Qfyilyi

I have a file with contents like so:

## this must go ##

## also

 this

 must go

##

hello world

##and this one

too##

I want to remove all between ##, including multi lines, so I am left just with hello world

This removes only part that is on one line:

sed -i.bak 's/##.*##//g' myfile

How to remove multi line stuff too?

P.S Im on MAC

You may use perl to achieve what you want:

perl -0pe 's/##.*?##R*//gs' file > newfile

See the online demo

The 0 argument makes it possible to find matches across lines.

The pattern matches

• 
  ## - two # symbols


• 
  .*? - any 0+ chars (even line break chars due to the s modifier) as few as possible


• 
  ## - two # symbols


• 
  R* - any 0+ line break sequences.

It should be very easy task for awk(in case you are ok with it). Could you please try following, will add explanation shortly too.

awk '/^##.*##$/{next} /^##$/{flag="";next} /^##/ && !/##$/{flag=1} flag{next} 1' Input_file

Adding a non-one liner form of solution too now.

awk '

/^##.*##$/{

  next

}

/^##$/{

  flag=""

  next

}

/^##/ && !/##$/{

  flag=1

}

flag{

  next

}

1

'   Input_file

Give a try to this:

sed -n '/^##/ { :1 ; /##$/ { d } ; n ; b 1 } ; p' myfile

Wise men read this excellent tutorial: Sed - An Introduction and Tutorial by Bruce Barnett

The test:

sed -n '/^##/ { :1 ; /##$/ { d } ; n ; b 1 } ; p' myfile



hello world

This might work for you (GNU sed):

sed -z 's/##[^#]*(#[^#][^#]*)*##n?//g' file

The -z option allows the whole of the file to be slurped into sed's pattern space. The regexp match is in three parts. The first part matches ## followed by zero or more non-#'s. The second part matches a zero or more group of characters consisting of a single # followed by a non-# followed by zero or more non-#'s. The third part matches ## and a possible newline. This regexp removes such matches globally throughout the file.

This can by shortened slightly by using the -r option to sweeten the final offering to:

sed -rz 's/##[^#]*(#[^#]+)*##n?//g' file

If the version of sed does not offer either options, then another solution is:

sed 'H;$!d;x;s/.//;s/##[^#]*(#[^#][^#]*)*##n?//g' file

It should by noted, that in the example above, all ##'s either start or end at the beginning or end of a line and so the solution below might also fit the bill:

sed 's/^##/,/##$/d' file