Process Substitution

In the previous section we've seen how to chain the output of one command to the next one. But what if you want to chain the output of two or more commands to another one? What if you have a command that takes a file as an argument but you would like to process whatever is sent to that file?

Process substitution allows a process's input or output to be referred to using a filename. It has two forms: output <(...), and input >(...).

Example: Output

Imagine you have two files for which you want to compare the content. Using diff file1 file2 could generate false positives in the case lines are not ordered. So if you want to compare those files you could create two new files, ordered, and compare those. It would look like:

sort file1 > sorted_file1
sort file2 > sorted_file2
diff sorted_file1 sorted_file2

With process substitution you can do it in one line:

diff <(sort file1) <(sort file2)

Example: Input

Imagine you want to store logs of an application into a file and at the same time print it on the console. A very handy command for that is tee.

echo "Hello, world!" | tee /tmp/hello.txt

Now let's say you want to have only lowercase characters in the file but keep the regular case on the output. You could use process substitution that way:

echo "Hello, world!" | tee >(tr '[:upper:]' '[:lower:]' > /tmp/hello.txt)

Exercise

Two files f1.txt and f2.txt are created below. Complete the line so that the files are sorted with sort and compared with diff using process substitution on both inputs.