Add a new set of page features for distillability testing

components/dom_distiller/core/page_features.cc

 // 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 "base/json/json_reader.h"
+#include "third_party/WebKit/public/platform/WebDistillability.h"
 #include "third_party/re2/re2/re2.h"
+#include "url/gurl.h"
 
 namespace dom_distiller {
 /* This code needs to derive features in the same way and order in which they
  * are derived when training the model. Parts of that code are reproduced in the
  * comments below.
  */
 
 namespace {
 
 std::string GetLastSegment(const std::string& path) {
@@ skipping 151 matching lines @@
   GURL parsed_url(url);
   if (!parsed_url.is_valid()) {
     return std::vector<double>();
   }
 
   return CalculateDerivedFeatures(isOGArticle, parsed_url, numElements,
                                   numAnchors, numForms, innerText, textContent,
                                   innerHTML);
 }
 
+std::vector<double> CalculateDerivedFeatures(
+    const blink::WebDistillabilityFeatures& f,
+    const GURL& url) {
+  const std::string& path = url.path();
+  std::vector<double> features;
+  // 'opengraph', opengraph,
+  features.push_back(f.openGraph);
+  // 'forum', 'forum' in path,
+  features.push_back(Contains("forum", path));
+  // 'index', 'index' in path,
+  features.push_back(Contains("index", path));
+  // 'search', 'search' in path,
+  features.push_back(Contains("search", path));
+  // 'view', 'view' in path,
+  features.push_back(Contains("view", path));
+  // 'archive', 'archive' in path,
+  features.push_back(Contains("archive", path));
+  // 'asp', '.asp' in path,
+  features.push_back(Contains(".asp", path));
+  // 'phpbb', 'phpbb' in path,
+  features.push_back(Contains("phpbb", path));
+  // 'php', path.endswith('.php'),
+  features.push_back(EndsWith(".php", path));
+  // 'pathLength', len(path),
+  features.push_back(path.size());
+  // 'domain', len(path) < 2,
+  features.push_back(path.size() < 2);
+  // 'pathComponents', CountMatches(path, r'\/.'),
+  features.push_back(CountMatches(path, "\\/."));
+  // 'slugDetector', CountMatches(path, r'[^\w/]'),
+  features.push_back(CountMatches(path, "[^\\w/]"));
+  // 'pathNumbers', CountMatches(path, r'\d+'),
+  features.push_back(CountMatches(path, "\\d+"));
+  // 'lastSegmentLength', len(GetLastSegment(path)),
+  features.push_back(GetLastSegment(path).size());
+  // 'formCount', numForms,
+  features.push_back(f.formCount);
+  // 'anchorCount', numAnchors,
+  features.push_back(f.anchorCount);
+  // 'elementCount', numElements,
+  features.push_back(f.elementCount);
+  // 'anchorRatio', float(numAnchors) / max(1, numElements),
+  features.push_back(
+      double(f.anchorCount) / std::max<double>(1, f.elementCount));
+  // 'mozScore'
+  features.push_back(f.mozScore);
+  // 'mozScoreAllSqrt'
+  features.push_back(f.mozScoreAllSqrt);
+  // 'mozScoreAllLinear'
+  features.push_back(f.mozScoreAllLinear);
+
+  return features;
+}
+
 }  // namespace dom_distiller