+IN_PROC_BROWSER_TEST_F(ShapeDetectionBrowserTest,
+                       MAYBE_DetectFacesOnImageWithNoFaces) {
+  const std::string image_path = "/blank.jpg";
+  RunDetectFacesOnImageUrl(image_path, "");
+}
+
+#if defined(OS_ANDROID)
+#define MAYBE_DetectFacesOnImageWithOneFace DetectFacesOnImageWithOneFace
+#else
+#define MAYBE_DetectFacesOnImageWithOneFace DISABLED_DetectFacesOnImageWithOneFace
+#endif
+IN_PROC_BROWSER_TEST_F(ShapeDetectionBrowserTest,
+                       MAYBE_DetectFacesOnImageWithOneFace) {
+  const std::string image_path = "/single_face.jpg";
+  RunDetectFacesOnImageUrl(image_path,
+                           "x=68.640625,y=102.96875,w=171.5625,h=171.5625#");

[mcasas, 2016/10/18 23:15:13]

Hmm I wonder if this would not produce failures due to 
numerical accuracy, i.e. perhaps device A says x=68.6...
and device B says x=68.7...

It's OK if you don't address this concern in this particular CL,
but chances are that MAC would return something slightly 
different, so keep that in mind, and optimally add a comment
TODO detailing that maybe in the future you'll need to pass 
the points as an array, sth like:
// TODO(xianglu): consider passing the coordinates as an
// array to allow for approximate comparisons.

[xianglu, 2016/10/19 23:26:10]

Done.

+}
+
+}  // namespace content