more_agent/app/models.py

"""
MACP 数据模型定义模块。

定义了平台所有核心数据实体的 Pydantic 模型，涵盖文档模板、文档、
GJB规则、验证报告、Skill、记忆条目、知识库片段、模型配置、多模态处理结果等。
"""

from __future__ import annotations

from datetime import datetime
from enum import Enum
from typing import Any, Optional

from pydantic import BaseModel, Field


# =============================================================================
# 枚举类型
# =============================================================================

class TemplateType(str, Enum):
    """模板类型枚举"""
    SRS = "需求规格说明书"
    SDD = "软件设计说明书"
    STP = "软件测试计划"
    STR = "软件测试报告"
    SUM = "软件用户手册"
    OTHER = "其他"


class GjbStandard(str, Enum):
    """GJB标准枚举"""
    GJB438C = "GJB438C"
    GJB437B = "GJB437B"
    GJB2786A = "GJB2786A"


class DocumentStatus(str, Enum):
    """文档状态枚举"""
    DRAFT = "草稿"
    SUBMITTED = "已提交"
    REVIEWED = "已审核"
    APPROVED = "已批准"


class MemoryType(str, Enum):
    """记忆类型枚举"""
    SHORT_TERM = "短期"
    LONG_TERM = "长期"


class KnowledgeLevel(str, Enum):
    """知识库层级枚举"""
    DOCUMENT = "文档"
    CHAPTER = "章节"
    PARAGRAPH = "段落"


class ModelStatus(str, Enum):
    """模型调用状态枚举"""
    ACTIVE = "主用"
    FAILOVER = "故障转移中"
    UNAVAILABLE = "不可用"


class SystemStatus(str, Enum):
    """系统运行状态枚举"""
    NORMAL = "正常运行状态"
    MAINTENANCE = "维护状态"


class EditMode(str, Enum):
    """文档编辑模式枚举"""
    MANUAL = "人工编辑"
    AGENT = "Agent对话编辑"


class FileType(str, Enum):
    """文件类型枚举"""
    IMAGE = "图片"
    AUDIO = "音频"
    VIDEO = "视频"


# =============================================================================
# 核心数据实体
# =============================================================================

class Template(BaseModel):
    """文档模板实体

    包含模板ID、名称、类型、所属GJB标准、版本号、内容结构、创建时间、作者。
    """
    id: str = Field(..., description="模板ID")
    name: str = Field(..., description="模板名称")
    template_type: TemplateType = Field(..., description="模板类型")
    gjb_standard: GjbStandard = Field(..., description="所属GJB标准")
    version: str = Field("1.0", description="版本号")
    content_structure: dict[str, Any] = Field(default_factory=dict, description="内容结构")
    created_at: datetime = Field(default_factory=datetime.now, description="创建时间")
    author: str = Field("admin", description="作者")


class Document(BaseModel):
    """文档实体

    包含文档ID、标题、关联模板ID、当前版本、编辑历史、状态、权限信息。
    """
    id: str = Field(..., description="文档ID")
    title: str = Field(..., description="文档标题")
    template_id: str = Field(..., description="关联模板ID")
    current_version: str = Field("1.0", description="当前版本")
    content: str = Field("", description="文档内容")
    edit_history: list[dict[str, Any]] = Field(default_factory=list, description="编辑历史")
    status: DocumentStatus = Field(DocumentStatus.DRAFT, description="文档状态")
    permissions: dict[str, list[str]] = Field(default_factory=dict, description="权限信息")
    created_at: datetime = Field(default_factory=datetime.now, description="创建时间")
    updated_at: datetime = Field(default_factory=datetime.now, description="更新时间")


class GjbRule(BaseModel):
    """GJB规则实体

    包含规则ID、对应章节、要求描述、是否必选、适用文档类型、修正建议。
    """
    id: str = Field(..., description="规则ID")
    section: str = Field(..., description="对应章节")
    requirement: str = Field(..., description="要求描述")
    mandatory: bool = Field(True, description="是否必选")
    applicable_types: list[TemplateType] = Field(..., description="适用文档类型")
    suggestion: str = Field("", description="修正建议")


