#!/bin/bash

# Data set name: Chess games from the Lichess Data Base for standard rated games
# in November 2023
data=lichess_db_standard_rated_2023-11

# Minimum "ELO" rating of black and white players
min_rating=2000
min_ply=20

# Check if file exists and download iff not
if [ -f "${data}.fen" ]; then
    echo "File '${data}.fen' already exists, assuming job already done."
    echo "To rerun delete (rename) the files '${data}.pgn.zst' and/or '${data}.fen'"
else
    # First, compile `png2fen`
    make pgn2fen

    # Download the PGN data base via `wegt` if not found.
    #   The flag `-q` suppresses `wget`s own output and `-O-` tells `wget` to
    #   stream the downloaded file to `stdout`.
    # Otherwise, use the file on disk directly.
    # Decompress the stream with `zstdcat` (no temporary files)
    # The uncompressed PGN data is then piped into `pgn2fen` which converts
    #   the PGN data base into a list of FEN strings while filtering only
    #   positions with evaluation. The `--scored` parameter specifies to extract
    #   a position evaluation from the PGN and ONLY write positions with scores.
    #   That is, positions without a score are removed! Parameter `--rating`
    #   filters games where both players have at least the minimum rating, and
    #   `--ply` specifies to only consider positions from `$min_ply` onward.
    if [ -f "${data}.pgn.zst" ]; then
        zstdcat ${data}.pgn.zst \
            | ./pgn2fen --scored --rating $min_rating --ply $min_ply \
            > ${data}.fen
    else
        wget -qO- https://database.lichess.org/standard/${data}.pgn.zst | zstdcat \
            | ./pgn2fen --scored --rating $min_rating --ply $min_ply \
            > ${data}.fen
    fi
fi