Catching errors early with assert

Already in the course, we’ve asked you to make many assumptions in your code: that input will be properly formed, that functions will only be called with certain values, and so on. Every piece of code makes assumptions about the world. Violating these assumptions leads to various error conditions—sometimes an exception, sometimes just bad behavior. We’ve seen a lot of exceptions, but not so much in the way of bad behavior. Let’s look at the letter example again:

What happens when you enter 0, and why? You’ll notice it picks the letter z, because 0-1 is -1, and in Python, indexing with -1 goes from the end of the list. A user would be confused by that output! It would be better if our program gave some kind of error message.

One approach would be to add an if, checking that n is between 1 and 26, inclusive. Another option is to simply state our assumption in code. A nice way to do that is with assert. Assertions are self checks in your code, saying, “Hey, this had better be true or something is really wrong! Here’s an example:

The new assertion on line 6 is the same as having written:

if not (1 <= n <= 26):
    raise AssertionError()

Using assert is more concise and looks better. In general, assert’s’ are a wonderful way to catch bugs, especially ‘logic’ bugs where your program behaves strangely but doesn’t crash.

Better errors with assert

By default, a failed assert just yields an AssertionError. You can easily write better error messages as follows:

Notice that we’ve used the form assert EXPR, MESSAGE, where EXPR should evaluate to True; if it doesn’t, then we’ll raise an AssertionError with MESSAGE contained in it.