Shell Programming and Scripting

Hi,

I have one file as:

Code:

$cat file1
abc  pqr  123
def  wxy  234
xyz  ghi   567

and another file as

Code:

$cat file2
345
456
987

I want the output of file2 to be appended in file1. And file 1 should look like:

Code:

$cat file1
abc  pqr  123  345
def  wxy  234  456
xyz  ghi   567  987

Please let me know how to move forward with this.
Thanks in advance.

Code:

paste file1 file2

Since you want two spaces between joined lines and want to overwrite one of your input files, try:

Code:

cp file1 tmp.$$ && paste -d " " tmp.$$ /dev/null file2 > file1 && rm tmp.$$

Sir , Can you please help me to understand the command that you have suggested?

1) Copy the contents of file1 to a file named tmp.$$ (note that this is Don Cragun's preference, you can change tmp.$$ to whatever name you want, myAux for example, see below), where $$ expands to the PID of your current shell (you can check this by running the following 2 commands and comparing their respective outputs echo $$ and ps aux | grep bash | grep -v grep)

Code:

gacanepa@debian:~$ echo $$
3428
gacanepa@debian:~$ ps aux | grep bash | grep -v grep
gacanepa  3428  0.0  1.3   6196  3380 pts/0    Ss   11:42   0:01 -bash
gacanepa@debian:~$

2) Merge files tmp.$$, /dev/null, and file2, line by line, and set the delimiter a space instead of a tab (that is what the -d option of the paste command is used for). The special file /dev/null is used here to add an extra space before the 4th column to match the same space separator as the other columns (which according to your example, seems to be 2 spaces).
3) Redirect the output to file1 and overwrite its contents.
4) Remove the auxiliary file tmp.$$ (or myAux, as seen below)
Without /dev/null:

Code:

gacanepa@debian:~/scripts/bash/tests$ cat file1
abc  pqr  123
def  wxy  234
xyz  ghi  567
gacanepa@debian:~/scripts/bash/tests$ cat file2
345
456
987
gacanepa@debian:~/scripts/bash/tests$ cp file1 myAux && paste -d " " myAux file2 > file1 && rm myAux
rm: remove regular file `myAux'? y
removed `myAux'
gacanepa@debian:~/scripts/bash/tests$ cat file1
abc  pqr  123 345
def  wxy  234 456
xyz  ghi  567 987
gacanepa@debian:~/scripts/bash/tests$

With /dev/null:

Code:

gacanepa@debian:~/scripts/bash/tests$ cat file1
abc  pqr  123
def  wxy  234
xyz  ghi  567
gacanepa@debian:~/scripts/bash/tests$ cat file2
345
456
987
gacanepa@debian:~/scripts/bash/tests$ cp file1 myAux && paste -d " " myAux /dev/null file2 > file1 && rm myAux
rm: remove regular file `myAux'? y
removed `myAux'
gacanepa@debian:~/scripts/bash/tests$ cat file1
abc  pqr  123  345
def  wxy  234  456
xyz  ghi  567  987
gacanepa@debian:~/scripts/bash/tests$

Hope it helped .

Hi sidpatil,
I believe gacanepa already explained how the paste command portion of the AND list:

Code:

cp file1 tmp.$$ && paste -d " " tmp.$$ /dev/null file2 > file1 && rm tmp.$$

works.

If I had known the name of the script that Bhrigu was creating (assume scriptname for demonstration purposes), I would have chosen the name scriptname.$$ instead of tmp.$$ to make it more obvious what script created the temp file. Note that using scriptname.$$ rather than something like scriptname.tmp (or myAux) allows multiple copies of this script to be run concurrently without interfering with each other (assuming that the files being processed were parameterized rather than hard-coded into the script).

The AND list:

Code:

command1 && command2 && command3

executes command1. If and only if it completes successfully (terminates with exit code 0), then command2 will be run. And, if and only if, command2 also completes successfully, then command3 will be run. In this case, if the cp command completes successfully, we have a backup copy of file1. If the paste completes successfully, file1 has been replaced with the contents of file2 appended to the ends of the lines from the original file1. And if both of them succeed, we remove the backup copy of the initial contents of file1. So, if the cp fails, file1 has not been changed. If cp succeeds and paste fails, file1 can be returned to its original contents by copying the backup file back into file1.

This could also be done by:

Code:

paste -d " " file1 /dev/null file2 > tmp.$$ && mv tmp.$$ file1

but the permissions and ownership of file1 might change, and if there were any hard links to file1 when this script started, they would no longer be linked to file1 if this script completes successfully.

To parameterize the script and make sure that it was invoked with two pathname operands, I would tend to write a slightly longer version of this script:

Code:

#!/bin/ksh
IAm=${0##*/}     # basename of the path used to invoke this script
if [ $# -ne 2 ]  # verify arg count
then    printf "Invalid number of operands.\nUsage: %s path1 path2\n" "$IAm" >&2
fi
tmpf="$IAm.$$"   # create temp file name

# Do the work...
cp "$1" "$tmpf" && paste -d " " "$tmpf" /dev/null "$2" > "$1" && rm "$tmpf"

I usually use the Korn shell, but any shell that accepts basic POSIX conforming shell syntax (including, but not limited to, bash and ksh) will be just fine for this simple script.

Don Cragun, thanks for that thorough explanation!