decode HTML entities manually.

Author: Tjatse

lib/article.js

@@ -15,7 +15,8 @@
 
 "use strict";
 
-var read = require('./reader');
+var entities = require('entities'),
+  read = require('./reader');
 
 module.exports = Article;
 
@@ -92,20 +93,23 @@ Article.prototype.getTitle = function() {
   if(!title){
     return '';
   }
+
+  title = entities.decode(title);
+
   // split title by separators.
-  var betterTitle = '', subTitles = title.split(/[\|\-_«»]/g), betterTitleExist = function(t){
+  var betterTitle = '', subTitles = title.split(/[\|\-–_«»]/g), betterTitleExist = function(t){
     return (t && t.length >= 10);
   };
+
   // when find better title, break the loop.
   subTitles.forEach(function(t){
     if(betterTitleExist(betterTitle)){
       return false;
-    }else{
-      betterTitle += ' ';
     }
+    console.log(betterTitle);
     t = t.trim();
     if(t){
-      betterTitle += t;
+      betterTitle += ' ' + t;
     }
   });
 
@@ -135,6 +139,7 @@ Article.prototype.contentWithTEXT = function(node, output){
   if(output.stripSpaces){
     text = text.replace(/([\s\r\t\n]+)/g, ' ');
   }
+  text = entities.decode(text);
   return text;
 };
 /**
@@ -151,6 +156,7 @@ Article.prototype.contentWithHTML = function(node, output){
   if(output.stripSpaces){
     html = html.replace(/([\r\t\n]+)/g, ' ');
   }
+  html = entities.decode(html);
   return html;
 };
 /**
@@ -187,7 +193,7 @@ Article.prototype.contentWithJSON = function(node, dataType){
       //find text.
       var text = self.contentWithTEXT(child, dataType);
       if(text){
-        retVal.push({type: 'text', value: text});
+        retVal.push({type: 'text', value: entities.decode(text)});
       }
     }
   });