Fix mozilla#585 - remove nodes with role=complementary

bgrins · Apr 7, 2020 · d5621f8 · d5621f8
1 parent 668a3a1
commit d5621f8
Show file tree

Hide file tree

Showing 4 changed files with 1,151 additions and 0 deletions.
diff --git a/Readability.js b/Readability.js
@@ -840,6 +840,12 @@ Readability.prototype = {
             node = this._removeAndGetNext(node);
             continue;
           }
+
+          if (node.getAttribute("role") == "complementary") {
+            this.log("Removing complementary content - " + matchString);
+            node = this._removeAndGetNext(node);
+            continue;
+          }
         }
 
         // Remove DIV, SECTION, and HEADER nodes without any content(e.g. text, image, video, or iframe).

diff --git a/test/test-pages/archive-of-our-own/expected-metadata.json b/test/test-pages/archive-of-our-own/expected-metadata.json
@@ -0,0 +1,8 @@
+{
+  "title": "Conversations with a Cryptid - Chapter 1 - AMournfulHowlInTheNight - 僕のヒーローアカデミア | Boku no Hero Academia",
+  "byline": "Organization for Transformative Works",
+  "dir": null,
+  "excerpt": "An Archive of Our Own, a project of the Organization for Transformative Works",
+  "siteName": null,
+  "readerable": true
+}