Tokeninput is a jQuery plugin which allows your users to select multiple items from a predefined list, using autocompletion as they type to find each item. You may have seen a similar type of text entry when filling in the recipients field sending messages on facebook.

This is a forked version of jquery-tokeninput with a couple of different things. I grouped a small list of changes between original loopj's version from mine.

In loopj's version, only a parsed text of what the user types goes to the server-side destination, without accents and all in lowercase. This version will send to server-side the intact version without parsing any character. This particularly allows to the server-side to know how to handle new custom entries and special items.

Original jquery-tokeniput used a simple regex expression to highlight the typed characters in the dropdown list. This caused to be a huge problem with accents and other special chars. After doing a major clean-up[1] on this, this updated version of tokeninput will properly highlight terms like “são paulo” when “sao paulo” is provided.

[1] Based on this great article http://www.alistapart.com/articles/accent-folding-for-auto-complete/

Originally jquery-tokeninput would only accept objects with 'id' and 'name' parameters, and would even ignore any other element in the object. With several hooks on this, I made token accept any kind of object, with options to choose which columns to use when searching (by using searchColumns) and what to display as the name of the item (with parseName param).

$("#demo").tokenInput([
    {id: 1, city: 'Vitória', state: 'ES'},
    {id: 5, name: 'São Paulo', state: 'SP'},
    {id: 2, city: 'Palo Alto', state: 'CA'},
    {id: 3, city: 'San Francisco', state: 'CA'},
],
{
    parseName: function(element) {
        return element.city + ', '+ element.state;
    },
    searchColumns: ['city', 'state']
});

Besides choosing predefined items from a list

TODO

Original jquery-tokeninput had some misterious bug that kept deleted elements when form was submited. Besides this, it had a hard-to-maintain way to handle actions to add and remove tokens.

With this new approach, I cleaned up and simplified the code, and added options to choose how to send the output to the server.

Originally output would be a list of numbers imploded with ,, like 16,742,58,254. With custom entries support, output becomes 16,42,'Subway',37,'McDonald's',734.

Alternatively, a function can be used to parse the output using tokensFormatter. With this, you can per instance return a json as output.