meijiaka-zy/python-api/app/api/v1/tasks.py

"""
统一任务管理 API
===============

提供任务创建和状态查询接口，支持：
- video: 视频生成
- image: 图片生成
- script: 脚本生成
- subtitle: 字幕对齐
- copy: 文案提取
- avatar_clone: 形象克隆
"""

import json
import logging
import uuid
from datetime import UTC, datetime
from typing import Literal

from fastapi import APIRouter, Depends, HTTPException
from pydantic import BaseModel, Field, field_validator

from app.api.deps import get_current_user
from app.core.redis_client import get_redis_client
from app.models.user import User
from app.scheduler.registry import JobRegistry
from app.schemas.enums import AvatarCloneStatus
from app.schemas.segment import Segment

logger = logging.getLogger(__name__)

router = APIRouter(tags=["Tasks"])


# ========== 请求/响应模型 ==========


class VideoParams(BaseModel):
    """视频生成参数"""

    segments: list[Segment] = Field(..., description="分镜列表")
    human_id: int | None = Field(None, description="数字人主体ID")

    @field_validator("segments")
    @classmethod
    def validate_segments(cls, v: list[Segment]) -> list[Segment]:
        if not v:
            raise ValueError("segments 不能为空列表")
        return v


class ImageParams(BaseModel):
    """图片生成参数"""

    prompt: str = Field(..., min_length=1, description="图片描述")
    image_type: str = Field(default="cover", description="图片类型: empty_shot/cover")
    reference_image: str | None = Field(None, description="参考图片URL（图生图）")
    human_id: int | None = Field(None, description="数字人主体ID（omni-image使用）")

    @field_validator("prompt")
    @classmethod
    def validate_prompt(cls, v: str) -> str:
        if not v or not v.strip():
            raise ValueError("prompt 不能为空")
        return v.strip()


class ScriptParams(BaseModel):
    """脚本生成参数"""

    category: str = Field(..., min_length=1, description="大类代码")
    subcategory: str = Field(..., min_length=1, description="小类代码")
    style: str = Field(default="default", description="脚本风格")
    duration: int = Field(default=60, ge=10, le=300, description="视频时长（秒）")

    @field_validator("category")
    @classmethod
    def validate_category(cls, v: str) -> str:
        if not v or not v.strip():
            raise ValueError("category 不能为空")
        return v.strip()

    @field_validator("subcategory")
    @classmethod
    def validate_subcategory(cls, v: str) -> str:
        if not v or not v.strip():
            raise ValueError("subcategory 不能为空")
        return v.strip()


class SubtitleParams(BaseModel):
    """字幕生成参数"""

    video_path: str = Field(..., min_length=1, description="视频文件路径")
    language: str = Field(default="zh", description="语言代码")
    mode: str = Field(default="caption", description="模式: caption/auto_align")
    audio_text: str | None = Field(default=None, description="打轴文本（auto_align 模式必填）")

    @field_validator("video_path")
    @classmethod
    def validate_video_path(cls, v: str) -> str:
        if not v or not v.strip():
            raise ValueError("video_path 不能为空")
        return v.strip()


class CopyParams(BaseModel):
    """文案提取参数"""

    video_url: str = Field(..., min_length=1, description="视频链接URL")

    @field_validator("video_url")
    @classmethod
    def validate_video_url(cls, v: str) -> str:
        if not v or not v.strip():
            raise ValueError("video_url 不能为空")
        if not v.startswith(("http://", "https://")):
            raise ValueError("video_url 必须是有效的URL")
        return v.strip()


class TTSParams(BaseModel):
    """TTS 语音合成参数"""

    segments: list[dict] = Field(..., description="分镜列表，每项包含 id, text/voiceover")
    voice_id: str = Field(default="zh_female_yizhi", description="音色 ID")
    speed: float = Field(default=1.0, ge=0.5, le=2.0, description="语速")

