Shell Programming and Scripting

Hi
my file data is like below

ramu,sony,"raju \n ravi \n ramya" \n
ravi,sarah,"sowmya \n sorry \n sangam" \n

i want replace new line characters in between double coats with single space.

for example
cat input_file

ramu,sony,"raju
ravi
ramya"
ravi,sarah,"sowmya
sorry
sangam"

which i dont want. i want out put like below.i want replace newline s in between double quoates.

ramu,sony,"raju ravi ramya"
ravi,sarah,"sowmya sorry sangam"

any body can help either perl script or any other script.

but JOEYG given follwoing solution.
it is not giving correct data

> cat new_lines | sed 's/" /" ~/' | tr "\n" " " | tr "~" "\n"

with above script iam getting out put in single line as below
ramu,sony,"raju ravi ramya" ravi,sarah,"sowmya sorry sangam"

but i want out put like below.

ramu,sony,"raju ravi ramya"
ravi,sarah,"sowmya sorry sangam"

Can you attach the datafile? I am having a difficult time undertanding the spacing of certain characters; also, trying to get at those floating <new-line> characters is difficult.

To attach a file here, use the paperclip icon.

Code:

 tr '\n' ' ' < file | sed 's/"  /"\
/g'

Hi JOYEG
please find attatchement as ZIP file.
Thanks

I think the issue from my script, and possibly other proposed solutions, has to do with the spacing around double-quotes and <new-line> characters.

See the following example coding. To begin with, made the file with a # in place of the <new-line> or \n character. Verify it is there, and then translate to get the charcter in the new file

Code:

> cat new_lines
ramu,sony,"raju # ravi # ramya" # 
ravi,sarah,"sowmya # sorry s\ sangam" # 
> cat new_lines | tr "#" "\n"
ramu,sony,"raju 
 ravi 
 ramya" 
 
ravi,sarah,"sowmya 
 sorry s\ sangam" 
 
> cat new_lines | tr "#" "\n" >new_lines2

Now, I enter commands to grab the " that is followed by a space. In your example and my file, there is a space character after the closing " character. Hence, why the script works.

Code:

> cat new_lines2 | sed 's/" /" ~/' | tr "\n" " " | tr "~" "\n" | tr -s " "
ramu,sony,"raju ravi ramya" 
 ravi,sarah,"sowmya sorry s\ sangam"

How about Perl:

Code:

vnix$ perl -0777 -pe '1 while s/("[^"\n]+)\n([^"]+"\r(?:$|\n))/\1 \2/' Book1.csv 
Mohan,Mohan,"Hello Hello Test"
Mohan,Mohan,"Hello Hello Test1"
Mohan,Mohan,"Hello Hello Test2"

Why does the attachment contain a lot of commentary? At least please put the test data in a separate file inside the zip so we can see what's what. As you can see, I assumed the test material was in the first three lines of the file.

The test data had DOS CRs (\r) where the "real" newlines are supposed to be, but not in other places; knowing that from the start would have simplified the discussion a lot. If that's not really so then take out the \r in the script above.