How to grep faster?

Ajit Gunge · 05-14-2012, 07:16 AM

I have a folder which has a couple of txt files.Now what I want is I want to grep for a particlualr pattern in say the latest 4-5 files in the folder.It must list me those files and the occurrences of the line number in those files.How can I do this ?

pan64 · 05-14-2012, 07:33 AM

Code:

grep -n 'pattern' `ls -1t *.txt | head 6`

would grep in the last 6 files.

David the H. · 05-14-2012, 01:21 PM

No, it's almost never a good idea to parse ls for filenames, particularly when using a command substitution. Any name it hits that has whitespace in it will be unreadable.

http://mywiki.wooledge.org/ParsingLs

Unfortunately, it's not particularly easy to safely grab "n" number of files, based on metadata like modification time, but bashFAQ 99 gives you some tips on how to do it. Doubtless google will give more.

http://mywiki.wooledge.org/BashFAQ/099

(I'd help to work out a solution myself, but unfortunately I don't have much time now. I'll come back to it later when I have more time, and if the question is still active. )

Tinkster · 05-14-2012, 03:45 PM

Something like this may work ...

Code:

stat --printf "%Y\t%n\n" * | sort -k1,1n|tail -n 5

allend · 05-14-2012, 07:32 PM

If you are trying to confine the search to recently created files then using find will do what you want.

Code:

find . -mtime -6 -print0 | xargs -0 grep -n <search string>

Run from within the directory containing your files, this will look for all files created in the last week and then pass those file names to grep. The use of xargs avoids problem with filenames containing spaces.

Tinkster · 05-14-2012, 07:45 PM

Quote:

Originally Posted by allend

If you are trying to confine the search to recently created files then using find will do what you want.

Code:

find . -mtime -6 -print0 | xargs -0 grep -n <search string>

Run from within the directory containing your files, this will look for all files created in the last week and then pass those file names to grep. The use of xargs avoids problem with filenames containing spaces.

That's not the last few files, that's all files from the last 6 days ...

allend · 05-14-2012, 09:22 PM

Well, the OP did say

Quote:

I want to grep for a particular pattern in say the latest 4-5 files in the folder.It must list me those files and the occurrences of the line number in those files.

Depending on the frequency of new file creation then '-mmin' might be a better option than '-mtime'.

Ajit Gunge · 05-15-2012, 12:58 AM

Hi All,
Thanks for all your responses.That helps me a lot.I am marking the thread as solved as I think I have what I need.Thanks again all of you.

Ajit Gunge · 05-18-2012, 05:00 AM

Sorry guys I am marking this thread as unsolved again.When I run the command given by Thinkster I am getting errors as below

find: 0652-017 -print0 is not a valid option.
xargs: The -0 flag is not valid.
Usage: xargs [-p][-t] [-e[EndOfFileString]] [-E EndOfFileString]
[-I ReplacementString] [-i[ReplacementString]] [-L Number]
[-l[Number]] [-n Number [-x]] [-s Size] [Command [Argument ...]]
The uname -a command on my server gives the follwoing output.Can you please tell what command should I run to find all the files that have the pattern "xyz" fast.

whizje · 05-18-2012, 05:57 AM

Did you use the zero after print or the letter O it should be zero. As well as the -0 option.