4

I have two files fileA and fileB.

I have to extract column1 from fileA like awk '{print $1}' and then the output will be searched into other fileB and will save the matched records into a new file fileC in simple words like:

fileA:

seg1     rec1
seg2     rec2
seg3     rec3 

I need to retrieve column 1 by using awk command and this column 1 is searched into fileB to retrieve the records like:

fileB:

seg1     one
seg2     two
seg3     three
seg4     four
seg5     five

From fileA, column1 data is extracted and and this data is used to search in fileB and matched record is saved to a test file. My output should be like this:

fileC:

seg1       one
seg2       two
seg3       three
8

Can be achieved easily with awk as follows:

awk 'NR==FNR{inFileA[$1]; next} ($1 in inFileA)' fileA fileB > write_to_fileC

result,

seg1       one
seg2       two
seg3       three

at above, first we are reading the fileA and holds the entire column1 from into an array named inFileA, then look in fileB for its first column and if it's matched with the saved column1 from fileA then goes to print entire row of fileB.

4

If the columns to be compared are sorted, you can use join:

join -o 2.1,2.2 file1 file2

join matches sorted columns from input files and prints them. -o 2,1,2.2 restricts the output to first and second columns of the second input file.

3

You can use the following one-liner:

cut -f1 fileA | grep -f - fileB > fileC
  • the cut command will extract the first column of fileA (assuming tab separation. use -d to specify something else)
  • the grep command takes the output of cut and searches fileB for all strings.
  • the output will be written to fileC
  • Note that this assumes the fields are tab-separated. – terdon Dec 19 '16 at 15:10
  • @terdon yes, I know. I mentioned this in the first bullet point ;-) – Wayne_Yux Dec 19 '16 at 15:15
  • Argh. So you did. Sorry! – terdon Dec 19 '16 at 15:22
2

An attempt with bash script. (Remember to make executable.)

fileA and fileB should exist in the same folder as the script.

A general script which will work for any two files described with script and generate the file with matching text as <fa>_<fb>_match.txt:

To use this, run ./script_name.sh fileA fileB

#!/bin/bash
fa="$1"  # first file- which has  columns
fb="$2"  # second file - which has  raw data to be searched
# file with name <fa>_<fb>_match.txt will be generated.

myarr=($(awk 'NR>1 {print $1}' "$fa")) # NR makes awk to ignore first row.

for index in ${!myarr[@]}; do
    #echo $index/${#myarr[@]}
    #echo    "${myarr[index]}"
text="${myarr[index]}"
grep -w -F "$text" $fb  >>  $fa"_"$fb"_match".txt
done

# file with name <fa>_<fb>_match.txt will be generated.
  • This will also match substrings. For example, if fileA contains fooseg1 and file2 contains seg1, it will be considered a match. – terdon Dec 19 '16 at 15:13
  • 1
    You could fix that with grep's -w option. You could also do the whole thing in one step, there's no need for the array: awk 'NR>1 {print $1}' "$fa" | while read index; do .... Oh, and you should always quote your variables. That said, using the shell for this sort of thing is really not a good idea. It is hard to write, easy to get wrong and will always be very slow. – terdon Dec 19 '16 at 15:46
  • I am relatively new to shell scripting, and as mentioned by the links, I will study and improve my style. Thanks for the links. Really helpful ! – ankit7540 Dec 19 '16 at 16:41
  • You're very welcome. We're all here to learn! – terdon Dec 19 '16 at 16:51
2

You have already received some excellent answers. Just to add to the mix, here's a Perl approach:

$ perl -ane '$i ? $k{$F[0]} && print : { $k{$F[0]}++ }; $i++ if eof' fileA fileB
seg1     one
seg2     two
seg3     three

And a golfed version of KasiyA's answer:

$ awk 'NR==FNR ? a[$1] : $1 in a' fileA fileB 
seg1     one
seg2     two
seg3     three

And here's a kinda convoluted grep solution:

$ grep -Ff <(grep -oP '^\S+' fileA) fileB
seg1     one
seg2     two
seg3     three

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.