components/dom_distiller/core/distiller.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_H_
   6 #define COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_H_
   7
   8 #include <map>
   9 #include <string>
  10
  11 #include "base/callback.h"
  12 #include "base/containers/hash_tables.h"
  13 #include "base/memory/ref_counted.h"
  14 #include "base/memory/scoped_ptr.h"
  15 #include "base/memory/scoped_vector.h"
  16 #include "base/memory/weak_ptr.h"
  17 #include "components/dom_distiller/core/article_distillation_update.h"
  18 #include "components/dom_distiller/core/distiller_page.h"
  19 #include "components/dom_distiller/core/distiller_url_fetcher.h"
  20 #include "components/dom_distiller/core/proto/distilled_article.pb.h"
  21 #include "net/url_request/url_request_context_getter.h"
  22 #include "url/gurl.h"
  23
  24 namespace dom_distiller {
  25
  26 class DistillerImpl;
  27
  28 class Distiller {
  29  public:
  30   typedef base::Callback<void(scoped_ptr<DistilledArticleProto>)>
  31       DistillationFinishedCallback;
  32   typedef base::Callback<void(const ArticleDistillationUpdate&)>
  33       DistillationUpdateCallback;
  34
  35   virtual ~Distiller() {}
  36
  37   // Distills a page, and asynchronously returns the article HTML to the
  38   // supplied |finished_cb| callback. |update_cb| is invoked whenever article
  39   // under distillation is updated with more data.
  40   // E.g. when distilling a 2 page article, |update_cb| may be invoked each time
  41   // a distilled page is added and |finished_cb| will be invoked once
  42   // distillation is completed.
  43   virtual void DistillPage(const GURL& url,
  44                            scoped_ptr<DistillerPage> distiller_page,
  45                            const DistillationFinishedCallback& finished_cb,
  46                            const DistillationUpdateCallback& update_cb) = 0;
  47 };
  48
  49 class DistillerFactory {
  50  public:
  51   virtual scoped_ptr<Distiller> CreateDistillerForUrl(const GURL& url) = 0;
  52   virtual ~DistillerFactory() {}
  53 };
  54
  55 // Factory for creating a Distiller.
  56 class DistillerFactoryImpl : public DistillerFactory {
  57  public:
  58   DistillerFactoryImpl(
  59       scoped_ptr<DistillerURLFetcherFactory> distiller_url_fetcher_factory,
  60       const dom_distiller::proto::DomDistillerOptions& dom_distiller_options);
  61   ~DistillerFactoryImpl() override;
  62   scoped_ptr<Distiller> CreateDistillerForUrl(const GURL& url) override;
  63
  64  private:
  65   scoped_ptr<DistillerURLFetcherFactory> distiller_url_fetcher_factory_;
  66   dom_distiller::proto::DomDistillerOptions dom_distiller_options_;
  67 };
  68
  69 // Distills a article from a page and associated pages.
  70 class DistillerImpl : public Distiller {
  71  public:
  72   DistillerImpl(
  73       const DistillerURLFetcherFactory& distiller_url_fetcher_factory,
  74       const dom_distiller::proto::DomDistillerOptions& dom_distiller_options);
  75   ~DistillerImpl() override;
  76
  77   void DistillPage(const GURL& url,
  78                    scoped_ptr<DistillerPage> distiller_page,
  79                    const DistillationFinishedCallback& finished_cb,
  80                    const DistillationUpdateCallback& update_cb) override;
  81
  82   void SetMaxNumPagesInArticle(size_t max_num_pages);
  83
  84  private:
  85   // In case of multiple pages, the Distiller maintains state of multiple pages
  86   // as page numbers relative to the page number where distillation started.
  87   // E.g. if distillation starts at page 2 for a 3 page article. The relative
  88   // page numbers assigned to pages will be [-1,0,1].
  89
  90   // Class representing the state of a page under distillation.
  91   struct DistilledPageData {
  92     DistilledPageData();
  93     virtual ~DistilledPageData();
  94     // Relative page number of the page.
