operating-room-monitor-server/app/services/voice_file_log.py

"""语音确认（ASR/解析/审计）的终端 loguru 行 + 每手术 TSV 落盘，与 `consumption_tsv_log` 并列。"""

from __future__ import annotations

import re
import threading
from datetime import datetime, timezone
from pathlib import Path
from zoneinfo import ZoneInfo, ZoneInfoNotFoundError

from loguru import logger

from app.config import Settings

_lock = threading.Lock()

HEADER = (
    "时间戳(ISO,UTC)\t来源\t状态\tconfirmation_id\tasr/识别文本\t"
    "resolved_label\trejected\terror\taudio_object_key\n"
)


def _ts_iso_utc() -> str:
    return datetime.now(timezone.utc).isoformat(timespec="milliseconds")


def _encode_cell(value: str) -> str:
    return (value or "").replace("\r", " ").replace("\n", " ").replace("\t", " ")


def _log_tz_info(settings: Settings) -> object:
    raw = (settings.consumption_log_timezone or "").strip()
    if not raw:
        lt = datetime.now().astimezone().tzinfo
        return lt if lt is not None else timezone.utc
    try:
        return ZoneInfo(raw)
    except ZoneInfoNotFoundError:
        return timezone.utc


def _ts_local_for_display(settings: Settings) -> str:
    tz = _log_tz_info(settings)
    return datetime.now(tz).isoformat(timespec="milliseconds")


def _safe_surgery_path_segment(surgery_id: str) -> str:
    s = (surgery_id or "unknown").strip() or "unknown"
    s = re.sub(r"[^\w\-.@]", "_", s)
    return s[:200] if len(s) > 200 else s


def resolved_voice_log_path(surgery_id: str, settings: Settings) -> Path:
    raw = (settings.voice_file_log_path or "logs/voice_{surgery_id}.txt").strip()
    safe = _safe_surgery_path_segment(surgery_id)
    if "{surgery_id}" in raw:
        raw = raw.replace("{surgery_id}", safe)
    else:
        p0 = Path(raw)
        if p0.suffix:
            raw = str(p0.with_name(f"{p0.stem}_{safe}{p0.suffix}"))
        else:
            raw = f"{raw.rstrip('/')}_{safe}.txt"
    p = Path(raw).expanduser()
    if not p.is_absolute():
        p = Path.cwd() / p
    return p


def init_voice_log_file(surgery_id: str, settings: Settings) -> None:
    """与 `init_consumption_log_file` 同生命周期：`start_surgery` 时截断并写表头。"""
    if not settings.voice_file_log_enabled:
        return
    path = resolved_voice_log_path(surgery_id, settings)
    path.parent.mkdir(parents=True, exist_ok=True)
    with _lock:
        with path.open("w", encoding="utf-8") as f:
            f.write(HEADER)


def append_voice_tsv_line(surgery_id: str, line: str, settings: Settings) -> None:
    if not settings.voice_file_log_enabled:
        return
    path = resolved_voice_log_path(surgery_id, settings)
    path.parent.mkdir(parents=True, exist_ok=True)
    with _lock:
        with path.open("a", encoding="utf-8") as f:
            f.write(line)


def emit_voice_event(
    settings: Settings,
    *,
    surgery_id: str,
    source: str,
    status: str,
    confirmation_id: str,
    asr_text: str | None = None,
    resolved_label: str | None = None,
    rejected: str | bool | None = None,
    error_message: str | None = None,
    audio_object_key: str | None = None,
) -> None:
    """
    终端：单条可 grep 的 VoiceConfirm 行；文件：TSV 一行（与启用的 `voice_file_log_enabled` 一致）。

    :param source: `wav` | `text` | `n/a`
    :param status: 与审计 `status` 或 `minio_not_configured` 等说明型状态一致
    """
    rj: str
    if rejected is None:
        rj = ""
    elif isinstance(rejected, bool):
        rj = "true" if rejected else "false"
    else:
        rj = str(rejected)

    ts_utc = _ts_iso_utc()
    local_hint = _ts_local_for_display(settings)
    if status in ("recognized", "rejected"):
        logger.info(
            "VoiceConfirm local_ts={!r} surgery_id={} source={} status={} "
            "confirmation_id={} asr_text={!r} resolved_label={!r} rejected={} "
            "error={!r} audio_key={!r}",
            local_hint,
            surgery_id,
            source,
            status,
            confirmation_id,
            asr_text,
            resolved_label,
            rj,
            error_message,
            audio_object_key,
        )
    else:
        logger.warning(
            "VoiceConfirm local_ts={!r} surgery_id={} source={} status={} "
            "confirmation_id={} asr_text={!r} resolved_label={!r} rejected={} "
            "error={!r} audio_key={!r}",
            local_hint,
            surgery_id,
            source,
            status,
            confirmation_id,
            asr_text,
            resolved_label,
            rj,
            error_message,
            audio_object_key,
        )

    if not settings.voice_file_log_enabled:
        return
    row = [
        _encode_cell(ts_utc),
        _encode_cell(source),
        _encode_cell(status),
        _encode_cell(confirmation_id),
        _encode_cell("" if asr_text is None else asr_text),
        _encode_cell("" if resolved_label is None else resolved_label),
        _encode_cell(rj),
        _encode_cell("" if error_message is None else error_message),
        _encode_cell("" if audio_object_key is None else audio_object_key),
    ]
    line = "\t".join(row) + "\n"
    append_voice_tsv_line(surgery_id, line, settings)