Skip to main content

GUARDS: Handle Exceptions Like Never Before

# Before
def main():
    key_str: str = input("Insert key -> ")
    try:
        key: int = int(key_str)
    except ValueError:
        print("Key is not an integer")
        return
    try:
        value = my_dict[key]
    except KeyError:
        value = "not found"
    print(f"Value at key {key} is {value}")

# After
from guards import *
from operator import getitem
def main():
    safe_int = guard(int, ValueError)
    key_maybe: Outcome[int, ValueError] = safe_int(input("Insert key -> "))
    if iserror(key_maybe):
        print("Key is not an integer")
        return
    key: int = key_maybe.ok
    value = guard(getitem, KeyError)(my_dict, key)
    print(f"Value at key {key} is {value.ok if isok(value) else 'not found'}")

# More compact
from guards import *
from operator import getitem
def main():
    safe_int = guard(int, ValueError)
    key_maybe = safe_int(input("Insert key -> "))
    if let_not_ok(key := key_maybe.let):
        print("Key is not an integer")
        return
    value = guard(getitem, KeyError)(my_dict, key).or_else("not found")
    print(f"Value at key {key} is {value}")

guards is a Python library implementing an alternative to the classic try/except/else statements. You still throw errors with raise, but can catch them in a sweeter and more functional approach. Requires Python 3.10+.

Guards let you handle errors in expressions, not just statements.

safe_int = guard(int, ValueError)
# Impossible with try/except (needs indentation and blocks)
result = [safe_int(x).or_else(0) for x in user_inputs]

This means:

  • Chain operations without nested try blocks
from operator import getitem
l = [6, 2, 5]
safe_get = guard(getitem, IndexError)
text = outcome_do(
    x1 + x2 + x3
    for x1 in safe_get(l, 0)
    for x2 in safe_get(l, 1)
    for x3 in safe_get(l, 2)
).or_else(0)
  • Pass error-handling logic as values
def assert_raises(func, exception, *args, **kwargs):
    match guard(func, exception)(*args, **kwargs):
        Ok(value): raise AssertionError(f"Expected a raised {exception}, but got value {value}")
        Error(exc): return
  • Type-check your error handling
def f(x: str) -> int | None:
    outcome = guard("Hello World".index, ValueError)(x)
    if isok(outcome):
        reveal_type(outcome) # Ok[int]
        return outcome.ok
    reveal_type(outcome) # Error[ValueError]
    #return outcome.ok # Would raise an issue by the type checker
  • And more!
my_list = ["4.2", "2.7", "pizza"]
numbers, errors = outcome_partition(guard(float, ValueError)(x) for x in my_list)

Installation

Install this library with pip like usual:

pip install guards

Summary

The guard() function blocks another function from raising a set of errors. It takes a function f to guard and one or more BaseException types to guard against, and returns a new function.

open_safe = guard(open, FileNotFoundError, PermissionError)
age_outcome = guard(int, ValueError)(input("Insert age -> "))

The returned function calls the original function f with the same arguments passed to it. The difference comes after the function was called:

  • If f returned a value, return an Ok(value) object.
  • If f raised an exception exc it is guarded against, return an Error(exc) object.
  • If f raised an exception it is not guarded against, the exception is propagated.
safe_float = guard(float, ValueError)
print(safe_float("25"))
print(safe_float("ten"))
print(safe_float([4, 2]))

Outputs:

Ok(25.0)
Error(ValueError("could not convert string to float: 'ten'"))
Traceback (most recent call last):
  File ".../script.py", line 5, in <module>
    print(safe_float([4, 2]))
          ^^^^^^^^^^^^^^^^^^
  File ".../guards.py", line 368, in inner_func
    ok = f(*args, **kwargs)
         ^^^^^^^^^^^^^^^^^^
TypeError: float() argument must be a string or a real number, not 'list'

The union of the return types of a guarded function Ok | Error is called Outcome, which is similar to "result" types in other programming languages.

The simplest way to handle exceptions is to use isok() and iserror() functions to check the type of an outcome, then access their internal values with .ok and .error properties.

file_outcome = guard(open, FileNotFoundError, PermissionError)(PATH)
if isok(file_outcome):
    with file_outcome.ok as file:
        print("File contents:")
        print(file.read())
else: # elif iserror(file_outcome):
    os_error = file_outcome.error
    if isinstance(os_error, PermissionError):
        print("Cannot read the file.")
    else: # elif isinstance(os_error, FileNotFoundError):
        print("The file doesn't exist!")

This is just the basic of error-handling. See the documentation or examples.md for more.

Try/Except VS Guards

Try/Except Guards
❌Needs blocks and indentation ✅Can be used inside expressions
❌Strictly procedural ✅Multi-paradigm
❌Encourages coarse-grained error handling ✅Encourages fine-grained error handling (coarse-grained is still possible)
❌Terrible when used often ✅Multiple uses are not a problem
❌Hard to pass around ✅Returned outcome can be passed around
❌Bare except antipattern ✅Specifying no types to guard against is a runtime warning
❌Risk of unbound variables ✅No need to worry about variables
❌Try/Except is all you get ✅Methods and functions for the most common use-cases

Exceptions and Guards VS Pure Result Values

Exceptions and Guards Pure Result Values
Propagating errors is the default Handling errors is the default
✅Raised exceptions are all you need ❌Needs panics to work well
✅Only the exceptions you care about are handled ❌Needs unwrapping the error even if you only care about a specific one
✅Can be learned whenever you have to handle errors ❌Must be learned when writing or using a fallible function
✅Great compatibility with Python ❌Hard to integrate in a language like Python
✅Exceptions automatically create stack traces ❌You often need libraries for backtraces
❌Raised errors cannot be typed (out of this module's scope) ✅Functions' errors are typed
❌Rarely clear whether a function can error ✅Often clear whether a function can error
❌Needs understanding of functions as values ✅Easier to understand
❌Uses a different control flow ✅Control flow is clearer
❌Error handling is easy to ignore ✅Errors must always be dealt with

Contributing

If you found a problem or want to contribute to this package, you can open a new Issue or Pull Request.

Metadata

Release files for guards 1.1.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for guards 1.1.0
File Size Uploaded
guards-1.1.0.tar.gz 21.5 kB Details

Built distribution (wheel)

Table of built distributions (wheels) for guards 1.1.0
File Interpreter ABI Platform
guards-1.1.0-py3-none-any.whl Python 3 none any Details

Total release size: 36.9 kB

Release files / guards-1.1.0.tar.gz

Download URL guards-1.1.0.tar.gz
Size 21.5 kB
Tags Source
SHA-256 checksum
How to use checksums
2bfda5812802287d3826b4b62e85d99169f29444c7e899994643f29a227913b4
BLAKE2b-256 checksum
How to use checksums
47efdec148209a6ff2ed4fb48c8fdffec229335dd80f5669a926a0e754e0759f
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.14.0

Release files / guards-1.1.0-py3-none-any.whl

Download URL guards-1.1.0-py3-none-any.whl
Size 15.3 kB
Tags Python 3
SHA-256 checksum
How to use checksums
3ccaef32a5ee949b23f84bd768dc9180010751bdea6f82d7198ef94261219640
BLAKE2b-256 checksum
How to use checksums
23f12dde6f1aa588784034b61d02ae54b4049ec5c9943554f916ba2987538ff3
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.2.0 CPython/3.14.0

Release history Release notifications | RSS feed

This release

1.1.0 This release

2 release files

1.0.4

2 release files

1.0.3

2 release files

1.0.2

2 release files

1.0.1

2 release files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page