You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Instead of using sourcemod plugins and log parsers, I want to parse STV demos directly. Pulling out many more stats and fixing some bugs with the current system.
@unclamped Yeah i'm aware of that project already and probably makes more sense for their project to do it in python if they are trying to (eventually...) feed it into some ml models.
Currently the project im working on that I link to does a very similar thing, implementing the parsing using the same underlying rust library, but it also exposes it over a HTTP API service so i don't really gain anything by using a python interface. Additionally i want to keep the CPU usage as low as possible as we process a lot of demos, and i want to eventually open it up for the public to use.
Additionally i want to keep the CPU usage as low as possible as we process a lot of demos, and i want to eventually open it up for the public to use.
It's not upstreamed as its still pretty hacky, but I was able to get a pretty huge (~10x) speedup on some custom tf2-demo-parser analyzers by being very careful about handling entity props: demostf/parser@master...Gcommer:parser:perf
I'd be down to contribute some profiling work if this is a pain point for you.
Instead of using sourcemod plugins and log parsers, I want to parse STV demos directly. Pulling out many more stats and fixing some bugs with the current system.
The work for this is being done @ https://github.com/leighmacdonald/tf2_demostats
This project will implement a simple HTTP api that will allow the upload of a demo and return the json formatted parsed stats from it.
This api may eventually be opened up for the publics use as well, possibly paid.
The text was updated successfully, but these errors were encountered: