components/dom_distiller/content/distillable_page_utils.cc

   1 // Copyright 2015 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include "components/dom_distiller/content/distillable_page_utils.h"
   6
   7 #include "base/bind.h"
   8 #include "base/location.h"
   9 #include "base/single_thread_task_runner.h"
  10 #include "base/strings/utf_string_conversions.h"
  11 #include "base/thread_task_runner_handle.h"
  12 #include "base/values.h"
  13 #include "components/dom_distiller/content/distiller_javascript_utils.h"
  14 #include "components/dom_distiller/core/distillable_page_detector.h"
  15 #include "components/dom_distiller/core/experiments.h"
  16 #include "components/dom_distiller/core/page_features.h"
  17 #include "content/public/browser/render_frame_host.h"
  18 #include "grit/components_resources.h"
  19 #include "ui/base/resource/resource_bundle.h"
  20
  21 namespace dom_distiller {
  22 namespace {
  23 void OnOGArticleJsResult(base::Callback<void(bool)> callback,
  24                          const base::Value* result) {
  25   bool success = false;
  26   if (result) {
  27     result->GetAsBoolean(&success);
  28   }
  29   callback.Run(success);
  30 }
  31
  32 void OnExtractFeaturesJsResult(const DistillablePageDetector* detector,
  33                                base::Callback<void(bool)> callback,
  34                                const base::Value* result) {
  35   callback.Run(detector->Classify(CalculateDerivedFeaturesFromJSON(result)));
  36 }
  37 }  // namespace
  38
  39 void IsOpenGraphArticle(content::WebContents* web_contents,
  40                         base::Callback<void(bool)> callback) {
  41   content::RenderFrameHost* main_frame = web_contents->GetMainFrame();
  42   if (!main_frame) {
  43     base::ThreadTaskRunnerHandle::Get()->PostTask(FROM_HERE,
  44                                                   base::Bind(callback, false));
  45     return;
  46   }
  47   std::string og_article_js = ResourceBundle::GetSharedInstance()
  48                                   .GetRawDataResource(IDR_IS_DISTILLABLE_JS)
  49                                   .as_string();
  50   RunIsolatedJavaScript(main_frame, og_article_js,
  51                         base::Bind(OnOGArticleJsResult, callback));
  52 }
  53
  54 void IsDistillablePage(content::WebContents* web_contents,
  55                        bool is_mobile_optimized,
  56                        base::Callback<void(bool)> callback) {
  57   switch (GetDistillerHeuristicsType()) {
  58     case DistillerHeuristicsType::ALWAYS_TRUE:
  59       base::ThreadTaskRunnerHandle::Get()->PostTask(FROM_HERE,
  60                                                     base::Bind(callback, true));
  61       return;
  62     case DistillerHeuristicsType::OG_ARTICLE:
  63       IsOpenGraphArticle(web_contents, callback);
  64       return;
  65     case DistillerHeuristicsType::ADABOOST_MODEL:
  66       // The adaboost model is only applied to non-mobile pages.
  67       if (is_mobile_optimized) {
  68         base::ThreadTaskRunnerHandle::Get()->PostTask(
  69             FROM_HERE, base::Bind(callback, false));
  70         return;
  71       }
  72       IsDistillablePageForDetector(
  73           web_contents, DistillablePageDetector::GetDefault(), callback);
  74       return;
  75     case DistillerHeuristicsType::NONE:
  76     default:
  77       base::ThreadTaskRunnerHandle::Get()->PostTask(
  78           FROM_HERE, base::Bind(callback, false));
  79       return;
  80   }
  81 }
  82
  83 void IsDistillablePageForDetector(content::WebContents* web_contents,
  84                                   const DistillablePageDetector* detector,
  85                                   base::Callback<void(bool)> callback) {
  86   content::RenderFrameHost* main_frame = web_contents->GetMainFrame();
  87   if (!main_frame) {
  88     base::ThreadTaskRunnerHandle::Get()->PostTask(FROM_HERE,
  89                                                   base::Bind(callback, false));
  90     return;
  91   }
  92   std::string extract_features_js =
  93       ResourceBundle::GetSharedInstance()
  94           .GetRawDataResource(IDR_EXTRACT_PAGE_FEATURES_JS)
  95           .as_string();
  96   RunIsolatedJavaScript(
  97       main_frame, extract_features_js,
  98       base::Bind(OnExtractFeaturesJsResult, detector, callback));
  99 }
 100
 101 }  // namespace dom_distiller