fix: bug that stopped proper attr cleaning in certain cases

dist/mercury.js

@@ -1498,6 +1498,8 @@ function removeAllButWhitelist($article) {
       return acc;
     }, {});
   });
+
+  return $article;
 }
 
 // function removeAttrs(article, $) {
@@ -1508,9 +1510,11 @@ function removeAllButWhitelist($article) {
 
 // Remove attributes like style or align
 function cleanAttributes($article) {
-  removeAllButWhitelist($article);
-
-  return $article;
+  // Grabbing the parent because at this point
+  // $article will be wrapped in a div which will
+  // have a score set on it.
+  console.log('HMM', $article.parent().length);
+  return removeAllButWhitelist($article.parent().length ? $article.parent() : $article);
 }
 
 function removeEmpty($article, $) {
@@ -2603,9 +2607,6 @@ function extractCleanNode(article, _ref) {
   // Make links absolute
   makeLinksAbsolute(article, $, url);
 
-  // Remove unnecessary attributes
-  cleanAttributes(article);
-
   // We used to clean UL's and OL's here, but it was leading to
   // too many in-article lists being removed. Consider a better
   // way to detect menus particularly and remove them.
@@ -2615,6 +2616,9 @@ function extractCleanNode(article, _ref) {
   // Remove empty paragraph nodes
   removeEmpty(article, $);
 
+  // Remove unnecessary attributes
+  cleanAttributes(article, $);
+
   return article;
 }
 

src/cleaners/content.js

@@ -51,9 +51,6 @@ export default function extractCleanNode(
   // Make links absolute
   makeLinksAbsolute(article, $, url);
 
-  // Remove unnecessary attributes
-  cleanAttributes(article);
-
   // We used to clean UL's and OL's here, but it was leading to
   // too many in-article lists being removed. Consider a better
   // way to detect menus particularly and remove them.
@@ -63,5 +60,8 @@ export default function extractCleanNode(
   // Remove empty paragraph nodes
   removeEmpty(article, $);
 
+  // Remove unnecessary attributes
+  cleanAttributes(article, $);
+
   return article;
 }

src/mercury.test.js

@@ -18,7 +18,16 @@ describe('Mercury', () => {
     });
 
     it('does the whole thing', async function() {
-      const result = await Mercury.parse('http://theconcourse.deadspin.com/phyllis-schlafly-finally-croaks-1786219220');
+      const result =
+        await Mercury.parse('http://deadspin.com/remember-when-donald-trump-got-booed-for-butchering-ta-1788216229');
+
+      assert.equal(typeof result, 'object');
+      assert.equal(result.content.indexOf('score="') === -1, true);
+    });
+
+    it('does blogger', async function() {
+      const result =
+        await Mercury.parse('https://googleblog.blogspot.com/2016/08/onhub-turns-one-today.html');
 
       assert.equal(typeof result, 'object');
     });

src/utils/dom/clean-attributes.js

@@ -1,7 +1,6 @@
 import { WHITELIST_ATTRS_RE } from './constants';
 
 function removeAllButWhitelist($article) {
-  // $('*', article).each((index, node) => {
   $article.find('*').each((index, node) => {
     node.attribs = Reflect.ownKeys(node.attribs).reduce((acc, attr) => {
       if (WHITELIST_ATTRS_RE.test(attr)) {
@@ -11,6 +10,8 @@ function removeAllButWhitelist($article) {
       return acc;
     }, {});
   });
+
+  return $article;
 }
 
 // function removeAttrs(article, $) {
@@ -21,7 +22,11 @@ function removeAllButWhitelist($article) {
 
 // Remove attributes like style or align
 export default function cleanAttributes($article) {
-  removeAllButWhitelist($article);
-
-  return $article;
+  // Grabbing the parent because at this point
+  // $article will be wrapped in a div which will
+  // have a score set on it.
+  return removeAllButWhitelist(
+    $article.parent().length ?
+      $article.parent() : $article
+  );
 }