read-art - npm Package Compare versions

Comparing version 0.0.6 to 0.0.7

examples/simple.js

 var read = require('../');
 
-read('http://auto.rednet.cn/html/tupian/20145/20145124642220.html', function(err, art){
+read('http://www.tianjinwe.com/rollnews/201405/t20140512_8874908.html', {
+  dataType: 'json'
+}, function(err, art){
   if(err){
@@ -5,0 +7,0 @@ console.log('[ERROR]', err.message);

index.js

@@ -27,6 +27,3 @@ // Copyright 2014 Tjatse
  * @param uri uri or html
- * @param options including:
- *  cacheable: false as default.
- *  killBreaks: true as default.
- *  lowerCaseTags: true as default.
+ * @param options reference to https://github.com/Tjatse/node-readability#options
  * @param callback callback, have two arguments been passed:
@@ -49,4 +46,5 @@ * 1: error
   }
-  defineBoolean(options, 'killBreaks', true);
-  defineBoolean(options, 'lowerCaseTags', true);
+  defineOption(options, 'killBreaks', true);
+  defineOption(options, 'lowerCaseTags', true);
+  defineOption(options, 'dataType', 'html');
 
@@ -95,3 +93,3 @@ // indicating uri is html or url.
     o.html = o.html.replace(/(<br\s*\/?>(\s|&nbsp;?)*){1,}/g,'<br />');
-    // remove formats like \r\t\n
+    // remove tab symbols like \r\t\n
     o.html = o.html.replace(/([\n\r\t]*){2,}/gi, '');
@@ -109,3 +107,3 @@ }
 /**
- * Define property of object to default boolean value.
+ * Define property of object to default value.
  * @param options option object
@@ -115,3 +113,3 @@ * @param k key
  */
-function defineBoolean(options, k, v){
+function defineOption(options, k, v){
   if(typeof options[k] == 'undefined'){
@@ -118,0 +116,0 @@ options[k] = v;

lib/article.js

@@ -62,4 +62,17 @@ // Copyright 2014 Tjatse
   // else read it by article reader.
-  var node = read(this.$, this.options);
-  var content = ((node && node.length > 0) ? node.html() : '' );
+  var node = read(this.$, this.options),
+      content;
+  if((node && node.length > 0)){
+    switch(this.options.dataType){
+      case 'text':
+        content = node.text();
+        break;
+      case 'html':
+      default:
+        content = node.html();
+        break;
+    }
+  }else{
+    content = '';
+  }
   // if cacheable, cache it.
@@ -66,0 +79,0 @@ if(this.caches){

lib/reader.js

@@ -129,3 +129,2 @@ // Copyright 2014 Tjatse
     node.data(scoreKey, score);
-
     if(!topCandidate || score > topCandidate.data(scoreKey)){
@@ -148,3 +147,3 @@ topCandidate = node;
     parent, siblings;
-  if((parent == topCandidate.parent()) && parent.length > 0 && parent.get(0).name.toLowerCase() != 'body'){
+  if((parent = topCandidate.parent()) && parent.length > 0 && parent.get(0).name.toLowerCase() != 'body'){
     siblings = parent.children();
@@ -158,2 +157,3 @@ }else{
       append = false;
+    
     if(node.is(topCandidate) || (node.data(scoreKey) || 0) > siblingScoreThreshold){
@@ -160,0 +160,0 @@ append = true;

package.json

 {
   "name": "read-art",
-  "version": "0.0.6",
+  "version": "0.0.7",
   "description": "Scrape article from any page, automatically, make web page readability.",
@@ -5,0 +5,0 @@ "main": "index.js",

README.md

@@ -69,2 +69,5 @@ # read-art -- readability reference to Arc90's
 ## Options
+### dataType
+The data type of article content, including: html, text.
+
 ### cacheable
@@ -74,3 +77,3 @@ A value indicating whether cache body && title.
 ### killBreaks
-Kill breaks in the HTML, and convert them to simple `<br />`.
+Kill breaks, blanks, tab symbols(\r\t\n) into one <br />.
 
@@ -77,0 +80,0 @@ ###options from [cheerio](https://github.com/cheeriojs/cheerio)

No alert changes

Improved metrics

Total package byte prevSize

30484

increased by0.95%

Lines of code

740

increased by1.79%

Number of lines in readme file

156

increased by1.96%

No dependency changes