refactor(ai): 合并ai_document表到vector_store表，切换文档ID类型为字符串
- 删除旧的ai_document表及相关索引，新增vector_store表兼容Spring AI PgVectorStore结构 - 调整实体类AiDocument映射到vector_store表，使用字符串ID代替Long类型 - 修改Mapper接口及XML中所有ID相关SQL使用字符串类型，并替换表名为vector_store - 修改服务接口与实现类，文档ID参数类型统一为字符串 - 处理文档分块时改用UUID生成chunk ID，确保唯一且格式正确 - 禁用Spring Ai PgVectorStore的自动schema初始化，使用手动创建的表结构 - 更新配置文件OpenAI模型API key及基础URL配置，支持多模型与聊天功能 - 优化日志输出，增加分块文档ID和父文档ID显示，方便调试追踪
This commit is contained in:
@@ -8,6 +8,122 @@
|
||||
CREATE EXTENSION IF NOT EXISTS vector;
|
||||
CREATE EXTENSION IF NOT EXISTS "uuid-ossp";
|
||||
|
||||
-- =====================================================
|
||||
-- Spring AI PgVectorStore 向量存储表
|
||||
-- 用于RAG文档向量存储和相似度搜索
|
||||
-- 兼容PgVectorStore默认结构，同时支持完整文档管理
|
||||
-- =====================================================
|
||||
DROP TABLE IF EXISTS vector_store CASCADE;
|
||||
CREATE TABLE vector_store (
|
||||
-- PgVectorStore 核心字段
|
||||
id VARCHAR(255) PRIMARY KEY,
|
||||
content TEXT,
|
||||
metadata JSONB,
|
||||
embedding vector(1536),
|
||||
|
||||
-- 关联关系
|
||||
project_id BIGINT,
|
||||
timeline_node_id BIGINT,
|
||||
kb_id BIGINT,
|
||||
|
||||
-- 文档来源
|
||||
source_type VARCHAR(50),
|
||||
source_id BIGINT,
|
||||
|
||||
-- 文档扩展信息
|
||||
title VARCHAR(500),
|
||||
content_raw TEXT,
|
||||
summary TEXT,
|
||||
|
||||
-- 文档元数据
|
||||
doc_type VARCHAR(50),
|
||||
language VARCHAR(10) DEFAULT 'zh',
|
||||
file_type VARCHAR(50),
|
||||
file_size BIGINT,
|
||||
file_path VARCHAR(500),
|
||||
|
||||
-- 时间信息 (用于时间维度检索)
|
||||
doc_date DATE,
|
||||
doc_datetime TIMESTAMP,
|
||||
|
||||
-- 分块信息(大文档分块存储)
|
||||
chunk_index INT DEFAULT 0,
|
||||
chunk_total INT DEFAULT 1,
|
||||
chunk_parent_id VARCHAR(255),
|
||||
|
||||
-- 标签和分类
|
||||
tags JSONB,
|
||||
category VARCHAR(100),
|
||||
|
||||
-- 使用统计
|
||||
view_count INT DEFAULT 0,
|
||||
query_count INT DEFAULT 0,
|
||||
last_queried_at TIMESTAMP,
|
||||
|
||||
-- 状态
|
||||
status VARCHAR(20) DEFAULT 'active',
|
||||
error_message TEXT,
|
||||
|
||||
-- 创建信息
|
||||
create_by BIGINT,
|
||||
create_time TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
|
||||
update_by BIGINT,
|
||||
update_time TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
|
||||
deleted SMALLINT DEFAULT 0,
|
||||
|
||||
-- 外键约束
|
||||
CONSTRAINT fk_vs_project FOREIGN KEY (project_id) REFERENCES project(id) ON DELETE SET NULL,
|
||||
CONSTRAINT fk_vs_timeline FOREIGN KEY (timeline_node_id) REFERENCES project_timeline(id) ON DELETE SET NULL,
|
||||
CONSTRAINT fk_vs_kb FOREIGN KEY (kb_id) REFERENCES ai_knowledge_base(id) ON DELETE SET NULL,
|
||||
CONSTRAINT fk_vs_create_by FOREIGN KEY (create_by) REFERENCES sys_user(id) ON DELETE SET NULL
|
||||
);
|
||||
|
||||
-- 创建向量索引（使用IVFFlat算法，适合中等数据量）
