feat(video-analysis): 抖音免浏览器解析链路 + 快手支持 + 小红书加固 + 发送防重复
- 抖音（2026-10-13）：删除 SSR + playwright 旧链路（douyin_ssr.py 移除， douyin_api.py 精简为下载工具），新增 douyin_web.py（x-tt-argus 网关头 + web detail 接口 + 最高码率梯/多 CDN 回退下载）与 douyin_sign.py （纯 Python a_bogus 签名，vendor 自 f2（Apache-2.0），依赖 gmssl SM3）； 实测 10/10 成功率、单作品 <1s（旧浏览器链路 5-15s 且随机撞验证码） - 快手（2026-10-08）：kuaishou_content.py 分享页 INIT_STATE 静态解析（无登录态）， 清晰度梯最佳档 + 图集 URL 组装，入口/策略/Web 面板同步接入 - 小红书：按链接来源定国际/国内站优先序、视频 backupUrls 逐候选回退、 图片 urlDefault→infoList WB_DFT→url 三档回退 - 长视频防重复发送（2026-10-13）：媒体发送带 per-call _timeout（默认 300s）， 超时=结果未知不再降级重试；auto/active 两条 handler 互斥 - Web 解析封面抽帧 ffmpeg 加 CREATE_NO_WINDOW（pythonw 下不再弹 cmd 黑窗） - 依赖：+gmssl；测试：新增 test_douyin_web / test_kuaishou_parse / test_video_cleanup， 扩充 test_video_policy / test_video_sender_link / test_video_group_file / test_video_media_naming Co-Authored-By: Claude Code <noreply@anthropic.com>
This commit is contained in:
sansenhoshiandClaude Code committed
1 parent
a31aab00e8
commit
a789a44f5d
25 files changed
+3536
-895
No files matched your search
@@ -17,6 +17,7 @@ hexi/data/temp 是下载媒体中转区(见 utils.get_temp_root),发送成功后
|
||||
|
||||
import asyncio
|
||||
import time
|
||||
from collections.abc import Callable
|
||||
from pathlib import Path
|
||||
from typing import Literal
|
||||
|
||||
@@ -25,9 +26,11 @@ from nonebot.adapters.onebot.v11 import Bot, MessageEvent
|
||||
from pydantic import BaseModel
|
||||
|
||||
from hexi.core.custom_utils import check_manage
|
||||
from hexi.core.task_slots import TASKS, Task
|
||||
|
||||
require("nonebot_plugin_apscheduler")
|
||||
from nonebot_plugin_apscheduler import scheduler
|
||||
from hexi.core.scheduler import add_job
|
||||
|
||||
from nonebot_plugin_alconna import UniMessage
|
||||
|
||||
@@ -101,33 +104,62 @@ def _prune_empty_dirs(root: Path, max_age: float) -> int:
|
||||
return removed
|
||||
|
||||
|
||||
def clean_temp_files(sub: str = "", days: int | None = None) -> tuple[int, int]:
|
||||
def clean_temp_files(
|
||||
sub: str = "",
|
||||
days: int | None = None,
|
||||
on_progress: Callable[..., None] | None = None,
|
||||
) -> tuple[int, int]:
|
||||
"""清理 temp[/sub] 下超过期限的文件与随之空掉的目录。
|
||||
|
||||
Args:
|
||||
on_progress: 进度回调 `(done, total, phase, message, warning)`（关键字可省）——
|
||||
扫描阶段 total 未知，之后按"处理过的文件数"上报。跑在哪个线程调用都行
|
||||
（任务槽那边会切回事件循环）。**返回 False 表示调用方请求取消**，
|
||||
本函数会在处理完当前文件后尽早收尾（清理天然可中断，不硬杀线程）。
|
||||
|
||||
Returns:
|
||||
(removed, total) — 删除的文件数、统计到的文件总数
|
||||
(removed, total) — 删除的文件数、统计到的文件数（被取消时是"取消前"的进度）
|
||||
"""
|
||||
|
||||
def report(**kwargs) -> bool:
|
||||
if on_progress is None:
|
||||
return True # 没有槽（纯函数调用）时不存在取消
|
||||
return bool(on_progress(**kwargs))
|
||||
|
||||
root = get_temp_root(sub)
|
||||
if not root.is_dir():
