Use grep as a Substitute for cat

I came across a fascinating and educative article, Unix Recovery Legend, when someone posted a link to it on Reddit. The article presents a story of the writer, Mario Wolczko, and some other colleagues attempting to recover a system partially destructed by rm -rf /.

The rm command was interrupted before it scraped off directories like /usr, but it was too late to save /bin, /dev and /etc, which were respectively the directory for basic executable programs, device files, and system configuration files. Commands like ls and ps were now on strike since their executables were in /bin, the backup tape could not be mounted because the tape deck’s device entries were gone with /dev, and absence of network configuration files in /etc denied any possibility to rescue the system through the network.

Fortunately, those people still had access to some vital utility programs on the system. The writer had an instance of GNU Emacs running on his terminal; even though the executable might have been deleted, the binary had already been loaded into memory so the program could still run. Therefore, they had a text editor available. In addition, because the destructive command was interrupted before it could touch /usr, /usr/bin was saved. There were a few additional programs installed to that location instead of /bin, and the team of people rescuing the system was able to use some of them to complete the recovery.

At the end of the article, there was one thing that intrigued me: the writer suggested using /usr/bin/grep as a substitute for /bin/cat in case of deletion of the cat program. How could this be done?

grep searches for occurrences of strings matching a regular expression in some text and prints lines that contain any of those occurrences. cat, on the other hand, prints all lines from the sources of input specified in the command. The major difference between output of grep and cat is that grep only gives a subset of the lines, whereas cat prints all lines unconditionally. If we can prevent grep from filtering out lines so it prints all of them, then it works effectively the same as cat.

The most elegant way to let grep give every line of the input is to specify the empty string as the pattern to be searched for. The empty string regular expression matches everything, so every line is matched, hence be in the output of grep.

Mission complete: for the common and simple usage of cat where you just need to view a single file’s contents, grep "" can do the same job.

Create and Write Files Using grep

For the cat program, besides the basic usage of viewing a file’s contents from terminal output, it can also be used to write to files straight from the shell, without using any text editor. If you just run cat without any arguments, it will read from the standard input (in other words, your keyboard) instead of any file. When you enter something in the terminal, cat will just repeat what you type by copying the input to its output.

Who would like such a program that would only stupidly repeat what you tell it? The beauty of this behavior of the cat program could never have been seen without Unix shells’ output redirection: you can let cat write its output to a file rather than the terminal, so now it will transfer the text you enter from the keyboard to the file on disk.

Tip: When you have entered everything, press Ctrl-D to send the end-of-file (EOF) character to the input. The EOF character is the common way to tell Unix programs that you are done, so they can stop listening to your input and know they can start processing what you have given to them. For cat, once it receives EOF from the input, it will exit after it has transferred everything to the output.

Can grep "" replace cat for the purpose of writing files? Surely it can, because grep also has the similar behavior where if you do not specify files, it will read from the standard input, though command-line options for grep, like -r, can override this behavior.

Searching for Blank Lines With grep

If the empty regular expression is defined to match not only empty strings but everything, then how should we write the regular expression that is just for real empty strings, and use grep to find blank lines in a file?

In regular expression syntax, there are two special characters, ^ and $, for start and end of a line respectively. For example, the regular expression ^grep$ only matches grep; none of egrep, grep -E, fgreping is matched. Thus, the regular expression ^$ can be used to select empty lines. To find empty lines with grep, use the command grep "^$".

The grep program also supports a -x option, which implicitly surrounds the specified regular expression with ^ and $. This means grep -x "" is effectively the same as grep "^$". You can use either of them to search for blank lines.