Add support for providing an external file for extracting content.

components/dom_distiller/standalone/content_extractor.cc

Index: components/dom_distiller/standalone/content_extractor.cc
diff --git a/components/dom_distiller/standalone/content_extractor.cc b/components/dom_distiller/standalone/content_extractor.cc
index 6c0e75bd5884391e189dc628cddddaf1b7a2fea8..d093f3ac69dc3d68e54f1d04c8de89014764fa1d 100644
--- a/components/dom_distiller/standalone/content_extractor.cc
+++ b/components/dom_distiller/standalone/content_extractor.cc
@@ -5,6 +5,8 @@
 #include <sstream>
 
 #include "base/command_line.h"
+#include "base/files/file_path.h"
+#include "base/files/file_util.h"
 #include "base/files/scoped_temp_dir.h"
 #include "base/id_map.h"
 #include "base/message_loop/message_loop.h"
@@ -109,6 +111,11 @@ const char* kOriginalDomains = "original-domains";
 // Maximum number of concurrent started extractor requests.
 const int kMaxExtractorTasks = 8;
 
+// A path to a script for extracting content (domdistiller.js). If this argument
+// is passed in, the script will be used instead of using the bundled version
+// of the script.
+const char* kExternalDomDistillerJs = "external-dom-distiller-js";
+
 scoped_ptr<DomDistillerService> CreateDomDistillerService(
     content::BrowserContext* context,
     const base::FilePath& db_path,
@@ -125,8 +132,24 @@ scoped_ptr<DomDistillerService> CreateDomDistillerService(
   scoped_ptr<DomDistillerStore> dom_distiller_store(
       new DomDistillerStore(db.Pass(), db_path));
 
-  scoped_ptr<DistillerPageFactory> distiller_page_factory(
-      new DistillerPageWebContentsFactory(context));
+  scoped_ptr<DistillerPageFactory> distiller_page_factory;
+  if (base::CommandLine::ForCurrentProcess()->HasSwitch(
+          kExternalDomDistillerJs)) {
+    std::string external_script_path =
+        base::CommandLine::ForCurrentProcess()->GetSwitchValueASCII(
+            kExternalDomDistillerJs);
+    std::string script_content;
+    if (!base::ReadFileToString(base::FilePath(external_script_path),
+                                &script_content)) {
+      ADD_FAILURE() << "Failed to read external script for distillation.";
+      return nullptr;
+    }
+    distiller_page_factory.reset(
+        new DistillerPageWebContentsFactory(context, script_content));
+  } else {
+    distiller_page_factory.reset(new DistillerPageWebContentsFactory(context));
+  }
+
   scoped_ptr<DistillerURLFetcherFactory> distiller_url_fetcher_factory(
       new DistillerURLFetcherFactory(context->GetRequestContext()));