Shell Programming and Scripting

I have log file that is incremented every second and is rather big. I need monitor error in this file. I do not want to parse it all every time.

I want parse it first time then remember last position (as byte for example)
All consequent parsing I want to start from last saved position to the end of file.

Up to now I use this method

COUNT_OLD=`awk '$1>=0 {print $1}' file.seq`
COUNT_NEW=`cat file | wc –l` #get file lines count
cat file | grep -nf file.search | tr ':' ' ' | awk '$1 > '$COUNT_OLD' && $1 < ‘$COUNT_NEW’{print "FOUND: "$0}' > errors
echo $COUNT_NEW > file.seq

I.e. using “grep -n” feature manually select only new lines.So it does not really “incremental”.
Other approach is to use for example “head” or “tail -n” command to copy new line some where and parse only that file but I think it is inefficient…

I want to read directly from middle of the file. For example if file 1000 byte. I want to read directly from 100 byte to 999 byte.

Is it possible by standard shell commands? Or I need use some of programming languages?

To read bytes from a file:

Code:

## Command-line info:
file=$1
firstbyte=$2
lastbyte=$3

## Calculate bytes to skip and no. of bytes wanted
skip=$(( $firstbyte - 1 ))
count=$(( $lastbyte - $firstbyte + 1 ))

## Do it!
dd bs=1 skip=$skip count=$count < "$file" 2>/dev/null

To read lines from a file:

Code:

sed -n "$firstline,$lastline p" "$file"

can you please tell me what actually the values $1,$2 and $3..etc., does?

They are the command-line arguments.