Eu usaria perl ou awk para ler os dados um parágrafo por vez e remover todos, exceto a primeira nova linha:
perl -00 -pe '$\="\n\n"; s/\n/perl -00 -pe ' # each record is separated by blank lines (-00)
# read the file a record at a time and auto-print (-p)
$\="\n\n"; # auto-append 2 newlines to each record
s/\n/awk -v RS= -F'\n' '{print $1; for (i=2; i<=NF; i++) printf "%s", $i; print ""; print ""}' file
/; # turn the first newline into a null byte
s/\n//g; # remove all other newlines
s/perl -00 -pe '$\="\n\n"; s/\n/perl -00 -pe ' # each record is separated by blank lines (-00)
# read the file a record at a time and auto-print (-p)
$\="\n\n"; # auto-append 2 newlines to each record
s/\n/awk -v RS= -F'\n' '{print $1; for (i=2; i<=NF; i++) printf "%s", $i; print ""; print ""}' file
/; # turn the first newline into a null byte
s/\n//g; # remove all other newlines
s/%pre%/\n/ # restore the first newline
' file
/; s/\n//g; s/%pre%/\n/' file
/\n/ # restore the first newline
' file
/; s/\n//g; s/%pre%/\n/' file
Comentou
%pre%Similarmente
%pre%