Shell Programming and Scripting

Hi All,

I am working on one of the script developement for my project, where I need to find the distinct types of files from given directory based on pattern provided.

for e.g. directory listing is :

Code:

abc123.dat.20141212_021012
abc123.dat.20141312_041012
abc123.dat.20141112_031012
cdf023.dat.20141112_031012
cdf023.dat.20141112_031012
cdf023.dat.20141112_031012

Run the script : ksh find_distinct_files.ksh -Pattern *.dat.????????_??????

output should be :

Code:

1) abc123.dat
2) cdf023.dat

similarly, there can be any other pattern too for file in given directory , below are some more examples
e.g.
1) input pattern : ???_*.dat ==>
files in directory :

Code:

       123_abc.dat
        234_abc.dat

output : abc.dat



1) input pattern : *_????.dat ==>
files in directory :

Code:

        abc_2014.dat
        abc_2013.dat

output : abc.dat

My point is, there can be any pattern of the file may exists in the directory.. so I need to find the all files matching input pattern and keep distinct files type. May be in another worlds, from given String, remove portion matching the pattern and keep rest..


Can anybody give me some pointers / ideas of how this can be done.

Thanks in Advance ..

Thanks
Abhijeet R

Your specifications are vague, so I will make an assumption:
file names have 3 dot-separated fields in them - the first two fields are to be considered.

Code:

cd /some/directory
ls | awk -F '.' '{print $1 "." $2}' | sort -u

If you want a better answer, please provide better specifications.

Thanks for quick reply, my bad that I did not clarify myself . . .

My point was, there can be any pattern of the file may exists in the directory.. so I need to find the all files matching input pattern and keep distinct files type. May be in another worlds, from given String, remove portion matching the pattern and keep rest..


e.g.
1) input pattern : ???_*.dat ==>
files in directory :
123_abc.dat
234_abc.dat
output : abc.dat

1) input pattern : *_????.dat ==>
files in directory :
abc_2014.dat
abc_2013.dat
output : abc.dat

Please tell me if this explains the problem statement. updating above in the original post

Thanks
Abhijeet R

So - you want to ls the files matching the pattern and then remove the portinon matching one char wildcards ?, but keep the more generic wildcard(s) * matches? And you want to keep chars given literally, but remove single chars pre-, postfixed or interspersed in the ?s?
First, you'll have to quote the parameter when calling the script to prevent the shell from expanding it. find . -name "$1" might provide you with the desired file names. Let me think a while for the removal of the name portions.

Thats appropriate requirement.. I could not think of any solution to this prob.

For the first question, a possible solution is not too hard. Put

Code:

Y=\\\(.${1//\?/.} 
Y=${Y/../\\\)..}  
ls $1 | sed "s#$Y#\1#"

into a script and run it with your (quoted!) pattern as parameter 1. For above files, it will yield

Code:

abc123.dat
abc123.dat
abc123.dat
cdf023.dat
cdf023.dat
cdf023.dat

; you may want to pipe the result through a sort -u.

---------- Post updated at 14:58 ---------- Previous update was at 14:47 ----------

The second question is driving me crazy; I think it can't be resolved without making further assumptions on the structure of the filenames to be selected. It does not discriminate the two types given from each other. Run (combined with question 1 solution from above)

Code:

Y=${1//\?/.}
[[ "${1:1:1}" == "?" ]] &&
        Y=${Y%${1#*_}}\\\(.${1#*_}\\\) ||
        Y=\\\(.${Y/../\\\)..}

ls $1 | sed "s#$Y#\1#"

as a script with "???_*.dat" as parameter 1, and it will yield

Code:

abc.dat
abc.dat
2013.dat
2014.dat

The third one I don't even dare to tackle...

---------- Post updated at 15:41 ---------- Previous update was at 14:58 ----------

Simplified version anchoring solutions for 2 and 3 at the "_" char, still not resolving above issue:

Code:

case $1 in
        (*\?)   Y=${1#"${1%%.\?\?*}"}$ ;;
        (\?*)   Y=^${1%"${1#*_}"} ;; # ||
        (*)     Y=${1//[^\?_]/} ;;
esac
Y=${Y//\?/.}
ls $1 | sed "s#$Y##"