I want to write a script that merges contents of several .csv files in one .csv file, i.e appends columns of all other files to the columns of first file. I had tried doing so using a "for" loop but was not able to proceed with it.

Does anyone know how to do this in Linux?

  • By merge, you mean that you want to combine lines from file A and file B if, for example, their first entry is the same? In that case, join might be for you, otherwise, errr, cat? – Ulrich Schwarz Jul 4 '16 at 12:42
  • You need to explain the type of merge you want; adding file2's columns as new columns to file1? Or ading file2's rows after file1? Or matching on an index key? Or...? – Stephen Harris Jul 4 '16 at 12:42
  • Appending columns of all other files to the columns of first file. – rmb Jul 4 '16 at 12:44
  • 1
    paste may be what you're looking for – Stephen Harris Jul 4 '16 at 13:26
  • Please edit your question and show us an example of your input file(s) and desired output file(s). There are many ways to "merge" csv files. – terdon Jul 4 '16 at 13:51

Here's a perl script that reads in each line of each file specified on the command line and appends it to elements in the array (@csv). When there's no more input, it prints out each element of @csv.

The .csv files will be appended in the order that they are listed on the command line.

WARNING: This script assumes that all input files have the same number of lines. Output will likely be unusable if any file has a different number of lines from any of the others.


use strict;

my @csv=();

foreach (@ARGV) {
  my $linenum=0;

  open(F,"<",$_) or die "couldn't open $_ for read: $!\n";

  while (<F>) {
    $csv[$linenum++] .= "," . $_;


foreach (@csv) {
  s/^,//;   # strip leading comma from line
  print $_,"\n";

Given the following input files:

==> 1.csv <==

==> 2.csv <==

==> 3.csv <==

it will produce the following output:

$ ./mergecsv.pl *.csv

OK, now that you've read this far it's time to admit that this doesn't do anything that paste -d, *.csv doesn't also do. So why bother with perl? paste is quite inflexible. If your data is exactly right for what paste does, you're good - it's perfect for the job and very fast. If not, it's completely useless to you.

There are any number of ways a perl script like this could be improved (e.g. handling files of different lengths by counting the number of fields for each file and adding the correct number of empty fields to @csv for each of the file(s) which are missing lines. or at least detecting different lengths and exiting with an error) but this is a reasonable starting point if more sophisticated merging is required.

BTW, this uses a really simple algorithm and stores the entire contents of all input files in memory (in @csv) at once. For files up to a few MB each on a modern system, that's not unreasonable. If, however, you are processing HUGE .csv files, a better algorithm would be to:

  • open all the input files and, while there's still input to read:
    • read a line from each file
    • append the lines (in @ARGV order)
    • print the appended line

The simplest approach for achieving that would be typing the following command

cat *csv > combined.csv

This file would contain the contents of all your csv files just in the way you mentioned.

  • 1
    Wouldn't this copy the rows in a common file instead of the columns? – fschmitt Jul 4 '16 at 13:53
  • @fschmitt Copying the rows is appending to the columns, right? – Kusalananda Jul 4 '16 at 13:55
  • @Kusalananda no, it's just concatenating the files. – cas Jul 4 '16 at 14:00
  • @cas Matter of interpretation. If I have a row and append values to it, they go on the same row. If I have a column and append values to it, they go in the same column. If I want to append the columns to columns, do they go as new columns, or are the appended to the end of the existing ones? – Kusalananda Jul 4 '16 at 14:19
  • 4
    In this particular case, what the OP wants is: for each individual line of all input files, append the fields line to make one long line with all the columns, and then repeat the process for all subsequent lines. This is hard to describe unambiguously in plain English - the algorithm in my perl script (or the more memory-efficient algorithm described after it) explains it better and is easier to follow. – cas Jul 4 '16 at 14:34

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.