Create a regular expression to match any of the phrases added to the trie (inspired by Dan Kogai's Regexp::Trie Perl module.)


Create a regular expression to match any of the phrases added to the trie (inspired by Dan Kogai's Regexp::Trie Perl module.

  1. npm install regex-trie
  2. require and use (see the Usage section for more comprehensive usage instructions.)
var RegexTrie = require('regex-trie'),
    trie      = new RegexTrie(),
    regex     = trie.add('foo').add('bar').toRegExp();
  1. npm install regex-trie
  2. create your application using RegexTrie:
// app.js 
var RegexTrie = require('regex-trie'),
    trie      = new RegexTrie(),
    regex     = trie.add('foo').add('bar').toRegExp();
  1. Use browserfy to create the the browser-safe package, e.g.: browserify app.js -o bundle.js.
var RegexTrie = require('regex-trie');
// Create a new RegexTrie instance 
var trie = new RegexTrie();
// Add phrases to the trie 
// You can use an array to add phrases if you'd rather (duplicate 
// pharses are ignored.) 
trie.add(['foo', 'bar', 'baz']);
// Fetch a RegExp to represent all the phrases in the trie 
var regex = trie.toRegExp(); // regex => /(?:foo|ba[rz])/ 
// What matches? 
var things_to_match = ['foo', 'bar', 'baz', 'bat', 'fun', 'food'],
    match_results   =, regex);
// => [ true, true, true, false, false, true ] 

Creates a new instance of RegexTrie. Currently doesn't accept any options however this will likely change as the module evolves.

Adds a new phrase to the trie. Accepts singleton arguments, or an array of phrases. Ignores any values which aren't literals (objects, bools, arrays, etc).

        .add(['who', 'what', 'when', 'where'];

All numbers (except NaN) are coerced into strings before being added.

Before adding new phrases, the trie is checked to see whether or not that phrase already exists (using contains).

Will check to see if the trie contains a phrase which matches phrase, and return true or false if the phrase does or does not exist.

Returns a RegExp instance which should match each individual phrase in the tree. The trie will escape any character that matches: /([^A-Za-z0-9_])/. For example, if the following values are added, the pipe (OR) will be escaped:

    trie.add(['foo', '|', 'bar'].toRegExp();
    // => (?:foo|\||bar) 

The RegExp returned by regex() is a non-capturing, un-anchored regular expression meaning it'll never capture its matches and all of the following phrases will still match:

    var regex = trie.add(['foo', 'bar', 'car']).toRegExp();
    ['fool', 'afool', 'bart', 'abart', 'acar', 'acard'].forEach( function (word) {
    // Output => true, true, true, true, true, true 

regex-trie uses Gulp as its build system. Currently gulpfile defines a few tasks:

  • lint -- JSHint (see .jshintrc for this project's settings)
  • test -- runs mocha from gulp
  • docs -- yuidocjs to produce development documentation
  • watch -- watches for changes to JS files in ./test/ and ./lib/ and runs the lint task
  • default -- by default the watch task runs (which runs lint)
  • continuous -- runs watch (which runs lint) and test on every JS file change.

Please see package.json for the latest development dependencies. At the time of writing, you'll need:

    "mocha": "~1.17.1"
    "should": "~3.1.2"
    "gulp-jshint": "~1.4.0"
    "gulp-util": "~2.2.14"
    "gulp": "~3.5.2"
    "gulp-watch": "~0.5.0"
    "blanket": "~1.1.6"
    "gulp-yuidoc": "~0.1.0"

The tests within regex-trie use mocha with should.js assertions. To test the module, just run mocha from your terminal.

List of things to add aren't in any specific order.

  1. Regex options to configure capturing and anchoring
  2. Cache compiled trie branches (to speed up RegExp generation)

See LICENSE.txt for license rights and limitations (MIT).