SimpleSearch

SimpleSearch is becoming a powerful JavaScript tool to filter an array for matched (and approximately matched) strings.

Installation and Setup

We use UMD to maintain compatibility with several popular module systems.

CommonJS Setup (for environments like Node.js and Browserify)

Install SimpleSearch...

npm install --save simplesearch

...and require it in your project

var SimpleSearch = require('simplesearch');
SimpleSearch.filter(...);

AMD Setup (for Require.JS)

Disclaimer: it's been awhile since I've used Require.JS, and I'm no expert. This method is currently untested, but it should work.

Install SimpleSearch by downloading a tagged release. Or you could install with npm and dig it out of the node_modules folder... I don't really know the preferred way to do package management in AMD environments.

Then, start using it like this:

require(['path/to/simplesearch.js'], function(SimpleSearch) {
  SimpleSearch.filter(...);
});

No module loader? No problem.

If you're just looking to dump a script into your page, we can work with that. Start by downloading the latest release.

<script src="/path/to/simplesearch.js"></script>
<script>
  window.SimpleSearch.filter(...);
</script> 

Usage

SimpleSearch comes with two public methods. The full documentation also covers private methods, but you shouldn't use those. The following should be enough to get you started.

Filter(data, search)

Given an array as the first paramater and a string of search text as the second, filter will return a new array that is a subset of the original. The results are filtered according to the match criteria listed in the next section.

var result = SimpleSearch.filter(["foo", "bar", "baz"], "ba");
// result: ["bar", "baz"]

Matches(item, search)

Internally, the filter function passes each item through SimpleSearch.matches(), which is publicly available.

var matches = SimpleSearch.matches('foobar', 'fbr');
// matches: true

This function returns true or false for matches according to the following rules:

• Matches can occur anywhere in the string, so SimpleSearch.matches('bar', 'a'); // true
• The search string can leave out characters, so SimpleSearch.matchs('bar', 'br'); // true
• Search string cannot switch character order, so SimpleSearch.matchs('bar', 'rb'); // false
• Non-alphanumeric characters in the search string (including spaces) are ignored, so SimpleSearch.matches('bar', '(b a.r)'); //true
• Word order of the target doesn't matter, so SimpleSearch.matches('bar foo', 'foobar'); // true

Roadmap

The matches function has some hard-coded rules about how spaces, special characters, text case, and word order work. All of that will soon be configurable.

Filter currently only handles arrays of strings. Filtering an array of objects is coming next.

// data as an array of objects instead of strings
var data = [{
  name: "berry",
  alias: ["strawberry", "acai", "blueberry"]
}, {
  name: "chickpea",
  alias: "garbanzo bean"
}, {
  name: "apple, raw"
}];

Instead of passing an array and a search string to filter, you will be able to pass a big configuration object.

// filter(obj)
SimpleSearch.filter({
  data: data,
  fields: ["name", "alias"], // fields to be searched, ordered by preference.
                             // we only search "alias" if "name" doesn't match
  exactMatch: false, // if set to true, "ba" would not match "bean"
  splitWords: true, // if set to false, "bean garb" wouldn't match "garbanzo bean"
  punctuation: false, // if set to true, "(raw)" wouldn't match "raw"
  success: function (match) {
    // a callback function to be run for each matching item
    // receives a `match` object with the following properties
    // - `field` -- the field in which the match was found
    // - `original` -- original object (from data array) in which match was found
    // - `charPos` -- array of indexes of matched characters
    //                e.g. [0, 1, 3] for "bean" given search string "ben"
  },
  fail: function (original) {
    // callback to be run if match fails
    // receives the original object that was tested for a match
  }
});

Farther future, we're looking into the Levenshtein algorithm to do fuzzy matching. This way, we'll be able to:

• Give each match a score, based on similarity to the search term
• Sort by score before returning the filtered array (and before calling success/fail callback functions)