1.18 Exceptions

This lesson shows the usage of exceptions in python

An exception is basically an error which is raised when a statement or command fails in python.

a = 2

assert isinstance(a, int)

The function isinstance is being used here to check the type of a python object. The above statement did not return anything because the assertion was successful. However if the assertion fails, we get a special kind of error called AssertionError.

AssertionError

If we run the code given below, it will result in AssertionError because

the variable a is int type and not float type.

# uncomment following 1 line
# assert isinstance(a, float)

We can write more useful error messages instead of throwing just AssertionError as follows,

# uncomment following 1 line
# assert isinstance(a, float), f"a is not float but is of {type(float)} type"

Similar to AssertionError, there are many other types of errors/exceptions in python

TypeError

This error is raised when an operation is failed due to wrong type of a variable.

# uncomment following line
# 4 + 'a'  # -> TypeError: unsupported operand type(s) for +: 'int' and 'str'

Since we can not add a string into an integer, we got TypeError above.

ModuleNotFoundError

This error is raised when a specific module that we are importing is not available/installed.

# uncomment following 1 line
# import ai4water  # -> ModuleNotFoundError: No module named 'ai4water'

Since ai4water package is not installed, we got ModuleNotFoundError above.

Whenever we import a variable, function or class or any python object, python searches in all the directories (folders) which are in its path. So when we say a package is not installed or available, it means, this package or the object that we are importing is not available in any of the directories in the path. In other words we also say that, this package/object is not available in python’s path. If we want to check what are the lists of directories/folders in python’s path, we can do this by printing the value of sys.path.

sys.path returns a list so we can iterate over it.

import sys

for p in sys.path:
    print(p)

/home/docs/checkouts/readthedocs.org/user_builds/python-seekho/checkouts/latest/docs/source
/home/docs/.asdf/installs/python/3.9.18/lib/python39.zip
/home/docs/.asdf/installs/python/3.9.18/lib/python3.9
/home/docs/.asdf/installs/python/3.9.18/lib/python3.9/lib-dynload
/home/docs/checkouts/readthedocs.org/user_builds/python-seekho/envs/latest/lib/python3.9/site-packages
/home/docs/checkouts/readthedocs.org/user_builds/python-seekho/checkouts/latest/scripts/basics

KeyError

This error is raised when a specific key is not available in a dictionary.

human = {"name": "Ali"}

# uncomment following 1 line
# print(human['age'])  # KeyError: 'age'

Since the dictionary human does not have age key, we will get KeyError, if we will run the above cell.

PermissionError

This error is raised when the user does not have permission to access the file or to do an operation on such a file for which he/she does not have corresponding rights.

f = open('NewFile.txt', 'w')

import os

# uncomment the following line
# os.remove('NewFile.txt')  # -> PermissionError

since the file NewFile.txt was already opened, we can not delete it and got PermissionError.

f.close()
os.remove('NewFile.txt')

Once we close the file using f.close(), then we can delete the file.

PermissionError is also raised when you are trying to open/create a file but the first argument to open function is folder instead of file.

FileNotFoundError

This error is raised when the the file that we are trying to open does not exist in the directory/path.

# uncomment following 1 line
# open("NonExistingFile", "r")  # -> FileNotFoundError: [Errno 2] No such file or directory: 'NonExistingFile'

IndexError

This error is raised when try to access a value from a sequence based upon the index which is not available for that sequence. For example, for a list with three members, if we try to access its fourth member, we will get IndexError.

my_list = [1, 2, 3]
# uncomment following line
# my_list[3]  # -> IndexError: list index out of range

Same is true for tuples.

my_tuple = (1, 2, 3)

# uncomment following line
# my_tuple[3]  # -> IndexError: tuple index out of range

There is another commonly encountered error i.e., AttributeError. We are not introducing it here. It will discussed later in 3.15 __getattr__.

Error handling

So what is the purpose of knowing all of these errors?

Sometimes we would like to forecast/predict an error and bypass the error by applying the solution. In the function below, we would like to add a value in 2. However, if due to TypeError we can not do so, then we would first convert it to integer before adding it into 2.

def add_number(val):
    try:
        val = val + 2
    except TypeError:
        print('TypeError was encountered but bypassed')
        val = int(val) + 2
    return val


add_number(2)

4

add_number('2')

TypeError was encountered but bypassed

4

Sometimes we want to handle/catch multiple exceptions. We can do this in one line by writing all the exceptions inside a tuple after except keyword.

def multiple_exceptions(wname):
    try:
        out = float(wname.split('_')[2])
    except (ValueError, IndexError) as e:
        raise Exception(f"{wname} raised ", e)
    return out

Above, we want to catch ValueError and IndexError at the same time.

# uncomment following line
# multiple_exceptions("11_2.5.h5")  # -> IndexError

# uncomment following line
# multiple_exceptions("11_2_5.h5")  # -> ValueError

multiple_exceptions("11_2_5")

5.0

Error handling is also used by printing out more useful error messages to the user. Suppose the package ai4water is not installed which may be required. Then instead of just throwing ModuleNotFoundError, we can help the user by giving more helpful error message

def better_error_message():
    try:
        import ai4water
    except ModuleNotFoundError as e:
        raise ModuleNotFoundError(f"You must install ai4water using pip install ai4water \n{e}")
    return

# uncomment following line
# better_error_message()

Custom Errors

We can create our own errors based upon our needs. Any new error must inherit from Exception class or its base classes. If at this stage you are uncomfortable with the concept of class or base class, you can jump to lessons of 3.1 introduction and 3.13 inheritance in 3. oop chapter to know about them.

Below we create an error which will be raised if a value is above 256.

class IllegalValue(Exception):
    def __init__(self, val):
        self.val = val

    def __str__(self):
        return f"val must be below 256 but it is {self.val}"


value = 1000

We can invoke this as given below.

# Uncomment following two lines
# if value>256:
#     raise IllegalValue(value)