    @field_validator("segments")
    @classmethod
    def validate_segments(cls, v: list[dict]) -> list[dict]:
        if not v:
            raise ValueError("segments 不能为空列表")
        return v


class TaskCreateRequest(BaseModel):
    """创建任务请求"""

    project_id: str | None = Field(None, description="项目ID（可选）")
    params: dict = Field(default_factory=dict, description="任务参数")


class TaskCreateResponse(BaseModel):
    """创建任务响应"""

    task_id: str = Field(..., description="任务ID")
    status: str = Field("pending", description="任务状态")
    message: str = Field("任务已创建", description="状态消息")


class TaskStatusResponse(BaseModel):
    """任务状态响应"""

    task_id: str = Field(..., description="任务ID")
    type: str | None = Field(None, description="任务类型")
    status: str = Field(..., description="任务状态: pending/running/waiting/completed/failed")
    progress: int = Field(0, description="进度百分比 (0-100)")
    message: str = Field("", description="状态描述")
    completed: int = Field(0, description="已完成子任务数")
    total: int = Field(0, description="总子任务数")
    result: dict | None = Field(None, description="任务结果（完成时）")
    error: str | None = Field(None, description="错误信息（失败时）")
    created_at: str = Field("", description="任务创建时间（ISO格式）")


# ========== 辅助函数 ==========


def _generate_task_id() -> str:
    """生成任务ID"""
    return f"task_{uuid.uuid4().hex[:16]}"


# ========== API 路由 ==========


@router.post("/{task_type}", response_model=TaskCreateResponse)
async def create_task(
    task_type: Literal["video", "image", "script", "subtitle", "copy", "avatar_clone", "tts"],
    request: TaskCreateRequest,
    current_user: User = Depends(get_current_user),
) -> TaskCreateResponse:
    """
    创建新任务

    根据任务类型写入 Redis，由 Async Engine Scheduler 统一调度。
    """
    task_id = _generate_task_id()
    user_id = str(current_user.id)
    project_id = request.project_id or request.params.get("project_id", "")

    redis = get_redis_client()
    registry = JobRegistry(redis)

    try:
        await registry.create(task_id, task_type, user_id)
    except Exception as e:
        logger.error(f"[API] Failed to create registry entry: {e}")
        raise HTTPException(status_code=500, detail="创建任务失败：Redis连接错误")

    try:
        if task_type == "video":
            # 字段适配：前端 shots/element_id → 后端 segments/human_id
            import re

            video_params = dict(request.params)
            if "shots" in video_params:
                shots = video_params.pop("shots")
                for s in shots:
                    # 清洗 id：前端可能发送数字，Segment 模型要求 str
                    if "id" in s and not isinstance(s["id"], str):
                        s["id"] = str(s["id"])
                    # 清洗 duration：前端可能发送 "5s"，Segment 模型要求 int
                    duration = s.get("duration")
                    if isinstance(duration, str):
                        m = re.search(r"\d+", duration)
                        s["duration"] = int(m.group()) if m else None
                video_params["segments"] = shots
            if "element_id" in video_params:
                video_params["human_id"] = video_params.pop("element_id")
            validated = VideoParams(**video_params)
            segments = validated.segments
            human_id = validated.human_id

            normalized_segments = []
            for s in segments:
                normalized_segments.append(
                    {
                        "id": str(s.id),
                        "type": s.type,
                        "scene": s.scene,
                        "voiceover": s.voiceover,
                        "duration": s.duration,
                        "human_id": (human_id if s.type == "segment" else None),
                        "voice_id": s.voice_id,
                        "provider_task_id": None,
                        "status": "pending",
                        "video_url": None,
                        "local_path": None,
                        "qiniu_url": None,
                        "error_message": None,
                    }
                )

            await registry.update(
                task_id,
                status="running",
                message=f"开始生成视频，共 {len(normalized_segments)} 个镜头...",
                completed=0,
                total=len(normalized_segments),
                params={
                    "project_id": project_id,
                    "user_id": user_id,
                    "human_id": human_id,
                    "shots": json.dumps(normalized_segments, ensure_ascii=False),
                },
            )
            await registry.add_running(task_id)

        elif task_type == "image":
            image_validated = ImageParams(**request.params)
            await registry.update(
                task_id,
                status="running",
                message="准备生成图片...",
                completed=0,
                total=1,
                params={
                    "project_id": project_id,
                    "user_id": user_id,
                    "prompt": image_validated.prompt,
                    "image_type": image_validated.image_type,
                    "reference_image": image_validated.reference_image,
                    "human_id": image_validated.human_id,
                },
            )
            await registry.add_running(task_id)

