Constrain which encodings are supported by response.text.

[tomchristie]

• Initially raised as discussion AssertionError #2881

---

Currently when accessing response.text any installed codec may be loaded, depending on the Content-Type header of the response. This is problematic partly because not all codecs are text codecs. It also feels too open, as custom codecs might be installed with arbitrary behaviours.

May suggestion would be that we support the same set of encodings as the chromium browser... https://chromium.googlesource.com/chromium/chromium/+/refs/heads/trunk/chrome/browser/character_encoding.cc#36

We can effect this change by having a hardcoded set of supported codecs, here...

httpx/httpx/_utils.py

Lines 71 to 79 in e63b659

	def is_known_encoding(encoding: str) -> bool:
	"""
	Return `True` if `encoding` is a known codec.
	"""
	try:
	codecs.lookup(encoding)
	except LookupError:
	return False
	return True