Skip to main content

No project description provided

Project description

DrResult

More radical approach to Rust's std::result in Python.

Motivation

I do not want exceptions in my code. Rust has this figured out quite neatly by essentially revolving around two pathways for errors: A possible error condition is either one that has no prospect of being handled -- then the program should terminate -- or it is one that could be handled -- then it has to be handled or explicitly ignored.

This concept is replicated here by using mapping all unhandled exceptions to Panic and providing a Rust-like result type to signal error conditions that do not need to terminate the program.

Documentation

Concept

At each point in your code, there are exceptions that are considered to be expected and there are exceptions that are considered unexpected.

In general, an unexpected exception will be mapped to Panic. No part of DrResult will attempt to handle a Panic exception. And you should leave all exception handling to DrResult, i.e. have no try/except blocks in your code. In any case, you should never catch Panic. An unexpected exception will therefore result in program termination with stack unwinding.

You will need to specify which exceptions are to be expected. There are two modes of operation here: You can explicitly name the exceptions to be expected in a function. Or you can skip that and basically expect all exceptions.

Basically means: By default only Exception is expected, not BaseException. And even of type Exception that are some considered to be never expected:

AssertionError, AttributeError, ImportError, MemoryError, NameError, SyntaxError, SystemError, TypeError

If you do not explicitly expect these, they will be implicitly unexpected. (Obviously, the exact list may be up for debate.)

Basic Usage

@noexcept

If your function knows no errors, you mark it as @noexcept():

from drresult import noexcept

@noexcept()
def sum(a: list) -> int:
    result = 0
    for item in a:
        result += item
    return result

result = func([1, 2, 3])

This will do what you expect it does. But if you screw up...

@noexcept()
def sum(a: list) -> int:
    result = 0
    for item in a:
        result += item
    print(a[7])   # IndexError
    return result

result = func([1, 2, 3])    # Panic!

... then it will raise Panic preserving the stack trace and the original exception.

This way all unexpected exceptions are normalised to Panic.

Please note that a @noexecpt function does not return a result but just the return type itself.

@returns_result() and expects

Marking a function as @returns_result will wrap any exceptions thrown in an Err result. But only those exceptions that are expected. As noted above, if you do not explicitly specify exceptions to expect, most runtime exceptions are expected by default.

@returns_result()
def read_file() -> Result[str]:
    with open('/some/path/that/might/be/invalid') as f:
        return Ok(f.read())

result = read_file()
if result.is_ok():
    print(f'File content: {result.unwrap()}')
else:
    print(f'Error: {str(result.unwrap_err())}')

This will do as you expect.

You can also explicitly specify the exception to expect:

@returns_result(expects=[FileNotFoundError])
def read_file() -> Result[str]:
    with open('/some/path/that/might/be/invalid') as f:
        return Ok(f.read())

result = read_file()
if result.is_ok():
    print(f'File content: {result.unwrap()}')
else:
    print(f'Error: {str(result.unwrap_err())}')

This also will do as you expect.

If fail to specify an exception that is raised as expected...

from drresult import returns_result

@returns_result(expects=[IndexError, KeyError])
def read_file() -> Result[str]:
    with open('/this/path/is/invalid') as f:
        return Ok(f.read())

result = read_file()    # Panic!

.. it will be re-raised as Panic.

If you are feeling fancy, you can also do pattern matching:

@returns_result(expects=[FileNotFoundError])
def read_file() -> Result[str]:
    with open('/this/path/is/invalid') as f:
        return Ok(f.read())

result = read_file()
match result:
    case Ok(v):
        print(f'File content: {v}')
    case Err(e):
        print(f'Error: {e}')

And even fancier:

data = [{ 'foo': 'value-1' }, { 'bar': 'value-2' }]

@returns_result(expects=[IndexError, KeyError, RuntimeError])
def retrieve_record_entry_backend(index: int, key: str) -> Result[str]:
    if key == 'baz':
        raise RuntimeError('Cannot process baz!')
    return Ok(data[index][key])

def retrieve_record_entry(index: int, key: str):
    match retrieve_record_entry_backend(index: int, key: str):
        case Ok(v):
            print(f'Retrieved: {v}')
        case Err(IndexError()):
            print(f'No such record: {index}')
        case Err(KeyError()):
            print(f'No entry `{key}` in record {index}')
        case Err(RuntimeError() as e):
            print(f'Error: {e}')

retrieve_record_entry(2, 'foo')    # No such record: 2
retrieve_record_entry(1, 'foo')    # No entry `foo` in record 1
retrieve_record_entry(1, 'bar')    # Retrieved: value-2
retrieve_record_entry(1, 'baz')    # Error: Cannot process baz!

