Home

Awesome

Incremental Regular Expression Matcher

A node package for incremental regular expression matching in JavaScript - matching one character at a time. This is a feature not available in the powerful and fast RegEx capability of JavaScript. When you use a regular expression to match a string, you have to give it the entire string and the matcher will either match or fail. But there is a usecase where that is not sufficent, processinng user input as it arrives one character at a time. Fot his we need a regular expression that can match on charcter at a time as it arrives and tells you if the match is good so far, and if it is good if more characters could be matched. Once you have this basic capability you can do some amazing things:

Capabilities provided by the package

Demo here play with the code here - for developers

  1. Input validation support as you type.
  2. Automatically supply input for fixed values, e.g. /Phone: [0-9]+/ it can tell you that 'Phone: ' is the only acceptable input at the begining
  3. Provide masked input, for example a phone number (___)-___-____
  4. Provide multiple alternative masks phone number with and without extension (___)-___-____ Ext: ___
  5. Provide masks that support variable input, for example email: /[a-zA-Z_.-]+@[a-zA-Z_.-]+/
  6. Auto complete /Yes|No|Maybe/ - once you type 'Y' the only thing the input can be is 'Yes' and so on
  7. Show a dropdown list of possible input, /Alabama|Arizona|.../ input hint shows a list of all the states

But I am getting ahead of myself, but the package has buil-in support for all those capabilities. But lets start from the simplest capability, incremental matching.

This is an example of the incr regex in action

import {incrRegEx, DONE} from 'incr-regex-package';

//  var incrRegEx = require('incr-regex-package').incrRegEx;

// example match a US phone number
  var rx = incrRegEx( /\\d{3}-\\d{3}-\\d{4}/ );

// we are trying to match '212-409-5123'

  rx.match('2'); // => true,  matched "2..."
  rx.match('-'); // => false , i.e. it did not accept the character

  rx.match('1'); // => true ,  matched '21...'
  rx.match('2'); // =>  true   matched '212..'
  rx.match('4'); // =>  false , expected '-'

  rx.match('-'); // => true , matched '212-...'
  ....

  rx.match('3'); // => true,  matched '212-409-5123'

  rx.state() === DONE; // true

Clearly, this is useful for input validation. In particular it was to provide a more standard way to support masked input. Firstly, I would like to say that the ''Masked Input'' package is very useful and easy to use.

What I did not like as the method for defining the mask. It works very well I only found the method of defining input mask rather limited. The package has nice documentation and I highly recommend you look at it. It would have been nice if the mask could be defined using a regular expression. That turns out to be hard, becuase regular expression will match the entire input after you have entered it, but I have found no way to check the input as you are typing it in, check if the partial string matches the regular expression.

Installation

npm install incr-regex-package --save

Sample index.js file

var ir = require("incr-regex-package");

var incrRegEx = ir.incrRegEx;
var rx = ir.incrRegEx("ab[0-9]c");
var DONE = ir.DONE;
var MORE = ir.MORE;
var MAYBE = ir.MAYBE;
var FAILED = ir.FAILED;

console.log("Initial input mask: ", rx.minChars()); // Print the pseudo input msak for the regexp
console.log("matchStr:", rx.matchStr("ab"),
	        '\nState: '+rx.state(), 
	         '\nInput Mask left: "'+rx.minChars()+'"');

The output:

Initial input mask: ab_c<br> matchStr: [ true, 2, 'ab' ]<br> State: MORE<br> Input Mask left: "_c"<br>

What is supported

Almot everything supported by JavaScript regexp. With some notable differences:

Note: All regular expressions have an implicit:

Although ( ) and (?: ) are supported, but neither do input binding (number) e.g \1. This feature is not required for the main usecase for the package.

