Regex to parse docker tag? - regex

Question

'registry/rabbit',
'registry/rabbit:3',
'rabbit',
'rabbit:3'
Trying to come up with a regex that will match rabbit in the four cases above. Seems easy enough, but my regex-fu is failing me.

Answer 1

The format is a little under-specified, but this seems to work:
^(?:(?=[^:\/]{1,253})(?!-)[a-zA-Z0-9-]{1,63}(?<!-)(?:\.(?!-)[a-zA-Z0-9-]{1,63}(?<!-))*(?::[0-9]{1,5})?/)?((?![._-])(?:[a-z0-9._-]*)(?<![._-])(?:/(?![._-])[a-z0-9._-]*(?<![._-]))*)(?::(?![.-])[a-zA-Z0-9_.-]{1,128})?$
From the docs:
An image name is made up of slash-separated name components, optionally prefixed by a registry hostname. The hostname must comply with standard DNS rules, but may not contain underscores. If a hostname is present, it may optionally be followed by a port number in the format :8080. If not present, the command uses Docker’s public registry located at registry-1.docker.io by default. Name components may contain lowercase characters, digits and separators. A separator is defined as a period, one or two underscores, or one or more dashes. A name component may not start or end with a separator.
A tag name may contain lowercase and uppercase characters, digits, underscores, periods and dashes. A tag name may not start with a period or a dash and may contain a maximum of 128 characters.
Tests are here.

Answer 2

You can try:
(?:[a-z]+/)?([a-z]+)(?::[0-9]+)?
But you need to read the spec to replace a-z and 0-9 to all possible characters.
Alternatively, this regex will capture the container name without regard for the specification other than / and ::
(?:.+/)?([^:]+)(?::.+)?
Sample data and test
def s = [
'registry.example.com/org/image-name',
'registry/org/image-name',
'registry/image-name',
'image-name',
'registry.example.com/org/image-name:version',
'registry/org/image-name:version',
'registry/image-name:version',
'image-name:version',
]
s.forEach {
def image = (s =~ "(?:.+/)?([^:]+)(?::.+)?")[0][1]
println image
assert image == 'image-name'
}
Will output:
image-name
image-name
image-name
image-name
image-name
image-name
image-name
image-name

Answer 3

There is another solution to this problem which contains all varieties and does not use lookaheads.
// image/tag:v1.0.0
// 123.123.123.123:123/image/tag:v1.0.0
// your-domain.com/image/tag
// your-domain.com/image/tag:v1.1.1-patch1
// image/tag
// image
// image:v1.1.1-patch
// ubuntu#sha256:45b23dee08af5e43a7fea6c4cf9c25ccf269ee113168c19722f87876677c5cb2
// etc...
const dockerImageVerify = "^(([a-z0-9]|[a-z0-9][a-z0-9\\-]*[a-z0-9])\\.)*([a-z0-9]|[a-z0-9][a-z0-9\\-]*[a-z0-9])(:[0-9]+\\/)?(?:[0-9a-z-]+[/#])(?:([0-9a-z-]+))[/#]?(?:([0-9a-z-]+))?(?::[a-z0-9\\.-]+)?$"
This covers all cases.

Answer 4

To be precise, the question and almost all the answers have nothing to do with the docker image tag, but rather with the entire docker image name.
The regular expression for the image tag is much simpler:
/[a-zA-Z\d_][a-zA-Z\d-_.]{0,127}/
From the docs:
A tag name may contain lowercase and uppercase characters, digits, underscores,
periods and dashes. A tag name may not start with a period or a dash and may contain a
maximum of 128 characters.