|
||||
return 0, 0
|
||||
|
||||
max_age = _max_age_seconds(days)
|
||||
report(phase="扫描中", unit="个文件", message="正在统计文件…")
|
||||
files = _walk_files(root)
|
||||
removed = 0
|
||||
for path in files:
|
||||
if not _file_is_stale(path, max_age):
|
||||
continue
|
||||
try:
|
||||
path.unlink(missing_ok=True)
|
||||
removed += 1
|
||||
logger.info(f"temp 清理: 删除 {path}")
|
||||
except OSError as e:
|
||||
# 文件被占用(如发送中)等场景,留待下轮
|
||||
logger.warning(f"temp 清理: 跳过 {path} ({e})")
|
||||
report(
|
||||
done=0, total=len(files), phase="清理中",
|
||||
unit="个文件", message=f"共 {len(files)} 个文件",
|
||||
)
|
||||
|
||||
dirs = _prune_empty_dirs(root, max_age)
|
||||
if dirs:
|
||||
logger.info(f"temp 清理: 同时收掉 {dirs} 个空目录")
|
||||
removed = 0
|
||||
canceled = False
|
||||
for index, path in enumerate(files, 1):
|
||||
if _file_is_stale(path, max_age):
|
||||
try:
|
||||
path.unlink(missing_ok=True)
|
||||
removed += 1
|
||||
logger.info(f"temp 清理: 删除 {path}")
|
||||
except OSError as e:
|
||||
# 文件被占用(如发送中)等场景,留待下轮 —— 也算"部分失败",报给任务槽
|
||||
logger.warning(f"temp 清理: 跳过 {path} ({e})")
|
||||
report(warning=f"跳过 {path.name}（被占用？）：{e}")
|
||||
if not report(done=index, total=len(files), message=f"已删除 {removed} 个"):
|
||||
logger.info("temp 清理: 收到取消请求，提前结束")
|
||||
canceled = True
|
||||
break
|
||||
|
||||
if not canceled:
|
||||
report(phase="回收空目录", done=len(files), total=len(files))
|
||||
dirs = _prune_empty_dirs(root, max_age)
|
||||
if dirs:
|
||||
logger.info(f"temp 清理: 同时收掉 {dirs} 个空目录")
|
||||
return removed, len(files)
|
||||
|
||||
|
||||
@@ -165,7 +197,27 @@ async def _handle_clean(bot: Bot, event: MessageEvent):
|
||||
except ValueError:
|
||||
await UniMessage.text("天数参数不合法,示例:/清理temp 3").send()
|
||||
return
|
||||
removed, total = await _run_clean(days)
|
||||
task, reused = submit_clean(
|
||||
days,
|
||||
title="清理 temp（命令）",
|
||||
trigger="command",
|
||||
actor=event.get_user_id(),
|
||||
group_id=getattr(event, "group_id", "") or "",
|
||||
)
|
||||
if reused:
|
||||
await UniMessage.text(
|
||||
"已经有一个清理任务在跑了,等它跑完再来吧~（进度可在网页端看）"
|
||||
).send()
|
||||
return
|
||||
await TASKS.wait(task) # 命令要"跑完再回复",与网页端的即提交即返回不同
|
||||
|
||||
if task.status == "canceled":
|
||||
await UniMessage.text("清理被取消了（网页端点过取消）。").send()
|
||||
return
|
||||
if task.status != "done":
|
||||
await UniMessage.text(f"清理失败了:{task.error}").send()
|
||||
return
|
||||
removed, total = task.result or (0, 0)
|
||||
if total == 0:
|
||||
await UniMessage.text("temp 目录目前是空的,没什么可清理的~").send()
|
||||
else:
|
||||
@@ -174,7 +226,7 @@ async def _handle_clean(bot: Bot, event: MessageEvent):
|
||||
|
||||
@stats_cmd.handle()
|
||||
async def _handle_stats(event: MessageEvent):
|
||||
st = await _run_stats()
|
||||
st = await run_stats()
