How to split a long regular expression into multiple lines in JavaScript?

Question

I have a very long regular expression, which I wish to split into multiple lines in my JavaScript code to keep each line length 80 characters according to JSLint rules. It's just better for reading, I think. Here's pattern sample:

var pattern = /^(([^<>()[\]\\.,;:\s@\"]+(\.[^<>()[\]\\.,;:\s@\"]+)*)|(\".+\"))@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\])|(([a-zA-Z\-0-9]+\.)+[a-zA-Z]{2,}))$/;

Answer 1

Extending @KooiInc answer, you can avoid manually escaping every special character by using the source property of the RegExp object.

Example:

var urlRegex = new RegExp(
  /(?:(?:(https?|ftp):)?\/\/)/.source       // protocol
  + /(?:([^:\n\r]+):([^@\n\r]+)@)?/.source  // user:pass
  + /(?:(?:www.)?([^/\n\r]+))/.source       // domain
  + /(\/[^?\n\r]+)?/.source                 // request
  + /(\?[^#\n\r]*)?/.source                 // query
  + /(#?[^\n\r]*)?/.source                  // anchor
);

or if you want to avoid repeating the .source property you can do it using the Array.map() function:

var urlRegex = new RegExp([
  /(?:(?:(https?|ftp):)?\/\/)/,     // protocol
  /(?:([^:\n\r]+):([^@\n\r]+)@)?/,  // user:pass
  /(?:(?:www.)?([^/\n\r]+))/,       // domain
  /(\/[^?\n\r]+)?/,                 // request
  /(\?[^#\n\r]*)?/,                 // query
  /(#?[^\n\r]*)?/,                  // anchor
].map(function (r) { return r.source; }).join(''));

In ES6 the map function can be reduced to: .map(r => r.source).

Answer 2

[Edit 2022/08] Created a small github repository to create regular expressions with spaces, comments and templating.

---

You could convert it to a string and create the expression by calling new RegExp():

var myRE = new RegExp (['^(([^<>()[\]\\.,;:\\s@\"]+(\\.[^<>(),[\]\\.,;:\\s@\"]+)*)',
                        '|(\\".+\\"))@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.',
                        '[0-9]{1,3}\])|(([a-zA-Z\-0-9]+\\.)+',
                        '[a-zA-Z]{2,}))$'].join(''));

Notes:

1. when converting the expression literal to a string you need to escape all backslashes as backslashes are consumed when evaluating a string literal. (See Kayo's comment for more detail.)

2. RegExp accepts modifiers as a second parameter

   /regex/g => new RegExp('regex', 'g')

[Addition ES20xx (tagged template)]

In ES20xx you can use tagged templates. See the snippet.

Note:

• Disadvantage here is that you can't use plain whitespace in the regular expression string (always use \s, \s+, \s{1,x}, \t, \n etc).

(() => {
  const createRegExp = (str, opts) => 
    new RegExp(str.raw[0].replace(/\s/gm, ""), opts || "");
  const yourRE = createRegExp`
    ^(([^<>()[\]\\.,;:\s@\"]+(\.[^<>()[\]\\.,;:\s@\"]+)*)|
    (\".+\"))@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\])|
    (([a-zA-Z\-0-9]+\.)+[a-zA-Z]{2,}))$`;
  console.log(yourRE);
  const anotherLongRE = createRegExp`
    (\byyyy\b)|(\bm\b)|(\bd\b)|(\bh\b)|(\bmi\b)|(\bs\b)|(\bms\b)|
    (\bwd\b)|(\bmm\b)|(\bdd\b)|(\bhh\b)|(\bMI\b)|(\bS\b)|(\bMS\b)|
    (\bM\b)|(\bMM\b)|(\bdow\b)|(\bDOW\b)
    ${"gi"}`;
  console.log(anotherLongRE);
})();

Answer 3

Using strings in new RegExp is awkward because you must escape all the backslashes. You may write smaller regexes and concatenate them.

Let's split this regex

/^foo(.*)\bar$/

We will use a function to make things more beautiful later

function multilineRegExp(regs, options) {
    return new RegExp(regs.map(
        function(reg){ return reg.source; }
    ).join(''), options);
}

And now let's rock

var r = multilineRegExp([
     /^foo/,  // we can add comments too
     /(.*)/,
     /\bar$/
]);

Since it has a cost, try to build the real regex just once and then use that.

