from typing import Any, Literal

from pydantic import BaseModel, Field


class SceneAnalyzeRequest(BaseModel):
    timestamp: int
    audio_blob: str = Field(description="Base64 encoded audio or placeholder")
    device_id: str
    context_hints: dict[str, Any] = Field(default_factory=dict)


class SceneAnalyzeResponse(BaseModel):
    scene: str
    confidence: float
    greeting_strategy: Literal["contextual", "generic"]
    suggested_intent: str


class DialogueStreamPayload(BaseModel):
    dialogue_id: str
    text: str
    audio_url: str | None = None
    intent_type: str
    difficulty_level: str


class MqttHeader(BaseModel):
    version: str | None = None
    device_id: str
    req_id: str | None = None
    timestamp: int
    ttl: int | None = None


class MqttCommand(BaseModel):
    header: MqttHeader
    action: str
    payload: dict[str, Any] = Field(default_factory=dict)


class MqttReport(BaseModel):
    header: MqttHeader
    msg_type: Literal["state", "event", "log", "ack", "progress"]
    payload: dict[str, Any] = Field(default_factory=dict)


class DeviceState(BaseModel):
    device_id: str
    cpu: float | None = None
    mem: float | None = None
    battery: float | None = None
    net: str | None = None
    mode: str | None = None
    last_seen: int


class InitTaskRequest(BaseModel):
    device_id: str
    course_id: str | None = None
    force: bool = False