|
||||
if st["files"] == 0:
|
||||
if st.get("dirs"):
|
||||
await UniMessage.text(
|
||||
@@ -191,13 +243,37 @@ async def _handle_stats(event: MessageEvent):
|
||||
).send()
|
||||
|
||||
|
||||
async def _run_clean(days: int | None = None) -> tuple[int, int]:
|
||||
"""执行清理并记录日志(供命令与定时任务共用)"""
|
||||
removed, total = await asyncio.to_thread(clean_temp_files, "", days)
|
||||
return removed, total
|
||||
#: 任务槽名:命令 / Web / 每日定时三条路共用同一个槽,同槽只跑一个(防重复触发)
|
||||
CLEANUP_SLOT = "video_analysis:temp_cleanup"
|
||||
|
||||
|
||||
async def _run_stats() -> dict:
|
||||
def submit_clean(
|
||||
days: int | None,
|
||||
*,
|
||||
title: str,
|
||||
trigger: str = "",
|
||||
actor: str = "",
|
||||
group_id: str = "",
|
||||
) -> tuple[Task, bool]:
|
||||
"""把清理提交到任务槽,立即返回 (任务, 是否复用了正在跑的任务)。
|
||||
|
||||
任务体跑在线程池里(见 hexi.core.task_slots),进度实时推给网页端;
|
||||
同槽已有任务时不会重复清理 —— 服务端这一层才是"防连点"的真正保证。
|
||||
trigger / actor / group_id 是审计信息(谁点的:web 用户名 / command 的 QQ 号)。
|
||||
"""
|
||||
return TASKS.submit(
|
||||
slot=CLEANUP_SLOT,
|
||||
title=title,
|
||||
plugin_id="video_analysis",
|
||||
trigger=trigger,
|
||||
actor=actor,
|
||||
group_id=group_id,
|
||||
func=lambda report: clean_temp_files(days=days, on_progress=report),
|
||||
)
|
||||
|
||||
|
||||
async def run_stats() -> dict:
|
||||
"""异步统计(命令 / Web 缓存面板共用)"""
|
||||
return await asyncio.to_thread(temp_stats)
|
||||
|
||||
|
||||
@@ -213,7 +289,7 @@ def _register_auto_job() -> None:
|
||||
from apscheduler.jobstores.base import JobLookupError # noqa: PLC0415
|
||||
|
||||
if cleanup_config.video_analysis_temp_cleanup_mode == "auto":
|
||||
scheduler.add_job(
|
||||
add_job(
|
||||
_scheduled_cleanup,
|
||||
"cron",
|
||||
hour=cleanup_config.video_analysis_temp_cleanup_hour,
|
||||
@@ -239,13 +315,18 @@ def reload_cleanup_config() -> None:
|
||||
|
||||
|
||||
async def _scheduled_cleanup() -> None:
|
||||
"""每日定时清理 temp 目录(auto 模式)"""
|
||||
"""每日定时清理 temp 目录(auto 模式);同样走任务槽(与手动清理互斥)"""
|
||||
logger.info("temp 自动清理: 开始")
|
||||
try:
|
||||
removed, total = await asyncio.to_thread(clean_temp_files)
|
||||
logger.info(f"temp 自动清理完成:删除 {removed} 个,当前共 {total} 个文件")
|
||||
except Exception as e:
|
||||
logger.exception(f"temp 自动清理失败: {e}")
|
||||
task, reused = submit_clean(None, title="清理 temp（每日定时）", trigger="schedule")
|
||||
if reused:
|
||||
logger.info("temp 自动清理: 已有清理任务在跑，跳过本次")
|
||||
return
|
||||
await TASKS.wait(task)
|
||||
if task.status != "done":
|
||||
logger.error(f"temp 自动清理未完成（{task.status}）: {task.error}")
|
||||
return
|
||||
removed, total = task.result or (0, 0)
|
||||
logger.info(f"temp 自动清理完成:删除 {removed} 个,当前共 {total} 个文件")
|
||||
|
||||
|
||||
# 插件加载时按模式注册(auto)或提示(manual)
|
||||
|
||||
Reference in new issue
Block a user