Answer 4

Thanks to the wonderous world of template literals you can now write big, multi-line, well-commented, and even semantically nested regexes in ES6.

//build regexes without worrying about
// - double-backslashing
// - adding whitespace for readability
// - adding in comments
let clean = (piece) => (piece
    .replace(/((^|\n)(?:[^\/\\]|\/[^*\/]|\\.)*?)\s*\/\*(?:[^*]|\*[^\/])*(\*\/|)/g, '$1')
    .replace(/((^|\n)(?:[^\/\\]|\/[^\/]|\\.)*?)\s*\/\/[^\n]*/g, '$1')
    .replace(/\n\s*/g, '')
);
window.regex = ({raw}, ...interpolations) => (
    new RegExp(interpolations.reduce(
        (regex, insert, index) => (regex + insert + clean(raw[index + 1])),
        clean(raw[0])
    ))
);

Using this you can now write regexes like this:

let re = regex`I'm a special regex{3} //with a comment!`;

Outputs

/I'm a special regex{3}/

Or what about multiline?

'123hello'
    .match(regex`
        //so this is a regex

        //here I am matching some numbers
        (\d+)

        //Oh! See how I didn't need to double backslash that \d?
        ([a-z]{1,3}) /*note to self, this is group #2*/
    `)
    [2]

Outputs hel, neat!
"What if I need to actually search a newline?", well then use \n silly!
Working on my Firefox and Chrome.

---

Okay, "how about something a little more complex?"
Sure, here's a piece of an object destructuring JS parser I was working on:

regex`^\s*
    (
        //closing the object
        (\})|

        //starting from open or comma you can...
        (?:[,{]\s*)(?:
            //have a rest operator
            (\.\.\.)
            |
            //have a property key
            (
                //a non-negative integer
                \b\d+\b
                |
                //any unencapsulated string of the following
                \b[A-Za-z$_][\w$]*\b
                |
                //a quoted string
                //this is #5!
                ("|')(?:
                    //that contains any non-escape, non-quote character
                    (?!\5|\\).
                    |
                    //or any escape sequence
                    (?:\\.)
                //finished by the quote
                )*\5
            )
            //after a property key, we can go inside
            \s*(:|)
      |
      \s*(?={)
        )
    )
    ((?:
        //after closing we expect either
        // - the parent's comma/close,
        // - or the end of the string
        \s*(?:[,}\]=]|$)
        |
        //after the rest operator we expect the close
        \s*\}
        |
        //after diving into a key we expect that object to open
        \s*[{[:]
        |
        //otherwise we saw only a key, we now expect a comma or close
        \s*[,}{]
    ).*)
$`

It outputs /^\s*((\})|(?:[,{]\s*)(?:(\.\.\.)|(\b\d+\b|\b[A-Za-z$_][\w$]*\b|("|')(?:(?!\5|\\).|(?:\\.))*\5)\s*(:|)|\s*(?={)))((?:\s*(?:[,}\]=]|$)|\s*\}|\s*[{[:]|\s*[,}{]).*)$/

And running it with a little demo?

let input = '{why, hello, there, "you   huge \\"", 17, {big,smelly}}';
for (
    let parsed;
    parsed = input.match(r);
    input = parsed[parsed.length - 1]
) console.log(parsed[1]);

Successfully outputs

{why
, hello
, there
, "you   huge \""
, 17
,
{big
,smelly
}
}

Note the successful capturing of the quoted string.
I tested it on Chrome and Firefox, works a treat!

