I am trying to create a data endpoint that streams either the entirety of a file or responds appropriately to range requests. Streaming the whole file seems understandable, but it's not clear to me how to deal with range requests. Particularly, I can't see how aiohttp.MultipartWriter
can write to a StreamResponse
.
Here's an abstracted form of my code, so far:
from aiohttp.web import Request, StreamResponse
from aiohttp.multipart import MultipartWriter
async def data_handler(req:Request) -> StreamResponse:
is_range_request = "Range" in req.headers
with open("my_big_file", "rb") as f:
if is_range_request:
status_code = 202
content_type = "multipart/bytes"
else:
status_code = 200
content_type = "application/octet-stream"
resp = SteamResponse(status=status_code, headers={"Content-Type": content_type})
resp.enable_chunked_encoding()
resp.enable_compression()
await resp.prepare(req)
if is_range_request:
# _parse_range_header :: str -> List[ByteRange]
# ByteRange = Tuple[int, int] i.e., "from" and "to", inclusive
ranges = _parse_range_header(req.headers["Range"])
mpwriter = MultipartWriter("bytes")
for r in ranges:
range_from, range_to = r
range_size = (range_to - range_from) + 1
range_header = {"Content-Type": "application/octet-stream"}
# FIXME Won't this block?
f.seek(range_from)
mpwriter.append(f.read(range_size), range_header)
# TODO Write to response. How?...
else:
while True:
data = f.read(8192)
if not data:
await resp.drain()
break
resp.write(data)
return resp
This also doesn't return the response until it gets to the end. This doesn't seem correct to me: How does an upstream call know what's going on until the response is returned; or is the asyncio
stuff doing this for me automagically?