Originally Posted by Corona688

It fulfilled your specifications and worked on the data you gave. If you wanted dates from the future instead of today's, you could have informed the program of that, or at least informed me of that. Perhaps I was being too "clever" and automating it when all you wanted was something easy to plug values you wanted into?

What you were expecting was -- and remains -- unclear. Tell me exactly what you wish to match for, and exactly what output you expect from it, don't tell me my code is bad for not reading your mind.

That's exactly what I did. The sed chops it into blocks so awk knows where records begin and end, and beyond that, you can do whatever you please.

"grep" doesn't work that way of course -- it matches lines and only lines. If you want something which understands "if line 1 matches something, do something with line 2, print part of line 3, delete part of line 4" that's exactly what awk is for.

I'll keep my example simpler this time, less chance of misunderstanding, and it's easier to improve a working script than figure out what a broken one's supposed to do.

Code:

sed 's/^"/\n"/' np_log.txt | awk '$1 ~ FILE { print }' RS="" FS="\n" OFS="\n" ORS="\n" FILE="abcd.job" -

Red is the actual awk program -- in this case, just "print the entire record if the first field matches FILE". I had it as index() before to make it more robust, this is a simpler illustration of what I'm doing.

Green defines variables. I moved them to the back to make it clearer. RS, FS, OFS, ORS are special variables, tailored to your data here so a "record" is exactly one data block and a "field" is exactly one line( so line 1 is $1, line 2 is $2, etc, $NF is the last line, $0 is the entire data block, etc).

Blue is the input file to read. There can be one or more. - reads from standard input.