Need some help cutting a file up

jakev383 · 10-22-2009, 08:45 PM

I have a file that has one line in it, and looks like this:

Code:

-a test.example.com -a next.example.com -a last.example.com

I'm trying to code a bash script that will cut the file up, and only give me the *.example.com portions. I imagine I'll need the cut command (or multiple cut commands). Something like this:

Code:

list=`cat thefile | cut -d"-"

I know that's not a working example, but I wanted to see if anyone else had attempted something similar before and could give me an example. The number of entries can be variable and the domain names can of course be anything, but it will always be in the same format and on one line.
Thanks in advance!

ghostdog74 · 10-22-2009, 09:00 PM

say for example

Code:

$ a="-a test.example.com -a next.example.com -a last.example.com"
$ set -- $(echo ${a//-a/})
$ echo $@
test.example.com next.example.com last.example.com
$ echo $1
test.example.com
$ echo $2
next.example.com
$ echo $3
last.example.com

in a file

Code:

while read -r line
do
    set -- ${line//-a/}
    echo $1 
    echo $2
    echo $3
done < "file"

jakev383 · 10-22-2009, 09:03 PM

Hmm. That gets me running in the right direction I think. I do not know the number of entries in the file, but I can play and get that from "echo $@" I think.
Thanks!

lumak · 10-23-2009, 12:06 AM

I would use 'sed' to find all the lines containing the "-a" parameters then substitute the "-a" with nothing

Code:

sed -n -e 's/-a//p' file.txt

that's 'sed' '-n' (print only addressed lines specified with p) '-e'(xpression) 's'(ubstitute) '-a' with '' p(rint) for file.txt

there is also a way to use two expressions to first find a particular type of line (maybe it always starts with the same command) then process only those lines with the substitution you want.

ghostdog74 · 10-23-2009, 12:16 AM

Quote:

Originally Posted by lumak

I would use 'sed' to find all the lines containing the "-a" parameters then substitute the "-a" with nothing

Code:

sed -n -e 's/-a//p' file.txt

you forgot the "g" modifier. also, in OP's case, if he wants to use each example.com individually, then sed is not the correct tool for this purpose. A more "flexible" approach is to bring each example.com into an array of some sort so that each example.com may be used individually. Eg maybe OP wants to ping each host etc etc.

chrism01 · 10-23-2009, 02:06 AM

Perl's split() fn http://perldoc.perl.org/functions/split.html does the job in one

Code:

# split on string, not just single char
$var1 = "-a test.example.com -a next.example.com -a last.example.com";
@arr = split(/-a /, $var1);
print "@arr\n";

test.example.com  next.example.com  last.example.com

ghostdog74 · 10-23-2009, 03:12 AM

so is awk

Code:

awk 'BEGIN{FS="-a "}{print $1,$2,$3,$4}'  file

lumak · 10-24-2009, 10:50 AM

@ghostdog74, as you have it, awk would need some kind of internal loop, not to mention it would print a blank line as the first option. AND print every line, as is, before and after that. Also... the ideal solution would be to find the most generalized script that will work on any file with a '-a' formatted line. At the moment, we don't even know if the line starts with a command or if it follows a line that has a '\'.

catkin · 10-24-2009, 11:39 AM

How about this

Code:

#!/bin/bash

while read line
do
    array=($line)
    for ((i=1; i<${#array[*]}; i=i+2))
    do
        name="${array[$i]}"
        suffix="${name#*.}"
        echo "DEBUG: suffix is '$suffix'"
    done
done < input.txt

ghostdog74 · 10-24-2009, 07:01 PM

Quote:

Originally Posted by lumak

@ghostdog74, as you have it, awk would need some kind of internal loop, not to mention it would print a blank line as the first option.

that's not a problem at all. Can be easily fixed. The crux is, whether OP wants to use each domain individually. If he wants to use each domain found individually, he can always grab them from the arrays/fields. Doing only a substitution serves nothing, except if OP only wants to make output nice.

lumak · 10-24-2009, 08:40 PM

my point was that both of our solutions were incomplete and further education on the OP's end would be required to make it usable. That and sending my first solution (yes yes with the 'g' option) into a variable would allow further processing later in the script anyway. the -n prints only the lines with -a in it then it substitutes -a with nothing. Web addresses don't have spaces or return characters. The variable storing the output of sed can then be processed anyway imaginable. bash will treat both the return character and a space as white space. The only flaw in my solution is if there is something before and after on the same line with the '-a' but then you could do other tests to figure that out. So no. substituting doesn't only make the output nice. It's part of the process of an incomplete solution.