The accepted answer is incomplete/wrong.
The regex pattern;
should NOT validate domains such as:
-domain.com
, domain--.com
, -domain-.-.com
, domain.000
, etc...
should validate domains such as:
schools.k12
, newTLD.clothing
, good.photography
, etc...
After some further research; below is the most correct, cross-language and compact pattern I could come up with:
^(?!-)(?:(?:[a-zA-Zd][a-zA-Zd-]{0,61})?[a-zA-Zd].){1,126}(?!d+)[a-zA-Zd]{1,63}$
This pattern conforms with most* of the rules defined in the specs:
- Each label/level (splitted by a dot) may contain up to 63 characters.
- The full domain name may have up to 127 levels.
- The full domain name may not exceed the length of 253 characters in its textual representation.
- Each label can consist of letters, digits and hyphens.
- Labels cannot start or end with a hyphen.
- The top-level domain (extension) cannot be all-numeric.
Note 1: The full domain length check is not included in the regex. It should be simply checked by native methods e.g. strlen(domain) <= 253
.
Note 2: This pattern works with most languages including PHP, Javascript, Python, etc...
See DEMO here (for JS, PHP, Python)
More Info:
The regex above does not support IDNs.
There is no spec that says the extension (TLD) should be between 2 and 6 characters. It actually supports 63 characters. See the current TLD list here. Also, some networks do internally use custom/pseudo TLDs.
Registration authorities might impose some extra, specific rules which are not explicitly supported in this regex. For example, .CO.UK
and .ORG.UK
must have at least 3 characters, but less than 23, not including the extension. These kinds of rules are non-standard and subject to change. Do not implement them if you cannot maintain.
Regular Expressions are great but not the best effective, performant solution to every problem. So a native URL parser should be used instead, whenever possible. e.g. Python's urlparse()
method or PHP's parse_url()
method...
After all, this is just a format validation. A regex test does not confirm that a domain name is actually configured/exists! You should test the existence by making a request.
Specs & References:
UPDATE (2019-12-21): Fixed leading hyphen with subdomains.
与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…