More Efficient Regexp?

[jouell]

Hi All,

Can anyone make this regexp more efficient?

I am trying to grep for 2008:09:45 through 2008:10:45 am in a file:

egrep -e "2008:09:4[5-9]" -e "2008:09:5[0-9]" -e "2008:10:[0-3][0-9]" -e "2008:10:4[0-5]" FILE

other than just:

egrep -e "2008:09:4[5-9]|2008:09:5[0-9]|2008:10:[0-3][0-9]|2008:10:4[0-5]" FILE

Thanks!
-jouell

 

[feherke]

Hi

While you are trying to process numeric values with text utility, I see no nicer way.

But if the values are in chronological order without unwanted lines between them, you can get them as an interval :

sed -n '/2008:09:45/,/2008:10:45/p' FILE

Feherke.

http://rootshell.be/~feherke/

 

[Annihilannic]

Unfortunately the sed solution stops at the first occurrence of 2008:10:45. I presume you would want it to continue printing if there are multiple occurrences.

I don't think you can make the egrep/b] expression much more efficient than that. Is it performance you're trying to improve, or code brevity/readability?

Annihilannic.

 

[jouell]

Hi All

Thanks for the feedback. Good Question Annihilannic. I was looking to improve the code brevity and readability.

(Looks like this may be the best we can do with this setup)


Thanks!
-jouell

 

[Annihilannic]

Here are a couple of alternatives, but I don't think they're much clearer in the end:

egrep "2008:(09:(4[5-9]|5[0-9])|10:([0-3][0-9]|4[0-5]))" FILE
awk '{t=$1;gsub(":","",t)} t>=20080945 && t<=20081045' FILE
awk '$1>="2008:09:45" && $1<="2008:10:45"' FILE

Note that the awk solution is a lot less efficient than egrep, so if you're dealing with a significant amount of data, probably best avoided.

The second awk solution using string comparisons may be error-prone, you might want to test it more thoroughly than I have.

Annihilannic.