class ValidationReport(BaseModel):
    """验证报告实体

    包含报告ID、文档ID、验证时间、合规项数量、不合规项列表。
    """
    id: str = Field(..., description="报告ID")
    document_id: str = Field(..., description="文档ID")
    validated_at: datetime = Field(default_factory=datetime.now, description="验证时间")
    total_rules: int = Field(0, description="总规则数")
    compliant_count: int = Field(0, description="合规项数量")
    non_compliant_items: list[dict[str, Any]] = Field(default_factory=list, description="不合规项列表")


class Skill(BaseModel):
    """Skill实体

    包含Skill ID、名称、Prompt模板、关联工具列表、输入输出定义、编排顺序。
    """
    id: str = Field(..., description="Skill ID")
    name: str = Field(..., description="名称")
    prompt_template: str = Field("", description="Prompt模板")
    associated_tools: list[str] = Field(default_factory=list, description="关联工具列表")
    input_definition: dict[str, Any] = Field(default_factory=dict, description="输入定义")
    output_definition: dict[str, Any] = Field(default_factory=dict, description="输出定义")
    orchestration_order: int = Field(0, description="编排顺序")
    created_at: datetime = Field(default_factory=datetime.now, description="创建时间")


class MemoryEntry(BaseModel):
    """记忆条目实体

    包含记忆ID、会话ID、内容摘要、原始对话、嵌入向量、时间戳、有效期。
    """
    id: str = Field(..., description="记忆ID")
    session_id: str = Field(..., description="会话ID")
    summary: str = Field(..., description="内容摘要")
    original_dialogue: str = Field(..., description="原始对话")
    memory_type: MemoryType = Field(MemoryType.SHORT_TERM, description="记忆类型")
    created_at: datetime = Field(default_factory=datetime.now, description="时间戳")
    expiry_at: Optional[datetime] = Field(None, description="有效期")


class KnowledgeChunk(BaseModel):
    """知识库片段实体

    包含片段ID、来源文档、层级、内容文本、元数据标签、向量表示。
    """
    id: str = Field(..., description="片段ID")
    source_document: str = Field(..., description="来源文档")
    level: KnowledgeLevel = Field(..., description="层级")
    content: str = Field(..., description="内容文本")
    metadata_tags: list[str] = Field(default_factory=list, description="元数据标签")
    embedding: list[float] = Field(default_factory=list, description="向量表示")


class ModelProfile(BaseModel):
    """模型配置实体

    包含模型ID、厂商、API地址、认证密钥、健康状态、权重、是否为主用。
    """
    id: str = Field(..., description="模型ID")
    vendor: str = Field(..., description="厂商")
    api_url: str = Field(..., description="API地址")
    api_key: str = Field("******", description="认证密钥")
    health_status: ModelStatus = Field(ModelStatus.ACTIVE, description="健康状态")
    weight: int = Field(1, description="权重")
    is_primary: bool = Field(False, description="是否为主用")


class MultimodalResult(BaseModel):
    """多模态处理结果实体

    包含源文件ID、文件类型、OCR文本、ASR转录、关键帧图像列表、提取时间。
    """
    id: str = Field(..., description="源文件ID")
    file_type: FileType = Field(..., description="文件类型")
    file_name: str = Field(..., description="文件名")
    ocr_text: Optional[str] = Field(None, description="OCR文本")
    asr_transcript: Optional[str] = Field(None, description="ASR转录")
    key_frames: list[str] = Field(default_factory=list, description="关键帧图像列表")
    extracted_at: datetime = Field(default_factory=datetime.now, description="提取时间")


class McpTool(BaseModel):
    """MCP工具实体

    包含工具ID、名称、JSON Schema定义、描述、状态。
    """
    id: str = Field(..., description="工具ID")
    name: str = Field(..., description="工具名称")
    description: str = Field("", description="工具描述")
    json_schema: dict[str, Any] = Field(default_factory=dict, description="JSON Schema定义")
    enabled: bool = Field(True, description="是否启用")
    created_at: datetime = Field(default_factory=datetime.now, description="创建时间")


class Workflow(BaseModel):
    """工作流实体

    用于Skills流程编排，支持条件分支与循环逻辑。
    """
    id: str = Field(..., description="工作流ID")
    name: str = Field(..., description="工作流名称")
    skill_ids: list[str] = Field(default_factory=list, description="关联Skill ID列表")
    logic: dict[str, Any] = Field(default_factory=dict, description="流程逻辑定义")
    created_at: datetime = Field(default_factory=datetime.now, description="创建时间")