Long, long ago, people used to memorize phone numbers. To help people do this, telephone keypads included letters. Here is a typical encoding:
1 2 3
abc def
4 5 6
ghi jkl mno
7 8 9
pqrs tuv wxyz
* 0 #
+
1. Write a program that generates Boston vanity telephone numbers (1-617-XXX-XXXX), where XXX-XXXX encodes a person's name.
For example, the vanity number for 'Adriana' would be 1-617-237-4262, which is 1-617-ADRIANA.
But what to do with names longer or shorter than 7 letters? If a name is shorter, then the vanity number will begin with a padding of 7's. For example, the vanity number for 'Adam' would be 1-617-777-2326, which is 1-617-777-ADAM. If a name is longer than 7 letters, then the vanity number will drop all interior vowels starting from the end. For example, the vanity number for 'Elizabeth' would be 1-617-354-9284 (for 1-617-ELIZBTH). And for crazy long names, you just take the first 4 letters and the last 3. For example, 'Mahershalalhashbaz' (yes that's a real name) would be 1-617-624-3229 (for 1-617-MAHEBAZ).
2. Now write a program that does the reverse, taking a vanity telephone number and revealing the name. Impossible, you say? Not quite.
You could at least encode the first 3 letters of a name. A simple system for achieving this would convert the old keypad number-letter mapping into a list of tuples like this:
ADR = [(2,0), (3,0), (7,2)]
The first numbers of each tuple is the keypad number, and the second is the index of the correct letter under the keypad number.
But that system only uses 6 of your 7 telephone digits (wasting entropy), and can't we do better than a 3-letter name encoding? (Hint: yes we can, here's one way.)