data/memo/create.py

# /// script
# requires-python = "==3.12"
# dependencies = [
#     "datasets==3.2.0",
# ]
# ///
from datetime import datetime, timedelta
from pathlib import Path
from typing import cast

from datasets import Dataset, load_dataset

column_order = [
    "text",
    "source",
    "id",
    "added",
    "created",
    "license",
    "domain",
    "metadata",
]


def convert_sample(example: dict) -> dict:
    # from sample:
    # {
    #     "filename": "1894_Aagaard_UnderligeFyre",
    #     "full_firstnames": "Oscar",
    #     "auth_first": "Oscar",
    #     "auth_last_modern": "Aagaard",
    #     "pseudonym": None,
    #     "publ_date": 1894,
    #     "title_modern": "Underlige Fyre",
    #     "published_under_gender": "male",
    #     "real_gender": "male",
    #     "nationality": "no",
    #     "auth_id": 1,
    #     "auth_last": "Aagaard",
    #     "title": "Underlige Fyre",
    #     "surname": "Aagaard",
    #     "title.1": "Underlige Fyre",
    #     "subtitle": "Fortælling",
    #     "volume": None,
    #     "year": 1894,
    #     "pages": 263.0,
    #     "illustrations": "n",
    #     "typeface": "roman",
    #     "publisher": "Gyldendal",
    #     "price": 3.0,
    #     "source": "KB",
    #     "notes": None,
    #     "filepath": None,
    #     "historical": None,
    #     "period": "nan",
    #     "period_notes": "nan",
    #     "novel_start": 13.0,
    #     "novel_end": 275.0,
    #     "serialno": 854.0,
    #     "category": "O",
    #     "e_canon": 0,
    #     "ce_canon": 0,
    #     "lex_canon": 0,
    #     "text": "Første kapitel. Argus & co. Waterclerker — hvormange er der vel, ...",
    # }

    min_date = datetime.fromisoformat(f"{example["year"]}-01-01")
    max_date = datetime.fromisoformat(f"{example["year"]}-12-31")
    text = f"{example['title_modern']}\n\nSkrevet af {example['full_firstnames']} {example['auth_last_modern']}\nPubliceret {example['year']} af {example['publisher']}\n\n{example['text']}"

    new_example = dict(
        text_new=text,
        id=example["filename"],
        source="memo",
        domain="Wiki & Books",
        license="cc-by-sa-4.0",
        added="2025-03-08",
        created=f"{min_date.date()}, {max_date.date()}",
        metadata={"source-pretty": "MeMo Canonical Novels"},
    )

    return new_example


def main():
    ds = load_dataset("chcaa/memo-canonical-novels", split="train")
    ds = cast(Dataset, ds)
    dates = [datetime.fromisoformat(f"{date}-01-01").date() for date in ds["year"]]

    max_date = max(dates) + timedelta(days=364)
    print(str(min(dates)), ",", str(max_date))  # 1870-01-01 , 1899-12-31

    assert len(set(ds["filename"])) == len(ds), "IDs are not unique"
    assert len(set(ds["text"])) == len(ds), "Texts are not unique"

    ds = ds.map(convert_sample, num_proc=4)
    ds = ds.select_columns(column_order[1:] + ["text_new"])
    ds = ds.rename_columns({"text_new": "text"})
    # ensure order
    ds = ds.select_columns(column_order)

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


if __name__ == "__main__":
    main()