============
:code:`genericfuncs` allows you to create functions which execute different
implementations depending on the arguments.
This module can be seen as a powerful improvement over Python 3's :code:`singledispatch`:
* Allows dispatch over any boolean callable, not just type checks.
* Allows dispatch over any number of arguments, not just the first argument.
Basic usage
***********
from genericfuncs import generic
@generic
def func(a):
# default implementation
raise ValueError()
@func.when(lambda a: a.startswith('foo')):
def _impl_1(a):
return a.upper()
@func.when(lambda a: a.startswith('bar')):
def _impl_2(a):
return a.lower()
The first predicate that returns True has its mapped implementation invoked.
Predicates are checked in order of definition.
Installation
************
:code:`pip install genericfuncs`
Advanced
********
Arguments are injected into predicates and implementations by their name.
This means a predicate or implementation is able to specify only the arguments it needs. For example::
@generic
def multiple_params_func(a, b, c):
return a + b + c # default implementation
@multiple_params_func.when(lambda b: b > 10) # only inject argument `b` to the predicate
def _when_b_greater_than_10(a): # only inject `a` to the implementation
return a * 10
@multiple_params_func.when(lambda a, b: a % b == 0) # only inject `a` and `b`
def _when_a_divisible_by_c(a, b, c): # use all arguments
return a / b * c
However the call site must list all mandatory arguments, as usual in Python::
multiple_params_func(10, 20, 30) --> 100 # _when_b_great_than_10() invoked
multiple_params_func(4, 2, 'bla') --> 'blabla' # _when_a_divisible_by_c() invoked
multiple_params_func(0, 0, 0) --> 0 # default implementation invoked
When defining a predicate, one can list exception types that should not
propagate if raised inside the predicate. For example::
@my_generic_func.when(lambda a: a > 10, ignored_errors=[TypeError])
def _implementation(a):
...
When invoking :code:`my_generic_func(MyThing())`, a :code:`TypeError` will be raised inside the predicate
because :code:`MyThing` doesn't support :code:`>` operator.
Normally, the error would propagate and crash the program.
Specifying :code:`ignored_errors=[TypeError]` makes the error be silently ignored,
moving on to the next predicate.
Metadata
Release files for genericfuncs 0.2.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| genericfuncs-0.2.0.zip | 10.9 kB | Without provenance | Details for genericfuncs-0.2.0.zip |
Release files / genericfuncs-0.2.0.zip
| Download URL | genericfuncs-0.2.0.zip |
|---|---|
| Size | 10.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
fdafbd29be4c4594c92d4676ee932893b706b38fbd658f1b80ab06cd03081919
|
|
BLAKE2b-256 checksum How to use checksums |
72a10871200a2f55a105a1aaebfe5441be75b98803c63f6f909f333003a89b8f
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
Security for genericfuncs 0.2.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.