Pipe grep result to awk with ": " as field separator

Question

I would like to filter the result of a grep command, eg :

myRepo/path/to/my/file.php:123:                             error_log(' - myError');

If I do the following, it works.

echo " myRepo/path/to/my/file.php:123: error_log(' - myError');" | awk -F': ' '{print $1}'

But when it's the result of a grep command, it outputs the whole line, why ?

grep -rn "myError" | awk -F': ' '{print $1}'

I have CentOS 6 with Awk 3.1.7, bash 4.1.2

score 1 · Accepted Answer · answered Aug 26 '16 at 16:19

1

Your awk delimiter is a : and a space, but I would guess that your code is using tab for indentation (and hence does not match just on a space).

You can try to change your grep to the following (which causes awk to match on : and the tab character).

grep -rn "myError"|awk -F':\t' '{ print $1 }'

This works for me using bash 4.3.46 and awk 4.1.3 on slackware linux.

answered Aug 26 '16 at 16:19

halldorhaf

Oh, the issue was that it's either a space or a tab. So I ended up with : awk -F': |:\t' – Bastien974 Aug 26 '16 at 17:00

score 1 · Answer 2 · answered Aug 26 '16 at 16:39

1

Try this:

grep -rn "myError" | awk -F':[ \t]*' '{print $1}'

answered Aug 26 '16 at 16:39

NuTTyX

Still getting the whole line. – Bastien974 Aug 26 '16 at 16:53
And changing \t for an actual tab character? – NuTTyX Aug 26 '16 at 16:54
?! tab button doesn't do that in a shell and in a text editor it depends on which column I am. – Bastien974 Aug 26 '16 at 16:58
Doesn't do what? Typing a tab does work both in a shell and in a text editor. If it doesn't work, just use `-F ':' ` and trim leading spaces with sed `| sed 's/^[ \t]*//'` – NuTTyX Aug 26 '16 at 17:05

2 Answers2