Hello Unix gurus,
I am new to Unix. I am working on some dummy project which involves unix scripting.

I have a query :

There is a file Number.dat which is of form say

AAA|123|4563|animal
AAA|1234|45634|animal2
BBB|123444|456312|bird
BBB|123445|456313|bird2

Here AAA,BBB are headers.
Now I want to sort Number.dat based on header and place them in different header.dat file.
How do I do that? I would be really greatful if some one helps me.


Thanks in advance
Vinay

What have you done to attempt to solve this problem yourself?
Post your sample script, and we'll see how we can assist.

Regards

using awk we can get the solution for your problem


if $1 ='AAA' we can redirect it into new file AAA.dat
and if $1 ='BBB' we can redirect it into new file BBB.dat


please use this way to get the solutions

Regards,
Siva.P
Bangalore

Hello Franklin / Unix gurus,

My script is as follows:


#!/bin/ksh
sys=/export/home/vinay/Number.data
mod1=AAA
mod2=BBB

grep -h 'AAA' $sys > /export/home/vinay/AAA_$mod1.data
grep -h 'BBB' $sys > /export/home/vinay/BBB_$mod2.data


But I am not happy with the script since it has to traverse twice imto the Number.dat file


Also I some questions,

I want the process to use the Number.dat once rather than traversing twice.
Also, say if memory is full while script is running, then how should I handle it.
Kindly reply back if any one has a suggestion/script/solution

Thanks in advance.

Regards,
Vinay

Both awk and sed have facilities for writing into multiple files as they read in a single pass down the file. psiva_arul is hinting in the same direction in the posting above.

hello era /Unix gurus,
If you don't mind, can you assist in which fashion I can make use of sed into my script.

Thanks and Regards,
Vinay

Googling is not an option?

Notice that writing doesn't have any side effect (such as, for example, finish this line and fetch the next) so if you have patterns which overlap, you can get the same line written to multiple files (or need to write a slightly more complex script).

Hello era / unix gurus,

I tried the following:
sed -e "x;/^AAA/b" -e "/^BBB/b" -e d number.dat

It works. But I need to divert it to different files .How do I do it. Kindly assist if possible

Thanks and Regards,
Vinay

Did you try the script I posted above? If so, what's wrong with it? It writes lines matching "^AAA" to AAA.dat and lines matching the regular expression "^BBB" to BBB.dat, is that not what you want?

Hello era,

The script u had sent did not work.
Its giving an error
sed: command garbled: /^AAA/wAAA.dat/^BBB/wBBB.dat

That is why I tried
sed -e "x;/^AAA/b" -e "/^BBB/b" -e d number.dat
I am getting answer properly, but I dont know how to route it to different files.

The script should be on two different lines. You could also try

The sed w command takes a file name to write the line to.

Hi era,
sed -e '/^AAA/w AAA.dat' -e '/^BBB/w BBB.dat' number.dat works..
Thanks a lot.
Can I specify the path where it should be stored:
sed -e '/^AAA/w /export/home/vinay/AAA1.dat' -e '/^BBB/w /export/home/vinay/BBB1.dat' a.dat
I hope there is some error

Also does sed provides a way to search specific fields, like the -f1 (1st field) option

Also does the sed -e '/^AAA/w AAA.dat' -e '/^BBB/w BBB.dat' number.dat traverses through the number.dat once or twice.

.. Kindly assist me if possible
Thanks and Regards,
Vinay

sed uses regular expressions only, you can create a regex to look at the first field only with regex constructs; the ^ is already halfway there, as it forces the match to happen at beginning of line. Suppose the field separator is a vertical bar; then you can just add that after the string you want to search for, to anchor it properly.

This finds "AAA" followed by vertical bar, but only if it is at beginning of line (because of the ^) and ditto for "BBB".

The vertical bar has special meaning to some regular expression engines, just like the ^ -- if you get erratic behavior (all lines matching all the time) then you need to backslash-quote it, like \|. It is unfortunate that there are different dialects of sed so that we can't know for sure whether or not this is an issue in your case.

Hello era / unix gurus,

I tried the above script but its giving me
sed: command garbled: /^AAA \|/w/export/home/cmohanku/vinay/AAA1.dat

"Also does the sed -e '/^AAA/w AAA.dat' -e '/^BBB/w BBB.dat' number.dat traverses through the number.dat once or twice."


Please help me if possible

Thanks and Regards,
Vinay

It reads number.dat once, examining each line in turn.

Can't help you with the "command garbled" error; it works for me here. Is it the addition of the field separator or the full path name which is causing it confusion?

You seem to have a space between AAA and \| -- this will only match on a sequence where there really is a space between "AAA" and the vertical bar.