aiopath 0.6.5

📁 Async pathlib for Python

📁 Async pathlib for Python

aiopath is a complete implementation of Python's pathlib that's compatible with asyncio and the async/await syntax.

All I/O performed by aiopath is asynchronous and awaitable.

Use case

If you're writing asynchronous Python code and want to take advantage of pathlib's conveniences, but don't want to mix blocking and non-blocking I/O, then you can reach for aiopath.

For example, if you're writing an asynchronous web scraping script, you might want to make several concurrent requests to websites and write the content in the responses to secondary storage:

from asyncio import run, gather

from aiohttp import ClientSession
from aiopath import AsyncPath


async def save_page(url: str, name: str):
  path = AsyncPath(name)

  if await path.exists():
    return

  async with ClientSession() as session:
    response = await session.get(url)
    content: bytes = await response.read()

  await path.write_bytes(content)


async def main():
  urls = [
    'https://example.com',
    'https://github.com/alexdelorenzo/aiopath',
    'https://alexdelorenzo.dev',
    'https://dupebot.firstbyte.dev'
  ]

  scrapers = (
    save_page(url, f'{index}.html')
    for index, url in enumerate(urls)
  )

  await gather(*scrapers)


run(main())

If you used pathlib instead of aiopath in the example above, some tasks would block upon writing to the disk, and the other tasks making network connections would be forced to pause while the disk is accessed.

By using aiopath in the example above, the script can access the network and disk concurrently.

Implementation

aiopath is a direct reimplementation of CPython's pathlib.py and shares some of its code. aiopath's class hierarchy directly matches the one from pathlib, where Path inherits from PurePath, AsyncPath inherits from AsyncPurePath, and so on.

With aiopath, methods that perform I/O are asynchronous and awaitable, and methods that perform I/O and return iterators in pathlib now return async generators. aiopath goes one step further, and wraps os.scandir() and DirEntry to make AsyncPath.glob() completely async.

aiopath is typed with Python type annotations, and it takes advantage of libaio for async I/O on Linux.

Usage

aiopath's API directly matches pathlib, so check out the standard library documentation for PurePath and Path.

Running examples

To run the following examples with top-level await expressions, launch an asynchronous Python REPL using python3 -m asyncio or an IPython shell.

You'll also need to install asynctempfile via PyPI, like so python3 -m pip install asynctempfile.

Replacing pathlib

All of pathlib.Path's methods that perform synchronous I/O are reimplemented as asynchronous methods. PurePath methods are not asynchronous because they don't perform I/O.

from pathlib import Path

from asynctempfile import NamedTemporaryFile
from aiopath import AsyncPath


async with NamedTemporaryFile() as temp:
  path = Path(temp.name)
  apath = AsyncPath(temp.name)

  # check existence
  ## sync
  assert path.exists()
  ## async
  assert await apath.exists()

  # check if file
  ## sync
  assert path.is_file()
  ## async
  assert await apath.is_file()

  # touch
  path.touch()
  await apath.touch()

  # PurePath methods are not async
  assert path.is_absolute() == apath.is_absolute()
  assert path.as_uri() == apath.as_uri()

  # read and write text
  text: str = 'example'
  await apath.write_text(text)
  assert await apath.read_text() == text

assert not path.exists()
assert not await apath.exists()

You can convert pathlib.Path objects to aiopath.AsyncPath objects, and vice versa:

from pathlib import Path
from aiopath import AsyncPath


home: Path = Path.home()
ahome: AsyncPath = AsyncPath(home)
path: Path = Path(ahome)

assert isinstance(home, Path)
assert isinstance(ahome, AsyncPath)
assert isinstance(path, Path)

# AsyncPath and Path objects can point to the same file
assert str(home) == str(ahome) == str(path)

# but AsyncPath and Path objects are not equivalent
assert not home == ahome

AsyncPath is a subclass of Path and PurePath, and a subclass of AsyncPurePath:

from pathlib import Path, PurePath
from aiopath import AsyncPath, AsyncPurePath


assert issubclass(AsyncPath, Path)
assert issubclass(AsyncPath, PurePath)
assert issubclass(AsyncPath, AsyncPurePath)
assert issubclass(AsyncPurePath, PurePath)

path: AsyncPath = await AsyncPath.home()

assert isinstance(path, Path)
assert isinstance(path, PurePath)
assert isinstance(path, AsyncPurePath) 

Check out tests.py for more examples of how aiopath compares to pathlib.

Opening a file

You can get an asynchronous file-like object handle by using asynchronous context managers.

AsyncPath.open()'s async context manager yields an aiofile.AIOFile object.

from asynctempfile import NamedTemporaryFile
from aiopath import AsyncPath


text: str = 'example'

# you can access a file with async context managers
async with NamedTemporaryFile() as temp:
  path = AsyncPath(temp.name)

  async with path.open(mode='w') as file:
    await file.write(text)

  async with path.open(mode='r') as file:
    result: str = await file.read()

  assert result == text

# or you can use the read/write convenience methods
async with NamedTemporaryFile() as temp:
  path = AsyncPath(temp.name)

  await path.write_text(text)
  result: str = await path.read_text()
  assert result == text

  content: bytes = text.encode()

  await path.write_bytes(content)
  result: bytes = await path.read_bytes()
  assert result == content

Globbing

aiopath implements pathlib globbing using async I/O and async generators.

from aiopath import AsyncPath


home: AsyncPath = await AsyncPath.home()

async for path in home.glob('*'):
  assert isinstance(path, AsyncPath)
  print(path)

downloads: AsyncPath = home / 'Downloads'

if await downloads.exists():
  # this might take a while
  paths: list[AsyncPath] = \
    [path async for path in downloads.glob('**/*')]

Installation

Dependencies

• A POSIX compliant OS, or Windows
• Python 3.7+
• requirements.txt

Linux dependencies

If you're using a 4.18 or newer kernel and have libaio installed, aiopath will use it via aiofile. You can install libaio on Debian/Ubuntu like so:

$ sudo apt install libaio1 libaio-dev

PyPI

$ python3 -m pip install aiopath

Python 3.9 and older

aiopath for Python 3.9 and older is available on PyPI under versions 0.5.x and lower.

Python 3.10 and newer

aiopath for Python 3.10 and newer is available on PyPI under versions 0.6.x and higher.

GitHub

$ python3 -m pip install -r requirements.txt
$ python3 setup.py install

Python 3.9 and older

aiopath for Python 3.9 and older is developed on the Python-3.9 branch.

Python 3.10 and newer

aiopath for Python 3.10 and newer is developed on the Python-3.10 branch.

Support

Want to support this project and other open-source projects like it?

License

See LICENSE. If you'd like to use this project with a different license, please get in touch.

Credit

See CREDIT.md.