🗝 Fix and improve `typer` 🗝
Project description
🗝 Fix and improve typer
🗝
What is dtyper
, in one sentence?
Using import dtyper as typer
instead of import typer
would make your
typer.command
s directly callable.
(There's also a neat way to make dataclasses from typer commands, but that would be two sentences.)
Why dtyper
?
typer
is a famously clear and useful system
for writing Python CLIs but it has two issues that people seem to run into a
lot:
-
You can't call the
typer.command
functions it creates directly because they have the wrong defaults. -
As you add more arguments to your CLI, there is no easy way to break up the code sitting in one file without passing around long, verbose parameter lists.
dtyper
is a tiny, single-file library that adds to an existing installation
of typer
to solve these two problems without changing existing code at all.
-
dtyper.command
executestyper.command
then fixes the defaults. -
dtyper.function
decorates an existingtyper.command
to have correct defaults. -
dtyper.dataclass
automatically makes adataclass
from atyper.command
.
How to use dtyper
?
Install as usual with poetry add dtyper
, pip install dtyper
, or your
favorite package manager.
dtyper
is a drop-in replacement for typer
- it copies all typer
s
properties - so you can even write
import dtyper as typer
to experiment with it before deciding.
dtyper
has two new functions that typer
doesn't, and overrides a typer
class:
-
@dtyper.function
is a decorator that takes atyper
command and returns a callable function with the correct defaults. It is unncessary if you usedtyper.Typer
(below) -
@dtyper.dataclass
is a decorator that takes an existingtyper
ordtyper
command and makes adataclass
from it. -
dtyper.Typer
is a class identical totyper.Typer
, except it fixesTyper.command
functions so you can call them directly.
None of the typer
functionality is changed to the slightest degree - adding
dtyper
will not affect how your command line program runs at all.
Example 1: using dtyper
instead of typer
from dtyper import Argument, Option, Typer
app = Typer()
@app.command(help='test')
def get_keys(
bucket: str = Argument(
'buck', help='The bucket to use'
),
keys: bool = Option(
False, help='The keys to download'
),
):
print(bucket, keys)
You can call get_keys()
from other code and get the right defaults.
Without regular typer
, you sometimes get a typer.Argument
or
typer.Option
in place of an expected str
or bool
.
Example 2: a simple dtyper.dataclass
Here's a simple CLI in one Python file with two Argument
s and an Option
:
@command(help='test')
def get_keys(
bucket: str = Argument(
..., help='The bucket to use'
),
keys: str = Argument(
'keys', help='The keys to download'
),
pid: Optional[int] = Option(
None, '--pid', '-p', help='process id, or None for this process'
),
):
get_keys = GetKeys(**locals())
print(get_keys.run())
@dtyper.dataclass(get_keys)
class GetKeys:
site = 'https://www.some-websijt.nl'
def run(self):
return self.url, self.keys, self.pid
def __post_init__(self):
self.pid = self.pid or os.getpid()
def url(self):
return f'{self.site}/{self.url}/{self.pid}'
Example: splitting a large typer.command
into multiple files
Real world CLIs frequently have dozens if not hundreds of commands, with hundreds if not thousands of options, arguments, settings or command line flags.
The natural structure for this is the "big ball of mud", a popular anti-pattern known to cause misery and suffering to maintainers.
dtyper.dataclass
can split the user-facing definition of the API from its
implementation and then split that implementation over multiple files in a
natural and convenient way.
The example has three Python files.
interface.py
contains the Typer CLI definitions for this command.
@command(help='test')
def big_calc(
bucket: str = Argument(
..., help='The bucket to use'
),
more: str = Argument(
'', help='More information'
),
enable_something: boolean = Option(
False, help='Turn on one of many important parameters'
),
# [dozens of parameters here]
):
d = dict(locals()) # Capture all the command line arguments as a dict
from .big_calc import BigCalc # Lazy import to avoid a cycle
bc = BigCalc(**d)
bc.run()
big_calc.py
contains the dtyper.dataclass
implementation
from .interface import big_calc
from . import helper
import dtyper
@dtyper.dataclass(big_calc)
class BigCalc:
def run(self):
# Each argument in `big_calc` becomes a dataclass field
print(self.bucket, self.more)
print(self) # dataclass gives you a nice output of all fields
if helper.huge_thing(self) and self._etc():
self.stuff()
helper.more_stuff(self)
...
def _etc(self):
...
# Dozens more methods here perhaps!
Some of the code is offloaded to helper files like helper.py
:
def huge_thing(big_calc):
if has_hole(big_calc.bucket):
fix_it(big_calc.bucket, big_calc.more)
def more_stuff(big_calc):
# even more code
API Documentation
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file dtyper-2.5.1.tar.gz
.
File metadata
- Download URL: dtyper-2.5.1.tar.gz
- Upload date:
- Size: 6.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/1.7.1 CPython/3.10.11 Darwin/21.6.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 7644462564c99e7eca8a1635c4b5841c118c16443d72237fc0e2c992a111fba0 |
|
MD5 | 8c792bc48cccd4e64532ce18870389fa |
|
BLAKE2b-256 | b33c4202f1c087524dde9240fe95f7cbd1181c21949f2e8f5634dd3ac413200d |
File details
Details for the file dtyper-2.5.1-py3-none-any.whl
.
File metadata
- Download URL: dtyper-2.5.1-py3-none-any.whl
- Upload date:
- Size: 8.5 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/1.7.1 CPython/3.10.11 Darwin/21.6.0
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 50d2257dae1dfca1380f8a1c1495e88e096b9fa55ca829555a9b4e7c78ce8873 |
|
MD5 | 8cbf55d1e5beb22e29728dafbd71bb23 |
|
BLAKE2b-256 | 7b291887e294502992a04852c7b6d860c1e586917cf1eb3c14aec1bad6c4eea9 |