#!/bin/bash

# Pobieranie i rozpakowanie
echo "$KAGGLE_USERNAME"
echo "$KAGGLE_KEY"
kaggle datasets download -d thedevastator/1-5-million-beer-reviews-from-beer-advocate
unzip -o 1-5-million-beer-reviews-from-beer-advocate.zip
DATASET_FILE="beer_reviews.csv"

echo "------------------ Cut off top: ${1} rows ------------------"
head -n $1 $DATASET_FILE > cutoff_$DATASET_FILE

# Mieszanie i podział
echo "------------------ Split and shufle ------------------"
total_lines=$(tail -n +2 cutoff_$DATASET_FILE | wc -l)

train_lines=$((total_lines * 80 / 100))
dev_lines=$((total_lines * 10 / 100))
test_lines=$((total_lines - train_lines - dev_lines))

shuf cutoff_$DATASET_FILE -o shuffled.csv

head -n $train_lines shuffled.csv > train.csv
tail -n $((dev_lines + test_lines)) shuffled.csv | head -n $dev_lines > dev.csv
tail -n $test_lines shuffled.csv > test.csv

# Czyszczenie
echo "------------------ Clean ------------------"
rm cutoff_$DATASET_FILE shuffled.csv

mkdir -p data
mv train.csv dev.csv test.csv data/