src/upload_to_hf.py

"""
Upload UD dataset to HuggingFace Hub.
Dataset: undertheseanlp/UDD-v0.1

Loads train/dev/test JSONL splits and uploads as DatasetDict with domain field.

Usage:
    export $(cat .env | xargs) && python upload_to_hf.py
"""

import json
import os
from os.path import expanduser, join

from datasets import Dataset, DatasetDict
from huggingface_hub import HfApi, login


# Sent_id prefix -> domain mapping
DOMAIN_MAP = {
    "vlc-": "legal",
    "uvn-": "news",
    "uvw-": "wikipedia",
    "uvb-f-": "fiction",
    "uvb-n-": "non-fiction",
}


def get_domain(sent_id):
    """Extract domain from sent_id prefix."""
    for prefix, domain in DOMAIN_MAP.items():
        if sent_id.startswith(prefix):
            return domain
    return "unknown"


def load_jsonl(filepath):
    """Load JSONL file and add domain field."""
    data = []
    with open(filepath, "r", encoding="utf-8") as f:
        for line in f:
            row = json.loads(line)
            row["domain"] = get_domain(row.get("sent_id", ""))
            data.append(row)
    return data


def main():
    # Login with token from environment
    token = os.environ.get("HF_TOKEN")
    if token:
        print("Logging in with HF_TOKEN...")
        login(token=token)
    else:
        print("Warning: HF_TOKEN not set. Using cached credentials.")

    source_folder = expanduser("~/Downloads/UD_Vietnamese-UUD-v0.1")
    readme_file = join(source_folder, "README.md")

    # Load all splits
    splits = {}
    for split_name, filename in [("train", "train.jsonl"), ("validation", "dev.jsonl"), ("test", "test.jsonl")]:
        filepath = join(source_folder, filename)
        if os.path.isfile(filepath):
            print(f"Loading {split_name} from {filepath}...")
            data = load_jsonl(filepath)
            splits[split_name] = Dataset.from_list(data)
            print(f"  {split_name}: {len(data)} sentences")
        else:
            print(f"Warning: {filepath} not found, skipping {split_name} split")

    if not splits:
        print("Error: No data files found!")
        return

    # Create DatasetDict
    print("\nCreating HuggingFace DatasetDict...")
    dataset_dict = DatasetDict(splits)

    print(f"Dataset: {dataset_dict}")
    for split_name, ds in dataset_dict.items():
        print(f"  {split_name}: {len(ds)} rows, features: {list(ds.features.keys())}")

    # Print domain distribution
    for split_name, ds in dataset_dict.items():
        domains = {}
        for row in ds:
            d = row["domain"]
            domains[d] = domains.get(d, 0) + 1
        domain_str = ", ".join(f"{d}: {c}" for d, c in sorted(domains.items()))
        print(f"  {split_name} domains: {domain_str}")

    # Push to HuggingFace Hub
    repo_id = "undertheseanlp/UDD-v0.1"
    print(f"\nPushing to HuggingFace Hub: {repo_id}")

    dataset_dict.push_to_hub(
        repo_id,
        private=False,
        commit_message="Update: 40K sentences from 5 domains (legal, news, wikipedia, fiction, non-fiction)"
    )

    # Upload README.md
    if os.path.isfile(readme_file):
        print("Uploading README.md...")
        api = HfApi()
        api.upload_file(
            path_or_fileobj=readme_file,
            path_in_repo="README.md",
            repo_id=repo_id,
            repo_type="dataset",
            commit_message="Update README with dataset card"
        )

    print(f"\nDone! Dataset available at: https://huggingface.co/datasets/{repo_id}")


if __name__ == "__main__":
    main()