Regex file name with multiple digits

Question

Say I have the following file:

test_file-1234-master.tar.gz

I have tried to un tar using the following commands

 tar -xf test_file-[0-9]+-master.tar.gz
 tar -xf test_file-[:digit]-master.tar.gz

But no luck. How to match this pattern ?

NOTE: There will be always one file. I'm not trying to open multiple files.

Solution 1

With ksh, bash -O extglob and zsh -o kshglob only:

test_file-+([[:digit:]])-master.tar.gz

In bash, you have to set the extglob option first. This +(...) matches one or more occurrences of the given patterns. [:digit:] when inside a [...] bracket expression is a POSIX defined character class which includes Indo-Arabic decimal digits ([[:digit:]] is the same as [0123456789] or [0-9]).

It will match:

test_file-1234-master.tar.gz
test_file-1-master.tar.gz
test_file-123456789-master.tar.gz

It will not match:

test_file-1b-master.tar.gz
test_file--master.tar.gz
test_file-a1-master.tar.gz
test_file-abcd-master.tar.gz
test_file-Ⅵ-master.tar.gz # roman numeral
test_file-٨-master.tar.gz  # Eastern Arabic decimal digit

---

The tar command in your question should then be done like this (with a loop):

shopt -s extglob # bash
# setopt kshglob # zsh
for f in test_file-+([[:digit:]])-master.tar.gz; do
    tar xf "$f"
done

The more idiomatic short syntax in zsh is:

setopt extendedglob
for f (test_file-[0-9]##-master.tar.gz) tar xf $f

(# being the extendedglob equivalent of regexp *, and ## or +).

Solution 2

You're trying to use extended regular expression metacharacters and POSIX character classes (+ and [:digit:] respectively) in a globbing pattern;

Assuming bash or similiar, "basic" globbing only supports a handful of wildcards such as:

• ?: single charater
• *: zero or more characters
• []: character class
• {}: list
• [!]: negated character class

Differently from metacharacters in extended regular expressions, in "basic" globbing there's no way to modify a wildcard's behavior to match a fixed number of occurences;

So, using "basic" globbing, the shortest and tightest pattern would be:

tar -xf test_file-[0-9][0-9][0-9][0-9]-master.tar.gz

Comments

[Alessio]

shells do globbing, not regex. see unix.stackexchange.com/questions/64067/regex-match-in-cli

[Alessio]

to answer your question, you want: test_file-[0-9]*-master.tar.gz

[chaos]

@cas This will match one digit followed by anything.

[Alessio]

yes, so it will. test_file-[0-9][0-9][0-9][0-9]-master.tar.gz then.