Sorry, but you set yourself up for that: it usually works as soon as you do it right. ;-))

Seriously: grep expects its arguments in the form (i simplify to the relevant here): grep <pattern> <file>. What you fed it instead, though, is: grep <pattern> <variable, that is: string which to search for pattern>. This is why grep treated the string you wanted to be searched as a filename - and of course this filename is not found, which is why you got the error of "file not found".

There is in fact another form of grep-invocation, where you can feed grep something from <stdin>, like this:

Code:

some_process | grep <pattern>

To get your string there is possible but you first have to create a process with the string as output. Can you guess which process that could be?

Yes, of course: echo "$variable". Therefore, here is your solution:

Code:

echo "$variable" | grep <pattern>

Picture grep as a machinery which has two different intakes: a file or <stdin>. Just like a coffee machine might have two intakes, one for ground coffee and one for whole beans. Which one is selected depends on how it is invoked. But you cannot pour your data into the wrong intake and then expect grep to do it right: put your beans into the opening for ground coffee and the coffee machine will make all sorts of things but probably not coffee.

I hope this helps.

bakunin

PS: CR/LF line endings are indeed important and will confuse a lot of UNIX utilities. I suggest to correct them to UNIX newlines first thing. There are unix2dos and dos2unix, two utilities which change back and forth from/to UNIX-style line breaks from/to DOS-style line breaks, sed-scripts and many other ways of achieving the same.

PPS: you do not need to create a temporary file at all in your code: as i have shown you you can directly work with the variable content.