regular expression to capture the folder name with specfic structure

Question

what is the regular expression to capture the folder name with this structure

x.x.x.x-x ( x - integer number )

 ls /usr/hdp | .....
 2.6.0.3-8  current HG_MASTER 2.4.3 34.1 45-995

expected output

2.6.0.3-8



ls -lh /usr/hdp/
total 8.0K
drwxr-xr-x. 27 root root 4.0K Jan  1 15:24 2.6.0.3-8

As is traditional, http://mywiki.wooledge.org/ParsingLs is usually linked to about now. — Guy, Jan 22 '18 at 09:48
so what is your solution , if you not want to capture it with ls ? — jango, Jan 22 '18 at 09:50
both of the links above offer several alternatives. e.g. find with either a -name or -regex predicate (or the case-insensitive -iname and -iregex). — cas, Jan 22 '18 at 09:59

Egor Vasilyev · Answer 1 · 2018-01-23T07:15:21.030

1

You need to use find instead. For exact match use pattern:

find /usr/hdp -regextype sed -regex "\/usr\/hdp\/[0-9]*\.[0-9]*\.[0-9]*\.[0-9]*-[0-9]*"

If you need any files with only digits, . and - symbols in names try:

find /usr/hdp -regextype sed -regex "\/usr\/hdp\/[.0-9-]*"

edited Jan 23 '18 at 07:15

answered Jan 22 '18 at 10:14

Egor Vasilyev

2,513

not works on both way – jango Jan 22 '18 at 10:49
under /usr/hdp/ , we have 2.6.4.0-91 – jango Jan 22 '18 at 10:49
@jango, you did't say that filename can contain several digits consistently – Egor Vasilyev Jan 22 '18 at 10:53
@jango, can file contain other symbols? Can you complete your post with output of command ls -lh? – Egor Vasilyev Jan 22 '18 at 10:58
updated , see mu question again please – jango Jan 22 '18 at 11:22
so can you please update the answer ? – jango Jan 22 '18 at 11:32
Updated earlier. Works fine with GNU find 4.7.0 – Egor Vasilyev Jan 22 '18 at 12:13
close, but that won't actually work. you're running find /usr/udp but the pattern you're searching for starts with ./. I expect you tested with find .. either use /usr/hdp/ at the start of the search pattern, or use .*/. – cas Jan 23 '18 at 01:22
@cas, i edited my answer. Thank you very much! It's my mistake, i missed it in my test lab. I was looking for files in current directory with find . and forgot to replace. – Egor Vasilyev Jan 23 '18 at 07:21

Raman Sailopal · Answer 2 · 2018-01-22T11:08:44.403

1

find . -type d -regextype posix-egrep -regex '^./([0-9]+[.]){3}[0-9]+[-][0-9]+'

Use find and then the regextype flag to set the reg expression syntax to egrep. Use -regex to check against 3 digits and a "dot" and then a digit a "dash" and a digit.

If the digits are more than one character, you will need + after each digit.

edited Jan 22 '18 at 11:08

answered Jan 22 '18 at 10:15

Raman Sailopal

1,432

why - f ( this is folder ) – jango Jan 22 '18 at 10:48
Yeah sorry. It's a directory and so should be d and not f for file. I will change. – Raman Sailopal Jan 22 '18 at 10:49
also the same your syntax not capture the folder - 2.6.4.0-91 – jango Jan 22 '18 at 10:52
You will need to add + to each [0-9] to signify 1 or more digits between 0 and 9. I have amended my solution – Raman Sailopal Jan 22 '18 at 11:09
so can you please update the answer ? – jango Jan 22 '18 at 11:32
I have. That's what I said in my last sentence – Raman Sailopal Jan 22 '18 at 11:37

ilkkachu · Accepted Answer · 2018-01-22T11:52:08.597

1

Since you tagged bash, you can actually do this with the shell's filename matching, using extended globs:

$ shopt -s extglob
$ touch /some/path/{2.6.0.3-8,22.66.0.333-8,foobar}
$ printf "%s\n" /some/path/+([0-9]).+([0-9]).+([0-9]).+([0-9])-+([0-9])
/some/path/22.66.0.333-8
/some/path/2.6.0.3-8

The pattern there works like more common things like * or *.png, so you can just stick the fixed parts like paths in it. Add a trailing slash / to only match directories. (the slash does appear in the output too, though.)

In ERE regex, the pattern is the same as [0-9]+\.[0-9]+\.[0-9]+\.[0-9]+-[0-9]+. A fuzzier version would just check for the allowed characters, like +([-.0-9]) (or [-.0-9]+ in ERE), but that of course would also match e.g. 123..456.

If you want to just print the filenames, find might be easier, but if you want to run a command on them, then a shell loop is also an option (along with find -exec).

edited Jan 22 '18 at 11:52

answered Jan 22 '18 at 11:34

ilkkachu

138,973

nice but how to do it when I am from other folder for example , I am in /tmp and want to capture the folder from /usr/hdp/ – jango Jan 22 '18 at 11:44
OP mentioned matching directories.. but that should be okay if no filenames match.. :) – Sundeep Jan 22 '18 at 11:46
@jango, just stick the path in the front, as with * – ilkkachu Jan 22 '18 at 11:47
2

@Sundeep, ah, true. But you can add the trailing slash to only match directories. – ilkkachu Jan 22 '18 at 11:52

Burak Talha Bolat · Answer 4 · 2018-01-22T12:11:19.513

-1

ls /usr/hdp | egrep '[0-9]+.[0-9]+.[0-9]+.[0-9]+-[0-9]+'

edited Jan 22 '18 at 12:11

answered Jan 22 '18 at 09:48

Burak Talha Bolat

29
4

regular expression to capture the folder name with specfic structure

4 Answers4