UNIX for Dummies Questions & Answers

Hi,
i have a file where the end-of-file might be at the end of of a valid text line or on a new line

case a)
p q r
s t u <eof>

case b)
p q r
s t u
<eof>

case c)
p q r
s t u
<no data, only carriage return>
<eof>

I have a requirement where <eof> line should not be read if it's on a newline (in a grep command)

I don't want to use tail -1 because that will omit data in case a), and
don't want to ignore blank lines because will lose the blank line after 's t u' in case c)

How do i do this?

Thanks,
-srinivas yelamanchili

@ysrini: Did you try searching the forum? There're several posts on this one.

Code:

grep -v "^<eof>" inputfile

balajesuri, <eof> is not a string by itself, here eof stands for end-of-file and we use <...> notation for explaining something that's not obvious
In the data i provided since you cannot see where the end of file character is i denoted that by <eof>

My question was on how to avoid a newline that's the end of the file too, and not the string "<eof>"

Like newline is represented by "\n", how's eof character represented?

Code:

sed '$d' inputfile; [[ `sed '$!d' inputfile` != '' ]] && sed '$!d' inputfile

Another one:

Code:

[[ `tail -1 input` != '' ]] && cat input || sed '$d' input