Seize is light Node or Browser web-page content extractor inspired by arc90 readability and Safari Reader.

npm i --save seize

Seize can be used with DOM libraries such as jsdom for example. It only extracts and prepares certain DOM-node for further usage.

var Seize = require('seize'),
    jsdom = require('jsdom').jsdom;

var window = jsdom('<your html here>').defaultView,
    seize  = new Seize(window.document);

seize.content(); // returns DOM-node
seize.text();    // returns only text

For browser usage you shoud clone you DOM object or create it from HTML string:

/**
 * Converts html string to Document
 * @param  {String} html  html document string
 * @return {Node}         document
 */
function HTMLParser(html){
  var doc = document.implementation.createHTMLDocument("example");
  doc.documentElement.innerHTML = html;
  return doc;
};

Here is algorythm how it works:

• Getting html tags that we expect to be text or content container such as p, table, img, etc.
• Filtering unnesessary tags by content and tag names wich defenantly can't be in a content container
• Setting score for each container by containing tags
• Setting score by class name, id name, tag xPath score and text score
• Sorting canditates by score
• Taking first candidate
• Cleaning up article

Seize still in development, so you can use it at one's own risk. You always can help to improve it.

• Improve readme
• Improve text scoring
• Improve page detection wich can't be extracted
• More tests
• More examples

You are welcomed to improve this small piece of software :)

• Kir Peremenov