Implicit conversion to bool

If you are feeling more lazy than fancy, you can do this:

result = Ok('foo')
assert result

result = Err('bar')
assert not result

unwrap_or_raise

You can replicate the behaviour of Rust's ?-operator with unwrap_or_raise:

@returns_result()
def read_json(filename: str) -> Result[str]:
    with open(filename) as f:
        return Ok(json.loads(f.read()))

@returns_result()
def parse_file(filename: str) -> Result[dict]:
    content = read_file(filename).unwrap_or_raise()
    if not 'required_key' in content:
        raise KeyError('required_key')
    return Ok(content)

If the result is not Ok, unwrap_or_raise() will re-raise the contained exception. Obviously, this will lead to an assertion if the contained exception is not expected.

gather_result

When you are interfacing with other modules that use exceptions, you may want to react to certain exceptions being raised. To avoid having to use try/except again, you can transform exceptions from a part of your code to results:

@returns_result()
def parse_json_file(filename: str) -> Result[dict]:
    with gather_result() as result:
        with open(filename) as f:
            result.set(Ok(json.loads(f.read())))
    result = result.get()
    match result:
        case Ok(data):
            return Ok(data)
        case Err(FileNotFoundError()):
            return Ok({})
        case _:
            return result

Printing the Stack Trace

If you want to format the exception stored in Err, you can use Err.__str__() and Err.trace(). The former will just provide the error message itself where the latter will provide the entire stack trace. The trace is filtered to remove all intermediate frames for internal functions.

Also, DrResult overrides the expecthook to filter the stack trace in case of panic.

constructs_as_result

If you have a class that might raise an error in its constructor, you can mark it as constructs_as_result:

@constructs_as_result
class Reader:
    def __init__(self, filename):
        with open(filename) as f:
            self.data = json.loads(f.read())

Creating an instance of this class will yield a Result that has to be unwrapped first.

reader = Reader('/path/to/existing/file').unwrap() # Ok
reader = Reader('/invalid/path/to/non/existing/file').unwrap() # panic!

Similar Projects

For a less extreme approach on Rust's result type, see:

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

drresult-0.6.3.tar.gz (9.3 kB view details)

Uploaded Source

Built Distribution

drresult-0.6.3-py3-none-any.whl (9.3 kB view details)

Uploaded Python 3

File details

Details for the file drresult-0.6.3.tar.gz.

File metadata

  • Download URL: drresult-0.6.3.tar.gz
  • Upload date:
  • Size: 9.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.8.3 CPython/3.10.12 Linux/5.15.0-122-generic

File hashes

Hashes for drresult-0.6.3.tar.gz
Algorithm Hash digest
SHA256 f9987a779a7d72e9b56cf0cebbf3a76320eea7cae784794160ac90619b91d331
MD5 744231662d4564872ca5f80643f262db
BLAKE2b-256 5dce1e3e9d63760514f7824545e49c6ece842801f795c0ae9af82db3e684e644

See more details on using hashes here.

File details

Details for the file drresult-0.6.3-py3-none-any.whl.

File metadata

  • Download URL: drresult-0.6.3-py3-none-any.whl
  • Upload date:
  • Size: 9.3 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: poetry/1.8.3 CPython/3.10.12 Linux/5.15.0-122-generic

File hashes

Hashes for drresult-0.6.3-py3-none-any.whl
Algorithm Hash digest
SHA256 0ff182d79ad4320030a39cb8fc1f351db095fc8f223465a75f0d87a8f30e606b
MD5 356bffb8056f5bc6cb313ecd9816a110
BLAKE2b-256 b473f1cf7e2a3ee67acb9dad7baedc74e4f0f606e25a39cf764ac039555f5c5e

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page