7 Tips for Handling Python Exception

Introduction

Exception handling is commonly seen in Java, C++ or any other modern programming languages. Similarly, in Python, it has two types of errors – syntax errors and exceptions. Syntax errors refer to the syntactic problems which are detected by interpreter when translating the source code into byte code, while the exceptions are only detected during the runtime when the particular line is evaluated and executed. Python exceptions can be handled by application code but not the syntax errors.

In this article, we will discuss about the different ways to handle Python exceptions and when shall we use them.

Raise Exception

Python has a list of built-in exceptions which can be generated by interpreter or the built-in functions. Before we touch on the exception handling, let’s first look at how to raise a built-in exception.

Without implementing any new exception, you can use the raise keyword to throw out a built-in exception from your code. For instance, the below code checks the data type and value for mode variable, and raises built-in exceptions when wrong data type or value specified:

if not isinstance(mode, int):
    raise TypeError("mode must be numeric values")
        
if not mode in [0, 1]:
    raise ValueError("mode has to be 0 or 1")

When the mode variable is passed as a string “1”, you shall see exception raised as per below:

Catch Exception

To catch an exception, Python uses try-except which is similar to the try-catch syntax in Java/C++. In the except block, you can either write your own logic to handle the exception, or re-raise the exception to the callers.

For example, the below code would catch the KeyError exception and re-raise it as ValueError to the caller:

def read_config(config):
    
    try:
        mode = config["mode"]
    except KeyError as err:
        raise ValueError(f"No mode has been specified.")        
                
    return mode

When calling the read_config function with an empty dictionary:

read_config(config={})

You shall see the below exception raised:

The traceback shows both the cause of the exception and the last exception raised. If you want to hide the details of the cause (e.g. KeyError exception), you can use the raise exception from None to disable the chaining of the exceptions:

raise ValueError(f"No mode has been specified.") from None

When re-run the code, you shall see that only the last exception was showing in the trackback:

You can also specify a different exception when you think a explicit cause is needed rather than the original exception, e.g. instead of raise from KeyError, you can raise from AttributeError for better clarity to the caller.

Catch Multiple Exceptions

Catching multiple exceptions are as simple as adding multiple else if statements. Be cautious about the sequence of the exception classes as the derived exception classes must be placed before the base class, otherwise all your exceptions would be caught by the base exception class.

Below is an example to catch all the file operation exceptions, since the OSError is the base class, it shall be put as the last except block:

file_path = r"c://config.txt"

try:
    with open(file_path, "w") as f:
        config = f.write("password")
except FileNotFoundError:
    print(f"file {file_path} is not found")
except PermissionError:
    print(f"you have no permission to write to {file_path}")
except OSError as e:
    print("os error {e}")

When you run the above code, you shall see the output similar to below:

you have no permission to write to c://config.txt

You can also put multiple exceptions with parentheses to handle them within the same except block when there is no need to distinguish them:

except (OSError, FileNotFoundError, PermissionError) as e: 
    print(e)

This is similar to an IN operator, the exception will be caught as as long as it is an instance of any of these classes.

Suppress Exception with Pass Keyword

There are cases that you cannot do anything when the exception happens, but you do not wish your program to halt due to these exceptions. In this case, you shall just suppress the exception with pass keyword in the except block.

Below is an example to perform some clean-up in a class destructor method for a selenium automation project. When exception happens, you cannot do anything but letting the program exit peacefully:

def __del__(self):

    if self.driver:
        try:
            self.driver.quit()
        except:
            #suppress any exceptions
            pass

Use of the else block

The Python try-except statement also has a optional else clause, which allows you to execute some code when no exception is raised. This probably useful when you do not want some exceptions to be caught by the try-except which is not intended for them. For instance, when you put a few lines of codes in the try block, and intend to catch the exception for the first line, you may accidently catch some exceptions raised in the subsequent lines. To avoid that, you can split your code and put the rest of the code in the else block where it only get executed when no exception raised for the first line.

Another useful example would be, if you have a for-loop inside the try-except block, and you need to check if all iterations are successful without exception, instead of having a flag to keep track it, you can use the else block:

import math
from random import randint

def check_all_successful(num):
       
    while(True):
        result = {}
        try:
            #some dummy logic for demo purpose
            #exception would not raise if all random numbers generated are positive
            for i in [randint(-num, num) for _ in range(6)]:
                result[i] = math.sqrt(i)
                
        except Exception as e:
            print(f"failed to process: {i}, error: {e}")
        else:
            #when all iterations are successful
            print(result)
            break

When calling this function:

check_all_successful(10)

You may see the function ended when no exception raised, and below is the output:

Perform Clean-up action with Finally Keyword

If there is any operation you must perform regardless of whether the exception is raised or not, such as closing an IO object or network connection, you shall consider to use the finally statement. It is designed for any clean-up logic after the exception is raised, and make sure the code is executed before the exception raised to the caller.

For instance, below is some code snippet to connect to SQLite and update the data for a user table. When exception happens, the finally block shall close the connection properly before program exits:

import sqlite3
con = sqlite3.connect('example.db')

try:
    cur = con.cursor()
    cur.execute("UPDATE USER SET PASSWORD = 'password' WHERE NAME = 'admin'")
    con.commit()
finally:
    con.close()

One thing to be noted, if you have return statements in both try/except and finally block, only the result from finally block will be returned to the caller no matter the exception happens or not. For instance, the below will always return 9:

def test():
    try:
        return 1/0
    except:
        return 2
    finally:
        return 9

Create User Defined Exceptions

When writing your own application, you may want to define your own exceptions based on your business scenario. To create custom exceptions, you shall first create your base exception class which is derived from the Python base exception – Exception, then you can create your other exceptions deriving from this base exception class.

For instance, you may create below exception classes:

# exceptions.py

class AppException(Exception):
    pass

class CurrencyNotSupported(AppException):
    pass

And in your application code, you can raise your own exceptions whenever appropriate. E.g. :

def convert_to_usd(local_currency, local_amount):
    
    if local_currency.lower() not in ["sgd"]:
        raise CurrencyNotSupported(f"The currency {local_currency} is not supported")
        
    return local_amount * 1.32

When you passed in some invalid input to the function:

convert_to_usd("eur", 100)

You shall see the exception raised in the traceback:

Conclusion

In this article, we have reviewed through the different scenarios for raising and handling Python exceptions. As a general rule, you shall not suppress any exception when you are not 100% sure what to do with it. Because this may need to other system bugs or troubleshooting issues later on. The exceptions shall be handled by parties who has the best knowledge to handle it, so raise rather than suppress it when you are not sure how to handle it.