Bash在多行中匹配字符串并追加到前一行

Question

I'm trying to turn a text file with these lines:

policy rule {
    source-address host1;
    destination-address host2;
    application http;
policy someotherrule {
    source-address any;
    destination-address [ host2 host3];
    application ssh;

Into:

policy rule { source-address host1; destination-address host2; application http;
policy someotherrule { source-address any; destination-address [ host2 host3]; application ssh;

So basically, if a line ends with a semicolon, append it to the previous line.

I've been playing with sed/awk but haven't found a working solution to get the desired output.

Answer 1

Following gnu-awk should work:

awk -v RS='(^|\n)[^\n]*[^;]\n' 'NF{s=p $0; gsub(/(^|\n+)[[:blank:]]+/, " ", s); print s}
          {p=RT}' file
 policy rule { source-address host1; destination-address host2; application http;
 policy someotherrule { source-address any; destination-address [ host2 host3]; application ssh;

Answer 2

This simple one-liner will do:

perl -p0e 's/\n\s+/ /g' file
policy rule { source-address host1; destination-address host2; application http;
policy someotherrule { source-address any; destination-address [ host2 host3]; application ssh;

Answer 3

Here is an awk solution:

awk '{$1=$1;printf (/^policy/ &&NR>1?RS:"")"%s",$0} END {print ""}' file
policy rule {source-address host1;destination-address host2;application http;
policy someotherrule {source-address any;destination-address [ host2 host3];application ssh;

It sets the change the output of the line depending if line starts with policy or not.