|
||||
CREATE INDEX idx_vector_store_embedding ON vector_store
|
||||
USING ivfflat (embedding vector_cosine_ops)
|
||||
WITH (lists = 100);
|
||||
|
||||
-- 创建其他常用索引
|
||||
CREATE INDEX idx_vs_project ON vector_store(project_id) WHERE deleted = 0;
|
||||
CREATE INDEX idx_vs_timeline ON vector_store(timeline_node_id) WHERE deleted = 0;
|
||||
CREATE INDEX idx_vs_kb ON vector_store(kb_id) WHERE deleted = 0;
|
||||
CREATE INDEX idx_vs_source ON vector_store(source_type, source_id) WHERE deleted = 0;
|
||||
CREATE INDEX idx_vs_status ON vector_store(status);
|
||||
CREATE INDEX idx_vs_type ON vector_store(doc_type);
|
||||
CREATE INDEX idx_vs_tags ON vector_store USING GIN(tags);
|
||||
CREATE INDEX idx_vs_chunk_parent ON vector_store(chunk_parent_id) WHERE chunk_parent_id IS NOT NULL;
|
||||
|
||||
COMMENT ON TABLE vector_store IS '向量存储表 - 用于RAG文档检索和知识库管理';
|
||||
COMMENT ON COLUMN vector_store.id IS '文档ID（字符串类型，兼容PgVectorStore）';
|
||||
COMMENT ON COLUMN vector_store.content IS '文档内容文本';
|
||||
COMMENT ON COLUMN vector_store.metadata IS '文档元数据（JSONB格式，兼容PgVectorStore）';
|
||||
COMMENT ON COLUMN vector_store.embedding IS '向量嵌入（1536维）';
|
||||
COMMENT ON COLUMN vector_store.project_id IS '关联项目ID';
|
||||
COMMENT ON COLUMN vector_store.timeline_node_id IS '关联时间节点ID';
|
||||
COMMENT ON COLUMN vector_store.kb_id IS '关联知识库ID';
|
||||
COMMENT ON COLUMN vector_store.source_type IS '来源类型: project-项目文档, risk-风险文档, ticket-工单, report-日报, upload-上传文件, knowledge-知识库, chat-对话记录';
|
||||
COMMENT ON COLUMN vector_store.source_id IS '来源记录ID';
|
||||
COMMENT ON COLUMN vector_store.title IS '文档标题';
|
||||
COMMENT ON COLUMN vector_store.content_raw IS '原始内容(带格式)';
|
||||
COMMENT ON COLUMN vector_store.summary IS 'AI生成的摘要';
|
||||
COMMENT ON COLUMN vector_store.doc_type IS '文档类型: requirement-需求, design-设计, plan-计划, report-报告, contract-合同, photo-照片, other-其他';
|
||||
COMMENT ON COLUMN vector_store.language IS '语言: zh-中文, en-英文';
|
||||
COMMENT ON COLUMN vector_store.file_type IS '文件类型: pdf, doc, txt, md, jpg, png等';
|
||||
COMMENT ON COLUMN vector_store.file_size IS '文件大小(字节)';
|
||||
COMMENT ON COLUMN vector_store.file_path IS '文件存储路径';
|
||||
COMMENT ON COLUMN vector_store.doc_date IS '文档日期(如日报日期、照片拍摄日期)';
|
||||
COMMENT ON COLUMN vector_store.doc_datetime IS '文档时间戳';
|
||||
COMMENT ON COLUMN vector_store.chunk_index IS '分块序号';
|
||||
COMMENT ON COLUMN vector_store.chunk_total IS '总分块数';
|
||||
COMMENT ON COLUMN vector_store.chunk_parent_id IS '父文档ID(分块时使用)';
|
||||
COMMENT ON COLUMN vector_store.tags IS '标签数组';
|
||||
COMMENT ON COLUMN vector_store.category IS '分类';
|
||||
COMMENT ON COLUMN vector_store.view_count IS '查看次数';
|
||||
COMMENT ON COLUMN vector_store.query_count IS '被检索次数';
|
||||
