ansaurus

Question

Answer 1

+1 A:

grep "model" filename | head -n 5 > newfile

firedfly 2009-08-25 21:36:37

Answer 2

+5 A:

grep model [file] | head -n 5 > [newfile]

ghills 2009-08-25 21:36:50

Answer 3

A:

cat file | grep model | head -n 5 > outfile.txt

2009-08-25 21:37:56

This is horrific, using `cat` will load the whole file into memory first before running `grep` . Just call `grep` with a file parameter.

Matthew Scharley 2009-08-25 21:41:56

The other answers are better, but the above comment is not true. The shell will start all 3 commands without waiting for any output first.

mark4o 2009-08-25 21:47:35

On Windows, pipes are available on the command line, but the deal is that the first program is run and written to a file. Then, the next program is run on those results. That's not how Linux/Unix works. Unix isn't that brain damaged. And neither reads the whole file into memory first. Where did you get that idea?

xcramps 2009-08-25 21:51:55

I've been piping [z]cat to grep for many years, and I'm a creature of habit. That said, the poster did say "a giant file" and there is plenty of evidence that avoiding the call to cat can save significant time in that case

2009-08-25 21:55:04

@smalloy, how would cat "save significant time"? On my system, head(1) exits when it is done, and then cat(1) harmlessly terminates from a SIGPIPE having read only the beginning of its input.

pilcrow 2009-08-25 22:02:49

To memory, or to file, either will result in a significant time expenditure that results in no net gain

Matthew Scharley 2009-08-25 23:43:02

Answer 4

+11 A:

     grep model old_file_name.txt -m 5 > new_file_name.txt

No need for a pipe. grep supports almost everything you need on it's own.

scragar 2009-08-25 21:40:15

Thanks! Didn't know about `-m`.

seth 2009-08-25 21:43:59

ansaurus

tags:

views:

answers:

Grep and Pipe in Linux

related questions