linux - replace all double newlines with a space

Question

Is it possible to replace only double newlines in a text file using sed, awk, grep, tr, or whatever is needed?

this

is

data
this

is

more

data

What I need to do is to replace all double newlines (\n\n, there is no whitespace) with a space, but keep all single newlines, as to have all my data points on their own line.

this is data
this is more data

Is this possible?

Answer 1

Try

    sed ':a;N;$!ba;s/\n\n/ /g'  filename

This will first read the whole file, then replace the double newlines (and only those!) with a space. This trick (reading the whole file) is necessary because most GNU/Linux utilities process input one line at a time, which would give you the wrong result.

Answer 2

Just another variant: an awk version

awk 'BEGIN{RS="\n\n" ; ORS=" ";}{ print }' 

Answer 3

Adding to @MariusMatutiae's answer: If the file is too big to read to memory, you can use a classic but much slower way:

EMPTY=0
BUFFER=""
while read L; do
  if test -z "$L"; then
    EMPTY=$(($EMPTY+1))
    BUFFER="$BUFFER\n"
  else
    if test $EMPTY -lt 2; then
      echo -en "$BUFFER"
    else
      echo " "
    fi
    echo "$L"
    BUFFER=""
    EMPTY=0
  fi
done < filename
echo -ne "$BUFFER"