        elif task_type == "script":
            script_validated = ScriptParams(**request.params)
            await registry.update(
                task_id,
                status="running",
                progress=0,
                message="等待执行...",
                params={
                    "category": script_validated.category,
                    "subcategory": script_validated.subcategory,
                    "style": script_validated.style,
                    "duration": script_validated.duration,
                },
            )
            await registry.add_running(task_id)

        elif task_type == "subtitle":
            subtitle_validated = SubtitleParams(**request.params)
            await registry.update(
                task_id,
                status="running",
                message="准备字幕生成...",
                completed=0,
                total=1,
                params={
                    "project_id": project_id,
                    "video_path": subtitle_validated.video_path,
                    "language": subtitle_validated.language,
                    "mode": subtitle_validated.mode,
                    "audio_text": subtitle_validated.audio_text,
                },
            )
            await registry.add_running(task_id)

        elif task_type == "copy":
            copy_validated = CopyParams(**request.params)
            await registry.update(
                task_id,
                status="running",
                message="准备提取文案...",
                completed=0,
                total=1,
                params={"video_url": copy_validated.video_url},
            )
            await registry.add_running(task_id)

        elif task_type == "avatar_clone":
            name = request.params.get("name", "").strip()
            video_url = request.params.get("video_url", "").strip()
            if not name:
                raise ValueError("name 不能为空")
            if not video_url:
                raise ValueError("video_url 不能为空")
            if not video_url.startswith(("http://", "https://")):
                raise ValueError("video_url 必须是有效的URL")

            avatar_id = f"avt_{uuid.uuid4().hex[:16]}"
            now = datetime.now(UTC).isoformat()

            # avatar_clone 使用自己的 task_id（avt_xxx），不走通用的 task_xxx
            await registry.create(avatar_id, "avatar_clone", user_id)
            await registry.update(
                avatar_id,
                status="running",
                progress=5,
                message="开始形象克隆...",
                completed=0,
                total=1,
                params={
                    "avatar_id": avatar_id,
                    "name": name,
                    "video_url": video_url,
                    "user_id": user_id,
                },
                avatar_status=AvatarCloneStatus.PENDING.value,
                avatar_name=name,
                avatar_video_url=video_url,
                voice_id="",
                provider_element_id="",
                provider_voice_job_id="",
                provider_element_job_id="",
                trial_url="",
                fail_reason="",
                created_at=now,
                updated_at=now,
            )
            await registry.add_running(avatar_id)
            # 返回的任务 ID 用 avatar_id，保持前端兼容
            task_id = avatar_id

        elif task_type == "tts":
            tts_params = dict(request.params)
            raw_segments = tts_params.get("segments") or tts_params.get("texts", [])
            if isinstance(raw_segments, list):
                normalized = []
                for i, seg in enumerate(raw_segments):
                    if isinstance(seg, dict):
                        normalized.append({
                            "id": seg.get("id", f"tts_{i}"),
                            "text": seg.get("text") or seg.get("voiceover", ""),
                            "index": i,
                        })
                    else:
                        normalized.append({
                            "id": f"tts_{i}",
                            "text": str(seg),
                            "index": i,
                        })
                tts_params["segments"] = normalized
            else:
                raise ValueError("segments 必须为列表")

            await registry.update(
                task_id,
                status="running",
                message="准备语音合成...",
                completed=0,
                total=len(tts_params.get("segments", [])),
                params={
                    "project_id": project_id,
                    "user_id": user_id,
                    "segments": json.dumps(tts_params.get("segments", []), ensure_ascii=False),
                    "voice_id": tts_params.get("voice_id", "zh_female_yizhi"),
                    "speed": tts_params.get("speed", 1.0),
                },
            )
            await registry.add_running(task_id)

        else:
            raise HTTPException(status_code=400, detail=f"不支持的任务类型: {task_type}")

        logger.info(f"[API] Task created: {task_id}, type={task_type}, user={user_id}")
        return TaskCreateResponse(
            task_id=task_id,
            status="pending",
            message=f"{task_type} 任务已创建",
        )

