#!/bin/bash

pip install kaggle

kaggle datasets download -d thedevastator/1-5-million-beer-reviews-from-beer-advocate

unzip -o 1-5-million-beer-reviews-from-beer-advocate.zip

shuf beer_reviews.csv > shuffled_dataset.csv

split -l 80000 shuffled_dataset.csv train.csv
split -l 10000 train.csv dev.csv
mv shuffled_dataset.csv test.csv

head -n 1000 train.csv > train_head.csv
tail -n 1000 train.csv > train_tail.csv

if [ -n "$CUTOFF" ]; then
  head -n "$CUTOFF" train.csv > train_cutoff.csv
fi

tar -czf artifacts.tar.gz train.csv dev.csv test.csv train_head.csv train_tail.csv train_cutoff.csv

rm $DATASET_FILE shuffled_dataset.csv

echo "artifacts.tar.gz"