Add support for multipage distillation.

components/dom_distiller/core/distiller.h

Index: components/dom_distiller/core/distiller.h
diff --git a/components/dom_distiller/core/distiller.h b/components/dom_distiller/core/distiller.h
index adbd41b8cbb294a99f14f15740e6b40c839d767e..58513281ac7587921ea94b19cbcf2e82c4a7c4db 100644
--- a/components/dom_distiller/core/distiller.h
+++ b/components/dom_distiller/core/distiller.h
@@ -6,6 +6,7 @@
 #define COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_H_
 
 #include <map>
+#include <set>
 #include <string>
 
 #include "base/callback.h"
@@ -14,7 +15,7 @@
 #include "base/values.h"
 #include "components/dom_distiller/core/distiller_page.h"
 #include "components/dom_distiller/core/distiller_url_fetcher.h"
-#include "components/dom_distiller/core/proto/distilled_page.pb.h"
+#include "components/dom_distiller/core/proto/distilled_article.pb.h"
 #include "net/url_request/url_request_context_getter.h"
 #include "url/gurl.h"
 
@@ -24,8 +25,8 @@ class DistillerImpl;
 
 class Distiller {
  public:
-  typedef base::Callback<void(
-      scoped_ptr<DistilledPageProto>)> DistillerCallback;
+  typedef base::Callback<void(scoped_ptr<DistilledArticleProto>)>
+      DistillerCallback;
   virtual ~Distiller() {}
 
   // Distills a page, and asynchrounously returns the article HTML to the
@@ -74,24 +75,43 @@ class DistillerImpl : public Distiller,
   virtual void OnLoadURLDone() OVERRIDE;
   virtual void OnExecuteJavaScriptDone(const base::Value* value) OVERRIDE;
 
-  void OnFetchImageDone(const std::string& id, const std::string& response);
+  void OnFetchImageDone(DistilledPageProto* distilled_page_proto,
+                        const std::string& id,
+                        const std::string& response);
 
  private:
   virtual void LoadURL(const GURL& url);
-  virtual void FetchImage(const std::string& image_id, const std::string& item);
+  virtual void FetchImage(DistilledPageProto* distilled_page_proto,
+                          const std::string& image_id,
+                          const std::string& item);
 
   // Injects JavaScript to distill a loaded page down to its important content,
   // e.g., extracting a news article from its surrounding boilerplate.
   void GetDistilledContent();
 
+  // Adds url as a new page to the |article_proto_| and triggers distillation
+  // for the newly added page.
+  void AddAndDistillPage(const GURL& url);
+
+  // Distills the page.
+  void DistillNextPage(const GURL& url);
+
+  // Runs distillation callback when all distillation tasks are finished.
+  void DistillationTaskComplete();
+
+  // Returns the last page of |article_proto_|.
+  DistilledPageProto* GetLastPage() const;
+
   const DistillerPageFactory& distiller_page_factory_;
   const DistillerURLFetcherFactory& distiller_url_fetcher_factory_;
   scoped_ptr<DistillerPage> distiller_page_;
   DistillerCallback distillation_cb_;
 
   std::map<std::string, DistillerURLFetcher* > image_fetchers_;
-
-  scoped_ptr<DistilledPageProto> proto_;
+  scoped_ptr<DistilledArticleProto> article_proto_;
+  bool distillation_in_progress_;
+  // Set to keep track of which urls are already seen by the distiller.
+  std::set<std::string> processed_urls_;
 
   DISALLOW_COPY_AND_ASSIGN(DistillerImpl);
 };