Unix reading lines from a file

[supriyo banerjee]

Hi i am newbie to unix.

I am trying to read a single line from a file and i am using the following code:

#!/bin/ksh

i=1

if [ $i -le 45 ]
then
sed -n ''$i','$ip'' a.dat>>c.dat

i=i+1

fi


Now somehow i am missing out the sed options.

My basic purpose of these steps is to merge the content of 30 files whose locations are in the file a.dat. Can some one chip in with some help

[James R. Ferguson]

Hi:

# i=1; f=45
# sed -n "$i","$f"p a.dat > c.dat

...will write lines 1-45 of a.dat into c.dat

Regards!

...JRF...

[supriyo banerjee]

Hi jrf
Thanks...


Actually my fault i confused you with the post. The code which i have written is just to check if I am able to get the lines one by one written into c.dat...what i am looking ideally is for something which picks out each line one by one from a file.

[James R. Ferguson]

Hi (again):

If all you want to do is read lines from a file, the shell can do this easily:

#!/usr/bin/sh
typeset -i LINENBR
while read LINE
do
(( LINENBR=LINENBR+1 ))
if [ "${LINENBR}" -le 45 ]; then
echo ${LINE} >> c.dat
fi
done < a.dat
exit 0

Regards!

...JRF...

[supriyo banerjee]

I know this way thanks!

I am trying to do this using sed the stream editor...

[Peter Nikitka]

Hi,

- to get exactly one specific line out of a file, I use
li=45
sed $li!d filename

- to walk through a file line by line (my example puts the line number as a prefix):
typeset -i num=0
while read li
do print L$((num+=1)):$li
done
mfG Peter

[James R. Ferguson]

Hi:

> I am trying to do this using sed the stream editor...

That's wasteful if all that you are doing is using 'sed' to specify a line number and the line numbers represent a continuum. You are forcing 'sed' to read the input file n-times!

Hoever, your original code should have been:

# i=1;f=45;sed -n "$i","$f"p a.dat

...or for any line "i":

# sed -n "$i"p filein >> fileout

Regards!

...JRF...

[supriyo banerjee]

Thanks i got it done...

[Peter Nikitka · Accepted solution]

Hi,

to respond to your overlapping post:

> I am trying to do this using sed the stream editor...

I wonder, why one should do it that way, but ...

typeset -i num=45 i=0
while [ i -le num ]
do
sed -e $((i+=1))!d -e q filename >c.dat
done

The 'q' command terminates the sed as a speedup on large files.

mfG Peter

[James R. Ferguson]

Hi:

I see too, that you are a new member in our community. Welcome! Please see:

http://forums1.itrc.hp.com/service/forums/helptips.do?#28

Regards!

...JRF...