Trey Hunner

Python Tip #299 (of 365): Return Boolean expressions directly. 🧵 Do this: def is_even(n): return n % 2 == 0 Not this: def is_even(n): if n % 2 == 0: return True else: return False #Python #DailyPythonTip

But a condition that relies on truthiness might not be True or False: def is_number(value): if re.fullmatch(r"-?\d+(\.\d+)?", value): return True else: return False Returning that re.fullmatch call directly would return a re.Match object or None.

Python Tip #299 (of 365): Return Boolean expressions directly. 🧵 Do this: def is_even(n): return n % 2 == 0 Not this: def is_even(n): if n % 2 == 0: return True else: return False #Python #DailyPythonTip

Python Tip #298 (of 365): When creating lots of class instances, save memory with __slots__. 🧵 class Point: __slots__ = ("x", "y", "z") def __init__(self, x, y, z): (self.x, self.y, self.z) = (x, y, z) #Python #DailyPythonTip

With __slots__, Python allocates fixed spots for only the named attributes: no per-instance dictionary at all. The trade-offs: 1. You can't add attributes that aren't in __slots__ 2. Anything that inspects __dict__ will no longer work with your instances (like the built-in vars() function)

Python Tip #298 (of 365): When creating lots of class instances, save memory with __slots__. 🧵 class Point: __slots__ = ("x", "y", "z") def __init__(self, x, y, z): (self.x, self.y, self.z) = (x, y, z) #Python #DailyPythonTip

Python Tip #297 (of 365): Avoid making metaclasses if you can. 🧵 There's usually an easier tool for the job. Metaclasses customize the behavior of CLASSES the same way classes customize the behavior of their instances. #Python #DailyPythonTip

Metaclasses are powerful, but they can also make for code that's overly complex and hard to understand. Fortunately, many problems that COULD be solved with a metaclass often have a simpler solution.

Python Tip #296 (of 365): Always use functools.wraps when making a decorator. 🧵 A decorator REPLACES a decorated function with the function that was returned from calling the decorator. #Python #DailyPythonTip

functools.wraps borrows the name, signature, and docstring of the decorated function and updates the wrapper to match: >>> help(is_prime) Help on function is_prime in module __main__: is_prime(number) Return True if number is prime.

Python Tip #296 (of 365): Always use functools.wraps when making a decorator. 🧵 A decorator REPLACES a decorated function with the function that was returned from calling the decorator. #Python #DailyPythonTip

from functools import wraps def time_runs(func): @wraps(func) def wrapper(*args, **kwargs): start = perf_counter() return_value = func(*args, **kwargs) print(f"{func.__name__} ran in {perf_counter()-start} seconds") return return_value return wrapper

Python Tip #295 (of 365): Avoid needless deep copying. 🧵 Often a deepcopy call can be replaced by a comprehension, a slice, or some other operation that makes new objects. When you copy a list of dictionaries, you get a NEW list pointing to the SAME dictionaries. #Python #DailyPythonTip

We tend to prefer shallow copies in Python: list(items), dict(mapping), slicing, and comprehensions all make new containers without copying the CONTENTS of those containers (tip #223's list(...) call is one example).

Python Tip #295 (of 365): Avoid needless deep copying. 🧵 Often a deepcopy call can be replaced by a comprehension, a slice, or some other operation that makes new objects. When you copy a list of dictionaries, you get a NEW list pointing to the SAME dictionaries. #Python #DailyPythonTip

So if you plan to mutate those dictionaries, a shallow copy isn't enough. You could make a deep copy instead: from copy import deepcopy processed_data = deepcopy(tweet_data) for tweet in processed_data: tweet["date"] = datetime.strptime(tweet["date"], "%b %d %Y")

Python Tip #294 (of 365): Use exceptions to pass error messages to users. 🧵 Catch exceptions with messages users shouldn't see, and raise your own with a message they SHOULD see: class WeatherError(Exception): """Raised when we can't show the user a forecast.""" #Python #DailyPythonTip

Why not print and exit right where the problem was detected? Because exiting within a random function is messy. Imagine if load_forecast was called more than once, sometimes with data that didn't come from a user. Raising an exception lets EACH caller decide how to handle the exceptional case.

Python Tip #294 (of 365): Use exceptions to pass error messages to users. 🧵 Catch exceptions with messages users shouldn't see, and raise your own with a message they SHOULD see: class WeatherError(Exception): """Raised when we can't show the user a forecast.""" #Python #DailyPythonTip

Then, at the very edge of your program, catch JUST that exception type and show the user its message: def main(): try: forecast = load_forecast(sys.argv[1]) except WeatherError as error: sys.exit(f"Error: {error}") print(forecast["high"])

Python Tip #294 (of 365): Use exceptions to pass error messages to users. 🧵 Catch exceptions with messages users shouldn't see, and raise your own with a message they SHOULD see: class WeatherError(Exception): """Raised when we can't show the user a forecast.""" #Python #DailyPythonTip

Python Tip #293 (of 365): Write context managers with contextlib.contextmanager. 🧵 Instead of making a class with __enter__ and __exit__ methods, we can make a context manager out of a function: #Python #DailyPythonTip

The "try"/"finally" makes sure the environment variable is restored even if an exception is raised in the "with" block. The generator function you decorate must only "yield" one time: that "yield" shouldn't be in a loop and there shouldn't be more than one of them.

Python Tip #293 (of 365): Write context managers with contextlib.contextmanager. 🧵 Instead of making a class with __enter__ and __exit__ methods, we can make a context manager out of a function: #Python #DailyPythonTip

@contextmanager def set_env_var(name, new_value): original_value = os.environ.get(name) os.environ[name] = new_value try: yield finally: if original_value is None: del os.environ[name] else: os.environ[name] = original_value