Python 运维自动化系列（03）：磁盘清理的正确姿势——先 dry-run 再动手

from pathlib import Path

ROOT = Path(__file__).parent.parent


def dir_size(path: Path) -> int:
    return sum(f.stat().st_size for f in path.rglob("*") if f.is_file())


if __name__ == "__main__":
    entries = []
    for child in sorted(ROOT.iterdir()):
        if child.is_dir():
            entries.append((child.name + "/", dir_size(child)))
        else:
            entries.append((child.name, child.stat().st_size))

    total = sum(size for _, size in entries)
    print(f"项目根目录: {ROOT.name}  总计 {total / 1024**2:.1f} MB\n")
    for name, size in sorted(entries, key=lambda x: -x[1]):
        if size == 0:
            continue
        bar = "█" * max(1, round(size / total * 30))
        print(f"  {name:<26} {size / 1024**2:8.2f} MB  {bar}")

项目根目录: 新项目 (3)  总计 53.5 MB

  demos/                        53.46 MB  ██████████████████████████████
  logs/                          0.02 MB  █
  python-ops-monitoring-tutorial.md     0.02 MB  █
  python-ops-series-00-index.md     0.01 MB  █
  monitor.py                     0.01 MB  █
  .zwork/                        0.00 MB  █
  tests/                         0.00 MB  █
  .pytest_cache/                 0.00 MB  █
  pytest.ini                     0.00 MB  █

┌──────────┐      ┌─────────────┐      ┌──────────┐
│ 扫描器    │ ───→ │ dry-run 报告 │ ───→ │ 删除器    │
│ 只读不删  │      │ 人先看一眼   │      │ 只删报告里的│
└──────────┘      └─────────────┘      └──────────┘

import os
import time
from datetime import datetime
from pathlib import Path

SANDBOX = Path(__file__).parent / "fixtures" / "junkyard"
NOW = time.time()
DAYS_30 = 30 * 86400
JUNK_SUFFIXES = {".tmp", ".bak", ".old"}


def make_sandbox() -> None:
    """造一个沙箱：正常文件 + 过期垃圾混在一起。"""
    SANDBOX.mkdir(parents=True, exist_ok=True)
    files = [
        ("report_final.docx", 120_000, NOW - 3 * 86400),      # 3 天前，正常文件
        ("invoice_2026Q3.pdf", 300_000, NOW - 10 * 86400),    # 10 天前，正常文件
        ("export_001.tmp", 8 * 1024**2, NOW - 90 * 86400),    # 90 天前的垃圾
        ("export_002.tmp", 5 * 1024**2, NOW - 45 * 86400),    # 45 天前的垃圾
        ("db_snapshot.bak", 40 * 1024**2, NOW - 60 * 86400),  # 60 天前的垃圾
        ("config.yaml.old", 4_000, NOW - 120 * 86400),        # 120 天前的垃圾
        ("fresh_download.tmp", 2_000, NOW - 3600),            # 1 小时前，不算垃圾
    ]
    for name, size, mtime in files:
        p = SANDBOX / name
        with p.open("wb") as f:
            f.write(b"\0" * size)
        os.utime(p, (mtime, mtime))   # 把修改时间拨回过去


def scan(sandbox: Path, older_than_days: int = 30) -> list[Path]:
    """返回「命中清理规则」的文件列表。只扫描，不删除。"""
    cutoff = NOW - older_than_days * 86400
    hits = []
    for p in sandbox.rglob("*"):
        if p.is_file() and p.suffix.lower() in JUNK_SUFFIXES and p.stat().st_mtime < cutoff:
            hits.append(p)
    return hits


if __name__ == "__main__":
    make_sandbox()
    print(f"沙箱目录: {SANDBOX}")
    print(f"目录总占用: {sum(p.stat().st_size for p in SANDBOX.rglob('*') if p.is_file()) / 1024**2:.1f} MB\n")

    hits = scan(SANDBOX)
    print(f"== dry-run 清理报告（生成于 {datetime.now():%Y-%m-%d %H:%M:%S}，未删除任何文件）==")
    reclaim = 0
    for p in sorted(hits, key=lambda x: -x.stat().st_size):
        size = p.stat().st_size
        age_days = (NOW - p.stat().st_mtime) / 86400
        reclaim += size
        print(f"  [将删除] {p.name:<22} {size / 1024**2:6.1f} MB   {age_days:4.0f} 天未访问")
    print(f"\n共 {len(hits)} 个文件命中，可释放 {reclaim / 1024**2:.1f} MB")

    keep = [p.name for p in SANDBOX.rglob("*") if p.is_file() and p not in hits]
    print(f"以下文件未命中规则，保持不动: {keep}")

沙箱目录: demos\fixtures\junkyard
目录总占用: 53.4 MB

== dry-run 清理报告（生成于 2026-09-28 20:55:51，未删除任何文件）==
  [将删除] db_snapshot.bak          40.0 MB     60 天未访问
  [将删除] export_001.tmp            8.0 MB     90 天未访问
  [将删除] export_002.tmp            5.0 MB     45 天未访问
  [将删除] config.yaml.old           0.0 MB    120 天未访问

共 4 个文件命中，可释放 53.0 MB
以下文件未命中规则，保持不动: ['fresh_download.tmp', 'invoice_2026Q3.pdf', 'report_final.docx']

def delete(hits: list[Path]) -> None:
    for p in hits:
        p.unlink()