This repo contains code to perform distributed analysis on chess games (.pgn files). The analyses in old_version was used to create this article.
The data directory contains the raw data that was used to create this demo.
It also includes job specifications (chessMap and chessReduce) which will
launch the pfs jobs that actually perform the analysis and a script which will
send all of these to a running pfs instance using curl and kickoff the pipeline
with a commit.
The output contains the results of running the job that we got from pfs
(output/blunders) as well as a version of that data bucketed by rating
output/bucketed
map and reduce contain the actual implementation of the jobs used to do the
processing.
Dockerfile can be used to build the repo in to a container which you can give
to pfs. This container already exists in docker hub at pachyderm/chess but
you can use the Dockerfile to create a container with your own jobs. To use
your new jobs you'll need to change the image used in data/chessMap and
data/chessReduce.