COMMENT ON COLUMN vector_store.last_queried_at IS '最后被检索时间';
|
||||
COMMENT ON COLUMN vector_store.status IS '状态: active-可用, processing-处理中, error-错误, archived-归档';
|
||||
COMMENT ON COLUMN vector_store.error_message IS '错误信息';
|
||||
|
||||
-- 设置时区
|
||||
SET timezone = 'Asia/Shanghai';
|
||||
|
||||
@@ -969,118 +1085,8 @@ COMMENT ON COLUMN ai_knowledge_base.status IS '状态: active-可用, archived-
|
||||
-- 11. AI服务相关表
|
||||
-- =====================================================
|
||||
|
||||
-- AI文档向量表 (用于RAG知识库)
|
||||
DROP TABLE IF EXISTS ai_document CASCADE;
|
||||
CREATE TABLE ai_document (
|
||||
id BIGSERIAL PRIMARY KEY,
|
||||
doc_id UUID DEFAULT uuid_generate_v4(),
|
||||
|
||||
-- 关联关系
|
||||
project_id BIGINT,
|
||||
timeline_node_id BIGINT,
|
||||
kb_id BIGINT,
|
||||
-- 文档来源
|
||||
source_type VARCHAR(50) NOT NULL,
|
||||
source_id BIGINT,
|
||||
|
||||
-- 文档内容
|
||||
title VARCHAR(500),
|
||||
content TEXT NOT NULL,
|
||||
content_raw TEXT,
|
||||
summary TEXT,
|
||||
|
||||
-- 向量嵌入 (1536维适配OpenAI, 可调整为其他维度)
|
||||
embedding vector(1536),
|
||||
|
||||
-- 文档元数据
|
||||
doc_type VARCHAR(50),
|
||||
language VARCHAR(10) DEFAULT 'zh',
|
||||
file_type VARCHAR(50),
|
||||
file_size BIGINT,
|
||||
file_path VARCHAR(500),
|
||||
|
||||
-- 时间信息 (用于时间维度检索)
|
||||
doc_date DATE,
|
||||
doc_datetime TIMESTAMP,
|
||||
|
||||
-- 分块信息(大文档分块存储)
|
||||
chunk_index INT DEFAULT 0,
|
||||
chunk_total INT DEFAULT 1,
|
||||
chunk_parent_id BIGINT,
|
||||
|
||||
-- 标签和分类
|
||||
tags JSONB,
|
||||
category VARCHAR(100),
|
||||
|
||||
-- 使用统计
|
||||
view_count INT DEFAULT 0,
|
||||
query_count INT DEFAULT 0,
|
||||
last_queried_at TIMESTAMP,
|
||||
|
||||
-- 状态
|
||||
status VARCHAR(20) DEFAULT 'active',
|
||||
error_message TEXT,
|
||||
|
||||
-- 创建信息
|
||||
create_by BIGINT,
|
||||
create_time TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
|
||||
update_by BIGINT,
|
||||
update_time TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
|
||||
deleted SMALLINT DEFAULT 0,
|
||||
|
||||
-- 外键约束
|
||||
CONSTRAINT fk_ai_doc_project FOREIGN KEY (project_id) REFERENCES project(id) ON DELETE SET NULL,
|
||||
CONSTRAINT fk_ai_doc_timeline FOREIGN KEY (timeline_node_id) REFERENCES project_timeline(id) ON DELETE SET NULL,
|
||||
CONSTRAINT fk_ai_doc_kb FOREIGN KEY (kb_id) REFERENCES ai_knowledge_base(id) ON DELETE SET NULL
|
||||
);
|
||||
|
||||
-- 创建向量索引 (使用IVFFlat或HNSW)
|
||||
-- IVFFlat: 适合中等数据量, 内存占用小
|
||||
-- HNSW: 适合大数据量, 查询更快但内存占用大
|
||||
CREATE INDEX idx_ai_document_embedding ON ai_document
|
||||
USING ivfflat (embedding vector_cosine_ops)
|
||||
WITH (lists = 100);
|
||||
|
||||
-- 创建其他常用索引
|
||||
CREATE INDEX idx_ai_doc_project ON ai_document(project_id) WHERE deleted = 0;
|
||||
