data/nordjyllandnews/create.py

"""
This scripts download nordjylland news and converts it to the format of danish dynaword
"""

import random
from pathlib import Path
from typing import cast

from datasets import Dataset, load_dataset

schemas = [
    "{summary}\n\n{text}",
    "{text}\n\nOpsummering:\n{summary}",
    "{text}\n\nReferat:\n{summary}",
    "Lav et referat af nedenstående tekst:\n\nTekst:\n{text}\n\nReferat:\n{summary}",
]
source = "nordjyllandnews"


def convert_sample(example):
    schema = random.choice(schemas)
    new_example = dict(
        text_new=schema.format(text=example["text"], summary=example["summary"]),
        source=source,
        domain="News",
        license="Creative Commons Legal Code\n\nCC0 1.0 Universal",
        added="2024-12-16",
        created="2000-01-01, 2024-01-01",  # best guess
        metadata={"source-pretty": "Nordjylland News"},
    )

    return new_example


def main():
    ds = load_dataset("alexandrainst/nordjylland-news-summarization", split="train")
    ds = cast(Dataset, ds)

    ds = ds.map(convert_sample, remove_columns=ds.column_names)
    ds = ds.rename_columns({"text_new": "text"})
    ds = ds.add_column("id", [f"{source}_{i}" for i in range(len(ds))])  # type: ignore
    ds = ds.select_columns(
        ["text", "source", "id", "added", "created", "license", "domain", "metadata"]
    )

    save_path = Path(__file__).parent / f"{source}.parquet"
    ds.to_parquet(save_path)


if __name__ == "__main__":
    main()