app/audit.py

"""Audit row builder + fail-soft HF Dataset append."""
from __future__ import annotations
import json, os, threading


def build_row(*, session, text, read, mood_before, mood_after, deltas,
              trace, mood_word, ts, source="space:chat") -> dict:
    return {
        "ts": ts, "session": session, "text": text, "source": source,
        "read": list(read), "mood_before": list(mood_before),
        "mood_after": list(mood_after), "deltas": list(deltas),
        "mood_word": mood_word,
        "structures": trace.get("structures", []),
        "unknown_tokens": trace.get("unknown_tokens", []),
        "suspected_gap": trace.get("suspected_gap", False),
        "words": trace.get("words", []),
    }


class AuditLog:
    """Buffers rows to a local JSONL and flushes to a private HF Dataset.
    Fail-soft: any logging error is swallowed (never breaks /say)."""
    def __init__(self, repo_id: str, local_dir: str = "./data", flush_every: int = 25):
        self._repo = repo_id
        self._path = os.path.join(local_dir, "audit.jsonl")
        os.makedirs(local_dir, exist_ok=True)
        self._buf: list[dict] = []
        self._flush_every = flush_every
        self._lock = threading.Lock()

    def append(self, row: dict) -> None:
        try:
            with self._lock:
                self._buf.append(row)
                with open(self._path, "a") as f:
                    f.write(json.dumps(row) + "\n")
                if len(self._buf) >= self._flush_every:
                    rows, self._buf = self._buf, []
                    self._commit(rows)
        except Exception:
            pass  # fail-soft

    def pending(self) -> int:
        return len(self._buf)

    def _commit(self, rows: list[dict]) -> None:
        # Upload the cumulative JSONL to the HF Dataset. Requires HF_TOKEN.
        token = os.environ.get("HF_TOKEN")
        if not token:
            return
        from huggingface_hub import HfApi
        HfApi().upload_file(
            path_or_fileobj=self._path, path_in_repo="audit.jsonl",
            repo_id=self._repo, repo_type="dataset", token=token,
        )