chrisfarms / chrisfarms.github.com

HTML for projects

ObjectIndex.js

Overview

ObjectIndex is a little wrapper around Arrays of Objects that provides you with the ability to search through the array using a text query much like Gmail's search filters.

Examples

It's easist to explain with an example, so Say you have a collection of Objects in an Array like:

var data = [
    {name:'Peter', email:'[email protected]', description: 'He has a ball that is red'},
    {name:'Jane', email:'[email protected]', description: 'Jane likes Peter'},
    {name:'Paul', email:'[email protected]', description: 'Paul has a red ball'}
];

You could then create an ObjectIndex based on this array, and tell it what fields you will be querying against.

var index = new ObjectIndex({
    records: data,
    fields: ['name','email','description']
});

Once you have an index, you can run text-based queries against your collection like you would with a search engine.

results = index.query('peter');
=> [
    {name:'Peter', email:'[email protected]', description: 'Peter has a ball'},
    {name:'Jane', email:'[email protected]', description: 'Jane likes Peter'}
]

"peter" appeared in two of the records, so those two Objects were returned.

Queries can contain multiple terms and phrase-style searches too:

results = index.query('"red ball"');
=> [
    {name:'Paul', email:'[email protected]', description: 'Paul has a red ball'}
]

Even though both "Paul" and "Peter" records contain 'red' and 'ball', the phrase search (inside "quotes") limited it to just the "Paul" record.

You can limit your search to just a single property by writing columname:query:

results = index.query('name:peter');
=> [
    {name:'Peter', email:'[email protected]', description: 'He has a ball that is red'}
]

Other property types

So far we have just queried on string data, but you can also query against numeric and sub-array properties too.

var data = [
  {id: 1, name:'Banana', tags:['yellow','bendy]},
  {id: 2, name:'Apple', tags:['green','round]}
];

var index = new ObjectIndex({
    records: data,
    fields: ['id','tags']
});

results = index.query('tags:yellow');
=> [
  {id: 1, name:'Banana', tags:['yellow','bendy]}
]

Records that require getters

If your data collection is made up of more complex objects, you might find it useful to have the index fetch the propperty values via a getter method. Here's a slightly more complex example to show how this might work.

var MyRecord = function(defaults){   // a simple record class
  this.data = defaults;
}
MyRecord.prototype.get = function(key){ // this method will be used to fetch the properties
  return this.data[key];
}

var peter = new MyRecord({name:'Peter'});
var jane = new MyRecord({name:'Jane'});

var data = [peter, jane];

var index = new ObjectIndex({
    records: data,
    fields: ['name'],        // <-- each of these fields
    getter: 'get'            // <-- will be requested via this method
});

results = index.query('Jane');
=> [jane]

Lies, lies lies

Ok, ok, so ObjectIndex is stretching it a bit at the moment, right now each query is run over all the original records, no "index" is actually created just yet.

I just needed the query part of this first to get it working, the next part is to work on increasing performance of the queries by creating an index and word stemmer so acts like full-text search. That said performance isn't actually too bad for <10K records. Please fork and play with trying to get the graph as low as you can in test.html

What on earth is test.html about

Try to avert your eyes from test.html, I just wanted to write a little script so I can test and evaluate performance. If you open it in a browser it will run a tiny set of basic test queries over an ever increasing set of records, and draw a little graph of how long it took as the data set grows, the aim of the game is to keep those numbers down.

Browser Support

This version will not work in IE as I use indexOf, it would be a pretty minor fix, but I don't need it right now. Again, please fork if you want to help.

Licence & Author(s)

MIT "do what you want" Licenced Written by Chris Farmiloe © 2011 Tokenizer code based on a version written by Douglas Crockford © 2006