Explaining the Benefits of Re-Raising Exceptions in Python
The aim of this page📝is to explain the usefulness of re-raising exceptions with raise keyword via the 3 use cases.
2 min readSep 9, 2022
1. USECASE #1: BACKUP/ARCHIVE
- Imagine the following code.
- A little setup: You are responsible for maintaining a huge database of information for example, and any loss of data would be catastrophic!
- Now, if you wanted to perform logging or cleanup that only occurs on failure, you could put that between the
except:and theraise - → you’d do that work and then proceed as if the original exception was bubbling normally.
huge_dictionary = {'lots_of_important':['stuffs']}
try:
check_data(new_data) #make sure the data is in the correct format
huge_dictionary['lots_of_important'].append(new_data)
except:
data_writer.backup(huge_dictionary)
data_writer.close()
#and any other last-second changes
raise— from https://stackoverflow.com/a/39144452/11082684
2. USECASE #2: EXCEPTION CHAINING
- You get a confusing exception type and want to convert it into something more conventional
- See https://medium.com/p/36061c839985
- See python — Re-raise exception with a different type and message, preserving existing information — Stack Overflow
def multiplier(*args):
try:
i = iter(args)
result = next(i)
for the item in i:
result *= item
return result
except StopIteration as e:
raise TypeError("At least 1 argument is required") from e
>>> multiplier()
TypeError: Error: At least 1 argument required3. USECASE #3: LOGGING / PASSWORD BLURRING
- You can use query data in Redshift/another DB with Python and may want to log failures and replace credentials along the way
try:
redshift_host = "cluster01.redshift.foo.com"
redshift_db = "bar5"
redshift_port = 5348
redshift_user = "acme_108"
redshift_password = "password"
#... connect → execute → commit query against Redshift
except Exception as e:
logger.error(str(e).replace(redshift_password, "*"))
raise e
