join lines

s_linux · 04-08-2011, 04:32 PM

All,
I have following file which is output of ldapsearch command

Quote:

something:cn=someuser,cn=something,ou=some,ou=somea,
o= some#1#(22982374293742998374893)
something:cn=someotheruser,cn=something,ou=some,ou=s
om ea,o= some#1#(22982374293742998374893)
something:cn=someoneuser,cn=something,ou=some,ou=som
ea ,o= some#1#(22982374293742998374893)
something:cn=someoonetheruser,cn=something,ou=some,o
u= som ea,o= some#1#(22982374293742998374893)

actuvally 2nd, 4th..etc line starts with space.
so I wanted to use grep command in script to get one of those values in file.

Quote:

test=`grep "cn=someuser" "/tmp/test/dir.txt"`

i get only the first line as below

Quote:

something:cn=someuser,cn=something,ou=some,ou=somea,

but I wanted to have in one line

Quote:

something:cn=someuser,cn=something,ou=some,ou=somea,o= some#1#(22982374293742998374893)

also once I get the above value, I wanted to replace the value "#1#" to #0#. How can we do this.
Thanks

Ramurd · 04-08-2011, 05:14 PM

if you do "grep -A1" they'd be concatenated, however there's still this trick that the 2nd lines start with a space, which will be added in the concatenation and hence result in a bad ldap record.

so, you have to use a multi-trap-rocket-science thing:

0) make a backup of your output ;-)
1) remove the leading spaces on each line

Code:

sed -i 's/^ .*/.*/g' ldap_output_file

2) grep -A1 on the cn as you stated above
3) your $test now contains concatenated output, append that to a secondary output
4) apply sed like this sed -i 's/#1#/#0#/g'; however, given that this is ldap, you'll have to move #2# to #1# I guess, but I'm tired, so I am not going into that (yet).

Another approach is to use sed and print only the even lines in "file_even" and the odd lines in "file_odd", number them and join and remove line numbers:

Code:

1) sed -n 1~2p ldap_output | cat -n - > fileA
2) sed -n 2~2p ldap_output | cat -n - > fileB
3) join fileA fileB | cut -d ' ' -f 2- > new_ldap_output_file
4) sed -i "s/#1#/#0#/g' new_ldap_output_file

Nominal Animal · 04-08-2011, 06:58 PM

You can use tr and sed to merge the continued lines first:

Code:

tr '\a\n' ' \a' < input | sed -e 's|\a[\t ]||g' | tr '\a' '\n' > output

The first tr converts all BELs (\a) to spaces, and newlines (LF AKA \n) to BELs. The resulting output (which is a single long line) is fed to sed, which removes any BELs followed by a space or a tab (HT aka \t). Then, tr is used again to convert any BELs left to newlines. The output is saved to file output. Note that input and output must be different files; if they're the same file, you will just trash it.

Here is a wrapper around ldapsearch, which joins the continued lines together. Save it and run it just as if you would ldapsearch:

Code:

#!/bin/sh
ldapsearch "$@" | tr '\a\n' ' \a' | sed -e 's|\a[\t ]||g' | tr '\a' '\n'

kurumi · 04-08-2011, 08:53 PM

is this what you want?

Code:

$ ruby -ne 'print ($.%2==0)? $_: $_.chomp' file
something:cn=someuser,cn=something,ou=some,ou=somea,o= some#1#(22982374293742998374893)
something:cn=someotheruser,cn=something,ou=some,ou=som ea,o= some#1#(22982374293742998374893)
something:cn=someoneuser,cn=something,ou=some,ou=somea ,o= some#1#(22982374293742998374893)
something:cn=someoonetheruser,cn=something,ou=some,ou= som ea,o= some#1#(22982374293742998374893)

$ awk 'ORS=(NR%2==0)?"\n":"\0"' file
something:cn=someuser,cn=something,ou=some,ou=somea,o= some#1#(22982374293742998374893)
something:cn=someotheruser,cn=something,ou=some,ou=som ea,o= some#1#(22982374293742998374893)
something:cn=someoneuser,cn=something,ou=some,ou=somea ,o= some#1#(22982374293742998374893)
something:cn=someoonetheruser,cn=something,ou=some,ou= som ea,o= some#1#(22982374293742998374893)

grail · 04-09-2011, 01:31 AM

Assuming the space needs to be removed (not sure about ruby but kurumi will definitely fix it):

Code:

awk 'ORS=(gsub(/ /,""))?"\n":"\0"' file1
something:cn=someuser,cn=something,ou=some,ou=somea,o=some#1#(22982374293742998374893)
something:cn=someotheruser,cn=something,ou=some,ou=somea,o=some#1#(22982374293742998374893)
something:cn=someoneuser,cn=something,ou=some,ou=somea,o=some#1#(22982374293742998374893)
something:cn=someoonetheruser,cn=something,ou=some,ou=somea,o=some#1#(22982374293742998374893)

And assuming the search option is required, you can use the alternative:

Code:

awk '/cn=someuser/{line=$0;getline;gsub(/ /,"");print line $0}' file
something:cn=someuser,cn=something,ou=some,ou=somea,o=some#1#(22982374293742998374893)

s_linux · 04-11-2011, 10:00 AM

Thanks guys
It worked with sed command and wrapper around ldapsearch.

Thanks a lot.