_{If curious you can checkout what I was doing, and its demonstration.
Though it only works on Chrome, because Firefox doesn't support backreferences or named groups. So note the example given in this answer is actually a neutered version and might get easily tricked into accepting invalid strings.}

Answer 5

There are good answers here, but for completeness someone should mention Javascript's core feature of inheritance with the prototype chain. Something like this illustrates the idea:

RegExp.prototype.append = function(re) {
  return new RegExp(this.source + re.source, this.flags);
};

let regex = /[a-z]/g
.append(/[A-Z]/)
.append(/[0-9]/);

console.log(regex); //=> /[a-z][A-Z][0-9]/g

Answer 6

The regex above is missing some black slashes which isn't working properly. So, I edited the regex. Please consider this regex which works 99.99% for email validation.

let EMAIL_REGEXP = 
new RegExp (['^(([^<>()[\\]\\\.,;:\\s@\"]+(\\.[^<>()\\[\\]\\\.,;:\\s@\"]+)*)',
                    '|(".+"))@((\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.',
                    '[0-9]{1,3}\])|(([a-zA-Z\\-0-9]+\\.)+',
                    '[a-zA-Z]{2,}))$'].join(''));

Answer 7

To avoid the Array join, you can also use the following syntax:

var pattern = new RegExp('^(([^<>()[\]\\.,;:\s@\"]+' +
  '(\.[^<>()[\]\\.,;:\s@\"]+)*)|(\".+\"))@' +
  '((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\])|' +
  '(([a-zA-Z\-0-9]+\.)+[a-zA-Z]{2,}))$');

Answer 8

You can simply use string operation.

var pattenString = "^(([^<>()[\]\\.,;:\s@\"]+(\.[^<>()[\]\\.,;:\s@\"]+)*)|"+
"(\".+\"))@((\[[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\])|"+
"(([a-zA-Z\-0-9]+\.)+[a-zA-Z]{2,}))$";
var patten = new RegExp(pattenString);

Answer 9

I tried improving korun's answer by encapsulating everything and implementing support for splitting capturing groups and character sets - making this method much more versatile.

To use this snippet you need to call the variadic function combineRegex whose arguments are the regular expression objects you need to combine. Its implementation can be found at the bottom.

Capturing groups can't be split directly that way though as it would leave some parts with just one parenthesis. Your browser would fail with an exception.

Instead I'm simply passing the contents of the capture group inside an array. The parentheses are automatically added when combineRegex encounters an array.

Furthermore quantifiers need to follow something. If for some reason the regular expression needs to be split in front of a quantifier you need to add a pair of parentheses. These will be removed automatically. The point is that an empty capture group is pretty useless and this way quantifiers have something to refer to. The same method can be used for things like non-capturing groups (/(?:abc)/ becomes [/()?:abc/]).

This is best explained using a simple example:

var regex = /abcd(efghi)+jkl/;

would become:

var regex = combineRegex(
    /ab/,
    /cd/,
    [
        /ef/,
        /ghi/
    ],
    /()+jkl/    // Note the added '()' in front of '+'
);

If you must split character sets you can use objects ({"":[regex1, regex2, ...]}) instead of arrays ([regex1, regex2, ...]). The key's content can be anything as long as the object only contains one key. Note that instead of () you have to use ] as dummy beginning if the first character could be interpreted as quantifier. I.e. /[+?]/ becomes {"":[/]+?/]}

Here is the snippet and a more complete example:

function combineRegexStr(dummy, ...regex)
{
    return regex.map(r => {
        if(Array.isArray(r))
            return "("+combineRegexStr(dummy, ...r).replace(dummy, "")+")";
        else if(Object.getPrototypeOf(r) === Object.getPrototypeOf({}))
            return "["+combineRegexStr(/^\]/, ...(Object.entries(r)[0][1]))+"]";
        else 
            return r.source.replace(dummy, "");
    }).join("");
}
function combineRegex(...regex)
{
    return new RegExp(combineRegexStr(/^\(\)/, ...regex));
}

//Usage:
//Original:
console.log(/abcd(?:ef[+A-Z0-9]gh)+$/.source);
//Same as:
console.log(
  combineRegex(
    /ab/,
    /cd/,
    [
      /()?:ef/,
      {"": [/]+A-Z/, /0-9/]},
      /gh/
    ],
    /()+$/
  ).source
);

Answer 10

@Hashbrown's great answer got me on the right track. Here's my version, also inspired by this blog.

