When you use the $(command) and `command` forms of command substitution, the trailing <newline> character (if there is one) will be dropped from the output. There doesn't seem to be any need to assign the output of sed to a variable; just let sed write directly into your output file. The following would seem to do what you want:

Code:

sed 's/.*song=good.*/\#&/' $(cat /bin/user/input_file.txt) > /bin/user/new/output_file.txt

This should work on any Linux or UNIX System as long as you are using a shell (such as a Korn shell [ksh] or bash) that recognizes minimal requirements specified by the POSIX standards and the Single UNIX Specifications as long as the pathnames stored in input_file.txt do not contain any whitespace characters.

Note also that using:

Code:

echo $b

to print a line of output will produce different output when using different versions of echo if the variable b contains any tab characters, any occurrences of multiple adjacent spaces, starts with a minus sign, or contains any backslash characters. If you want to copy the contents of a variable to a file adding a <newline> character to the end of it (without altering spacing, leading minus signs, and backslash characters), you could replace the echo command with:

Code:

printf "%s\n" "$b"

And, finally, note that using:

Code:

echo $b > /bin/user/new/output_file.txt

will only save the output from the last file processed. Each time you use > as a redirection operator in the shell, all previous output in the destination file will be replaced.