Bash Hostname Completion

One of the more well known features in bash is command and filename tab completion. Installing the bash-completion package adds onto this. This package enables hostname completion and a lot more. After installing bash-completion, add the following to your .bashrc:

. /etc/bash_completion
complete -F _known_hosts ssh
I also suggest adding
complete -F _known_hosts ping
complete -F _known_hosts traceroute

Hostname completion relies on the ssh known_hosts file. Most modern distributions hash the ~/.ssh/known_hosts file for security reasons. This prevents hostname completion from working. If you are comfortable turning off hostname hashing, then add the following to your ~/.ssh/config:
HashKnownHosts no

If you had to turn off hostname hashing, you will need to re-populate your known_hosts file. I suggest creating a list of all of your hosts and logging into them in a for loop:

for i in $(cat hostlist)
   ssh -n -o StrictHostKeyChecking=no $i "uname -a"

This assumes that you are using ssh keys. If you are not, you will need to type your password for each host. At this point you can now use hostname tab completion. If you added the second two complete commands, the same hostname completion will work for ping and traceroute.

Matching a US Telephone Number With egrep Using Regular Expressions

This is the follow up to my post searching for social security numbers. US telephone numbers use the following format that can easily be matched with a regular expression.
(215) 555-1212
215 555 1212

The phone number can be broken down into a series of character classes. Using egrep, character classes are written inside of square brackets. The character class [0-9] represents a single number from 0-9. You can expand upon this and match a series of numbers from the character class by following it with a number inside of curly braces. [3-7]{3} matches exactly three numbers in the range of 3 through 7. We will use this notation to build the three parts of the phone number.

You can also build character classes containing specific characters or symbols. After the first three numbers of the phone number there are a few possibilities for the next character. It can be a right paren, a hyphen, a space, or a period. It can also be none of these. The ? operator matches exactly zero or one instance. Putting these two concepts together, we would build a character class and use the ? operator: [)- .]?

OK, let’s combine all of these concepts to build our regex. This is just one solution that I’ve come up with to match a phone number. The beauty of Unix is that there are many other solutions that are correct; some of which are probably better than my solution.

egrep “[(]?[2-9]{1}[0-9]{2}[)-. ]?[2-9]{1}[0-9]{2}[-. ]?[0-9]{4}” filename

This reads: zero or one left paren followed by a single number in the range two through nine, followed by two numbers in the range zero through nine, followed by zero or one right paren, hyphen, period, or space, followed by a single number in the range two through nine, followed by two numbers in the range zero through nine, followed by zero or one hyphen, period, or space followed by four numbers in the range zero through nine.

Until the explosion of cell phones, US area codes followed the format: number from 2-9, a 0 or 1, followed by a number from 0-9. When additional area codes were needed to accommodate the growing number of phone numbers, the requirement that the middle digit be a 0 or 1 was dropped.

If you have any questions about this article, please post in the comments section.

I updated the regex thanks to the feedback from Boris. The first and fourth digits cannot contain a zero or one so I created two separate character classes to accommodate that requirement.