    except ValueError as e:
        logger.warning(f"[API] Invalid params for {task_type}: {e}")
        try:
            await registry.update(task_id, status="failed", message=f"参数错误: {e}", error=str(e))
        except Exception as registry_err:
            logger.warning(f"[API] Failed to update registry for {task_id}: {registry_err}")
        raise HTTPException(status_code=422, detail=f"参数错误: {e}")

    except HTTPException:
        raise

    except Exception as e:
        logger.exception(f"[API] Failed to create task: {e}")
        try:
            await registry.update(task_id, status="failed", message=str(e), error=str(e))
        except Exception as registry_err:
            logger.warning(f"[API] Failed to update registry for {task_id}: {registry_err}")
        raise HTTPException(status_code=500, detail=f"创建任务失败: {str(e)}")


def _map_avatar_status(status: str) -> str:
    """将 AvatarCloneStatus 映射为统一任务状态"""
    mapping = {
        "succeed": "completed",
        "voice_failed": "failed",
        "element_failed": "failed",
        "timeout": "failed",
        "pending": "running",
        "voice_processing": "running",
        "element_pending": "running",
        "element_processing": "running",
    }
    return mapping.get(status, "running")


@router.get("", response_model=list[TaskStatusResponse])
async def list_tasks(
    project_id: str | None = None,
    current_user: User = Depends(get_current_user),
) -> list[TaskStatusResponse]:
    """
    查询当前用户所有进行中的任务

    从 Redis running 集合读取真实状态，支持按 project_id 过滤。
    """
    redis = get_redis_client()
    registry = JobRegistry(redis)

    try:
        jobs = await registry.list_running_by_user(str(current_user.id))
    except Exception as e:
        logger.error(f"[API] Redis error when listing tasks: {e}")
        raise HTTPException(status_code=503, detail="服务暂时不可用，请稍后重试")

    results: list[TaskStatusResponse] = []
    for job in jobs:
        # 按 project_id 过滤
        if project_id and job.project_id != project_id:
            continue
        results.append(
            TaskStatusResponse(
                task_id=job.job_id,
                type=job.job_type,
                status=job.status,
                progress=job.progress,
                message=job.message,
                completed=job.completed,
                total=job.total,
                result=None,  # 列表查询不返回 result，避免数据过大
                error=job.error,
                created_at=job.created_at,
            )
        )
    return results


@router.get("/{task_id}", response_model=TaskStatusResponse)
async def get_task_status(
    task_id: str,
    current_user: User = Depends(get_current_user),
) -> TaskStatusResponse:
    """
    查询任务状态

    前端通过轮询此接口获取任务进度。
    任务状态仅从 Redis 查询，记录过期后返回 404。
    """
    redis = get_redis_client()
    registry = JobRegistry(redis)

    try:
        job = await registry.get(task_id)
    except Exception as e:
        logger.error(f"[API] Redis error when getting task {task_id}: {e}")
        raise HTTPException(status_code=503, detail="服务暂时不可用，请稍后重试")

    if not job:
        raise HTTPException(status_code=404, detail="任务不存在或已过期")

    # 权限检查
    if job.user_id != str(current_user.id):
        raise HTTPException(status_code=403, detail="无权访问此任务")

    return TaskStatusResponse(
        task_id=task_id,
        type=job.job_type,
        status=job.status,
        progress=job.progress,
        message=job.message,
        completed=job.completed,
        total=job.total,
        result=job.result,
        error=job.error,
        created_at=job.created_at,
    )


@router.get("/{task_id}/result")
async def get_task_result(
    task_id: str,
    current_user: User = Depends(get_current_user),
) -> dict:
    """
    获取任务结果（简化接口，直接返回 result 字段）
    """
    redis = get_redis_client()
    registry = JobRegistry(redis)

    try:
        job = await registry.get(task_id)
    except Exception as e:
        logger.error(f"[API] Redis error when getting result {task_id}: {e}")
        raise HTTPException(status_code=503, detail="服务暂时不可用，请稍后重试")

    if not job:
        raise HTTPException(status_code=404, detail="任务不存在或已过期")

    if job.user_id != str(current_user.id):
        raise HTTPException(status_code=403, detail="无权访问此任务")

    if job.status != "completed":
        raise HTTPException(status_code=400, detail=f"任务未完成，当前状态: {job.status}")

    return job.result or {}