CODEWITHSUNDEEP
linuxbashsed
EngineSense
EngineSense

Reputation: 3616

Replace a word using sed from a file linux

My case is here.

I have file that contains #file_user

mark
sebastian
lewis
shumacher
hulkenburg

I have another file contains Add.ldif

dn: cn=salespartner,ou=Groups,o=company
changetype: modify
add: member
member: uid=user,ou=external,c=de,ou=People,o=company
member: uid=user,ou=external,c=de,ou=People,o=company
member: uid=user,ou=external,c=de,ou=People,o=company
member: uid=user,ou=external,c=de,ou=People,o=company

In the above i want to change the uid=user to the respective file user names

example 

dn: cn=typo3_frontend_global_salespartner,ou=Groups,o=company
changetype: modify
add: member
member: uid=mark,ou=external,c=de,ou=People,o=company
member: uid=sebastian,ou=external,c=de,ou=People,o=company
member: uid=lewis,ou=external,c=de,ou=People,o=company

I tried using this but i couldn't acheive the above example.

#!/bin/bash
set -x
for q in `cat user`
do  sed 's/user/'${q}'/g' add.ldif > out
done

Please suggest me how to do this. Meanwhile i'll understand advanced bash scripting

This very urgent for my work in ldap.

Thank in advance!!!!! Puspharaj

Upvotes: 0

Views: 1044

Answers (3)

micromoses
micromoses

Reputation: 7497

You have two problems with your script:

The first is that you use the > operator which will produce unwanted results for your use in a loop. It means that for each user in the file, you run sed and overwrite the contents of out, so I'm guessing that in the above example all the 'member:' lines will contain user=hulkenburg. You may want to use the >> operator instead, which appends to a file instead of overwriting it. But it is still useless, for the second reason:

You are using sed, which runs through the entire file and replaces all instances of user to whatever's in $q. Even if you run with the append operator, you'll end up with a out file which looks like this:

dn: cn=typo3_frontend_global_salespartner,ou=Groups,o=company
changetype: modify
add: member
member: uid=mark,ou=external,c=de,ou=People,o=company
member: uid=mark,ou=external,c=de,ou=People,o=company
member: uid=mark,ou=external,c=de,ou=People,o=company

dn: cn=typo3_frontend_global_salespartner,ou=Groups,o=company
changetype: modify
add: member
member: uid=sebastian,ou=external,c=de,ou=People,o=company
member: uid=sebastian,ou=external,c=de,ou=People,o=company
member: uid=sebastian,ou=external,c=de,ou=People,o=company

Which is probably not what you want. I suggest 2 solutions:

keep a ldif file for each user, by changing the sed line as follows:

do  sed 's/user/'${q}'/g' add.ldif > out.${q}

You'll have to change add.ldif as well to contain only 1 member: line.

Or, build the ldif file in a script. Have the contents of add.ldif look like this:

dn: cn=salespartner,ou=Groups,o=company
changetype: modify
add: member

Then, run the following script:

#!/bin/bash -x
for q in `cat user`
    do echo "member: uid=${q},ou=external,c=de,ou=People,o=company" >> add.ldif
done

I believe that the second solution is more elegant.

Upvotes: 0

Barmar
Barmar

Reputation: 781058

sed is probably not the best tool for this, since you don't want to make the same substitution on all lines. awk can do it better:

awk 'BEGIN { i = 0; j = 0; }
     FNR == NR { users[i++] = $0 }
     FNR != NR && /^member:/ && users[j] { sub("uid=user", "uid=" users[j++]); print }
     FNR != NR && !/^member:/ {print}' user add.ldif > out

Upvotes: 4

anubhava
anubhava

Reputation: 785186

This awk should work:

awk -F, 'FNR==NR{a[NR]=$1;next} $1~"user"{sub(/=user/, "=" a[++i], $1)} 1' OFS=, user add.ltif
changetype: modify
add: member
member: uid=mark,ou=external,c=de,ou=People,o=company
member: uid=sebastian,ou=external,c=de,ou=People,o=company
member: uid=lewis,ou=external,c=de,ou=People,o=company
member: uid=shumacher,ou=external,c=de,ou=People,o=company

Upvotes: 2

Related Questions