UNIX for Dummies Questions & Answers

I have 40 files with varying number of columns
They however do have identical column names for the overlapping columns
I would like to grep some info from these specific colums

e.g
file1

id beta se direction N
2 .5 .01 + 1000
5 -.6 .02 - 2000


file2

id beta se N
2 .6 .015 3000
5 .3 .01 2000

So as you can see file 1 has a column with direction of effect while file2 hasn't

I would like to grep all beta values for ID2 from the files and put them in one file

Any suggestions?
Thanks a lot in advance

Have a look at the paste command. That may have what you need.


Robin

Thanks,
I've tried the paste command, but I do not know the exact column numbers, so need to specify the column name somehow

If you just want the first two columns of all the files, you could use the cut command.

Code:

grep "^2 " file1 file2|cut -f-2 -d " "

The ^ in the grep signifies the start of line.



Is this better?

Robin

Thanks,

the problem with the files is that the column I'm looking for is sometimes the 4th column and sometimes the 6th or 2nd. So there is not regularity in the files

You might need a two step process in a loop then. Are the various files each the same format each time? If so, then you can build a case statement to collect it all together into one:-

Code:

#!/bin/ksh

for file in <list of files>
do
   case $file in
      file1)   cols="-2"    ;;
      file2)   cols="1,4"  ;;
      file3)   cols="3,6"   ;;
      *)       cols="1-"    ;;
   esac

   cut -f$cols -d " " $file
done

Obviously adjust the file names and columns required to suit your needs.

The wild-card option at the end of the case statement just completes the syntax to that cut will get fields from the first to the end, i.e. everything.

You can add a | grep after the done statement to select the rows you want, in this example above you would end up with:-

Code:

::
::
::
   esac

   cut -f$cols -d " " $file
done | grep "^2 "

Does this help?


Robin

Pls use code tags as required by forum rules.

And, I'm pretty sure you don't insist on using grep.

If you want the values identified by column header, and then by one column that contains a search value, try

Code:

awk     'FNR==1         {for (n=1; n<=NF; n++) {if ($n==ID) IDCOL=n; if ($n==RES) RESCOL=n}; next}
         $IDCOL==SRCH   {print $RESCOL}
        '  RES=beta  ID=id  SRCH=2   file1 file2
.5
.6

based on your sample files.