  95     int page_num;
  96     ScopedVector<DistillerURLFetcher> image_fetchers_;
  97     scoped_refptr<base::RefCountedData<DistilledPageProto> >
  98         distilled_page_proto;
  99
 100    private:
 101     DISALLOW_COPY_AND_ASSIGN(DistilledPageData);
 102   };
 103
 104   void OnFetchImageDone(int page_num,
 105                         DistillerURLFetcher* url_fetcher,
 106                         const std::string& id,
 107                         const std::string& original_url,
 108                         const std::string& response);
 109
 110   void OnPageDistillationFinished(
 111       int page_num,
 112       const GURL& page_url,
 113       scoped_ptr<proto::DomDistillerResult> distilled_page,
 114       bool distillation_successful);
 115
 116   virtual void FetchImage(int page_num,
 117                           const std::string& image_id,
 118                           const std::string& image_url);
 119
 120   // Distills the next page.
 121   void DistillNextPage();
 122
 123   // Adds the |url| to |pages_to_be_distilled| if |page_num| is a valid relative
 124   // page number and |url| is valid. Ignores duplicate pages and urls.
 125   void AddToDistillationQueue(int page_num, const GURL& url);
 126
 127   // Check if |page_num| is a valid relative page number, i.e. page with
 128   // |page_num| is either under distillation or has already completed
 129   // distillation.
 130   bool IsPageNumberInUse(int page_num) const;
 131
 132   bool AreAllPagesFinished() const;
 133
 134   // Total number of pages in the article that the distiller knows of, this
 135   // includes pages that are pending distillation.
 136   size_t TotalPageCount() const;
 137
 138   // Runs |finished_cb_| if all distillation callbacks and image fetches are
 139   // complete.
 140   void RunDistillerCallbackIfDone();
 141
 142   // Checks if page |distilled_page_data| has finished distillation, including
 143   // all image fetches.
 144   void AddPageIfDone(int page_num);
 145
 146   DistilledPageData* GetPageAtIndex(size_t index) const;
 147
 148   // Create an ArticleDistillationUpdate for the current distillation
 149   // state.
 150   const ArticleDistillationUpdate CreateDistillationUpdate() const;
 151
 152   const DistillerURLFetcherFactory& distiller_url_fetcher_factory_;
 153   scoped_ptr<DistillerPage> distiller_page_;
 154
 155   dom_distiller::proto::DomDistillerOptions dom_distiller_options_;
 156   DistillationFinishedCallback finished_cb_;
 157   DistillationUpdateCallback update_cb_;
 158
 159   // Set of pages that are under distillation or have finished distillation.
 160   // |started_pages_index_| and |finished_pages_index_| maintains the mapping
 161   // from page number to the indices in |pages_|.
 162   ScopedVector<DistilledPageData> pages_;
 163
 164   // Maps page numbers of finished pages to the indices in |pages_|.
 165   std::map<int, size_t> finished_pages_index_;
 166
 167   // Maps page numbers of pages under distillation to the indices in |pages_|.
 168   // If a page is |started_pages_| that means it is still waiting for an action
 169   // (distillation or image fetch) to finish.
 170   base::hash_map<int, size_t> started_pages_index_;
 171
 172   // The list of pages that are still waiting for distillation to start.
 173   // This is a map, to make distiller prefer distilling lower page numbers
 174   // first.
 175   std::map<int, GURL> waiting_pages_;
 176
 177   // Set to keep track of which urls are already seen by the distiller. Used to
 178   // prevent distiller from distilling the same url twice.
 179   base::hash_set<std::string> seen_urls_;
 180
 181   size_t max_pages_in_article_;
 182
 183   bool destruction_allowed_;
 184
 185   base::WeakPtrFactory<DistillerImpl> weak_factory_;
 186
 187   DISALLOW_COPY_AND_ASSIGN(DistillerImpl);
 188 };
 189
 190 }  // namespace dom_distiller
 191
 192 #endif  // COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_H_