function regexp(...args) {
  function cleanup(string) {
    // remove whitespace, single and multi-line comments
    return string.replace(/\s+|\/\/.*|\/\*[\s\S]*?\*\//g, '');
  }

  function escape(string) {
    // escape regular expression
    return string.replace(/[-.*+?^${}()|[\]\\]/g, '\\$&');
  }

  function create(flags, strings, ...values) {
    let pattern = '';
    for (let i = 0; i < values.length; ++i) {
      pattern += cleanup(strings.raw[i]);  // strings are cleaned up
      pattern += escape(values[i]);        // values are escaped
    }
    pattern += cleanup(strings.raw[values.length]);
    return RegExp(pattern, flags);
  }

  if (Array.isArray(args[0])) {
    // used as a template tag (no flags)
    return create('', ...args);
  }

  // used as a function (with flags)
  return create.bind(void 0, args[0]);
}

Use it like this:

regexp('i')`
  //so this is a regex

  //here I am matching some numbers
  (\d+)

  //Oh! See how I didn't need to double backslash that \d?
  ([a-z]{1,3}) /*note to self, this is group #2*/
`

To create this RegExp object:

/(\d+)([a-z]{1,3})/i

Answer 11

Expanding on @korun's extension of @KooiInc answer, you can use

1. /myRegExp/.source for stand-alone components of your RegExp,
2. "myFragment" strings for regions that won't parse (i.e. unbalanced parens).

Below is a less-eye-gouge-y representation of the FHIR composite "dateTime" datatype and some examples to show off the capture groups.

const parseDateTime = new RegExp(
      "^" // match whole string
    +   /([0-9](?:[0-9](?:[0-9][1-9]|[1-9]0)|[1-9]00)|[1-9]000)/.source // year
    +   "(?:"
    +     /-(0[1-9]|1[0-2])/.source                         // month
    +     "(?:"
    +       /-(0[1-9]|[1-2][0-9]|3[0-1])/.source // date
    +       "(?:T("
    +         /(?:[01][0-9]|2[0-3])/.source // hour
    +         /:[0-5][0-9]/.source          // minute
    +         /:(?:[0-5][0-9]|60)/.source   // second
    +         /(?:\.[0-9]{1,9})?/.source    // decimal second
    +       "))?"
    +     ")?"
    +   ")?"
    +   /(Z|(?:\+|-)(?:(?:0[0-9]|1[0-3]):[0-5][0-9]|14:00)?)?/.source // timezone
    + "$" // end match whole string
);

const fhirDateTimes = {
  gYear: ['2018', '2018-05:00', '2018+05:00', '2018Z'],
  gYearMonth: ['1973-06', '1973-06-05:00', '1973-06+05:00', '1973-06Z'],
  date: ['1905-08-23', '1905-08-23-05:00', '1905-08-23+05:00', '1905-08-23Z'],
  dateTime: ['2015-02-07T13:28:17-05:00', '2017-01-01T00:00:00.000Z'],
};

Object.entries(fhirDateTimes).forEach(([xsdType, lexicalForms]) => {
  console.log(xsdType)
  lexicalForms.map(lexicalForm => {
    const m = lexicalForm.match(parseDateTime);
    console.log(" ".repeat(27-lexicalForm.length), m.join(', '));
  });
});

Note that I decomposed the RegExp in the FHIR spec verbatim. I suspect it should actually be more nuanced in that I don't believe timezones are allowed on gYear or gYearMonth.

Answer 12

Personally, I'd go for a less complicated regex:

/\S+@\S+\.\S+/

Sure, it is less accurate than your current pattern, but what are you trying to accomplish? Are you trying to catch accidental errors your users might enter, or are you worried that your users might try to enter invalid addresses? If it's the first, I'd go for an easier pattern. If it's the latter, some verification by responding to an e-mail sent to that address might be a better option.

However, if you want to use your current pattern, it would be (IMO) easier to read (and maintain!) by building it from smaller sub-patterns, like this:

var box1 = "([^<>()[\]\\\\.,;:\s@\"]+(\\.[^<>()[\\]\\\\.,;:\s@\"]+)*)";
var box2 = "(\".+\")";

var host1 = "(\\[[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\.[0-9]{1,3}\\])";
var host2 = "(([a-zA-Z\-0-9]+\\.)+[a-zA-Z]{2,})";

var regex = new RegExp("^(" + box1 + "|" + box2 + ")@(" + host1 + "|" + host2 + ")$");