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/message_loop/message_loop.h"
   9 #include "base/strings/utf_string_conversions.h"
  10 #include "base/values.h"
  11 #include "components/dom_distiller/core/distillable_page_detector.h"
  12 #include "components/dom_distiller/core/experiments.h"
  13 #include "components/dom_distiller/core/page_features.h"
  14 #include "content/public/browser/render_frame_host.h"
  15 #include "grit/components_resources.h"
  16 #include "ui/base/resource/resource_bundle.h"
  17
  18 namespace dom_distiller {
  19 namespace {
  20 void OnOGArticleJsResult(base::Callback<void(bool)> callback,
  21                          const base::Value* result) {
  22   bool success = false;
  23   if (result) {
  24     result->GetAsBoolean(&success);
  25   }
  26   callback.Run(success);
  27 }
  28
  29 void OnExtractFeaturesJsResult(const DistillablePageDetector* detector,
  30                                base::Callback<void(bool)> callback,
  31                                const base::Value* result) {
  32   callback.Run(detector->Classify(CalculateDerivedFeaturesFromJSON(result)));
  33 }
  34 }  // namespace
  35
  36 void IsOpenGraphArticle(content::WebContents* web_contents,
  37                         base::Callback<void(bool)> callback) {
  38   content::RenderFrameHost* main_frame = web_contents->GetMainFrame();
  39   if (!main_frame) {
  40     base::MessageLoop::current()->PostTask(FROM_HERE,
  41                                            base::Bind(callback, false));
  42     return;
  43   }
  44   std::string og_article_js = ResourceBundle::GetSharedInstance()
  45                                   .GetRawDataResource(IDR_IS_DISTILLABLE_JS)
  46                                   .as_string();
  47   main_frame->ExecuteJavaScript(base::UTF8ToUTF16(og_article_js),
  48                                 base::Bind(OnOGArticleJsResult, callback));
  49 }
  50
  51 void IsDistillablePage(content::WebContents* web_contents,
  52                        base::Callback<void(bool)> callback) {
  53   switch (GetDistillerHeuristicsType()) {
  54     case DistillerHeuristicsType::NONE:
  55       base::MessageLoop::current()->PostTask(FROM_HERE,
  56                                              base::Bind(callback, false));
  57       return;
  58     case DistillerHeuristicsType::OG_ARTICLE:
  59       IsOpenGraphArticle(web_contents, callback);
  60       return;
  61     case DistillerHeuristicsType::ADABOOST_MODEL:
  62       IsDistillablePageForDetector(
  63           web_contents, DistillablePageDetector::GetDefault(), callback);
  64       return;
  65   }
  66 }
  67
  68 void IsDistillablePageForDetector(content::WebContents* web_contents,
  69                                   const DistillablePageDetector* detector,
  70                                   base::Callback<void(bool)> callback) {
  71   content::RenderFrameHost* main_frame = web_contents->GetMainFrame();
  72   if (!main_frame) {
  73     base::MessageLoop::current()->PostTask(FROM_HERE,
  74                                            base::Bind(callback, false));
  75     return;
  76   }
  77   std::string extract_features_js =
  78       ResourceBundle::GetSharedInstance()
  79           .GetRawDataResource(IDR_EXTRACT_PAGE_FEATURES_JS)
  80           .as_string();
  81   main_frame->ExecuteJavaScript(
  82       base::UTF8ToUTF16(extract_features_js),
  83       base::Bind(OnExtractFeaturesJsResult, detector, callback));
  84 }
  85
  86 }  // namespace dom_distiller