diff --git a/.gitignore b/.gitignore index 96affcd..76031fe 100644 --- a/.gitignore +++ b/.gitignore @@ -10,6 +10,9 @@ tmp/ # 检测缓存（断点续跑数据，含运行记录） runs/ +# 运行日志 +logs/ + # 检测输出（生成物） output/ 检测*/ diff --git a/README.md b/README.md index f654a3c..63b86a2 100644 --- a/README.md +++ b/README.md @@ -94,6 +94,20 @@ max_tokens = 5000 └── legacy/ # 早期脚本与实验数据 ``` +## 日志与排错 + +所有运行过程写入日志文件（exe 同目录 / 源码项目根目录下的 `logs/violation_detector_日期.log`）： + +- 文件记录 DEBUG 级全量：运行配置、缓存命中、每次请求的重试与 HTTP 响应体、异常完整堆栈； +- 控制台/GUI 只显示 INFO 级简洁信息，异常时同样附带完整 traceback； +- GUI 右下角「打开日志」按钮直达日志文件；CLI 出错时会在结尾打印日志文件路径。 + +```bash +# 跑测试（33 个单元测试，无需网络与 API Key） +uv run --with pytest --with aiohttp --with openpyxl --no-project \ + env PYTHONPATH=src pytest tests -q +``` + ## 检测规则与已知特性 - 分类按提示词 17 类标准输出唯一分类，自动归一化写法差异。 diff --git a/src/violation_detector/cli.py b/src/violation_detector/cli.py index 0e65fa5..70ff1bd 100644 --- a/src/violation_detector/cli.py +++ b/src/violation_detector/cli.py @@ -2,13 +2,17 @@ """CLI 入口：python -m violation_detector <图片文件夹> [选项]""" import argparse import asyncio +import logging import sys from . import __version__ from .config import DEFAULT_CONFIG, load_config, missing_fields +from .logutil import log_file, setup_logging from .pipeline import run_detection from .report import build_report, organize_output +logger = logging.getLogger("violation_detector.cli") + def build_parser() -> argparse.ArgumentParser: p = argparse.ArgumentParser( @@ -32,6 +36,8 @@ def build_parser() -> argparse.ArgumentParser: def main(argv=None) -> int: args = build_parser().parse_args(argv) + log_path = setup_logging(console=True) + logger.info("商品图合规检测工具 v%s 启动，日志文件：%s", __version__, log_path) cfg = load_config(args.config) if args.prompt: cfg.prompt_file = args.prompt @@ -53,13 +59,25 @@ def main(argv=None) -> int: rows, summary = asyncio.run(run_detection( args.folder, cfg, mode=args.mode, verify=not args.no_verify)) except FileNotFoundError as e: - print(f"错误：{e}") + logger.error("路径或文件不存在：%s", e) + return 1 + except KeyboardInterrupt: + logger.warning("用户中断") + return 130 + except Exception: # noqa: BLE001 顶层兜底：完整堆栈进日志与控制台 + logger.exception("检测过程发生未预期异常") + print(f"\n检测失败，完整日志：{log_file()}") return 1 - out_dir = args.output or args.folder - ts_dir = organize_output(rows, args.folder, out_dir) - xlsx = build_report(rows, str(ts_dir), {"mode": args.mode, **{ - k: summary.get(k) for k in ("ds_calls", "ds_peak_cost", "ds_idle_cost")}}) + try: + out_dir = args.output or args.folder + ts_dir = organize_output(rows, args.folder, out_dir) + xlsx = build_report(rows, str(ts_dir), {"mode": args.mode, **{ + k: summary.get(k) for k in ("ds_calls", "ds_peak_cost", "ds_idle_cost")}}) + except Exception: # noqa: BLE001 + logger.exception("输出组织/报表生成失败") + print(f"\n报表生成失败，完整日志：{log_file()}") + return 1 print(f"\n分类输出目录：{ts_dir}") print(f"报表已生成：{xlsx}") return 0 diff --git a/src/violation_detector/gui.py b/src/violation_detector/gui.py index b58e5ba..3ac0cfb 100644 --- a/src/violation_detector/gui.py +++ b/src/violation_detector/gui.py @@ -6,6 +6,7 @@ 检测在后台线程运行 asyncio 事件循环；日志与进度经线程安全队列由主线程刷新。 """ import asyncio +import logging import os import queue import threading @@ -14,9 +15,12 @@ from tkinter import filedialog, messagebox, scrolledtext, ttk from . import __version__ from .config import load_config, missing_fields, save_config +from .logutil import QueueHandler, log_file, setup_logging from .pipeline import run_detection from .report import build_report, organize_output +logger = logging.getLogger("violation_detector") + class FirstRunDialog: """首次配置对话框：补齐级联模式必填的四项模型配置并写入 config.ini。""" @@ -68,6 +72,8 @@ class App: root.geometry("880x640") self.cfg = load_config() self.q: queue.Queue = queue.Queue() + setup_logging(console=False) + logger.addHandler(QueueHandler(self.q)) self.stop_event = threading.Event() self.worker: threading.Thread | None = None self.last_xlsx = "" @@ -110,6 +116,7 @@ class App: self.btn_open.pack(side="left", padx=8) self.btn_folder = ttk.Button(ctrl, text="打开输出目录", command=self.open_outdir, state="disabled") self.btn_folder.pack(side="left", padx=8) + ttk.Button(ctrl, text="打开日志", command=self.open_log).pack(side="right") self.progress = ttk.Progressbar(frm, mode="determinate") self.progress.pack(fill="x", pady=(6, 2)) @@ -162,17 +169,23 @@ class App: xlsx, outdir, summary = payload self.btn_run.configure(state="normal") self.btn_stop.configure(state="disabled") - if xlsx: - self.btn_open.configure(state="normal") - self.btn_folder.configure(state="normal") - self.last_xlsx, self.last_outdir = xlsx, outdir + if not xlsx: + messagebox.showerror( + "检测失败", + f"检测未能完成，错误详情已在上方日志窗口显示，\n" + f"完整堆栈已写入日志文件：\n{log_file()}\n\n" + f"（也可点\"打开日志\"查看）") + continue + self.btn_open.configure(state="normal") + self.btn_folder.configure(state="normal") + self.last_xlsx, self.last_outdir = xlsx, outdir counts = summary.get("counts", {}) top = "\n".join(f" {c}：{n} 张" for c, n in counts.items()) or " （无）" messagebox.showinfo( "检测完成", f"图片 {summary.get('total', 0)} 张已处理，已按分类归档。\n\n分类分布：\n{top}\n\n" f"DeepSeek 成本≈¥{summary.get('ds_peak_cost', 0):.2f}（高峰）/ " - f"¥{summary.get('ds_idle_cost', 0):.2f}（空闲）\n输出目录：{outdir or '未生成'}") + f"¥{summary.get('ds_idle_cost', 0):.2f}（空闲）\n输出目录：{outdir}") except queue.Empty: pass self.root.after(100, self.poll_queue) @@ -186,6 +199,11 @@ class App: if d and os.path.isdir(d): os.startfile(d) # noqa: S606 + def open_log(self): + lf = log_file() + if lf and os.path.isfile(lf): + os.startfile(lf) # noqa: S606 + # ---------- 检测控制 ---------- def start(self): folder = self.var_folder.get().strip() @@ -217,12 +235,9 @@ class App: self.btn_open.configure(state="disabled") self.btn_folder.configure(state="disabled") self.stop_event.clear() - self._ui_log(f"===== 开始检测：{folder} =====") + logger.info("===== 开始检测：%s（模式 cascade，输出：%s）=====", folder, out_dir) self.var_progress_label.set("检测中…") - def log(msg): - self.q.put(("log", str(msg))) - def progress(done, total, _label=None): self.q.put(("progress", done, total)) @@ -232,19 +247,19 @@ class App: xlsx, outdir, summary = "", "", {} try: rows, summary = loop.run_until_complete(run_detection( - folder, cfg, log=log, progress=progress, + folder, cfg, progress=progress, stop=self.stop_event.is_set, mode="cascade", verify=True)) - self.q.put(("log", "正在整理输出（分类归档 + 生成报表）…")) + logger.info("正在整理输出（分类归档 + 生成报表）…") ts_dir = organize_output(rows, folder, out_dir) xlsx = build_report(rows, str(ts_dir), {"mode": "cascade", **{k: summary.get(k) for k in ("ds_calls", "ds_peak_cost", "ds_idle_cost")}}) outdir = str(ts_dir) - log(f"输出目录：{ts_dir}") - log(f"报表：{xlsx}") - except Exception as e: # noqa: BLE001 - log(f"检测失败：{type(e).