Liberal, Accurate Regex Pattern for Matching All URLs

Liberal Regex Pattern for All URLs

The regex patterns in this gist are intended to match any URLs,
including "mailto:foo@example.com", "x-whatever://foo", etc. For a
pattern that attempts only to match web URLs (http, https), see:
https://gist.github.com/gruber/8891611


# Single-line version of pattern:

(?i)\b((?:[a-z][\w-]+:(?:/{1,3}|[a-z0-9%])|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:'".,<>?«»“”‘’]))


# Multi-line commented version of same pattern:

(?xi)
\b
(							# Capture 1: entire matched URL
  (?:
    [a-z][\w-]+:				# URL protocol and colon
    (?:
      /{1,3}						# 1-3 slashes
      |								#   or
      [a-z0-9%]						# Single letter or digit or '%'
      								# (Trying not to match e.g. "URI::Escape")
    )
    |							#   or
    www\d{0,3}[.]				# "www.", "www1.", "www2." … "www999."
    |							#   or
    [a-z0-9.\-]+[.][a-z]{2,4}/	# looks like domain name followed by a slash
  )
  (?:							# One or more:
    [^\s()<>]+						# Run of non-space, non-()<>
    |								#   or
    \(([^\s()<>]+|(\([^\s()<>]+\)))*\)	# balanced parens, up to 2 levels
  )+
  (?:							# End with:
    \(([^\s()<>]+|(\([^\s()<>]+\)))*\)	# balanced parens, up to 2 levels
    |									#   or
    [^\s`!()\[\]{};:'".,<>?«»“”‘’]		# not a space or one of these punct char
  )
)

It seems the expression doesn't deal with a number of real-life updates to URI patterns in the last years, like internationalised domains, new top-level domains that are not between 2 and 4 characters long, IRIs etc. Some references: http://www.icann.org/en/topics/TLD-acceptance/ and http://www.ietf.org/rfc/rfc3987.txt. A domain like http://موقع.وزارة-الاتصالات.مصر/ is legal and functional today.

KJD: Have you actually tried it? The pattern matches "http://موقع.وزارة-الاتصالات.مصر/" in both PCRE and Perl. What makes you think it doesn't work?

You're of course correct — I jumped the gun in scanning through the expression. I guess the case where the [a-z]{2,4} pattern fails is in the following case, which is not a legal URL and therefore you could definitely argue is less important to catch:

#!/usr/bin/env python
# -*- coding: utf-8 -*-

import re

def matches(s):
    if re.match(r'(?i)\b((?:[a-z][\w-]+:(?:/{1,3}|[a-z0-9%])|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:\'".,<>?«»“”‘’]))', s):
        print "%s matches" % (s)
    else:
        print "%s doesn't match" % (s)

matches('موقع.وزارة-الاتصالات.مصر/ar/default.aspx')
matches('example.com/index.html')

John, since the last revision, your regex will also match stuff like http://#, http://## and http://## /. The previous version didn’t have that problem. I made a quick test case here: http://mathiasbynens.be/demo/url-regex

Hi.
Please check your regex with link bellow.

http://ddos-link.com/[test.......................................]

This link will burn my server :)

This above script will hang on Chrome&NodeJS at 100% CPU usage.

"http://www.ghislainproulx.net/Blog/2014/09/contributing-to-a-github-open-source-project-(from-a-visual-studio-developer-perspective)".replace(/\b((?:[a-z][\w-]+:(?:\/{1,3}|[a-z0-9%])|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}\/)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:'".,<>?«»“”‘’]))/gi, function(url){
  // this will never be executed on Chrome/Node
  console.log(url);
});

Yeah, I'm seeing a hang with the input:

"Ficheiro:Joseph_Ducreux_(French_-_Self-Portrait,_Yawning_-_Google_Art_Project.jpg"

as well. The balanced-paren rules seem to be blowing up.

