Creating temporary files in bash

Solution 1:

The mktemp(1) man page explains it fairly well:

Traditionally, many shell scripts take the name of the program with the pid as a suffix and use that as a temporary file name. This kind of naming scheme is predictable and the race condition it creates is easy for an attacker to win. A safer, though still inferior, approach is to make a temporary directory using the same naming scheme. While this does allow one to guarantee that a temporary file will not be subverted, it still allows a simple denial of service attack. For these reasons it is suggested that mktemp be used instead.

In a script, I invoke mktemp something like

mydir=$(mktemp -d "${TMPDIR:-/tmp/}$(basename $0).XXXXXXXXXXXX")

which creates a temporary directory I can work in, and in which I can safely name the actual files something readable and useful.

mktemp is not standard, but it does exist on many platforms. The "X"s will generally get converted into some randomness, and more will probably be more random; however, some systems (busybox ash, for one) limit this randomness more significantly than others

By the way, safe creation of temporary files is important for more than just shell scripting. That's why python has tempfile, perl has File::Temp, ruby has Tempfile, etc…

Solution 2:

Yes, use mktemp.

It will create a temporary file inside a folder that is designed for storing temporary files, and it will guarantee you a unique name. It outputs the name of that file:

> mktemp
/tmp/tmp.xx4mM3ePQY
>

Solution 3:

You might want to look at mktemp

The mktemp utility takes the given filename template and overwrites a portion of it to create a unique filename. The template may be any filename with some number of 'Xs' appended to it, for example /tmp/tfile.XXXXXXXXXX. The trailing 'Xs' are replaced with a combination of the current process number and random letters.

For more details: man mktemp

Solution 4:

Is there any advantage in creating a temporary file in a more careful way

The temporary files are usually created in the temporary directory (such as /tmp) where all other users and processes has read and write access (any other script can create the new files there). Therefore the script should be careful about creating the files such as using with the right permissions (e.g. read only for the owner, see: help umask) and filename should be be not easily guessed (ideally random). Otherwise if the filenames aren't unique, it can create conflict with the same script ran multiple times (e.g. race condition) or some attacker could either hijack some sensitive information (e.g. when permissions are too open and filename is easy to guess) or create/replacing the file with their own version of the code (like replacing the commands or SQL queries depending on what is being stored).

You could use the following approach to create the temporary directory:

TMPDIR=".${0##*/}-$$" && mkdir -v "$TMPDIR"

or temporary file:

TMPFILE=".${0##*/}-$$" && touch "$TMPFILE"

However it is still predictable and not considered safe.

As per man mktemp, we can read:

Traditionally, many shell scripts take the name of the program with the pid as a suffix and use that as a temporary file name. This kind of naming scheme is predictable and the race condition it creates is easy for an attacker to win.

So to be safe, it is recommended to use mktemp command to create unique temporary file or directory (-d).

Solution 5:

mktemp is probably the most versatile, especially if you plan to work with the file for a while.

You can also use a process substitution operator <() if you only need the file temporarily as input to another command, e.g.:

$ diff <(echo hello world) <(echo foo bar)