修复：恢复任务状态并精简新建流程

DEVELOPMENT_LOG.md

@@ -1,5 +1,17 @@
 # Development Log
 
+## 2026-08-25 任务 20 状态恢复与新建流程精简
+
+- 已确认任务 `20 - E1829 110128 舒淇,安以轩P29-29` 的第 2 次切片批次、12 条 active 成片和发送中心 12/12 关联完整，原“失败”来自完成后重复触发 AI 分析并尝试删除仍被切片引用的候选记录。
+- 修复前 SQLite Online Backup 为 `backups/workflow-before-task20-repair-20260824-233721.sqlite3`；只把任务 `3210d91ee1fb` 恢复为 `completed / 100%` 并清空误报错误，候选、切片、Workflow Job、AI run 和发布任务 ID 前后完全一致，修复依据已写入任务日志。
+- 手动 AI 入口新增原子重入保护：活跃全自动 Job、已有 active 切片或已有发送中心记录时返回 HTTP 409；候选替换在同一事务内先检查 `output_clip` 引用，改为可理解的业务冲突，不再暴露 SQLite 外键错误，也不会把已完成任务降级为失败。
+- 新建任务只保留本机视频上传；任务名称历史候选最多显示最近 5 个未删除且去重后的名称，JSON 已有文件创建入口、目录浏览 API 和 NAS 页面能力已移除。
+- `source_type`、`nas_file_path` 物理列继续保留兼容；账本迁移 `20260824_02_task_upload_only` 在写入前创建 SQLite Online Backup，将旧来源归一到 `original_video_path + upload`，不删除外部视频。
+- “每小时高光密度”和候选总上限仅在长直播高光模式显示和提交；通用与康熙模式不读取页面参数，服务端固定使用兼容默认值。
+- 新增 `POST /api/subtitles/tasks/{task_id}/skip-to-review`；明确跳过字幕后进入片段审核，不再自动续跑文案和发送中心。审核保存并完整关联后才标记完成；关联不完整时保留审核状态与错误。
+- 本轮未修改 AI Provider、Prompt、模型、候选算法或真实发布逻辑，也没有重跑任务 20 的 AI、切片或发布任务。
+- 活动库迁移备份为 `data/backups/workflow-before-task-upload-only-20260825-000702-586803-d14c09b5.sqlite3`；迁移后旧来源记录为 0，`integrity_check=ok`、`foreign_key_check=0`。完整测试 `672 passed`，Ruff、Python Compileall、JavaScript 语法和浏览器烟测通过；8001、Scheduler 与 8765 Windows Worker 健康。
+
 ## 2026-08-24 稳定 V1 P1.3c 数据库迁移账本与唯一索引 Fail-Closed
 
 - 新增 `schema_migrations` 账本；每条正式新迁移记录稳定 version、name、checksum 和完成时间，同版本定义漂移时拒绝启动。

NEXT_STEPS.md

@@ -1,5 +1,12 @@
 # Next Steps
 
+## 2026-08-25 任务 20 修复后检查
+
+1. 打开 `http://127.0.0.1:8001/`，确认任务 20 显示“已完成”、无外键错误，并仍为 12 条 active 切片和发送中心已关联 12/12。
+2. 打开“新建任务”：任务名候选最多 5 个，只能上传本机文件；通用/康熙模式不显示密度设置，长直播模式才显示。
+3. 对低风险全自动测试任务选择“跳过字幕，进入片段审核”，确认不会立即生成发送中心任务；在片段审核保存后才同步。
+4. 不需要点击“立即发送”；本轮不会执行真实投稿，也不改变 `NEED_REVIEW` 的人工确认边界。
+
 ## 2026-08-24 稳定 V1 整改路线
 
 - [x] P0.1：Pytest 与活动数据库、媒体目录彻底隔离，危险整表清理增加 fail-closed 校验。
@@ -12,6 +19,7 @@
 - [x] P1.3a：任务状态接口受控转换；Task/Job lease 代际一致；自动流水线取消恢复、READY 终态和关联发布清理明确；切片批次编号、输出目录、Task 终态和数据库激活原子化。
 - [x] P1.3b：自动流水线已增加版本化步骤 checkpoint、输入/产物证据、旧 Job 原位重试和重领恢复；已覆盖 AI、转写、切片、文案、排期与发布草稿的失效边界。
 - [x] P1.3c：已增加 `schema_migrations` 版本/checksum 账本、迁移前 Online Backup 和 fail-closed v2 活动发布唯一索引；重复数据、索引漂移或账本异常都会拒绝启动。本轮未直接迁移活动库。
+- [x] 任务 20 专项：完成状态恢复、重复 AI 409 保护、上传单入口、最近 5 个任务名、长直播参数显隐和“跳过字幕后进入片段审核”闭环。
 - [ ] P1.3d：处理字幕批准批次原子性、字幕/切片中断目录清理和剩余跨进程恢复边界。
 - [ ] P1.4：统一第三方 AI/FFmpeg 超时、错误 JSON、429/5xx 与重试幂等边界，并避免重复计费。
 - [ ] P1.5：在不扩大个人本地项目范围的前提下处理密钥日志、输入校验和本地管理员接口门禁。