The following version doesn't have the performance problem:

/\b((?:[a-z][\w\-]+:(?:\/{1,3}|[a-z0-9%])|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}\/)(?:[^\s()<>]|\((?:[^\s()<>]|(?:\([^\s()<>]+\)))*\))+(?:\((?:[^\s()<>]|(?:\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:'".,<>?«»“”‘’]))/i

Note that I've just removed three unnecessary/redundant + operators, so that the regexp ends with:

  (?:                           # One or more:
    [^\s()<>]                       # Non-space, non-()<>  (removed a + here)
    |                               #   or
    \(([^\s()<>]|(\([^\s()<>]+\)))*\)   # balanced parens, up to 2 levels (removed a + here)
  )+
  (?:                           # End with:
    \(([^\s()<>]|(\([^\s()<>]+\)))*\)   # balanced parens, up to 2 levels (removed a + here)
    |                                   #   or
    [^\s`!()\[\]{};:'".,<>?«»“”‘’]        # not a space or one of these punct char
  )
)

See https://gerrit.wikimedia.org/r/#/c/170329/1/lib/index.js for a colorized diff.

I've tested that this change fixes the problems noted by @FGRibreau and @chemist777 above.

After a little testing, both the original @gruber version and @cscott version, I've found that you can omit the domain extension, and it is still considered a valid URL. Surely that's a bit of a hole in the logic, or is there a reason for that?

I've come up with the follow version. This version requires that you begin with a protocol like http:// https:// and even mailto:

No I'm not a regex genius, but I've been plugging away at test this variation, and it seems to work so far.

_(?i)\b((?:(?:https?|ftps?)://|ftp\.|ftps\.|mailto:|www\d{0,3}[.])(?:[^\s()<>]|\((?:[^\s()<>]|(?:\([^\s()<>]+\)))*\))+(?:\((?:[^\s()<>]|(?:\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:'".,<>?«»“”‘’]))_iuS

Those experiencing hanging problems should try the pattern in a real regular expression engine, that is, one which does not backtrack.

Could anybody help me out with a version of this that also optionally allows the url to be enclosed like so: URL:thefullurl This is a format I come across rather often still in an old forum.

To add support for URIs with schemes like stratum+tcp and xmlrpc.beep or paths starting with + or ? (e.g. sms:, magnet:), I'm using a version with [a-z][\w.+-]+:(?:/{1,3}|[?+]?[a-z0-9%]) as the first section.

you can omit the domain extension, and it is still considered a valid URL. Surely that's a bit of a hole in the logic, or is there a reason for that?

Belated @mattauckland - guessing the reason is for URLs like http://localhost/ to be matched.

The two balanced parens parts use capturing groups, while the rest of the regex uses non-capturing groups (except for the outermost match, obviously). May i suggest to change ( into (?: in those four places?

Hi

To make the regex to match against

example.com
abv.bg
google.com

but also unfortunately also against

filename.txt

I added {0,1} at the end of the 'balanced parens, up to 2 levels'
and made backslash optional
(the second 2 groups was eating characters from matched string when match against domain.[a-z]{2,4}, and so {2,4} becomes incorrect in that case)

my final regex is:
\b((?:[a-z][\w-]+:(?:\/{1,3}|[a-z0-9%])|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}\/?)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\)){0,}(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s\!()\[\]{};:\'\"\.\,<>?«»“”‘’]){0,})

I have the same problem , i used
GRUBER_URLINTEXT_PAT = re.compile(ur'(?i)\b((?:https?://|www\d{0,3}[.]|[a-z0-9.-]+[.][a-z]{2,4}/)(?:[^\s()<>]+|(([^\s()<>]+|(([^\s()<>]+)))))+(?:(([^\s()<>]+|(([^\s()<>]+))))|[^\s`!()[]{};:'".,<>?\xab\xbb\u201c\u201d\u2018\u2019]))')

but it return some urls like :
https://t.co/h…

i need your help please !