Websocket recv() limited bytes returned

[johnraysimpson]

The question

I have an issue with the websocket implementation, it seems that the recv() function can only return a maximum of 8192 bytes but I have a connection where the receive message is larger than this, I was wondering if there was a parameter I could use to increase/ignore that maximum?

Versions

• OS: Windows 11
• curl_cffi version 0.7.1

[johnraysimpson]

Oh I do see the remaining turns up eventually, I can probably handle that but thought it was worth mentioning. Takes a good 10 seconds to appear

[lexiforest]

Hi, can you have a look at #487, is this the same kind of issue?

[johnraysimpson]

Hey, I don't think so, my script doesn't really have points where recv() hangs and I'm utilising Websocket.run_forever() (I could try to see what happens when I use Websocket.recv() in a for loop to see if it's the same issue). I've traced it back to Curl.ws_recv() in curl.py where it chunks data up into 1024 bytes and collates them, but I can't see anything that is restricting the whole message. It just seems to gather the first 8192 bytes (which would be 8 iterations of Curl.ws_recv()), waits a while then sends the rest of the same message.
Fortunately this only happens at the start of my script, I can send some of it but it's a UK site and I'm not sure if it has location blocking.

EDIT:
Strangely enough, the remainder of the message is larger than 8192 so that's confusing

[johnraysimpson]

Looking into this a bit further, what I've found is that if another, separate message comes through then the end of the last message is attached to the start of the new message, it comes through quicker but can be a bit irritating separating it all and making sure I have full messages