feat: 建立 SAM2 标注闭环基线

- 打通工作区真实标注闭环:支持手工多边形、矩形、圆形、点区域和线段生成 mask,并可保存、回显、更新和删除后端 annotation。

- 增强 polygon 编辑器:支持顶点拖动、顶点删除、边中点插入、多 polygon 子区域选择编辑,以及区域合并和区域去除。

- 接入 GT mask 导入:后端支持二值/多类别 mask 拆分、contour 转 polygon、distance transform seed point,前端支持导入、回显和 seed point 拖动编辑。

- 完善导出能力:COCO JSON 导出对齐前端,PNG mask ZIP 同时包含单标注 mask、按 zIndex 融合的 semantic_frame 和 semantic_classes.json。

- 打通异步任务管理:新增任务取消、重试、失败详情接口与 Dashboard 控件,worker 支持取消状态检查并通过 Redis/WebSocket 推送 cancelled 事件。

- 对接 Dashboard 后端数据:概览统计、解析队列和实时流转记录从 FastAPI 聚合接口与 WebSocket 更新。

- 增强 AI 推理参数:前端发送 crop_to_prompt、auto_filter_background 和 min_score,后端支持点/框 prompt 局部裁剪推理、结果回映射和负向点/低分过滤。

- 接入 SAM3 基础设施:新增独立 Python 3.12 sam3 环境安装脚本、外部 worker helper、后端桥接和真实 Python/CUDA/包/HF checkpoint access 状态检测。

- 保留 SAM3 授权边界:当前官方 facebook/sam3 gated 权重未授权时状态接口会返回不可用,不伪装成可推理。

- 增强前端状态管理:新增 mask undo/redo 历史栈、AI 模型选择状态、保存状态 dirty/draft/saved 流转和项目状态归一化。

- 更新前端 API 封装:补充 annotation CRUD、GT mask import、mask ZIP export、task cancel/retry/detail、AI runtime status 和 prediction options。

- 更新 UI 控件:ToolsPalette、AISegmentation、VideoWorkspace 和 CanvasArea 接入真实操作、导入导出、撤销重做、任务控制和模型状态。

- 新增 polygon-clipping 依赖,用于前端区域 union/difference 几何运算。

- 完善后端 schemas/status/progress:补充 AI 模型外部状态字段、任务 cancelled 状态和进度事件 payload。

- 补充测试覆盖:新增后端任务控制、SAM3 桥接、GT mask、导出融合、AI options 测试;补充前端 Canvas、Dashboard、VideoWorkspace、ToolsPalette、API 和 store 测试。

- 更新 README、AGENTS 和 doc 文档:冻结当前需求/设计/测试计划,标注真实功能、剩余 Mock、SAM3 授权边界和后续实施顺序。
This commit is contained in:
2026-05-01 15:26:25 +08:00
parent f020ff3b4f
commit 689a9ba283
48 changed files with 3280 additions and 176 deletions

104
backend/tests/test_tasks.py Normal file
View File

@@ -0,0 +1,104 @@
from models import ProcessingTask
def test_cancel_task_revokes_celery_and_updates_project(client, db_session, monkeypatch):
project = client.post("/api/projects", json={
"name": "Cancelable",
"video_path": "uploads/1/clip.mp4",
"status": "parsing",
}).json()
task = ProcessingTask(
task_type="parse_video",
status="running",
progress=35,
message="正在使用 FFmpeg/OpenCV 拆帧",
project_id=project["id"],
celery_task_id="celery-1",
payload={"source_type": "video"},
)
db_session.add(task)
db_session.commit()
db_session.refresh(task)
revoked = []
published = []
monkeypatch.setattr(
"routers.tasks.celery_app.control.revoke",
lambda celery_id, terminate, signal: revoked.append((celery_id, terminate, signal)),
)
monkeypatch.setattr("routers.tasks.publish_task_progress_event", lambda event_task: published.append(event_task.status))
response = client.post(f"/api/tasks/{task.id}/cancel")
assert response.status_code == 200
body = response.json()
assert body["status"] == "cancelled"
assert body["progress"] == 100
assert body["message"] == "任务已取消"
assert body["error"] == "Cancelled by user"
assert revoked == [("celery-1", True, "SIGTERM")]
assert published == ["cancelled"]
assert client.get(f"/api/projects/{project['id']}").json()["status"] == "pending"
def test_retry_task_creates_fresh_parse_task(client, db_session, monkeypatch):
project = client.post("/api/projects", json={
"name": "Retryable",
"video_path": "uploads/2/clip.mp4",
"source_type": "video",
"status": "error",
}).json()
task = ProcessingTask(
task_type="parse_video",
status="failed",
progress=100,
message="解析失败",
error="ffmpeg failed",
project_id=project["id"],
payload={"source_type": "video"},
)
db_session.add(task)
db_session.commit()
db_session.refresh(task)
class FakeAsyncResult:
id = "celery-retry"
queued = []
published = []
monkeypatch.setattr("routers.tasks.parse_project_media.delay", lambda task_id: queued.append(task_id) or FakeAsyncResult())
monkeypatch.setattr("routers.tasks.publish_task_progress_event", lambda event_task: published.append((event_task.id, event_task.status)))
response = client.post(f"/api/tasks/{task.id}/retry")
assert response.status_code == 202
body = response.json()
assert body["id"] != task.id
assert body["status"] == "queued"
assert body["progress"] == 0
assert body["celery_task_id"] == "celery-retry"
assert body["payload"]["retry_of"] == task.id
assert queued == [body["id"]]
assert published[0] == (body["id"], "queued")
assert published[-1] == (body["id"], "queued")
assert client.get(f"/api/projects/{project['id']}").json()["status"] == "parsing"
def test_task_actions_reject_invalid_states(client, db_session):
project = client.post("/api/projects", json={
"name": "Done",
"video_path": "uploads/3/clip.mp4",
}).json()
task = ProcessingTask(
task_type="parse_video",
status="success",
progress=100,
project_id=project["id"],
payload={"source_type": "video"},
)
db_session.add(task)
db_session.commit()
db_session.refresh(task)
assert client.post(f"/api/tasks/{task.id}/cancel").status_code == 409
assert client.post(f"/api/tasks/{task.id}/retry").status_code == 409