How to improve the way I handle greping in this script

Question

What my script does is:

1. Append data to a file on a continuous basis.
2. If an error pattern is observed, then do something.

I need feedback on the way I am grepping for patterns in the continuously appended file. For grep, I am creating a pattern file first and then using it with -f option to grep. Can I do this better? Any other better way I can write my over all logic.

Here's my script

#!/bin/bash
# List of error patterns to look for. Add more patterns to this list
p1='pattern1'
p2='pattern2'
p3='pattern3'
p4='pattern4'
PATTERNFILE=~/patternfile
FILETOCHECK=~/filetocheck
captureFileToCheck()
{
 //appends data into filetocheck &
}
createPatternFile()
{
 for i in `seq 100` # hopefully 100 is all we will need
 do
 if [ ! "x$p{$i}" = "x" ]
 then
 echo $p{$i} >> $PATTERNFILE
 else
 break
 fi
 done
}
createPatternFile
captureFileToCheck
while [ 1 ]
do
 if grep -f $PATTERNFILE $FILETOCHECK
 then
 echo "something starting"
 #doSometing
 echo "something complete"
 break
 fi
done

• l0b0: Any specific issue you're having? Security? Efficiency? Accuracy? Complexity?

Answer 1

Here is another go at your function

patterns() {
cat <<EOF
pattern1
pattern2
pattern3
pattern4
EOF
}

However, the $patternfile is not a necessity. I would go with just

while :
do 
 grep -f <(patterns) $filetocheck && {
 echo found;
 break
 }
done

• Ankur Agarwal: This is a good way to do it. I am concerned about portability!
• Rahul Gopinath: Ignore my comment about portability. Arrays are ok since this is explicitly bash.

Answer 2

Use more quotes, as the saying goes.

I find while true is more clear, but the idiomatic while : is also popular.

You might want to use an array for the patterns, then you can simply loop over it (for pattern in "${patterns[@]}") instead of using seq and the if statement.

Uppercase variables are usually used for exports; that is, stuff which is relevant outside your script.

You should avoid cluttering ~; use tmp_dir="$(mktemp -d)" instead.