components/dom_distiller/core/distiller_page.h

   1 // Copyright 2013 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 #ifndef COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_PAGE_H_
   6 #define COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_PAGE_H_
   7
   8 #include <string>
   9
  10 #include "base/callback.h"
  11 #include "base/memory/scoped_ptr.h"
  12 #include "base/memory/weak_ptr.h"
  13 #include "base/values.h"
  14 #include "third_party/dom_distiller_js/dom_distiller.pb.h"
  15 #include "ui/gfx/size.h"
  16 #include "url/gurl.h"
  17
  18 namespace dom_distiller {
  19
  20 struct DistilledPageInfo {
  21   struct MarkupArticle {
  22     std::string published_time;
  23     std::string modified_time;
  24     std::string expiration_time;
  25     std::string section;
  26     std::vector<std::string> authors;
  27
  28     MarkupArticle();
  29     ~MarkupArticle();
  30   };
  31
  32   struct MarkupImage {
  33     std::string url;
  34     std::string secure_url;
  35     std::string type;
  36     std::string caption;
  37     int width;
  38     int height;
  39
  40     MarkupImage();
  41     ~MarkupImage();
  42   };
  43
  44   struct MarkupInfo {
  45     std::string title;
  46     std::string type;
  47     std::string url;
  48     std::string description;
  49     std::string publisher;
  50     std::string copyright;
  51     std::string author;
  52     MarkupArticle article;
  53     std::vector<MarkupImage> images;
  54
  55     MarkupInfo();
  56     ~MarkupInfo();
  57   };
  58
  59   std::string title;
  60   std::string html;
  61   std::string next_page_url;
  62   std::string prev_page_url;
  63   std::vector<std::string> image_urls;
  64   MarkupInfo markup_info;
  65
  66   DistilledPageInfo();
  67   ~DistilledPageInfo();
  68
  69  private:
  70   DISALLOW_COPY_AND_ASSIGN(DistilledPageInfo);
  71 };
  72
  73 class SourcePageHandle {
  74  public:
  75   virtual ~SourcePageHandle() {}
  76 };
  77
  78 // Injects JavaScript into a page, and uses it to extract and return long-form
  79 // content. The class can be reused to load and distill multiple pages,
  80 // following the state transitions described along with the class's states.
  81 // Constructing a DistillerPage should be cheap, as some of the instances can be
  82 // thrown away without ever being used.
  83 class DistillerPage {
  84  public:
  85   typedef base::Callback<void(scoped_ptr<DistilledPageInfo> distilled_page,
  86                               bool distillation_successful)>
  87       DistillerPageCallback;
  88
  89   DistillerPage();
  90   virtual ~DistillerPage();
  91
  92   // Loads a URL. |OnDistillationDone| is called when the load completes or
  93   // fails. May be called when the distiller is idle. Callers can assume that,
  94   // for a given |url| and |options|, any DistillerPage implementation will
  95   // extract the same content.
  96   void DistillPage(const GURL& url,
  97                    const dom_distiller::proto::DomDistillerOptions options,
  98                    const DistillerPageCallback& callback);
  99
 100   // Called when the JavaScript execution completes. |page_url| is the url of
 101   // the distilled page. |value| contains data returned by the script.
 102   virtual void OnDistillationDone(const GURL& page_url,
 103                                   const base::Value* value);
 104
 105  protected:
 106   // Called by |DistillPage| to carry out platform-specific instructions to load
 107   // and distill the |url| using the provided |script|. The extracted content
 108   // should be the same regardless of the DistillerPage implementation.
 109   virtual void DistillPageImpl(const GURL& url, const std::string& script) = 0;
 110
 111  private:
 112   bool ready_;
 113   DistillerPageCallback distiller_page_callback_;
 114   DISALLOW_COPY_AND_ASSIGN(DistillerPage);
 115 };
 116
 117 // Factory for generating a |DistillerPage|.
 118 class DistillerPageFactory {
 119  public:
 120   virtual ~DistillerPageFactory();
 121
 122   // Constructs and returns a new DistillerPage. The implementation of this
 123   // should be very cheap, since the pages can be thrown away without being
 124   // used.
 125   virtual scoped_ptr<DistillerPage> CreateDistillerPage(
 126       const gfx::Size& render_view_size) const = 0;
 127   virtual scoped_ptr<DistillerPage> CreateDistillerPageWithHandle(
 128       scoped_ptr<SourcePageHandle> handle) const = 0;
 129 };
 130
 131 }  // namespace dom_distiller
 132
 133 #endif  // COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_PAGE_H_