__name__}: {e}") + logger.info("输出目录：%s", ts_dir) + logger.info("报表：%s", xlsx) + except Exception: # noqa: BLE001 完整堆栈进日志窗口与日志文件 + logger.exception("检测失败") finally: loop.close() self.q.put(("done", xlsx, outdir, summary)) diff --git a/src/violation_detector/logutil.py b/src/violation_detector/logutil.py new file mode 100644 index 0000000..2a0b3d9 --- /dev/null +++ b/src/violation_detector/logutil.py @@ -0,0 +1,68 @@ +# -*- coding: utf-8 -*- +"""统一日志：滚动文件（DEBUG 全量）+ 控制台（INFO 简洁）+ GUI 队列。 + +- 文件记录完整堆栈与调试细节，出错时按日志文件排查； +- 控制台/GUI 只看简洁信息，异常同样会附带完整 traceback。 +""" +import logging +import sys +from datetime import datetime +from pathlib import Path + +from .config import app_dir + +LOGGER_NAME = "violation_detector" +LOG_DIR = app_dir() / "logs" +_log_file: Path | None = None + + +def setup_logging(console: bool = True, log_dir: Path | None = None) -> Path: + """初始化日志（幂等），返回日志文件路径。 + + console=True 时向 stdout 输出 INFO 级简洁信息；文件始终记录 DEBUG 级全量。 + """ + global _log_file + target_dir = Path(log_dir) if log_dir else LOG_DIR + target_dir.mkdir(parents=True, exist_ok=True) + _log_file = target_dir / f"{LOGGER_NAME}_{datetime.now().strftime('%Y%m%d')}.log" + + root = logging.getLogger(LOGGER_NAME) + root.setLevel(logging.DEBUG) + root.handlers.clear() + root.propagate = False + + detail_fmt = logging.Formatter( + "%(asctime)s.%(msecs)03d %(levelname)-7s [%(name)s] %(message)s", + datefmt="%Y-%m-%d %H:%M:%S") + fh = logging.FileHandler(_log_file, encoding="utf-8") + fh.setLevel(logging.DEBUG) + fh.setFormatter(detail_fmt) + root.addHandler(fh) + + if console: + ch = logging.StreamHandler(sys.stdout) + ch.setLevel(logging.INFO) + ch.setFormatter(logging.Formatter("%(message)s")) + root.addHandler(ch) + return _log_file + + +def log_file() -> Path | None: + """当前日志文件路径（未初始化时为 None）。""" + return _log_file + + +class QueueHandler(logging.Handler): + """把日志记录推入线程安全队列（GUI 主线程轮询刷新到日志窗口）。""" + + def __init__(self, q): + super().__init__(level=logging.INFO) + self.q = q + self.setFormatter(logging.Formatter("%(asctime)s %(levelname)s %(message)s", + datefmt="%H:%M:%S")) + + def emit(self, record): + try: + self.q.put(("log", self.format(record))) + except Exception: # noqa: BLE001 日志通道绝不影响业务 + self.handleError(record) diff --git a/src/violation_detector/pipeline.py b/src/violation_detector/pipeline.py index 8abebc3..05458ff 100644 --- a/src/violation_detector/pipeline.py +++ b/src/violation_detector/pipeline.py @@ -7,11 +7,15 @@ mode: cascade 默认，豆包初筛 + DeepSeek 复检 doubao 仅豆包 deepseek 仅 DeepSeek（全量直接进 DeepSeek，含无违规两票复核） + +所有过程与异常通过 logging 输出（文件记录 DEBUG 级全量与完整堆栈）。 """ import asyncio import json +import logging import os import re +import time from datetime import datetime from pathlib import Path @@ -20,6 +24,8 @@ import aiohttp from .config import PRICE, AppConfig, RUNS_DIR, load_prompt from .providers import IMAGE_EXTS, detect_ark, detect_deepseek +logger = logging.getLogger("violation_detector.pipeline") + STATUS_LABEL = {"ok": "", "parse_fail": "模型返回无法解析为标准JSON", "error": "调用失败（网络/服务异常）"} @@ -75,12 +81,15 @@ def _load_cache(path: Path) -> dict: try: return json.loads(path.read_text(encoding="utf-8")) except json.JSONDecodeError: - pass + logger.exception("缓存文件损坏，忽略并重建：%s", path) return {} def _save_cache(path: Path, data: dict): - path.write_text(json.dumps(data, ensure_ascii=False, indent=1), encoding="utf-8") + try: + path.write_text(json.dumps(data, ensure_ascii=False, indent=1), encoding="utf-8") + except OSError: + logger.exception("缓存写入失败（不影响检测结果）：%s", path) def _is_peak_now() -> bool: @@ -109,12 +118,16 @@ def deepseek_cost(results: dict) -> tuple: async def _run_stage(detect, session, sem, cfg, prompt, folder, files, cache_path, - log, progress, stop): + progress, stop): """执行一个检测阶段：跳过缓存中已完成的结果，边跑边落盘。""" cache = _load_cache(cache_path) todo = [f for f in files if f not in cache or cache[f].get("status") != "ok"] total = len(files) done = total - len(todo) + if todo: + logger.info(" 缓存命中 %d/%d，本次需检测 %d 张", done, total, len(todo)) + else: + logger.info(" 全部 %d 张命中缓存，跳过请求", total) if progress: progress(done, total) @@ -125,19 +138,21 @@ async def _run_stage(detect, session, sem, cfg, prompt, folder, files, cache_pat for fut in asyncio.as_completed(tasks): try: res = await fut - except Exception as e: # noqa: BLE001 单任务异常不拖垮整批 + except Exception: # noqa: BLE001 单任务异常不拖垮整批，完整堆栈进日志 + logger.exception("检测任务抛出未捕获异常（按检测异常记录）") res = {"file": "?", "provider": "?", "status": "error", "attr": "", "logic": "", "category": "检测异常", - "raw": f"{type(e).__name__}: {e}", "usage": {}} + "raw": "未捕获异常（详见日志）", "usage": {}} cache[res["file"]] = res done += 1 cat = res["category"] if res["status"] == "ok" else f"[{res['status']}]" - log(f" ({done}/{total}) {res['file']} -> {cat}") + logger.info(" (%d/%d) %s -> %s", done, total, res["file"], cat) if progress: progress(done, total) if done % 5 == 0: _save_cache(cache_path, cache) if stop and stop(): + logger.warning("收到停止请求，取消剩余 %d 个任务", len(tasks) - done) for t in tasks: t.cancel() break @@ -146,7 +161,7 @@ async def _run_stage(detect, session, sem, cfg, prompt, folder, files, cache_pat async def _verify_clean(session, sem, prompt, folder, files, stage2, cfg, - log, progress, stop, cache_path): + progress, stop, cache_path): """对阶段二判为无违规的图片各追加两票独立复核。""" verify = [f for f in files if f in stage2 and stage2[f].get("status") == "ok" @@ -154,7 +169,7 @@ async def _verify_clean(session, sem, prompt, folder, files, stage2, cfg, and "votes" not in stage2[f]] if not verify: return - log(f"复核阶段：{len(verify)} 张 DeepSeek 无违规图片，各追加 2 票") + logger.info("复核阶段：%d 张 DeepSeek 无违规图片，各追加 2 票", len(verify)) async def run_verify(fname): out = [] @@ -168,10 +183,15 @@ async def _verify_clean(session, sem, prompt, folder, files, stage2, cfg, tasks = [asyncio.ensure_future(run_verify(f)) for f in verify] vdone = 0 for fut in asyncio.as_completed(tasks): - fname, extra = await fut + try: + fname, extra = await fut + except Exception: # noqa: BLE001 + logger.exception("复核任务抛出未捕获异常，跳过该图") + continue runs = [stage2[fname]] + [r for r in extra if