Reading input files by line using read command in shell scripting skips last line

Question

I usually use the read command to read an input file to the shell script line by line. An example code such as the one below yields a wrong result if a new line isn\'t inser

Answer 1

read reads until it finds a newline character or the end of file, and returns a non-zero exit code if it encounters an end-of-file. So it's quite possible for it to both read a line and return a non-zero exit code.

Consequently, the following code is not safe if the input might not be terminated by a newline:

while read LINE; do
  # do something with LINE
done

because the body of the while won't be executed on the last line.

Technically speaking, a file not terminated with a newline is not a text file, and text tools may fail in odd ways on such a file. However, I'm always reluctant to fall back on that explanation.

One way to solve the problem is to test if what was read is non-empty (-n):

while read -r LINE || [[ -n $LINE ]]; do
  # do something with LINE
done

Other solutions include using mapfile to read the file into an array, piping the file through some utility which is guaranteed to terminate the last line properly (grep ., for example, if you don't want to deal with blank lines), or doing the iterative processing with a tool like awk (which is usually my preference).

Note that -r is almost certainly needed in the read builtin; it causes read to not reinterpret \-sequences in the input.