Agent/examples/demand/pattern_builds/models2.py

"""Pattern Mining 数据模型"""
from sqlalchemy import Column, BigInteger, Integer, String, Float, Boolean, Text, DateTime, JSON, Index
from sqlalchemy.ext.declarative import declarative_base

Base = declarative_base()


class TopicPatternExecution(Base):
    __tablename__ = 'topic_pattern_execution'

    id = Column(BigInteger, primary_key=True, autoincrement=True)
    # 筛选条件快照
    cluster_name = Column(String(32), nullable=True)
    merge_leve2 = Column(String(100), nullable=True)
    platform = Column(String(50), nullable=True)
    account_name = Column(String(200), nullable=True)
    post_limit = Column(Integer, nullable=False, default=500)
    # 挖掘参数
    min_absolute_support = Column(Integer, nullable=False, default=2)
    classify_execution_id = Column(BigInteger, nullable=True)
    mining_configs = Column(JSON, nullable=True)  # [{dimension_mode, target_depths: [...]}, ...]
    # 结果摘要
    post_count = Column(Integer, nullable=True)
    itemset_count = Column(Integer, nullable=True)  # 所有 config 的项集总数
    # 状态
    status = Column(String(50), nullable=False, default='running')
    error_message = Column(Text, nullable=True)
    start_time = Column(DateTime, nullable=True)
    end_time = Column(DateTime, nullable=True)


class TopicPatternMiningConfig(Base):
    """一次执行中的一个 (dimension_mode, target_depth) 挖掘配置"""
    __tablename__ = 'topic_pattern_mining_config'

    id = Column(BigInteger, primary_key=True, autoincrement=True)
    execution_id = Column(BigInteger, nullable=False, index=True)
    dimension_mode = Column(String(50), nullable=False)
    target_depth = Column(String(50), nullable=False)
    # 结果摘要
    transaction_count = Column(Integer, nullable=True)
    itemset_count = Column(Integer, nullable=True)


class TopicPatternCategory(Base):
    """分类树节点快照 - 记录执行时的分类结构"""
    __tablename__ = 'topic_pattern_category'

    id = Column(BigInteger, primary_key=True, autoincrement=True)
    execution_id = Column(BigInteger, nullable=False, index=True)
    source_stable_id = Column(BigInteger, nullable=True)  # 原 GlobalCategory.stable_id
    source_type = Column(String(50), nullable=False)      # 实质/形式/意图
    name = Column(String(500), nullable=False)
    description = Column(Text, nullable=True)
    category_nature = Column(String(50), nullable=True)    # 内容/维度/领域
    path = Column(String(1000), nullable=True)             # 完整路径 如 /食品/水果
    level = Column(Integer, nullable=True)
    parent_id = Column(BigInteger, nullable=True)          # FK → 本表 id（快照内的父节点）
    parent_source_stable_id = Column(BigInteger, nullable=True)  # 原 parent_stable_id（方便建树）
    element_count = Column(Integer, default=0)             # 该分类下直属元素数


class TopicPatternElement(Base):
    """帖子级元素记录 - 每行 = 一个帖子的一个点下的一个元素，代替 data_cache JSON"""
    __tablename__ = 'topic_pattern_element'

    id = Column(BigInteger, primary_key=True, autoincrement=True)
    execution_id = Column(BigInteger, nullable=False, index=True)
    # 帖子 & 选题点上下文
    post_id = Column(String(100), nullable=False, index=True)
    point_type = Column(String(50), nullable=False)         # 灵感点/目的点/关键点
    point_text = Column(String(500), nullable=True)         # 选题点内容（"点" 字段）
    # 元素信息
    element_type = Column(String(50), nullable=False)       # 实质/形式/意图
    name = Column(String(500), nullable=False)
    description = Column(Text, nullable=True)
    # 分类关联
    category_id = Column(BigInteger, nullable=True, index=True)  # FK → TopicPatternCategory.id
    category_path = Column(String(1000), nullable=True)     # 冗余存储分类路径 如 "食品>水果"


class Post(Base):
    """帖子元数据 - 全局通用，跨 execution，通过 post_id 关联筛选"""
    __tablename__ = 'post'

    id = Column(BigInteger, primary_key=True, autoincrement=True)
    post_id = Column(String(100), nullable=False, unique=True, index=True)
    account_name = Column(String(200), nullable=True, index=True)
    merge_leve2 = Column(String(100), nullable=True, index=True)
    platform = Column(String(50), nullable=True, index=True)


class TopicPatternItemset(Base):
    """频繁项集"""
    __tablename__ = 'topic_pattern_itemset'

    id = Column(BigInteger, primary_key=True, autoincrement=True)
    execution_id = Column(BigInteger, nullable=False, index=True)
    mining_config_id = Column(BigInteger, nullable=False, index=True)  # FK → TopicPatternMiningConfig.id
    combination_type = Column(String(200), nullable=False, index=True)
    item_count = Column(Integer, nullable=False, index=True)
    support = Column(Float, nullable=False)
    absolute_support = Column(Integer, nullable=False, index=True)
    dimensions = Column(JSON, nullable=True)        # ["实质","形式"]
    is_cross_point = Column(Boolean, nullable=False, default=False)
    matched_post_ids = Column(JSON, nullable=True)  # ["post_id_1", "post_id_2", ...] 直接存 post_id


class TopicPatternItemsetItem(Base):
    """项集中的单个 item - 关联到分类节点"""
    __tablename__ = 'topic_pattern_itemset_item'

    id = Column(BigInteger, primary_key=True, autoincrement=True)
    itemset_id = Column(BigInteger, nullable=False, index=True)  # FK → TopicPatternItemset.id
    point_type = Column(String(50), nullable=True)      # 灵感点/目的点/关键点（substance_form_only 模式下为空）
    dimension = Column(String(50), nullable=True)       # 实质/形式/意图（point_type_only 模式下为空）
    category_id = Column(BigInteger, nullable=True)     # FK → TopicPatternCategory.id
    category_path = Column(String(1000), nullable=True) # 分类路径部分 如 "食品>水果"
    element_name = Column(String(500), nullable=True)   # 若为 name 层 item（含||），提取的元素名称