Answer 5

The best source for the official regular expression would be the Go implementation for OCI references used by Docker itself.
To get the full pattern, we can build and execute the following:
package main
import (
"fmt"
"github.com/distribution/distribution/reference"
)
func main() {
fmt.Printf("%q\n", reference.ReferenceRegexp)
}
go mod init foo.example
go mod tidy
go run ./main.go
This gives us:
^((?:(?:[a-zA-Z0-9]|[a-zA-Z0-9][a-zA-Z0-9-]*[a-zA-Z0-9])(?:(?:\\.(?:[a-zA-Z0-9]|[a-zA-Z0-9][a-zA-Z0-9-]*[a-zA-Z0-9]))+)?(?::[0-9]+)?/)?[a-z0-9]+(?:(?:(?:[._]|__|[-]*)[a-z0-9]+)+)?(?:(?:/[a-z0-9]+(?:(?:(?:[._]|__|[-]*)[a-z0-9]+)+)?)+)?)(?::([\\w][\\w.-]{0,127}))?(?:#([A-Za-z][A-Za-z0-9]*(?:[-_+.][A-Za-z][A-Za-z0-9]*)*[:][[:xdigit:]]{32,}))?$
The capture groups are image name, image tag, and digest respectively.

Answer 6

I think this covers all:
const ipRegexStr = '((([0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\.){3}([0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))';
const hostnameRegexStr = '((([a-zA-Z0-9]|[a-zA-Z0-9][a-zA-Z0-9\-]*[a-zA-Z0-9])\.)*([A-Za-z0-9]|[A-Za-z0-9][A-Za-z0-9\-]*[A-Za-z0-9]))';
const tagNameRegexStr = '([a-z0-9](\-*[a-z0-9])*)';
const tagVersionRegexStr = '([a-z0-9\_]([\-\.\_a-z0-9])*)';
const dockerRepoTagRegexStr = `^(${ipRegexStr}|${hostnameRegexStr}/)${tagNameRegexStr}(:${tagVersionRegexStr})?$`;
const dockerTagRegex = new RegExp(dockerRepoTagRegexStr);
console.log(dockerTagRegex.test('11.com/0-2:122--')); //true
console.log(dockerTagRegex.test('192.168.1.2/0---2:232-323-d-')); //true
console.log(dockerTagRegex.test('-0-2')); //false
console.log(dockerTagRegex.test('0-2-')); //false
console.log(dockerTagRegex.test('0-2:-23423')); //false

Answer 7

This is yet another iteration inspired by above answers. Ideally there should be a long-format in the docker-compose file so we don't have to break these up. I'm also not sure if I'm naming things the right way (registry/image/tag:label). I can't vouch for how much it follows spec, whether it is too strict or not strict enough, etc. YMMV.
this is Python (not JS) so may not apply to most in the devops tooling space
allows strict uppercase variable names for the image tag labels (${VAR_NAME})
enforces some sane character limits but I did not research what the spec allows
this code block is in VERBOSE mode and needs to be wrapped in re.compile("<block>", re.X) to allow for comments/whitespace to be removed
# FIRST we have the registry/image/tag:label format
(?:^
# registry
(?:(?P<registry>
(?:
# IPv4
(?P<reg_ip>(?:(?:[0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5])\.){3}([0-9]|[1-9][0-9]|1[0-9]{2}|2[0-4][0-9]|25[0-5]))
|
# domain
(?P<reg_domain>(?:(?:[a-zA-Z][a-zA-Z0-9\-]{,48}[a-zA-Z0-9])\.){,4}(?:[A-Za-z]|[A-Za-z][A-Za-z0-9\-]{,48}[A-Za-z0-9]))
|
# hostname
(?P<reg_host>([A-Za-z][A-Za-z0-9\-]*[A-Za-z0-9]))
)
# port
(?P<reg_port>:[0-9]+)?
)
# registry is optional, and not greedy
/)??
(?P<image>[0-9a-z_-]{1,40})
(?:/
(?P<tag>[0-9a-z_-]{1,40})
)?
# label can be a literal or a variable
(?::
(?P<label>[a-z0-9][a-z0-9._-]{1,38}[a-z0-9]|\${[A-Z][A-Z0-9_]{,38}[A-Z0-9]})
)?
# next we have the digest format
$)|(?:^
(?P<user>[a-zA-Z]\w{,25})
#sha256:
(?P<hashcode>[0-9a-f]{64})
$)
Here's a code-hint to render a new image string after changing parts of the parsed image string:
def __str__(self):
if self.image:
image_tag = '/'.join(filter(None, (self.image, self.tag)))
imag_tag_label = ':'.join(filter(None, (image_tag, self.label)))
return '/'.join(filter(None, (self.registry, imag_tag_label)))
return f'{self.user}#sha256:{self.hashcode}'