I have a text file title 'results' and it contains
[ PASSED ] 11 tests.
[ PASSED ] 590 tests.
[ PASSED ] 1231 tests.
[ FAILED ] 4 tests.
[ FAILED ] 500 tests.
I would like to add the PASSED tests and store into a variable. Add the FAILED tests, add them and store into another variable.
How can I do this?
One quick way using awk.
Assuming your test output is in a file called test.out:
#!/bin/bash
npasses=$(<test.out awk '
/ PASSED / { total += $4 }
END { print total }')
echo number of passing tests: $npasses
<test.out means awk reads from test.out.
/ PASSED / { total += $4 } appends the forth field to a variable called total, but only for lines matching the regex PASSED.
END { print total } runs at end of file, and prints the value stored in total.
If the log is in a file, you can use
regex='\[ (PASSED|FAILED) \] (\d+) tests.'
while read -r line; do
[[ $line =~ $regex ]] || continue
count=${BASH_REMATCH[2]}
case ${BASH_REMATCH[1]} in
PASSED) let passed += count ;;
FAILED) let failed += count ;;
esac
done < input.txt
To read directly from another process, replace the last line with
done < <( sourcecommand )
Don't pipe the output of sourcecommand in to the while loop; that will cause passed and failed to be updated in a subshell.
Using bash 4's associative arrays:
declare -A total
while read _ result _ n _; do
((total[$result]+=$n))
done < results
for key in "${!total[@]}"; do
printf "%s\t%d\n" "$key" ${total[$key]}
done
PASSED 1832
FAILED 504
Expanding from Mikel's answer,
You can use eval to set the variables directly.
eval $(awk '/PASSED/ {pass += $4} /FAILED/ {fail += $4} END {print "pass="pass";fail="fail}' FILE_WITH_DATA)
You now have the variables set for you.
echo $pass
echo $fail
awk.casestatement.bash:. Why do you want to do this inbash? It seems easier to write a smallawkscript, and run store the awk script's output in a bash variable.