Take the following contents of a file:
"52245" "528" "06156903" "52246" "530" "00584709"
What pattern would match both 52245 and 52246 but nothing else?
Technically correct. I'm almost sure he's looking for something else, but wasn't specific about what.
Max Schmeling
2009-04-23 17:00:47
You forgot the quotation marks.
Michael Myers
2009-04-23 17:11:13
+1 for making me laugh
Tomalak
2009-04-23 17:12:26
@mmyers Oops, didn't see the "grep" tag. I supplied a java regex where quoties would not be used.
Brian Knoblauch
2009-04-23 17:29:20
+15
A:
Something that can only match those two numbers and nothing else:
^\"5224[56]\"$
Now if you're looking for something a bit more general (for example, any number with 5 digits), you'll want something like
^\"\d{5}\"$
I'm assuming the quotation marks (") are part of the file. If they aren't, omit the \" parts from the expression.
The particular grep expression you want is this:
grep -E "^\"[[:digit:]]{5}\"$" filename
or to take a suggestion from the comments:
grep -P "^\"\d{5}\"$" filename
I've tested both and they work on my machine!
Welbog
2009-04-23 16:59:14
Yes the quotation marks are part of the file. I am using grep like this:grep ^\"\d{5}\"$ myfileIt returns nothing, I have also tried with -G and -E optionsAny ideas what is wrong?
2009-04-23 17:05:17
\d is a Perl style special character. Try grep -P ^\"\d{5}\"$ . I think there are other ways to match digits without -P and \d, but they're not worth it.
David Berger
2009-04-23 17:08:42
Take care to escape everything from your shell though. In bash, that means using single quotes around the regexp.
David Schmitt
2009-04-23 17:10:11
i'm not sure that the \d{} is correct...I use [[:digit:]]\{5\}..and you should put " around your expression (use grep with -e too).
LB
2009-04-23 17:10:54
I've included your suggestion into the answer, David Berger. Good tip. I didn't know about -P until now.
Welbog
2009-04-23 17:12:20
+7
A:
^"5224[56]"$
^"5224(5|6)"$
^"52{2}4[56]"$
^"(52245|52246)"$
...
You should base the regex you use on the semantic you want to express. If you are looking for two arbitrary numbers use ^"(52245|52246)"$. If the numbers have any meaning - a type code or something like that - I would stick with ^"5224(5|6)"$.
Daniel Brückner
2009-04-23 17:01:21