From f2729a5ee6db59c7bd2d8e0555bd2af738f88fea Mon Sep 17 00:00:00 2001
From: Adam Pash <adam.pash@gmail.com>
Date: Fri, 9 Sep 2016 12:00:09 -0400
Subject: [PATCH] improved wiki extractor

---
 src/extractor/custom/wikipedia.org/index.js | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/src/extractor/custom/wikipedia.org/index.js b/src/extractor/custom/wikipedia.org/index.js
index 41f3d006..73c07aca 100644
--- a/src/extractor/custom/wikipedia.org/index.js
+++ b/src/extractor/custom/wikipedia.org/index.js
@@ -7,9 +7,10 @@ const WikipediaExtractor = {
 
     // transform top infobox to an image with caption
     transforms: {
-      '.infobox img': ($node, $) => {
+      '.infobox img': ($node) => {
         $node.parents('.infobox').prepend($node)
       },
+      '.infobox caption': 'figcaption',
       '.infobox': 'figure',
     },
 
@@ -17,6 +18,7 @@ const WikipediaExtractor = {
     clean: [
       '.mw-editsection',
       'figure tr, figure td, figure tbody',
+      '#toc',
     ],
 
   },