pattern matching question

laopi · 11-05-2008, 10:18 AM

Hi,

I have a pattern file,

1111
2222
4444

and another working file

1111,abc
2222,dce
3333,ert
4444,dfg
5555,tuy

I only want the lines

1111,abc
2222,dce
4444,dfg

I can cat the working file line by line, but it's really slow, because my working file is huge, any idea?

peter

druuna · 11-05-2008, 10:30 AM

Hi,

Here are 2 that came to mind, there are other ways of doing it, but I think these are the best (speed wise).

Using egrep: egrep "1111|2222|4444" infile
Using sed: sed -n '/\(1111\|2222\|4444\)/p' infile

I think that the egrep version will be fastest, it does depend a bit how big the infile is.

Hope this helps.

laopi · 11-05-2008, 10:51 AM

my pattern file is big too, I think the command line will be oversized.

druuna · 11-05-2008, 10:57 AM

Hi,

That is an assumption or a fact?

laopi · 11-05-2008, 11:08 AM

my current pattern file is 5238 lines and the size is 240336.

druuna · 11-05-2008, 11:11 AM

Hi,

grep -f pattern.file in.file

One pattern per line in the pattern.file.

laopi · 11-05-2008, 11:34 AM

it's running some mins. ago, not finished, but I'm pretty sure that's what I want.

peter