#!/bin/bash
pip install kaggle
kaggle datasets download ruchi798/movies-on-netflix-prime-video-hulu-and-disney --unzip
sed -i '/^$/d' ./MoviesOnStreamingPlatforms_updated.csv

shuf ./MoviesOnStreamingPlatforms_updated.csv | tail -n +"${CUTOFF}" > ./MoviesOnStreamingPlatforms_updated.shuf
head -n 1674 ./MoviesOnStreamingPlatforms_updated.shuf > ./MoviesOnStreamingPlatforms_updated.test
head -n 3348 ./MoviesOnStreamingPlatforms_updated.shuf | tail -n 1674 > ./MoviesOnStreamingPlatforms_updated.dev
tail -n +3349 ./MoviesOnStreamingPlatforms_updated.shuf > ./MoviesOnStreamingPlatforms_updated.train
rm ./MoviesOnStreamingPlatforms_updated.shuf
rm ./MoviesOnStreamingPlatforms_updated.csv
wc -l ./MoviesOnStreamingPlatforms_updated*