How add to any `{` and `:` character in a file a newline

Question

I want to copy a given text file (on macOS or Unix in general) to stdout where to any { and : character a newline is added.

I tried

sed “s/{/{\n/g” myfile.txt 

Just for the curly bracket character, but this doesn’t seem to work.

Do I need to install the GNU version of sed for this? And if so, how can I add newlines to both characters { and : in one go?

Answer 1

Just backslash the literal newline.

catalina$ sed 's/[{:]/\
> /'g <<\_
> hello: this{is} an example{of}something: poo
> there
> _
hello
 this
is} an example
of}something
 poo
there

If you would like to replace the matched character with itself, include & in the replacement.

catalina$ sed 's/[{:]/&\
/'g <<\_
> hello: this is {another} example
> more: newlines!
> here
> _
hello:
 this is {
another} example
more:
 newlines!
here

There are many variations around how exactly this works in exactly which sed version, though I believe the above should work everywhere. GNU sed adds some conveniences, like the ability to use \n as an abbreviation for newline, but this is not portable. In general, I would suggest moving to Awk or Perl if you need to use non-portable sed features.

Sometimes you can also use shell features like sed $'s/[{:]/\\\n/g' but this (for now) is Bash-specific, and probably not an improvement in terms of legibility. (In brief, $'...' offers a single-quoted string with "C string" semantics, meaning the shell converts \n to a literal newline, \t to a literal tab, etc; you then also need to escape a literal backslash with another backslash to prevent the shell from interpreting it as something else. This shell feature is proposed to be included in a future POSIX version, so it will ultimately be portable to any POSIX-conformant shell, but don't hold your breath.)

In case it's not obvious, [{:] is a regex character class which matches a single character out of the enumeration between the square brackets.

Answer 2

This might work for you:

sed '/[{:]/G;//s/\([{:]\)\(.*\)\(.\)$/\1\3\2/;P;D' file

If a line contains either { or :, append an empty line to it and using pattern matching insert that newline between the matching pattern and the rest of the line.

Print/delete the first line in the pattern space and repeat.

N.B. The D command deletes up to and including the first newline in the pattern space and if the pattern space still contains text, inhibits the sed cycle from replacing the pattern space by the next line. Thus the remaining text can be processed.