CREATE INDEX idx_ai_doc_timeline ON ai_document(timeline_node_id) WHERE deleted = 0;
|
||||
CREATE INDEX idx_ai_doc_kb ON ai_document(kb_id) WHERE deleted = 0;
|
||||
CREATE INDEX idx_ai_doc_source ON ai_document(source_type, source_id) WHERE deleted = 0;
|
||||
CREATE INDEX idx_ai_doc_status ON ai_document(status);
|
||||
CREATE INDEX idx_ai_doc_type ON ai_document(doc_type);
|
||||
CREATE INDEX idx_ai_doc_tags ON ai_document USING GIN(tags);
|
||||
|
||||
COMMENT ON TABLE ai_document IS 'AI文档向量表 - 存储所有用于RAG的文档向量';
|
||||
COMMENT ON COLUMN ai_document.doc_id IS '文档唯一标识';
|
||||
COMMENT ON COLUMN ai_document.project_id IS '关联项目ID';
|
||||
COMMENT ON COLUMN ai_document.timeline_node_id IS '关联时间节点ID';
|
||||
COMMENT ON COLUMN ai_document.kb_id IS '关联知识库ID';
|
||||
COMMENT ON COLUMN ai_document.source_type IS '来源类型: project-项目文档, risk-风险文档, ticket-工单, report-日报, upload-上传文件, knowledge-知识库, chat-对话记录';
|
||||
COMMENT ON COLUMN ai_document.source_id IS '来源记录ID';
|
||||
COMMENT ON COLUMN ai_document.title IS '文档标题';
|
||||
COMMENT ON COLUMN ai_document.content IS '文档内容(纯文本)';
|
||||
COMMENT ON COLUMN ai_document.content_raw IS '原始内容(带格式)';
|
||||
COMMENT ON COLUMN ai_document.summary IS 'AI生成的摘要';
|
||||
COMMENT ON COLUMN ai_document.embedding IS '向量嵌入';
|
||||
COMMENT ON COLUMN ai_document.doc_type IS '文档类型: requirement-需求, design-设计, plan-计划, report-报告, contract-合同, photo-照片, other-其他';
|
||||
COMMENT ON COLUMN ai_document.language IS '语言: zh-中文, en-英文';
|
||||
COMMENT ON COLUMN ai_document.file_type IS '文件类型: pdf, doc, txt, md, jpg, png等';
|
||||
COMMENT ON COLUMN ai_document.file_size IS '文件大小(字节)';
|
||||
COMMENT ON COLUMN ai_document.file_path IS '文件存储路径';
|
||||
COMMENT ON COLUMN ai_document.doc_date IS '文档日期(如日报日期、照片拍摄日期)';
|
||||
COMMENT ON COLUMN ai_document.doc_datetime IS '文档时间戳';
|
||||
COMMENT ON COLUMN ai_document.chunk_index IS '分块序号';
|
||||
COMMENT ON COLUMN ai_document.chunk_total IS '总分块数';
|
||||
COMMENT ON COLUMN ai_document.chunk_parent_id IS '父文档ID(分块时使用)';
|
||||
COMMENT ON COLUMN ai_document.tags IS '标签数组';
|
||||
COMMENT ON COLUMN ai_document.category IS '分类';
|
||||
COMMENT ON COLUMN ai_document.view_count IS '查看次数';
|
||||
COMMENT ON COLUMN ai_document.query_count IS '被检索次数';
|
||||
COMMENT ON COLUMN ai_document.last_queried_at IS '最后被检索时间';
|
||||
COMMENT ON COLUMN ai_document.status IS '状态: active-可用, processing-处理中, error-错误, archived-归档';
|
||||
COMMENT ON COLUMN ai_document.error_message IS '错误信息';
|
||||
|
||||
-- AI对话记录表 (合并会话管理功能，无需单独的session表)
|
||||
-- 注：ai_document表已合并到vector_store表中
|
||||
DROP TABLE IF EXISTS ai_chat_history CASCADE;
|
||||
CREATE TABLE ai_chat_history (
|
||||
id BIGSERIAL PRIMARY KEY,
|
||||
|
||||
Reference in New Issue
Block a user