sed: match two lines

Question

I have a text file, where records are split on two lines. I would like to put them on one line. The records look something like this:

2013-07-29
 apt-get install git
2013-07-29
 apt-get install rsync
2013-07-31
 apt-get install nmap
2013-08-02
 apt-get install cifs-utils
2013-08-05
 apt-get install telnet

I would like to get the following:

2013-07-29 apt-get install git
2013-07-29 apt-get install rsync
2013-07-31 apt-get install nmap
2013-08-02 apt-get install cifs-utils
2013-08-05 apt-get install telnet

I tried with sed, but I am unable to match two lines. Is there some trick in sed, or is there some better tool ?

Answer 1

sed 'N;s/\n/ /'

• N joins next input line to the pattern space
• s/\n/ / replaces the new-line that separates the lines with a space.

or:

sed -n 'N;s/\n/ /;/regex/p'

• -n disables automatic printing of pattern space as last command
• /regex/p prints lines matching regex

Use the g modifier for the substitutions in case you are joining more than two lines, e.g. N;N;N;s/\n/ /g.

Answer 2

It seems that the apt-get lines have a preceding whitespace. Saying:

sed -e :a -e '$!N;s/\n / /;ta' -e 'P;D' inputfile

should produce the desired result.

Alternatively, you can also use paste:

paste -d' ' - - < inputfile