hugging-science/request-moderator

1"""Submissions queue backed by the existing hugging-science/feedback HF Dataset.2 3`requests.jsonl` holds every non-"feedback" type (dataset/model/organization/4blog/challenge/collaboration); `feedback.jsonl` holds "feedback". Both are5written by the feedback-api Space's /submit endpoint. Ingestion happens6there, not here — this module only reads those files and updates a row's7status when the reviewer approves/rejects/acknowledges it.8"""9 10import json11import threading12from datetime import datetime, timezone13 14from huggingface_hub import HfApi15from huggingface_hub.utils import EntryNotFoundError16 17from about import DATASET_REPO, FEEDBACK_FILE, REQUESTS_FILE, TOKEN18 19_api = HfApi(token=TOKEN)20_lock = threading.Lock()21 22ALL_FILES = (REQUESTS_FILE, FEEDBACK_FILE)23 24 25def _now() -> str:26    return datetime.now(timezone.utc).isoformat()27 28 29def _read_file(filename: str) -> list[dict]:30    try:31        path = _api.hf_hub_download(32            repo_id=DATASET_REPO, repo_type="dataset", filename=filename33        )34    except EntryNotFoundError:35        return []36    with open(path, encoding="utf-8") as f:37        return [json.loads(line) for line in f if line.strip()]38 39 40def _write_file(filename: str, rows: list[dict]) -> None:41    content = "\n".join(json.dumps(row) for row in rows) + "\n"42    _api.upload_file(43        path_or_fileobj=content.encode("utf-8"),44        path_in_repo=filename,45        repo_id=DATASET_REPO,46        repo_type="dataset",47        commit_message="Update requests queue",48    )49 50 51def _read_all() -> list[dict]:52    return [row for filename in ALL_FILES for row in _read_file(filename)]53 54 55def list_submissions(status: str | None = None, type_: str | None = None) -> list[dict]:56    rows = _read_all()57    if status is not None:58        rows = [r for r in rows if r["status"] == status]59    if type_ is not None:60        rows = [r for r in rows if r["type"] == type_]61    return rows62 63 64def get_submission(submission_id: str) -> dict | None:65    for row in _read_all():66        if row["id"] == submission_id:67            return row68    return None69 70 71def _update_row(submission_id: str, mutate_fn) -> None:72    """Find `submission_id` in whichever file its type lives in, mutate it in73    place, and rewrite only that file."""74    with _lock:75        for filename in ALL_FILES:76            rows = _read_file(filename)77            for row in rows:78                if row["id"] == submission_id:79                    mutate_fn(row)80                    _write_file(filename, rows)81                    return82 83 84def mark_approved(submission_id: str, pr_url: str) -> None:85    def mutate(row):86        row["status"] = "approved"87        row["reviewed_at"] = _now()88        row["pr_url"] = pr_url89 90    _update_row(submission_id, mutate)91 92 93def mark_acknowledged(submission_id: str) -> None:94    """For non-PR types (feedback, collaboration, challenge): close out a95    submission without opening a PR — e.g. once it's been read or the96    requester has been emailed."""97 98    def mutate(row):99        row["status"] = "acknowledged"100        row["reviewed_at"] = _now()101 102    _update_row(submission_id, mutate)103 104 105def mark_rejected(submission_id: str, reason: str | None) -> None:106    def mutate(row):107        row["status"] = "rejected"108        row["reviewed_at"] = _now()109        row["reject_reason"] = reason110 111    _update_row(submission_id, mutate)112