How do I match all lines not matching a particular pattern using grep? I tried this:
grep '[^foo]'
-
15[^error_log] would never ever work anyway, [] are char classes, regexp 's in general are not good at negative patterns (unless the engine implements negative lookaheads).Jaap– Jaap2011-11-09 15:49:54 +00:00Commented Nov 9, 2011 at 15:49
Add a comment
|
3 Answers
grep -v is your friend:
grep --help | grep invert
-v, --invert-match select non-matching lines
Also check out the related -L (the complement of -l).
-L, --files-without-match only print FILE names containing no match
14 Comments
Babken Vardanyan
Worth mentioning that for multiple (negative) matches
-e option can be used: grep -v -e 'negphrase1' -e 'negphrase2'
Nicholas Adams
Similar to the comment from @Babken-Vardanyan Also - able to use pipes to join multiple matches e.g.
grep -v 'negphrase1|negphrase2|negphrase3'
Evan Langlois
Last comment is NOT the same as it will search for things that don't match both rather than either. ie If it matches one but not the other its still printed. Try it both ways with non-similar strings
Zlemini
@EvanLanglois - forcing grep to interpret the pattern as an extended regular expression using
-E works, i.e. grep -vE 'negphrase1|negphrase2|negphrase3'
Motti
@OlleHärstedt, I think I misunderstood your scenario in my previous comment, the following may be what you're looking for
grep "" /dev/null * | grep foo | grep -v bar | cut -d: -f1 | sort -u (why the first grep?, there's always a way :)) |
You can also use awk for these purposes, since it allows you to perform more complex checks in a clearer way:
Lines not containing foo:
awk '!/foo/'
Lines containing neither foo nor bar:
awk '!/foo/ && !/bar/'
Lines containing neither foo nor bar which contain either foo2 or bar2:
awk '!/foo/ && !/bar/ && (/foo2/ || /bar2/)'
And so on.
7 Comments
Peter T.
That's actually quite cool. You don't even have to learn the complete awk language in order to group regexp with logical operators. Thanks for this answer!
Eduardo Pignatelli
The OP specifically asks for
grep. Why is this upvoted?
Ed S
It's possible that the OP wasn't aware of
awk capabilities and providing an alternate solution that provides the same result is helpful.
Old Uncle Ho
The awk command syntax you outline is very clear -- but I must be doing it wrong as it's not working for me! I'm trying to select certain lines of output from a locate command. Each line should not contain either of 'icon' or 'timeshift' or 'Papirus' to reduce output for clarity. This does not work: locate mime | awk '!timeshift && !icon && !Papirus' | wc -l , producing 58234 lines. A revised grep command produced 933 lines which is the correct output. Wonder where I strayed? Or maybe it's a change in or different version of awk? GNU Awk 5.1.0, API: 3.0 (GNU MPFR 4.1.0, GNU MP 6.2.1)
fedorqui
@OldUncleHo you need to enclose the string within
/ /, so that Awk sees is as a string. Otherwise, it will treat timeshift and the other values as variable names. All together, you need to use awk '!/timeshift/ && !/icon/ && !/Papirus/ or using some regex-fu awk '!/(timeshift|icon|Papirus)/'. |
In your case, you presumably don't want to use grep, but add instead a negative clause to the find command, e.g.
find /home/baumerf/public_html/ -mmin -60 -not -name error_log
If you want to include wildcards in the name, you'll have to escape them, e.g. to exclude files with suffix .log:
find /home/baumerf/public_html/ -mmin -60 -not -name \*.log
1 Comment
gawkface
while one is going to use
mmin to search for files modified within 60 mins, use -type f too as mentioned here stackoverflow.com/a/33410471/2361131