r.get("status") == "ok"] if len(runs) < 2: stage2[fname]["vote_note"] = "复核调用失败，维持单票结论" + logger.error("复核调用失败：%s", fname) else: final, note = decide_final(runs) final = dict(final) @@ -179,22 +199,26 @@ async def _verify_clean(session, sem, prompt, folder, files, stage2, cfg, final["vote_note"] = note stage2[fname] = final vdone += 1 - log(f" 复核 ({vdone}/{len(verify)}) {fname} -> {stage2[fname]['category']}｜" - f"{stage2[fname].get('vote_note', '')}") + logger.info(" 复核 (%d/%d) %s -> %s｜%s", vdone, len(verify), fname, + stage2[fname]["category"], stage2[fname].get("vote_note", "")) if progress: progress(vdone, len(verify)) _save_cache(cache_path, stage2) return stage2 -async def run_detection(folder, cfg: AppConfig, log=print, progress=None, stop=None, +async def run_detection(folder, cfg: AppConfig, progress=None, stop=None, mode="cascade", verify=True): """执行级联检测，返回 (rows, summary)。rows 已按文件名自然排序。""" + t0 = time.monotonic() prompt = load_prompt(cfg.prompt_path) files = list_images(folder) if not files: raise FileNotFoundError(f"文件夹中没有图片: {folder}") - log(f"共 {len(files)} 张图片，模式：{mode}，提示词：{cfg.prompt_path}") + logger.info("共 %d 张图片，模式：%s，提示词：%s", len(files), mode, cfg.prompt_path) + logger.debug("运行配置：ark=%s workers=%d | deepseek=%s workers=%d max_tokens=%d " + "| recheck=%s", cfg.ark_model, cfg.ark_workers, cfg.deepseek_model, + cfg.deepseek_workers, cfg.max_tokens, cfg.recheck_categories) ark_cache_path = _cache_path("ark", folder) ds_cache_path = _cache_path("deepseek", folder) @@ -206,10 +230,10 @@ async def run_detection(folder, cfg: AppConfig, log=print, progress=None, stop=N # ---- 阶段一：豆包初筛（cascade/doubao 模式）---- stage1 = {} if mode in ("cascade", "doubao"): - log(f"[阶段一] 豆包初筛（并发 {cfg.ark_workers}）") + logger.info("[阶段一] 豆包初筛（并发 %d）", cfg.ark_workers) sem1 = asyncio.Semaphore(cfg.ark_workers) stage1 = await _run_stage(detect_ark, session, sem1, cfg, prompt, folder, - files, ark_cache_path, log, progress, stop) + files, ark_cache_path, progress, stop) # ---- 阶段二：DeepSeek 复检 ---- stage2 = {} @@ -218,25 +242,25 @@ async def run_detection(folder, cfg: AppConfig, log=print, progress=None, stop=N if stage1.get(f, {}).get("category") in cfg.recheck_categories or stage1.get(f, {}).get("status") != "ok"] if recheck: - log(f"[阶段二] DeepSeek 复检 {len(recheck)} 张（初筛为" - f"{'/'.join(cfg.recheck_categories)}，并发 {cfg.deepseek_workers}）") + logger.info("[阶段二] DeepSeek 复检 %d 张（初筛为%s，并发 %d）", + len(recheck), "/".join(cfg.recheck_categories), + cfg.deepseek_workers) sem2 = asyncio.Semaphore(cfg.deepseek_workers) stage2 = await _run_stage(detect_deepseek, session, sem2, cfg, prompt, - folder, recheck, ds_cache_path, log, - progress, stop) + folder, recheck, ds_cache_path, progress, stop) else: - log("[阶段二] 无需复检的图片") + logger.info("[阶段二] 无需复检的图片") elif mode == "deepseek": - log(f"[阶段二] DeepSeek 全量检测（并发 {cfg.deepseek_workers}）") + logger.info("[阶段二] DeepSeek 全量检测（并发 %d）", cfg.deepseek_workers) sem2 = asyncio.Semaphore(cfg.deepseek_workers) stage2 = await _run_stage(detect_deepseek, session, sem2, cfg, prompt, folder, - files, ds_cache_path, log, progress, stop) + files, ds_cache_path, progress, stop) # ---- 无违规两票复核 ---- if stage2 and verify