Question

I have a regular expression that has several matches in a textfile.
I want to copy only the the matches to a second file. I dont want to copy the lines that contain the matches: I only want the matched text.

I dont find a way to do this in notepad++ (only copies complete lines, not only the matches). Also not in Visual Studio search.

Is there a way to copy only matches? Maybe in grepp or sed?

Answer 1

You can do it with both. Lets say I have a following file –

Sample file:

[jaypal:~/Temp] cat myfile 
this is some random number 424-555
and my cell is 111-222-3333
and 42555 is my zip code

And I want to capture only numbers from myfile

Using sed:

With sed you can use the combination of -n and p option along with grouped pattern.

sed -n 's/.[^0-9]*\([0-9-]\+\).*/\1/p'
   |   |          |          |  |  ||
    ---            ----------    -- |
     |                  |        |  ---------> `p` prints only matched output. Since
     V                  V        V              we are suppressing everything with -n
 Suppress       Escaped `(`      \1 prints      we use p to invoke printing.
 output        start the group   first matched   
               you can reference  group
               it with \1. If you
               have more grouped
               pattern then they can
               be called with \2 ...

Test:

[jaypal:~/Temp] sed -n 's/.[^0-9]*\([0-9-]\+\).*/\1/p' myfile 
424-555
111-222-3333
42555

You can simply re-direct this to another file.

Using grep:

You can use either –

egrep -o "regex" filename

or

grep -E -o "regex" filename

From the man page:

-E, --extended-regexp
    Interpret PATTERN as an extended regular expression (see below).

-o, --only-matching
    Show only the part of a matching line that matches PATTERN.

Test:

[jaypal:~/Temp] egrep -o "[0-9-]+" myfile
424-555
111-222-3333
42555

You can simply re-direct this to another file.

Note: Obviously these are simple examples but it conveys the point.