Python — Try Except

Crashes stop programs

number = int("abc")     # ValueError: invalid literal for int()
print("never runs")     # program is dead

try / except — catching the fall

try:
    number = int("abc")
    print("never reached")
except ValueError:
    print("That wasn't a number!")

print("program continues")    # ✓ alive

Catching specific types — always be specific

try:
    value = data["key"]
except KeyError:
    print("missing key")
except (TypeError, ValueError) as e:   # multiple types, alias the error
    print(f"bad input: {e}")

Common exceptions you'll meet:

Exception	Cause
ValueError	right type, bad content (int("x"))
TypeError	wrong type entirely ("a" + 1)
KeyError	missing dict key
IndexError	list position doesn't exist
FileNotFoundError	missing file
ZeroDivisionError	n / 0
AttributeError	method not on that object

The full form: else + finally

try:
    config = open("config.json")
except FileNotFoundError:
    config = default_config      # fallback
else:
    print("loaded fine")         # runs ONLY if no exception
finally:
    print("always runs")         # cleanup regardless

Raising your own

Fail loudly on bad input instead of producing garbage downstream:

def set_age(age):
    if not isinstance(age, int):
        raise TypeError("age must be an int")
    if age < 0:
        raise ValueError("age cannot be negative")

Callers decide how to handle it — separation of detection and response.

Custom exception classes

class PaymentError(Exception):
    pass

def charge(card, amount):
    if amount > card.limit:
        raise PaymentError("limit exceeded")

try:
    charge(card, 9999)
except PaymentError as e:
    show_user(e.message)

The anti-patterns

# ❌ swallows EVERYTHING silently
try:
    risky()
except:
    pass          # bugs become invisible ghosts

# ✓ minimum acceptable:
except Exception as e:
    log(e)        # at least leave a trace

Bare except: even catches Ctrl+C. Never ship one.

EAFP — Python's philosophy

"Easier to Ask Forgiveness than Permission":

# LBYL (look before you leap):
if "key" in data:
    x = data["key"]

# EAFP (pythonic): just try it
try:
    x = data["key"]
except KeyError:
    x = default

Both valid; Python culture leans EAFP.

Mini Practice

1. Guard an int(input()) loop until valid.
2. Handle KeyError, IndexError and ValueError distinctly in one block.
3. Write divide() raising ZeroDivisionError with a custom message.
4. Build ValidationError subclass; catch only it.
5. File reader with else/finally demonstrating all four blocks.

Next: user input →