Add calculation of derived features for distillable page model

components/dom_distiller/core/page_features_unittest.cc

Index: components/dom_distiller/core/page_features_unittest.cc
diff --git a/components/dom_distiller/core/page_features_unittest.cc b/components/dom_distiller/core/page_features_unittest.cc
new file mode 100644
index 0000000000000000000000000000000000000000..1bc59f2070006bcc4f40b661f3e9b0cb56da160c
--- /dev/null
+++ b/components/dom_distiller/core/page_features_unittest.cc
@@ -0,0 +1,90 @@
+// Copyright 2015 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 "components/dom_distiller/core/page_features.h"
+
+#include <string>
+#include <vector>
+
+#include "base/files/file_util.h"
+#include "base/json/json_reader.h"
+#include "base/memory/scoped_ptr.h"
+#include "base/path_service.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+namespace dom_distiller {
+
+// This test uses input data of core features and the output of the training
+// pipeline's derived feature extraction to ensure that the extraction that is
+// done in Chromium matches that in the training pipeline.
+TEST(DomDistillerPageFeaturesTest, TestCalculateDerivedFeatures) {
+  base::FilePath dir_source_root;
+  EXPECT_TRUE(PathService::Get(base::DIR_SOURCE_ROOT, &dir_source_root));
+  std::string input_data;
+  ASSERT_TRUE(base::ReadFileToString(
+      dir_source_root.AppendASCII(
+          "components/test/data/dom_distiller/core_features.json"),
+      &input_data));
+  std::string expected_output_data;
+  // This file contains the output from the calculation of derived features in
+  // the training pipeline.
+  ASSERT_TRUE(base::ReadFileToString(
+      dir_source_root.AppendASCII(
+          "components/test/data/dom_distiller/derived_features.json"),
+      &expected_output_data));
+
+  scoped_ptr<base::Value> input_json(base::JSONReader::Read(input_data));
+  ASSERT_TRUE(input_json);
+
+  scoped_ptr<base::Value> expected_output_json(
+      base::JSONReader::Read(expected_output_data));
+  ASSERT_TRUE(expected_output_json);
+
+  base::ListValue* input_entries;
+  ASSERT_TRUE(input_json->GetAsList(&input_entries));
+  ASSERT_GT(input_entries->GetSize(), 0u);
+
+  base::ListValue* expected_output_entries;
+  ASSERT_TRUE(expected_output_json->GetAsList(&expected_output_entries));
+  ASSERT_EQ(expected_output_entries->GetSize(), input_entries->GetSize());
+
+  // In the output, the features list is a sequence of labels followed by values
+  // (so labels at even indices, values at odd indices).
+  base::DictionaryValue* entry;
+  base::ListValue* derived_features;
+  ASSERT_TRUE(expected_output_entries->GetDictionary(0, &entry));
+  ASSERT_TRUE(entry->GetList("features", &derived_features));
+  std::vector<std::string> labels;
+  for (size_t i = 0; i < derived_features->GetSize(); i += 2) {
+    std::string label;
+    ASSERT_TRUE(derived_features->GetString(i, &label));
+    labels.push_back(label);
+  }
+
+  for (size_t i = 0; i < input_entries->GetSize(); ++i) {
+    base::DictionaryValue* core_features;
+    ASSERT_TRUE(input_entries->GetDictionary(i, &entry));
+    ASSERT_TRUE(entry->GetDictionary("features", &core_features));
+    std::vector<double> derived(
+        CalculateDerivedFeaturesFromJSON(core_features));
+
+    ASSERT_EQ(labels.size(), derived.size());
+    ASSERT_TRUE(expected_output_entries->GetDictionary(i, &entry));
+    ASSERT_TRUE(entry->GetList("features", &derived_features));
+    for (size_t j = 0; j < derived.size(); ++j) {
+      double expected_value;
+      if (!derived_features->GetDouble(j * 2 + 1, &expected_value)) {

[nyquist, 2015/04/01 01:19:14]

Nit: int value_index = j * 2 + 1;
Use here and below in assert?

[cjhopman, 2015/04/01 20:12:13]

Done.

+        bool bool_value;
+        ASSERT_TRUE(derived_features->GetBoolean(j * 2 + 1, &bool_value));
+        expected_value = bool_value ? 1.0 : 0.0;
+      }
+      std::string entry_url;

[nyquist, 2015/04/01 01:19:14]

Nit: Could this and filling it with the url string be moved outside of the inner
for-loop?

[cjhopman, 2015/04/01 20:12:13]

Done.

+      ASSERT_TRUE(entry->GetString("url", &entry_url));
+      EXPECT_DOUBLE_EQ(derived[j], expected_value)
+          << "incorrect value for entry with url " << entry_url
+          << " for derived feature " << labels[j];
+    }
+  }
+}
+}