Add candidatesFilters option

Author: hankliu62

README.md

@@ -81,6 +81,23 @@ read(url, {
   });
 ```
 
+- `candidatesFilters` which allow set your own filters for candidate tags.
+
+options.candidatesFilters = [callback(obj, index)]
+```javascript
+read(url, {
+  candidatesFilters: [
+    function (obj) {
+      if (obj.tagName === 'ARTICLE' && elem.getAttribute('type') === 'video') {
+        return false;
+      }
+      return true;
+    }
+  ]}, function(err, article, response) {
+    //...
+  });
+```
+
 - `preprocess` which should be a function to check or modify downloaded source before passing it to readability.
 
 options.preprocess = callback(source, response, contentType, callback);

src/helpers.js

@@ -22,11 +22,16 @@ exports.debug = function(debug) {
 };
 
 var cleanRules = [];
+var candidatesFilters = [];
 
 module.exports.setCleanRules = function(rules) {
   cleanRules = rules;
 };
 
+module.exports.setCandidatesFilters = function(filters) {
+  candidatesFilters = filters;
+};
+
 /**
  * Prepare the HTML document for readability to scrape it.
  * This includes things like stripping javascript, CSS, and handling terrible markup.
@@ -65,7 +70,7 @@ var prepDocument = module.exports.prepDocument = function(document) {
       }
     }
   }
-  
+
   // Strip out all <script> tags, as they *should* be useless
   var scripts = document.getElementsByTagName('script');
   [].forEach.call(scripts, function (node) {
@@ -182,6 +187,12 @@ var grabArticle = module.exports.grabArticle = function(document, preserveUnlike
     grandParentNode.readability.contentScore += contentScore / 2;
   }
 
+  if (candidatesFilters.length) {
+    candidatesFilters.forEach(function(filterBy) {
+      candidates = candidates.filter(filterBy);
+    });
+  }
+
 
   /**
    * After we've calculated scores, loop through all of the possible candidate nodes we found

src/readability.js

@@ -18,6 +18,7 @@ function Readability(window, options) {
   this.bodyCache = null;
   this._articleContent = '';
   helpers.setCleanRules(options.cleanRulers || []);
+  helpers.setCandidatesFilters(options.candidatesFilters || []);
 
   this.cache = {};
 

test/article-tests.js

@@ -16,7 +16,7 @@ describe('Regression Tests', function() {
     ],
     notInclude: [
       'Donate to Wikipedia'
-    ]
+    ],
   },
   {
     fixture: 'mediashift',
@@ -80,11 +80,48 @@ describe('Regression Tests', function() {
       '最赞回应',
       '最新话题',
       '北京豆网科技有限公司',
-    ]
+    ],
+  },
+  {
+    fixture: 'ifeng',
+    title: '熊玲:什么样的婚姻才是鸡肋婚姻？',
+    include: [
+      '沃尔沃“憋”不住了，最高狂降8万，性能不输BBA，白菜价愣没人',
+      '打开APP',
+    ],
+    notInclude: [
+      '它是“迷恋婚姻又排拒婚姻”的一种复杂婚姻情感心理状态。它意味着即便你有千条理由走出婚姻，背后却有万种吸引力把你留在围城里。',
+      '在婚姻十字路口的人，你若要想你们的关系和好如初，就必须有重修婚姻的姿态，即必须有妥协的态度。',
+      '重修婚姻的办法很多很多，但最简单也是最核心的办法只有一个，那就是接受。',
+    ],
+  },
+  {
+    fixture: 'ifeng',
+    title: '熊玲:什么样的婚姻才是鸡肋婚姻？',
+    include: [
+      '它是“迷恋婚姻又排拒婚姻”的一种复杂婚姻情感心理状态。它意味着即便你有千条理由走出婚姻，背后却有万种吸引力把你留在围城里。',
+      '在婚姻十字路口的人，你若要想你们的关系和好如初，就必须有重修婚姻的姿态，即必须有妥协的态度。',
+      '重修婚姻的办法很多很多，但最简单也是最核心的办法只有一个，那就是接受。',
+    ],
+    notInclude: [
+      '沃尔沃“憋”不住了，最高狂降8万，性能不输BBA，白菜价愣没人',
+      '打开APP',
+    ],
+    options: {
+      candidatesFilters: [
+        function (elem) {
+          if (elem.tagName === 'ARTICLE' && elem.getAttribute('type') === 'video') {
+            return false;
+          }
+
+          return true;
+        }
+      ],
+    },
   }].forEach(function(testCase) {
     it('can extract ' + testCase.fixture + ' articles', function(done) {
       var html = fs.readFileSync(articleFixtures + '/' + testCase.fixture + '.html').toString();
-      read(html, function(error, article) {
+      read(html, testCase.options || {}, function(error, article) {
         if(error) {
           done(error)
         } else {