sed one-liner to remove all single newlines? - regex

Question

So for example,
A paragraph's newlines would be removed let's say
it contained only single
newlines.
Then the things I would want to skip out:
However.
Our previous pair of newlines wouldn't.

Answer 1

It’s not a sed solution — although you can always run any sed through s2p of course — but a very easy solution using perl is:
% perl -i.orig -ne 'print unless /^$/' file1 file2 file3
That has the advantage of being extensible to any whitespace on the otherwise blank lines, like spaces and tabs:
% perl -i.orig -ne 'print unless /^\s*$/' file1 file2 file3
In the event that have files with various line endings, like CR or CRLF, you could also do this, assuming you are running perl 5.10 or better:
% perl -0777 -i.orig -ne 's/\R+/\n/' file1 file2 file3
which will normalize all sequences of one or more Unicode line separators into single newlines.
If you have UTF‑8 files that might have (for example) U+00A0 NON-BREAK SPACE in them on otherwise empty lines, you can handle them by telling perl that they are UTF‑8 using the ‑CSD command-line switch:
% perl -CSD -i.orig -ne 'print unless /^\s*$/' file1 file2 file3
UPDATE
I’m really unclear what you mean by removing a paragraph. I think you just mean joining up lines in a paragraph.
If so — if what you want to do is squeeze newlines from paragraphs, then you want to do this:
% perl -i.orig -00 -ple 's/\s*\n\s*/ /g' file1 file2 file3
It may not look like it works, but it does: try it.

Answer 2

Here's a sed solution.
$ sed -n -e '1{${p;b};h;b};/^$/!{H;$!b};x;s/\(.\)\n/\1 /g;p' 5751270.txt
A paragraph would be removed let's say it contained only single newlines.
However.
Our previous pair of newlines wouldn't.

Answer 3

You can try this bash script
#!/bin/bash
exec 8<"file"
while read -r line <&8
do
if (( ${#line} > 0 )); then
read -r next <&8
if (( ${#next} > 0 ));then
continue
else
echo "$line"
echo "$next"
fi
fi
done
exec <&8-