#!/bin/bash

pip install kaggle --upgrade

kaggle datasets download -d gulczas/spotify-dataset

shuf Spotify_Dataset.csv -o shuffled_spotify.csv

head -n 100 shuffled_spotify.csv > subset1.csv
tail -n 100 shuffled_spotify.csv > subset2.csv

cut -d ',' -f 1,2,3 shuffled_spotify.csv > trimmed_spotify.csv

cut -d ',' -f 1,2,4,5,6 shuffled_spotify.csv > processed_spotify.csv

echo "Shuffled dataset:" > results.txt
head shuffled_spotify.csv >> results.txt
echo "" >> results.txt

echo "Subset 1:" >> results.txt
head subset1.csv >> results.txt
echo "" >> results.txt

echo "Subset 2:" >> results.txt
head subset2.csv >> results.txt
echo "" >> results.txt

echo "Trimmed dataset:" >> results.txt
head trimmed_spotify.csv >> results.txt
echo "" >> results.txt

echo "Processed dataset:" >> results.txt
head processed_spotify.csv >> results.txt
echo "" >> results.txt

mkdir -p artifacts
mv shuffled_spotify.csv subset1.csv subset2.csv trimmed_spotify.csv processed_spotify.csv results.txt artifacts/