Bash substitution -- Fairly beginner stuff!

ipguru99 · 12-31-2007, 09:37 PM

Hello all,

Doing very tiny bash scripts, I seem to get down ok. But for some reason, the 'for x in `grep something... | cut .. ' stuff just doesn't stick. I have looked all over at samples and I always learn something new about sed or awk, or something new about piping stuff, but never seem to actually get it... so I thought I would post an example here of something I was trying to do the other day and see what people think. Thanks in advance!

I have a file with 200 names in it. I have manipulated it enough that I have almost all the fields filled with data. Either by using cut from another file or continuing to sed the thing to death, one field at a time. All of the fields are the same for all 200 users, so this works. My problem comes in where I need to pull a field (below, number 2.. the emailid) for all 200 lines and place it as the value for the 6th field (the quotes after displayname).

Code:

ca user1@email.com  '12345' default displayName  '' givenName xxx sn yyyy

(CreateAccount emailid 'password' default-class displayName-goes-between-the quotes... I am not concerned with anything after displayname yet)

So as an example, the script runs through the file, it pulls user1@email.com and I get a line that looks like this

Code:

ca user1@email.com  '12345' default displayName  'user1@email.com' givenName xxx sn yyyy

I hope I am explaining this correctly. I am sure there are a ton of ways to do it, and if they are easier, please let me know. I have messed around with

Code:

for x in (cat file1.txt | grep *???@*); do sed 's/\'\'/$x/g'

(I am doing this quickly and it was a couple of days ago.. so I am sure my regex stuff is wrong ;-), but at the time it was right... I just kept getting results like the same line 200 times, for every person).

Again, I am sure there are a couple of ways to do stuff like this... Thanks!

ghostdog74 · 12-31-2007, 09:46 PM

Quote:

Originally Posted by ipguru99

Code:

for x in (cat file1.txt | grep *???@*); do sed 's/\'\'/$x/g'

the for loop is redundant and is not the best way to do things, at least for your case.
If you want, provide a sample input files and show your desired output.

angrybanana · 01-01-2008, 04:43 AM

Is this what you want?

Code:

$ awk -v q="'" '{print $1,$3,$4,$5,q$2q,$7,$8,$9,$10}' file
ca '12345' default displayName 'user1@email.com' givenName xxx sn yyyy

if you don't mind the extra space (field 2 will turn into a space) then.

Code:

$ awk -v q="'" '{$6=q$2q;$2=""}1' file
ca  '12345' default displayName 'user1@email.com' givenName xxx sn yyyy

ipguru99 · 01-01-2008, 08:19 AM

angrybanana, thanks! Once again, something new I learned about awk! Is the 'q' used because of the single quotes.. or can q be just a place holder for anything?

Ghostdog74, the first code snippet is the sample input file. The second code snippet is the desired result. In real life, the first one has about 200 lines, not 1. Each line of the file is almost identical, just the second field changes (user1@email.com, userA@email.com...) and that is the one I want to insert into the 5th field.

Thanks guys!

angrybanana · 01-01-2008, 04:31 PM

Quote:

Originally Posted by ipguru99

angrybanana, thanks! Once again, something new I learned about awk! Is the 'q' used because of the single quotes.. or can q be just a place holder for anything?

-v variable = data

The reason I set a variable to single quote is to make the code more readable. Had I not done that, it'd be something like this.

Code:

$ awk '{print $1,$3,$4,$5,"'\''"$2"'\''",$7,$8,$9,$10}' file