UNIX for Dummies Questions & Answers

I want to substitute a charactor "PAN" with "TAN" in a shell, I used sed command in shell, it wo'nt work but the same is run from command prompt it was successful. the command is

sed ' s/PAN/TAN/g ' <i/p> > <o/p>
sed ' s/^M/^M/g ' <i/p> > <o/p> (1st ^M is Ctrl+V+M, 2nd should be line feed/next line)

the same problem is with delete option in sed

Please suggest



actually file consist of single line with multiple ^M character, i am trying to convert all ^M character into line feed i.e new line (so that single line file become multiple line sequential file), the same works from command line but not from shell program.

thanks

Try specifying the full path of sed in your shell script

There are errors in the samples posted which would perhaps be more visible if the lines were posted in code tags.
1) There is no "output filename" parameter to "sed". See "man sed".
2) There are excess space characters in the commands posted.

Code:

First command should probably be:
sed -e "s/PAN/TAN/g" input_filename > output_filename

Second command.
I have never been able to make this conversion work with "sed".
Part of the problem is that "sed" does not recognise a line at all unless it ends with a line-feed character.
If you just want to convert carriage-return to line-feed, use the unix translate command "tr".

cat input_filename | tr '[\r]' '[\n]' > output_filename

Most of these carriage-return issues arise when a text file is created on a M$ platform and the copied to a
unix platform using a file transfer method which does not convert the file (e.g. binary mode ftp).
The normal repair involves running unix command "dos2unix" or "dos2ux" to convert the text file.
When foreign character sets are involved it can be better to convert the file with "tr" or an "awk" program.