Replacing newline after a pattern in unix

Question

I want to replace a newline with space after a pattern.
For example my text is:
1.
good movie
(2006)
This is a world class movie for music.
Dir:
abc
With:
lan
,
cer
,
cro
Comedy
|
Drama
|
Family
|
Musical
|
Romance
120 mins.

53,097 I want above text to become something like this

1. good movie (2006)
This is a wold class movie fo music.
Dir: abc
With: lan, cer, cro
comedy | Drama | Family | Musical | Romance
120 mins

Possible duplicate: http://stackoverflow.com/questions/11048973/replace-new-line-return-with-space-using-regex — MiiinimalLogic, Jul 06 '15 at 16:52
First remove all the new lines, than do a regex replace on capitol letters or numbers followed by a . with themselves and a preceding newline. — kpie, Jul 06 '15 at 16:52

score 1 · Answer 1 · edited May 23 '17 at 11:43

After the question update, the requirements for the solution changed:

cat test.txt | tr '\n' ' '  | perl -ne 's/(?<!\|) ([A-Z])/\n\1/g; print' | sed 's/ ,/,/g' | sed 's/ \([0-9]\+\)/\n\1/g'; echo

output:

1. good movie (2006)
This is a world class movie for music.
Dir: abc
With: lan, cer, cro
Comedy | Drama | Family | Musical | Romance
120 mins.

Explanation:

First I replace all newline characters using tr.
Second I replace every capital letter by a preceding newline and itself unless it is preceeded by a pipe "| "symbol.
The third one corrects the comma spacings.
The last moves the duration declaration to a new line

The echo at the very end is to append a 'newline' to the output.

Deprecated:
Building on kpie's comment, I suggest you the following solution:

cat test.txt | sed ':a;N;$!ba;s/\n//g' | sed 's/\([A-Z]\)/\n\1/g'

I pasted your input into test.txt.
The first sed replacement is explained here: https://stackoverflow.com/a/1252191/1863086
The second one replaces every captial letter by a preceding newline and itself.

EDIT: Another possibility using tr:

cat test.txt | tr -d '\n' | sed 's/\([A-Z]\)/\n\1/g'; echo

This works fine on my ubuntu, but on other machine, it shows as 'Label too long: :a;N;$!ba;s/\n//g' — user2509229, Jul 08 '15 at 00:43
what kind of 'other machine'? I tested it on Fedora, Ubuntu and CentOS. Nevertheless, I edited my answer and added another possibility using `tr` instead of `sed` — bro, Jul 08 '15 at 06:03
well, it is hard to find a working answer when you are changing the requirements.. check the updated answer. — bro, Jul 09 '15 at 08:25

Replacing newline after a pattern in unix

1 Answers1