app/db/database.py

@@ -41,6 +41,27 @@
         f"DROP INDEX {PUBLISH_ACTIVE_UNIQUE_INDEX_LEGACY_NAME}"
     ).encode("utf-8")
 ).hexdigest()
+TASK_UPLOAD_ONLY_MIGRATION_VERSION = "20260824_02_task_upload_only"
+TASK_UPLOAD_ONLY_MIGRATION_NAME = "任务视频来源归一为本机上传"
+TASK_UPLOAD_ONLY_MIGRATION_SQL = """
+UPDATE tasks
+SET original_video_path = CASE
+        WHEN original_video_path IS NULL OR TRIM(original_video_path) = '' THEN nas_file_path
+        ELSE original_video_path
+    END,
+    source_type = 'upload',
+    nas_file_path = NULL
+WHERE source_type != 'upload'
+   OR source_type IS NULL
+   OR (nas_file_path IS NOT NULL AND TRIM(nas_file_path) != '')
+""".strip()
+TASK_UPLOAD_ONLY_MIGRATION_CHECKSUM = hashlib.sha256(
+    (
+        f"{TASK_UPLOAD_ONLY_MIGRATION_VERSION}\n"
+        f"{TASK_UPLOAD_ONLY_MIGRATION_NAME}\n"
+        f"{TASK_UPLOAD_ONLY_MIGRATION_SQL}"
+    ).encode("utf-8")
+).hexdigest()
 
 
 class SchemaMigrationError(RuntimeError):
@@ -79,6 +100,7 @@ def init_db() -> None:
     needs_subtitle_editor_backup = _requires_subtitle_editor_schema_migration(settings.database_path)
     needs_subtitle_auto_backup = _requires_subtitle_auto_schema_migration(settings.database_path)
     needs_publish_index_backup = _requires_publish_active_index_migration(settings.database_path)
+    needs_task_upload_only_backup = _requires_task_upload_only_migration(settings.database_path)
     if needs_long_live_backup:
         create_schema_migration_backup(
             settings.database_path,
@@ -116,6 +138,12 @@ def init_db() -> None:
             settings.data_dir / "backups",
             "publish-active-unique-index-v2",
         )
+    if needs_task_upload_only_backup:
+        create_schema_migration_backup(
+            settings.database_path,
+            settings.data_dir / "backups",
+            "task-upload-only",
+        )
 
     with get_connection() as connection:
         connection.executescript(
@@ -721,6 +749,31 @@ def _requires_publish_active_index_migration(database_path) -> bool:
             connection.close()
 
 
+def _requires_task_upload_only_migration(database_path) -> bool:
+    """只有确实存在旧 NAS 来源数据时才在归一化前备份。"""
+    if not database_path.exists() or database_path.stat().st_size == 0:
+        return False
+    connection = None
+    try:
+        connection = sqlite3.connect(f"{database_path.resolve().as_uri()}?mode=ro", uri=True, timeout=10)
+        columns = {row[1] for row in connection.execute("PRAGMA table_info(tasks)").fetchall()}
+        if not {"source_type", "nas_file_path"} <= columns:
+            return False
+        row = connection.execute(
+            """
+            SELECT 1 FROM tasks
+            WHERE source_type != 'upload'
+               OR source_type IS NULL
+               OR (nas_file_path IS NOT NULL AND TRIM(nas_file_path) != '')
+            LIMIT 1
+            """
+        ).fetchone()
+        return row is not None
+    finally:
+        if connection is not None:
+            connection.close()
+
+
 def _get_table_columns(connection: sqlite3.Connection, table_name: str) -> set[str]:
     rows = connection.execute(f"PRAGMA table_info({table_name})").fetchall()
     return {row["name"] for row in rows}
@@ -856,6 +909,24 @@ def _apply_publish_active_unique_index_migration(connection: sqlite3.Connection)
     connection.execute(f"DROP INDEX IF EXISTS {PUBLISH_ACTIVE_UNIQUE_INDEX_LEGACY_NAME}")
 
 
+def _apply_task_upload_only_migration(connection: sqlite3.Connection) -> None:
+    connection.execute(TASK_UPLOAD_ONLY_MIGRATION_SQL)
+
+
+def _verify_task_upload_only_migration(connection: sqlite3.Connection) -> None:
+    row = connection.execute(
+        """
+        SELECT 1 FROM tasks
+        WHERE source_type != 'upload'
+           OR source_type IS NULL
+           OR (nas_file_path IS NOT NULL AND TRIM(nas_file_path) != '')
+        LIMIT 1
+        """
+    ).fetchone()
+    if row is not None:
+        raise SchemaMigrationError("仍存在未归一化的 NAS 视频来源记录")
+
+
 def _registered_schema_migrations() -> tuple[SchemaMigration, ...]:
     return (
         SchemaMigration(
@@ -865,6 +936,13 @@ def _registered_schema_migrations() -> tuple[SchemaMigration, ...]:
             apply=_apply_publish_active_unique_index_migration,
             verify=_verify_publish_active_unique_index_migration,
         ),
+        SchemaMigration(
+            version=TASK_UPLOAD_ONLY_MIGRATION_VERSION,
+            name=TASK_UPLOAD_ONLY_MIGRATION_NAME,
+            checksum=TASK_UPLOAD_ONLY_MIGRATION_CHECKSUM,
+            apply=_apply_task_upload_only_migration,
+            verify=_verify_task_upload_only_migration,
+        ),
     )
 
 

app/main.py

@@ -9,7 +9,7 @@
 
 from app.core.config import settings
 from app.db.database import init_db
-from app.routers import ai_prompts, files, media, pages, publish, settings as settings_router, subtitles, tasks
+from app.routers import ai_prompts, media, pages, publish, settings as settings_router, subtitles, tasks
 from app.services.publish_scheduler import start_scheduler_background
 from app.services.storage_service import configure_runtime_media_storage
 from app.services.job_worker import WorkflowJobRunner
@@ -143,7 +143,6 @@ async def security_middleware(request: Request, call_next):
 app.include_router(ai_prompts.router)
 app.include_router(tasks.router)
 app.include_router(subtitles.router)
-app.include_router(files.router)
 app.include_router(media.router)
 app.include_router(publish.router)
 app.include_router(settings_router.router)

app/models/task.py

@@ -43,10 +43,8 @@ class TaskStatus(str, Enum):
 
 class TaskCreate(BaseModel):
     task_name: str = Field(..., min_length=1, max_length=120)
-    source_type: Literal["upload", "nas"] = "upload"
     platform: Literal["douyin", "bilibili", "general"] = "general"
     original_video_path: Optional[str] = None
-    nas_file_path: Optional[str] = None
     max_clip_duration: int = Field(default=10, ge=1, le=60)
     candidate_clip_count: int = Field(default=12, ge=1, le=50)
     selection_profile: Literal["general", "variety_comedy", "long_live_talk"]

app/routers/subtitles.py

@@ -19,6 +19,7 @@
     enqueue_task_subtitle_render,
     prepare_task_subtitle_review,
     skip_task_subtitles_and_resume,
+    skip_task_subtitles_to_review,
 )
 from app.services.subtitle_data_service import (
     SubtitleRevisionConflict,
@@ -72,6 +73,11 @@ def skip_and_resume(task_id: str) -> dict:
     return _call(lambda: skip_task_subtitles_and_resume(task_id))
 
 
+@router.post("/tasks/{task_id}/skip-to-review")
+def skip_to_review(task_id: str) -> dict:
+    return _call(lambda: skip_task_subtitles_to_review(task_id))
+
+
 @router.get("/tasks/{task_id}/jobs")
 def list_subtitle_jobs(task_id: str) -> dict:
     jobs = [

app/routers/tasks.py

@@ -37,17 +37,6 @@ async def list_tasks() -> list[dict]:
     return task_service.list_tasks()
 
 
-@router.post("")
-async def create_task(payload: TaskCreate, background_tasks: BackgroundTasks) -> dict:
-    try:
-        result = task_service.create_task_record(payload)
-        if payload.auto_mode:
-            result["auto_pipeline"] = start_auto_pipeline(result["id"], background_tasks=background_tasks)
-        return result
-    except ValueError as exc:
-        raise HTTPException(status_code=400, detail=str(exc)) from exc
-
-
 @router.post("/upload")
 async def create_upload_task(
     background_tasks: BackgroundTasks,
@@ -87,15 +76,14 @@ async def create_upload_task(
         )
         payload = TaskCreate(
             task_name=task_name,
-            source_type="upload",
             platform=platform,
             original_video_path=str(saved_path),
             max_clip_duration=max_clip_duration,
             candidate_clip_count=candidate_clip_count,
             selection_profile=selection_profile,
             final_clip_target=final_clip_target,
-            highlight_density_per_hour=highlight_density_per_hour,
-            highlight_total_limit=highlight_total_limit,
+            highlight_density_per_hour=(highlight_density_per_hour if selection_profile == "long_live_talk" else 4),
+            highlight_total_limit=(highlight_total_limit if selection_profile == "long_live_talk" else 30),
             ai_preference=ai_preference,
             auto_mode=auto_mode,
             auto_clip_count=auto_clip_count,
@@ -297,6 +285,8 @@ async def process_ai_analysis(
 ) -> dict:
     try:
         return await run_in_threadpool(task_service.process_task_ai_analysis, task_id, provider=provider)
+    except task_service.AIAnalysisConflictError as exc:
+        raise HTTPException(status_code=409, detail=str(exc)) from exc
     except ValueError as exc:
         raise HTTPException(status_code=400, detail=str(exc)) from exc
 
@@ -317,6 +307,8 @@ async def get_ai_analysis_runs(task_id: str) -> dict:
 async def restore_ai_analysis_run(task_id: str, run_id: str) -> dict:
     try:
         return task_service.restore_ai_analysis_run(task_id, run_id)
+    except task_service.AIAnalysisConflictError as exc:
+        raise HTTPException(status_code=409, detail=str(exc)) from exc
     except ValueError as exc:
         raise HTTPException(status_code=400, detail=str(exc)) from exc