Look ahead is not supported

Regular JavaScript

	var rx = /\d{3}-\d{3}-\d{4}/

In the incremental regex package

	var irx = incrRegEx("\\d{3}-\\d{3}-\\d{4}");

Future extensions

Some of the other future possibilities with the package:

  1. Generate string that will match a regular expression. This is very useful for generating sample strings from a regular expression
  2. Support for a smart regex tester
  3. Create railroad diagram Perhaps something like this:
   var exp = RX.or("hello","goodbye","hi")
                .opt(" my friend")
                .zeroOrMore(
                   RX.chars(
                        "[a-zA-Z]", 
                        RX.except("[aeiouAEIO]")
                        )
                  );
   // same ass from a regular expression
4. Creating regex not from strings but programatically, for example using a fluent API
 
   var rx1 = /(hello|goodbye|hi)(my friend)?[bcdfghjklmnpqrstvwxyzBCDFGHJKLMNPQRSTUVWXYZ]*/

Testing

I have an initial series of test, but they are by no means comprehensive. I am hoping to significantly augment the tests in the near future. So far all the tests are successful but are mostly small. I have done one simple rather pathological test with 40,000 characters. But this is way beyond the use case for the package. But it does demonstrate that the package is fairly robust. I will write up the design of the package in the wiki.

Although regular expression processing looks simple at first glance, bit I have been written several in my career including one as an intern at college. It is very easy to introduce subtle bugs. I cannot guarantee that this package has no major bugs, but I have included pretty extensive test scripts, and past experience suggests that the remaining bugs are come extreme corner cases. So far the experience is good. If someone take all the JavaScript RegEx test cases and try with this package it would be most helpful.

The package is relatively small and it makes use of javascript's regular expression processing under the hood, mostly in the parser and for single character tests. As I said before, JavaScript regular expression is the method of choice for text processing

import { DONE, MORE, FAILED, incrRegEx } from 'incr-regex-package'
// Example of use from the test cases
//
		it("incrRegEx  /abc/", () =>{
 			expect(incrRegEx !== undefined).to.be.true;
 			var r =  incrRegEx("abc");
			var res = [ r.matchStr('abc') , r.state() ]
 			expect(res).to.deep.equal([[true, 3,'abc'],DONE]);
 			// r.matchStr('abc') === [true, 3, 'abc']
 			// r.state()         === DONE

 			 			
 		});
 		it("Test minChars functionality /abd|abd?/ new",() =>{
 			var r =  incrRegEx("abd|abd?");
 			expect(r.minChars()).to.equal('ab');   // minimal string that will match the regexp
 			 			
 		});
 		it("incrRegEx test incremental matching => /abc|abd/ ", () =>{
 			expect(incrRegEx !== undefined).to.be.true;
 			var r =  incrRegEx("abc|abd");
			var res = [ r.match('a') , r.match('b'),  r.state() ]
 			expect(res).to.deep.equal([true,true,MORE]);
 			//
 			// Match some more
			res = [ r.match('d'),  r.state() ]
 			expect(res).to.deep.equal([true,DONE]);
 			 			
 		});
 		it("incrRegEx or /abc|abdc*/ new", () =>{
 			expect(incrRegEx !== undefined).to.be.true;
 			var r =  incrRegEx("abc|abdc*");
 			expect(r.minChars()).to.equal("ab_");

Justification for the project

JavaScript has excellent regular expression support, so why create another one? It definitely will not be as fast or as feature rich as the built-in support. Well, there is one compelling reason, input validation; and in particular how can we use a regular expression to validate the input as the user is typing in the text. It is easy to use standard regular expression to validate input after you have filled in the input field. Let me give an example, suppose you want to check an email address; so you use a regular expression for the input field, the regular expression will only validate after you have entered the entire email address, but will not tell you is valid as you are typing it in. For example if you have entered nurulc@abc a second @, for example: nurulc@abc@ is not allowed. Another example may help, a phone number

/\d{3}-\d{3}-\d{4}/ is the validation regexp.

if you type: 1234 I should not be able to enter that, since the input would never validate. While if I enter 123- this is valid up to that point in the input. Of course if you leave the input box, what you have entered is not a complete match and will show that this is not a valid phone number. The second case is handled perfectly by the standard JavaScript regular expression, but not while you are entering the data.

How it works

The following link will explain how the matching works: How It Works

The API