CODEWITHSUNDEEP
bash
Blomex
Blomex

Reputation: 417

Loop for all files with certain name in directory in bash

I'm trying to make a script that would test whether my file is exactly as it should be, but I haven't been using bash before:

 #!/bin/bash
./myfile <test.in 1>>test.out 2>>testerror.out
if cmp -s "test.out" "pattern.out"
    then
        echo "Test matches pattern"
    else
        echo "Test does not match pattern"
    fi
if cmp -s "testerror.out" "pattern.err"
    then
        echo "Errors matches pattern"
    else
        echo "Errors does not match pattern"
    fi

Can I write it in such way that after calling ./script.sh myfile pattern my scripts would run over all files named pattern*.in and check if myfile gives same files as pattern*.out and pattern*.err ? e.g there are files pattern1, pattern2, pattern4 and i want to run test for them, but not for pattern3 that doesn't exist.

Can I somehow go around creating new files? (Assuming i don't need them) If I were doing it from command line, I'd go with something like

< pattern.in ./myfile | diff -s ./pattern.out

but I have no idea how to write it in script file to make it work.

Or maybe i should just use rm everytime?

Upvotes: 3

Views: 311

Answers (1)

cxw
cxw

Reputation: 17041

If I understand you correctly:

for infile in pattern*.in ; do  
    outfile="${infile%.in}.out"
    errfile="${infile%.in}.err"

    echo "Working on input $infile with output $outfile and error $errfile"
    ./myfile <"$infile" >>"$outfile" 2>>"$errfile"
    # Your `if`..`fi` blocks here, referencing infile/outfile/errfile
done

The % replacement operator strips a substring off the end of a variable's value. So if $infile is pattern.in, ${infile%.in} is that without the trailing .in, i.e., pattern. The outfile and errfile assignments use this to copy the first part (e.g., pattern1) of the particular .in file being processed.

Upvotes: 1

Related Questions