system-backu/backend/app/core/executor.py

"""备份执行编排：解密配置 → 流式产出 → 流式上传 → 应用保留策略 → 写日志。"""
import asyncio
import hashlib
import io
import logging
import os
import tempfile
import time
from datetime import datetime
from pathlib import Path
from typing import Optional

from sqlalchemy.orm import Session

from app.core.backup.base import BaseBackup
from app.core.backup.directory import DirectoryBackup
from app.core.backup.influxdb import InfluxDBBackup
from app.core.backup.mysql import MySQLBackup
from app.core.retention import apply_retention
from app.core.storage.factory import create_storage
from app.database import SessionLocal
from app.models.job import Job
from app.models.run import Run, RunLog
from app.utils.crypto import decrypt_dict
from app.utils.logging import get_logger

log = get_logger(__name__)


def _make_backup(job: Job) -> BaseBackup:
    src = decrypt_dict(job.source_config_json)
    if job.type == "mysql":
        return MySQLBackup(src, job.name)
    if job.type == "directory":
        return DirectoryBackup(src, job.name)
    if job.type == "influxdb":
        return InfluxDBBackup(src, job.name)
    raise ValueError(f"Unsupported job type: {job.type}")


async def run_backup(run_id: int, trigger: str = "manual") -> None:
    """执行一次备份。所有副作用都在自己 new 的 Session 上完成。"""
    started = datetime.utcnow()
    db: Session = SessionLocal()
    try:
        run: Run | None = db.get(Run, run_id)
        if not run:
            log.error("run_id=%s not found", run_id)
            return

        job: Job | None = db.get(Job, run.job_id)
        if not job:
            run.status = "failed"
            run.error_message = f"Job id={run.job_id} disappeared"
            run.finished_at = datetime.utcnow()
            db.commit()
            return

        log_buffer = io.StringIO()
        tee = _TeeLogger(log, log_buffer)

        run.status = "running"
        run.trigger = trigger
        run.started_at = started
        db.commit()

        tee.info("Backup started: run_id=%s job=%s trigger=%s", run_id, job.name, trigger)

        try:
            backup = _make_backup(job)
            meta = backup.metadata()
            tee.info("Metadata: filename=%s content_type=%s", meta.suggested_filename, meta.content_type)

            storage = create_storage(job.storage)
            ok, msg = await storage.test_connection()
            tee.info("Storage test: ok=%s msg=%s", ok, msg)
            if not ok:
                raise RuntimeError(f"Storage not reachable: {msg}")

            # 关键路径：备份文件名以 {job_name}/{YYYY-MM-DD}/{filename} 组织
            date_prefix = datetime.utcnow().strftime("%Y-%m-%d")
            object_key = f"{job.name}/{date_prefix}/{meta.suggested_filename}"

            tee.info("Streaming backup to storage key=%s ...", object_key)

            async def _consume_with_progress():
                async for chunk in backup.produce():
                    yield chunk

            t0 = time.monotonic()
            size, sha = await storage.upload_stream(
                _consume_with_progress(),
                object_key,
                content_type=meta.content_type,
            )
            elapsed = int(time.monotonic() - t0)

            run.artifact_path = object_key
            run.artifact_size = size
            run.artifact_checksum = sha
            run.duration_seconds = elapsed
            run.status = "success"
            run.finished_at = datetime.utcnow()
            tee.info(
                "Backup success: size=%s bytes sha256=%s elapsed=%ss",
                size, sha[:12], elapsed,
            )

            # 应用保留策略
            try:
                removed = apply_retention(db, job)
                if removed:
                    tee.info("Retention cleanup removed %s old runs", removed)
            except Exception as e:
                tee.warning("Retention cleanup failed: %s", e)

        except Exception as e:
            log.exception("Backup run_id=%s failed", run_id)
            tee.error("Backup failed: %s: %s", type(e).__name__, e)
            run.status = "failed"
            run.error_message = f"{type(e).__name__}: {e}"[:2000]
            run.finished_at = datetime.utcnow()
        finally:
            log_text = log_buffer.getvalue()
            run.log_excerpt = log_text[-4000:] if len(log_text) > 4000 else log_text
            run_log = RunLog(run_id=run.id, content=log_text)
            db.add(run_log)
            db.commit()
            log.info("Backup run_id=%s finished status=%s", run_id, run.status)
    finally:
        db.close()


class _TeeLogger:
    """同时写标准 logging 和内存 StringIO（用于存数据库）。"""

    def __init__(self, real: logging.Logger, buf: io.StringIO):
        self._real = real
        self._buf = buf

    def _emit(self, level: int, msg: str) -> None:
        ts = datetime.utcnow().strftime("%H:%M:%S")
        line = f"[{ts}] {msg}"
        self._real.log(level, line)
        self._buf.write(line + "\n")

    def info(self, msg: str, *args) -> None:
        self._emit(logging.INFO, msg % args if args else msg)

    def warning(self, msg: str, *args) -> None:
        self._emit(logging.WARNING, msg % args if args else msg)

    def error(self, msg: str, *args) -> None:
        self._emit(logging.ERROR, msg % args if args else msg)