LookupError: unknown encoding: windows-874

In a previous version, I just passed the entire buffer to encoding_rs and had it handle decoding entirely in rust, but that was a little confusing.

Need more robust aliases

Here's the current supported architecture/os matrix.

Information about which architectures and operating systems Python supports

https://pythondev.readthedocs.io/platforms.html

Architectures

Well supported architectures:

• Intel x86 (32-bit) and x86_64 (64-bit, aka AMD64)

Best effort support architectures:

• ppc64le: should be well supported in practice
• AArch64
• ARMv7: should be well supported in practice
• s390x

Operating systems

• Linux
• Windows 8 and newer for Python 3.9
• FreeBSD 10 and newer
• macOS Snow Leopard (macOS 10.6, 2008) and newer

In compat.detect, the confidence is always 0.99:

It’s possible I’m misunderstanding things, but it seems like you could use guess_assess() instead of guess() to at least determine high vs. low confidence. From the chardetng docs:

Same as guess(), but also returns a Boolean indicating whether the guessed encoding had a higher score than at least one other candidate. If this method returns false, the guessed encoding is likely to be wrong.

On a related note, it would be nice if shortcuts.detect() returned the boolean from guess_assess, too (or had another method that did so), since the aliases buried in the shortcuts module seem pretty important, and possibly prone to change. Otherwise a user has to know about them if they want to safely use guess_assess instead. (Or the aliases could be documented and moved to a more accessible place.)

Currently, can be difficult for new users to find docs. Add clearer links to documentation

• Add link on pypi to documentation
• Call out written documentation in readme

Originally suggested here: #109 (comment)