On curlFS, webFS

The Vision

My plans for curlFS isn’t fully defined at this point, but here’s my overarching vision and thoughts right now.

• I want to be able to cd into my favorite websites, and ls to see updates.
• I want to be able to cat a URL and see the page in my terminal.
• I want to be able to echo data into a URL to POST to it.

To expand upon this, I want to be able to read updates / pages from my favorite websites in a clean, readable format that works well with existing tools.

Unfortunately, not too many websites out there have a text/plain option!

I’m thinking of solving the problem in the following way:

• Define your favorite websites in terms of their API schema
• Define translators / transformers for the data. Looking at unfomatted JSON can {"get":["visually","exhausting"]}

What if you could just see a markdown formatted post in your terminal? What if you could see images from the article processed with libcaca or aalib?

Well, maybe that’s a bit too far.

I’m going to try and get a rudimentary, all-inclusive curlFS up and running. Later, I will try to split it out into several smaller, focused libraries.

The Stack

The stack would have the following components:

libcpptsb: C++11/C++14 thread-safe buffers.

Read and write at the same time as long as the read won’t overlap the writing point, at which point it will block.

curlcpppp: an extended curlcpp.

I’d like to add a bit to the existing curlcpp library. It’s overall a nice, simple, small library, but I think it could be made a little nicer to work with. Whether this would be an additional wrapper on top, a fork, or changes contributed upstream remains to be seen.

Coincidentally, while looking for the link to curlcpp, I found the simlarly-named curlpp, which I’ll need to investigate.

curlFS: A FUSE interface for CURL.

Picture ls or stat as a HEAD request, reading from the file a GET, and writing to the file a POST.

transformator: Transform content, transparently.

Picks translation programs based on path, configuration, and analyzing the content. These programs would take the file contents at the given path, determine the transforming program, write to its standard input, and read back its standard output. This would be an additional FUSE filesystem layered over curlFS. However, I’d like to look into plan9’s plumber a bit more– it might be the perfect tool for this job.