In sed, how to represent "alphanumeric or _ or -"

Question

Tried

[a-zA-Z0-9-_]
[a-zA-Z0-0\-\_]
[[[:alnum:]]-_]
...

What is the correct way to represent that in regular expression?

It seems like [a-zA-Z0-9-] works for alphanumeric or dash. But when I add another underscore, it breaks.

Solution 1

That will be this character class:

[[:alnum:]_-]

Which means allow one of these:

1. Alpha numeric
1. Underscore
1. Hyphen

It is important to keep hyphen at 1st or last position in character class to avoid escaping.

Solution 2

All of these variations will work:

[a-zA-Z0-9_\-]
[a-zA-Z0-9_-]
[-_a-zA-Z0-9]
[-a-z_A-Z0-9]
[-a-zA-Z_0-9]
[-a-zA-Z0-9_]
...

The hyphen needs to be on one end. It can be escaped or not if at the end, and must be un-escaped if at the beginning. The underscore can be almost anywhere, as long as it's not in the middle of one of your ranges, and doesn't need to be escaped.

Comments

[Phil Perry]

To elaborate, if an un escaped hyphen - shows up anywhere but the very first or last position, the regex processor thinks you want a range rather than a lone character.

[Amitesh Bharti]

To fetch alphanumeric characters of a file sed '/^[^-[:alnum:]]/d' file

[soMuchToLearnAndShare]

for some reason, the [[:alnum:]_-] somehow match the forward slash too and followed by 1 more letter. meaning, it somehow match things like this: abcdef/somethingelse as abcdef/s

[anubhava]

No that cannot match a /. That would be because of absence of anchors.