Grep log for certain word only if followed by another word

Question

I have a log file with entries like the following:

Input

php notice bob
php notice winner
php notice blah winner
php notice asfsdf test
php erro)r bob                                                                                                
php error winner
php error trash winner
php error junk test
print this line

Output

php notice winner
php notice blah winner                                                                                        
php error winner
php error trash winner
print this line

What I'm trying to do is match all lines from a log file unless - it contains php followed by error or notice. If it contains php followed by error or notice, then it should only match if winner is also in the string. My regex-foo is weak. Thanks.

Answer 1

Use this: grep 'php\s\+\(error\|notice\).*winner' mylogfile

Answer 2

There might be a way to get grep to pull that off in one string but I think personally I'd make it script to handle that in either perl or bash. In bash it would be something like

while read line
do
# If line contains "php error" or "php notice" and "winner", print it, else skip it
   if echo $line | grep "php error" >/dev/null  || echo $line | grep -q "php notice" >/dev/nulll; then
       if echo $line | grep "winner" > /dev/null; then
           echo $line
       fi
   else 
   # If line does not contain "php error" or "php notice", print it.
       echo $line
   fi
done

I don't have a system handy to test/debug this right now, so you may have to tweak it a bit.