Here's a Python async function. What would the Elixir version be like?

Questions / Help
, ,
1

I have a function repair_file_in_place(path) that makes HTTP connections to check links for 404 errors.

Here’s the Python code to run repair_file_in_place() concurrently on all the markdown files in a directory:

async def repair_files_in_directory(path: str):
    """
    Repair all markdown files recursively in a directory.
    """
    async with aiohttp.ClientSession() as session:
        async with asyncio.TaskGroup() as group:
            for root, _, files in os.walk(path):
                for file in files:
                    if file.endswith(".md"):
                        group.create_task(repair_file_in_place(os.path.join(root, file), session))
  • How would the code be structured in Elixir / OTP?
  • Is there some kind of semaphore-like ability to limit the concurrency? My use case is not blowing the ulimit on open files.
2

Probably something like:

path
|> Path.join("**/*.md")
|> Path.wildcard()
|> Task.async_stream(&repair_file_in_place/1)
|> Stream.run()

Concurrency can be limited using options passed to async_stream. Docs: Task — Elixir v1.16.0

(Note: untested and possibly not optimal, but this is the first that came to mind!)

9 Likes
3

You are almost there, you forgot the if part:

paths
|> Enum.filter(&String.ends_with?(&1, ".md"))
...
4

This is handled by the wildcard, which expands the glob "your/path/**/*.md" into a list of all the .md files that live anywhere in your/path. :smiling_face:

2 Likes
5

It’s amazing that this actually works, and is so simple. Especially compared with the code I showed.

6

People dig Elixir for a reason, my dude. :smiley:

That kind of expressive and readable code hooked me to Elixir almost 8 years ago.

1 Like
7

Yeah, my bad. Good job.