add more type checking in params/form-data

[trim21]

• Initially raised as discussion bytes in form is handled in a very strange way #3168

Problem

we should check input value types. Current behavoir which simply convert input value to str(v) is not very ideal.

for example:

import httpx


r = httpx.post(
    "https://httpbin.org/post",
    data={
        "content": bytes.fromhex(
            "4f5f13d16da9f10d013933673a1ef4e7b974dbe3"
        ),  # random non-printable bytes
    },
)
print(r.json()["form"])

will get {'content': "b'O_\\x13\\xd1m\\xa9\\xf1\\r\\x0193g:\\x1e\\xf4\\xe7\\xb9t\\xdb\\xe3'"}

Proposal

add type checking and more coerce.

for example:

form/query

• convert int/float to str with str(v)
• bool (Represent as JSON-style "true"/"false")
• None (Represent as empty string)
• Enum (Use .value and then as above.)
• TypeError for anything else.

multipart_data

not sure how to handle this....

This will cause breaking changes, so we should consider do a minor version bump.

---

related: #1500 #1539 #1608

httpx/httpx/_content.py

Lines 136 to 149 in 4b85e6c

	def encode_urlencoded_data(
	data: RequestData,
	) -> tuple[dict[str, str], ByteStream]:
	plain_data = []
	for key, value in data.items():
	if isinstance(value, (list, tuple)):
	plain_data.extend([(key, primitive_value_to_str(item)) for item in value])
	else:
	plain_data.append((key, primitive_value_to_str(value)))
	body = urlencode(plain_data, doseq=True).encode("utf-8")
	content_length = str(len(body))
	content_type = "application/x-www-form-urlencoded"
	headers = {"Content-Length": content_length, "Content-Type": content_type}
	return headers, ByteStream(body)

httpx/httpx/_utils.py

Lines 56 to 68 in 4b85e6c

	def primitive_value_to_str(value: PrimitiveData) -> str:
	"""
	Coerce a primitive data type into a string value.
	Note that we prefer JSON-style 'true'/'false' for boolean values here.
	"""
	if value is True:
	return "true"
	elif value is False:
	return "false"
	elif value is None:
	return ""
	return str(value)