Add support for providing an external file for extracting content.

components/dom_distiller/core/distiller_page.h

 // Copyright 2013 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #ifndef COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_PAGE_H_
 #define COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_PAGE_H_
 
 #include <string>
 
 #include "base/callback.h"
@@ skipping 15 matching lines @@
 // content. The class can be reused to load and distill multiple pages,
 // following the state transitions described along with the class's states.
 // Constructing a DistillerPage should be cheap, as some of the instances can be
 // thrown away without ever being used.
 class DistillerPage {
  public:
   typedef base::Callback<
       void(scoped_ptr<proto::DomDistillerResult> distilled_page,
            bool distillation_successful)> DistillerPageCallback;
 
-  DistillerPage();
+  // The passed in JavaScript should contain the string $$OPTIONS, which will be
+  // replaced with options passed in as proto::DomDistillerOptions before it is
+  // executed in the page context.
+  explicit DistillerPage(const std::string& distiller_js_script);
   virtual ~DistillerPage();
 
   // Loads a URL. |OnDistillationDone| is called when the load completes or
   // fails. May be called when the distiller is idle. Callers can assume that,
   // for a given |url| and |options|, any DistillerPage implementation will
   // extract the same content.
   void DistillPage(const GURL& url,
                    const proto::DomDistillerOptions options,
                    const DistillerPageCallback& callback);
 
   // Called when the JavaScript execution completes. |page_url| is the url of
   // the distilled page. |value| contains data returned by the script.
   virtual void OnDistillationDone(const GURL& page_url,
                                   const base::Value* value);
 
  protected:
   // Called by |DistillPage| to carry out platform-specific instructions to load
-  // and distill the |url| using the provided |script|. The extracted content
-  // should be the same regardless of the DistillerPage implementation.
+  // and distill the |url|. The extracted content should be the same regardless
+  // of the DistillerPage implementation.
   virtual void DistillPageImpl(const GURL& url, const std::string& script) = 0;

[noyau (Ping after 24h), 2015/03/11 14:30:59]

The comment should specify that the options have been replaced in the passed on
script.

 
  private:
   bool ready_;
+
+  // Contains the script which will be injected into the page which is
+  // being distilled.  It should contain the string $$OPTIONS, which will be
+  // replaced with options passed in as proto::DomDistillerOptions before
+  // it is executed in the page context.
+  std::string distiller_js_script_;
+
   DistillerPageCallback distiller_page_callback_;
+
   DISALLOW_COPY_AND_ASSIGN(DistillerPage);
 };
 
 // Factory for generating a |DistillerPage|.
 class DistillerPageFactory {
  public:
   virtual ~DistillerPageFactory();
 
   // Constructs and returns a new DistillerPage. The implementation of this
   // should be very cheap, since the pages can be thrown away without being
   // used.
   virtual scoped_ptr<DistillerPage> CreateDistillerPage(
       const gfx::Size& render_view_size) const = 0;
   virtual scoped_ptr<DistillerPage> CreateDistillerPageWithHandle(
       scoped_ptr<SourcePageHandle> handle) const = 0;
 };
 
 }  // namespace dom_distiller
 
 #endif  // COMPONENTS_DOM_DISTILLER_CORE_DISTILLER_PAGE_H_