and not (stop and stop()): sem2 = asyncio.Semaphore(cfg.deepseek_workers) await _verify_clean(session, sem2, prompt, folder, files, stage2, cfg, - log, progress, stop, ds_cache_path) + progress, stop, ds_cache_path) # ---- 合并结果 ---- rows = [] @@ -263,7 +287,8 @@ async def run_detection(folder, cfg: AppConfig, log=print, progress=None, stop=N _save_cache(_cache_path("final", folder), {"rows": rows, "summary": summary, "generated_at": datetime.now().isoformat(timespec="seconds")}) - log(f"完成：{json.dumps(summary['counts'], ensure_ascii=False)}") - log(f"DeepSeek 用量：{summary['ds_calls']} 次调用，输出 {comp} token，" - f"成本≈¥{peak:.2f}（高峰）/¥{idle:.2f}（空闲）") + logger.info("完成：%s", json.dumps(summary["counts"], ensure_ascii=False)) + logger.info("DeepSeek 用量：%d 次调用，输出 %d token，成本≈¥%.2f（高峰）/¥%.2f（空闲）", + summary["ds_calls"], summary["ds_output_tokens"], peak, idle) + logger.info("总耗时 %.1f 秒", time.monotonic() - t0) return rows, summary diff --git a/src/violation_detector/providers.py b/src/violation_detector/providers.py index 2c2236e..24e963e 100644 --- a/src/violation_detector/providers.py +++ b/src/violation_detector/providers.py @@ -8,6 +8,7 @@ import asyncio import base64 import json +import logging import os import re @@ -15,6 +16,8 @@ import aiohttp from .config import AppConfig +logger = logging.getLogger("violation_detector.providers") + MIME = {".png": "image/png", ".jpg": "image/jpeg", ".jpeg": "image/jpeg", ".webp": "image/webp", ".gif": "image/gif", ".bmp": "image/bmp", ".tiff": "image/tiff"} @@ -114,6 +117,7 @@ async def detect_ark(session, sem, cfg: AppConfig, prompt: str, path: str, stop= last_err = "" for attempt in range(1, cfg.retries + 1): if stop and stop(): + logger.info("豆包检测 %s：用户停止", fname) return {"file": fname, "provider": "doubao", "status": "error", "attr": "", "logic": "", "category": "检测异常", "raw": "用户停止", "usage": {}} try: @@ -123,11 +127,17 @@ async def detect_ark(session, sem, cfg: AppConfig, prompt: str, path: str, stop= data = await r.json() content = data["choices"][0]["message"]["content"] return _result(fname, "doubao", content, data) - last_err = f"HTTP {r.status}: {(await r.text())[:300]}" - except Exception as e: # noqa: BLE001 - last_err = f"{type(e).__name__}: {e}" + body = (await r.text())[:2000] + last_err = f"HTTP {r.status}: {body[:300]}" + logger.warning("豆包检测 %s 第%d/%d次尝试失败：HTTP %s\n响应体：%s", + fname, attempt, cfg.retries, r.status, body) + except Exception: # noqa: BLE001 网络层异常：记录完整堆栈后重试 + last_err = "网络/协议异常（详见日志）" + logger.exception("豆包检测 %s 第%d/%d次尝试抛出异常", fname, attempt, cfg.retries) if attempt < cfg.retries: + logger.info("豆包检测 %s：%.0f 秒后重试", fname, 3 * attempt) await asyncio.sleep(3 * attempt) + logger.error("豆包检测 %s 最终失败：%s", fname, last_err) return {"file": fname, "provider": "doubao", "status": "error", "attr": "", "logic": "", "category": "检测异常", "raw": last_err, "usage": {}} @@ -156,6 +166,7 @@ async def detect_deepseek(session, sem, cfg: AppConfig, prompt: str, path: str, last_err = "" for attempt in range(1, cfg.retries + 1): if stop and stop(): + logger.info("DeepSeek 检测 %s：用户停止", fname) return {"file": fname, "provider": "deepseek", "status": "error", "attr": "", "logic": "", "category": "检测异常", "raw": "用户停止", "usage": {}} try: @@ -167,12 +178,22 @@ async def detect_deepseek(session, sem, cfg: AppConfig, prompt: str, path: str, if content and content.strip(): return _result(fname, "deepseek", content, data) # 推理烧尽预算会得到空正文，按可重试错误处理 - last_err = "empty content (reasoning exhausted max_tokens?)" + fr = data["choices"][0].get("finish_reason") + last_err = f"empty content (finish_reason={fr}, max_tokens={cfg.max_tokens})" + logger.warning("DeepSeek 检测 %s 第%d/%d次尝试返回空正文" + "（finish_reason=%s，推理可能耗尽 max_tokens=%s）", + fname, attempt, cfg.retries, fr, cfg.max_tokens) else: - last_err = f"HTTP {r.status}: {(await r.text())[:300]}" - except Exception as e: # noqa: BLE001 - last_err = f"{type(e).__name__}: {e}" + body = (await r.text())[:2000] + last_err = f"HTTP {r.status}: {body[:300]}" + logger.warning("DeepSeek 检测 %s 第%d/%d次尝试失败：HTTP %s\n响应体：%s", + fname, attempt, cfg.retries, r.status, body) + except Exception: # noqa: BLE001 网络层异常：记录完整堆栈后重试 + last_err = "网络/协议异常（详见日志）" + logger.exception("DeepSeek 检测 %s 第%d/%d次尝试抛出异常", fname, attempt, cfg.retries) if attempt < cfg.retries: + logger.info("DeepSeek 检测 %s：%.0f 秒后重试", fname, 3 * attempt) await asyncio.sleep(3 * attempt) + logger.error("DeepSeek 检测 %s 最终失败：%s", fname, last_err) return {"file": fname, "provider": "deepseek", "status": "error", "attr": "", "logic": "", "category": "检测异常", "raw": last_err, "usage": {}} diff --git a/tests/test_logutil.py b/tests/test_logutil.py new file mode 100644 index 0000000..9397932 --- /dev/null +++ b/tests/test_logutil.py @@ -0,0 +1,54 @@ +# -*- coding: utf-8 -*- +"""logutil 模块测试：文件落盘、级别过滤、异常堆栈、队列处理器。""" +import logging +import queue + +from violation_detector.logutil import LOGGER_NAME, QueueHandler, setup_logging + + +def get_logger(): + return logging.getLogger(LOGGER_NAME) + + +def test_setup_creates_file_and_logs(tmp_path): + path = setup_logging(console=False, log_dir=tmp_path) + assert path.exists() + get_logger().info("你好日志") + for h in get_logger().handlers: + h.flush() + text = path.read_text(encoding="utf-8") + assert "你好日志" in text + assert "[violation_detector]" in text # 详细格式含模块名 + + +def test_exception_logs_full_traceback(tmp_path): + path = setup_logging(console=False, log_dir=tmp_path) + try: + raise ValueError("爆炸点") + except ValueError: + get_logger().exception("捕获到异常") + for h in get_logger().handlers: + h.flush() + text = path.read_text(encoding="utf-8") + assert "Traceback (most recent call last)" in text + assert "ValueError: 爆炸点" in text + assert "test_exception_logs_full_traceback" in text # 堆栈含调用位置 + + +def test_debug_goes_to_file_not_queue(tmp_path): + """文件记录 DEBUG 全量；队列处理器（GUI）只收 INFO 以上。""" + path = setup_logging(console=False, log_dir=tmp_path) + q = queue.Queue() + get_logger().addHandler(QueueHandler(q)) + get_logger().debug("调试细节") + get_logger().warning("告警信息") + msgs = [] + while not q.empty(): + msgs.append(q.get_nowait()[1]) + # 队列里只有 WARNING，DEBUG 只进文件 + assert any("告警信息" in m for m in msgs) + assert not any("调试细节" in m for m in msgs) + for h in get_logger().handlers: + if isinstance(h, logging.FileHandler): + h.stream.flush() + assert "调试细节" in path.read_text(encoding="utf-8")