ssimple shell script to parse a file ~sed or awk

stevie_velvet · 07-13-2006, 12:31 PM

i have a file called myfile

A 1
B 2
C 3

i need to use sed or awk file take a variable (as a ext input), as a number & search each line for that number(pattern) & extracts the letter
for example, If I type 2, the result will be b

Caveat ~There is a TAB between the letters & numbers ~ not spaces, also there's no editng of myfile

whose good at scripting ?

raskin · 07-13-2006, 12:47 PM

Do you really need to do this in pure sed? I guess you want

read;
grep "^[^ ] $REPLY$" <file | sed -e 's/ .*//'

where " " should be written as <Tab> and file should not contain tabs except one in a line.

stevie_velvet · 07-13-2006, 02:17 PM

hi raskin

that looks good!
I guess i'm really need for it to be combined with awk
i.e.
| awk '{print $1}'

should be included

..just need a example to remind me of sed1

tq!!!!!!!!

s

stevie_velvet · 07-13-2006, 02:35 PM

..note sure waht you mena ~as " " isn't visible in your example

also where would i used read?

laslty...doens't s/ used as a substitution ?

spooon · 07-13-2006, 06:12 PM

how about

Code:

read REPLY
grep "$REPLY$" myfile | awk '{print $1}'

jlinkels · 07-13-2006, 06:48 PM

That can be shorter

Code:

awk -v n=my_number '$0 ~ n {print $1}' myfile

jlinkels

stevie_velvet · 07-14-2006, 12:24 AM

yep

i realise sed is overkill & its a question of GREP \ EGREP & AWK

i'll digest this

thanks!

s

spirit receiver · 07-14-2006, 03:41 AM

Actually, sed is more lightweight than awk (64K against 243K), so it's probably not appropriate to call sed overkill. The following command will do as well:

Code:

sed -nre "s/^(.*)\t$NUMBER\$/\1/p" myfile