HI

How can we concatenate files vertically ?
eg:
file a
aaaaaaaaaaaa
aaaaaaaaaaaa

file b
bbbbbbbbbbbb
bbbbbbbbbbbb
bbbbbbbbbbbb

file c
cccccccccccccc
cccccccccccccc
cccccccccccccc

file d should look like
aaaaaaaaaaaabbbbbbbbbbbbbcccccccccccccc
aaaaaaaaaaaabbbbbbbbbbbbbcccccccccccccc
aaaaaaaaaaaabbbbbbbbbbbbbcccccccccccccc

Can somebody help me in this aspect ?

Thanks
Bollin99

Try this:


paste -d'\0' file_a file_b file_c




note: files have to have the same number of records.

Here's a little trick I learnt using the file descriptors. It will allow you to format your output as you want and so, you can overcome instances where your files have different numbers of records.

# first define the file descriptors for your input files...
exec 3< file1
exec 4< file2
exec 5< file3

#Now read in from all 3 file descriptors simultaneously...
while read file1Line <&3 ; read file2Line <&4; read file3Line <&5
do
# format the output as required
# (I'm assuming that each input line has a fixed width of 15 chars)
printf "%-15s%-15s%-15s\n" "$file1Line" "$file2Line" "$file3Line"
done > newFile # redirect your output to a new file

The "" around the variables read in are important in order that at least a blank string is formatted!

Damian

HI
Firstly thank you for your replies. Bot work excellently.
but there is a constraint about reading the files.

I am not sure of the number of files that have to be concatenated.
The filenames will be read from a file and then will be concatenated.

so obviously we cannot give all the filenames if there are 10 or 20 file to be concatenated. Definig the file descriptors will be a problem.

So is there a mechanism to put the file names and the exec commands in a loop/sequence and execute the next set of command ?

eg: there are files 1-10 then
read the file names 1-10 from a file then
define the exec for 1-10 and then concatenate ?

Please let me know about the structre of the program.

Thank you
Bollin99

Try this:

Hi 'Bollin99'.

Obviously you didn't spot the deliberate mistake in my previous post. The method I gave you would require the file with the most rows to be assigned to the last file descriptor read in the 'while' logic test (because only the result of the final test is used).

I've attached a script that gets round this and which takes between 1 and 7 files as parameters.

If you need to work with greater numbers of files, you would just call the script within a loop. Something like...
------------------------------
touch tempFile
while read fileName
do
fpaste tempFile ${fileName} > newTempFile &&
mv newTempFile tempFile
done < fileListFile
------------------------------

This kind of defeats the object of dynamically allocating the file descriptors in the 'fpaste' script but I couldn't think of an easy way to get round the limitation of 9 file descriptors (of which 2 are used for stderr and stdout).

Damian

HI
Thank you, to both of you for sparing your time to find the best solution for my situation.
I have tested LKbrwn_dba's script, and it has been functioning well.
But i am skeptical whether it would work to concatenate 7-10 file with higher voulmes of date ......say a million rows in each file.
Other wise the script works well with no problem

THANK YOU
Bollin99

I don't see a problem. The only limitation I can envisage is the length of the input and concatanated rows. I could be proved wrong - it wouldn't be the first time!

There is a minor bug in the fpaste script.

Line 56 wants to be:

fileWidth=$( awk 'NR==1{print length($0)"."length($0); exit}' < ${fileName} )