List all entries in the current directory whose names contain only uppercase letters

Question

In one of my lab questions I was asked to list all entries(both files and folders) in the current directory whose names contain only uppercase letters using piping and grep command. Folders must have a forward slash displayed. Initially, my solution looked like this:

ls -p | grep -e "^[[:upper:]]*\/?$"

I thought that I can protect the '/' sign with the backward slash(as in echo \* vs echo *). However, it doesn't work as intended. One of my friends suggested the following solution:

ls -p | grep -e '^[[:upper:]]*/\?$'

I have no idea why the backward slash is used like this. What is even more weird,for me, that it actually produces the right result. I was wondering if someone could explain where the mistake in my solution is and why the second version actually works.

Answer 1

By default, grep uses basic regular expressions (BRE), in which ? matches a literal question mark. Adding the backslash made it have your desired interpretation as a quantifier meaning "zero or one occurrences of the previous expression" (i.e. optionally matching the trailing /).

From man grep:

Basic vs Extended Regular Expressions
           In  basic  regular expressions the meta-characters ?, +, {, |, (, and )
           lose their special meaning; instead use the  backslashed  versions  \?,
           \+, \{, \|, \(, and \).

You could achieve the same by switching to grep into extended regular expression (ERE) mode by adding the -E or --extended-regexp flag:

ls -p | grep -E -e '^[[:upper:]]*/?$'

Note that / is literal in both BRE and ERE - so doesn't need to be escaped in either case.

You could also consider using -x to enforce a whole-line match without the explicit ^ and $ anchors.

---

Note that (although it's fine as an exercise), piping ls to grep is generally not a recommended way to select files. In bash, you could instead consider using extended globbing:

shopt -s extglob nullglob

ls -pd *([[:upper:]])