Index: components/dom_distiller/core/page_features.h |
diff --git a/components/dom_distiller/core/page_features.h b/components/dom_distiller/core/page_features.h |
index 236796b542aecb860e3cb03311872183e50aec4f..cdab239191412081164e3c71f59445e3f12ff1a8 100644 |
--- a/components/dom_distiller/core/page_features.h |
+++ b/components/dom_distiller/core/page_features.h |
@@ -7,35 +7,27 @@ |
#include <vector> |
-#include "base/values.h" |
-#include "url/gurl.h" |
+class GURL; |
+ |
+namespace blink { |
+ struct WebDistillabilityFeatures; |
+} |
namespace dom_distiller { |
// The length of the derived features vector. |
-extern int kDerivedFeaturesCount; |
+extern unsigned kDerivedFeaturesCount; |
// The distillable page detector is a model trained on a list of numeric |
-// features derived from core more complex features of a webpage (like the |
-// body's .textContent). This derives the numeric features for a set of core |
-// features. |
+// features derived from features of a webpage (like body's number of elements |
+// ). This derives the numeric features form a set of core features. |
// |
// Note: It is crucial that these features are derived in the same way and are |
// in the same order as in the training pipeline. See //heuristics/distillable |
-// in the external DomDistillerJs repo. |
-std::vector<double> CalculateDerivedFeatures(bool isOGArticle, |
- const GURL& url, |
- double numElements, |
- double numAnchors, |
- double numForms, |
- const std::string& innerText, |
- const std::string& textContent, |
- const std::string& innerHTML); |
- |
-// Calculates the derived features from the JSON value as returned by the |
-// javascript core feature extraction. |
-std::vector<double> CalculateDerivedFeaturesFromJSON( |
- const base::Value* stringified_json); |
+// in the external DomDistiller repo. |
+std::vector<double> CalculateDerivedFeatures( |
+ const blink::WebDistillabilityFeatures& f, |
+ const GURL& url); |
} // namespace dom_distiller |