feat: 实现 BM25 稀疏 + 稠密向量混合检索功能
Some checks failed
构建并部署 AI Agent 服务 / deploy (push) Has been cancelled
Some checks failed
构建并部署 AI Agent 服务 / deploy (push) Has been cancelled
This commit is contained in:
22
tools/download_bm25.py
Normal file
22
tools/download_bm25.py
Normal file
@@ -0,0 +1,22 @@
|
||||
"""
|
||||
BM25模型预下载脚本
|
||||
执行后将模型缓存到 ./models/fastembed_cache 目录，打包进Docker镜像
|
||||
"""
|
||||
import os
|
||||
from fastembed.sparse.sparse_text_embedding import SparseTextEmbedding
|
||||
|
||||
if __name__ == "__main__":
|
||||
# 指定缓存目录
|
||||
cache_dir = "./models/fastembed_cache"
|
||||
os.makedirs(cache_dir, exist_ok=True)
|
||||
|
||||
print("正在下载BM25稀疏向量模型...")
|
||||
model = SparseTextEmbedding(
|
||||
model_name="Qdrant/bm25",
|
||||
cache_dir=cache_dir
|
||||
)
|
||||
|
||||
# 触发一次推理，确保模型文件完整下载
|
||||
list(model.embed(["init trigger"]))
|
||||
print(f"✅ BM25模型已成功缓存到: {cache_dir}")
|
||||
print("请将该目录提交到项目仓库，打包进Docker镜像")
|
||||
306
tools/test/test_backend.py
Normal file
306
tools/test/test_backend.py
Normal file
@@ -0,0 +1,306 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
完整后端测试 - 验证 Agent 所有功能
|
||||
包括：短期记忆、长期记忆、工具调用、流式对话、历史查询
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
import os
|
||||
import sys
|
||||
import uuid
|
||||
from dotenv import load_dotenv
|
||||
|
||||
# 添加项目根目录和 backend 目录到 Python 路径
|
||||
project_root = os.path.join(os.path.dirname(__file__), "..")
|
||||
backend_dir = os.path.join(project_root, "backend")
|
||||
sys.path.insert(0, project_root)
|
||||
sys.path.insert(0, backend_dir)
|
||||
load_dotenv()
|
||||
|
||||
from backend.app.config import DB_URI
|
||||
from langgraph.checkpoint.postgres.aio import AsyncPostgresSaver
|
||||
from backend.app.agent.agent_service import AIAgentService
|
||||
from backend.app.agent.history import ThreadHistoryService
|
||||
from backend.app.logger import info, warning, error
|
||||
|
||||
# PostgreSQL 连接字符串
|
||||
|
||||
async def print_section(title):
|
||||
"""打印测试区块标题"""
|
||||
print("\n" + "=" * 70)
|
||||
print(f" {title}")
|
||||
print("=" * 70)
|
||||
|
||||
async def test_short_term_memory(agent_service):
|
||||
"""测试短期记忆（同一 thread_id 继续对话）"""
|
||||
await print_section("测试 1: 短期记忆（Short-term Memory）")
|
||||
|
||||
thread_id = str(uuid.uuid4())
|
||||
user_id = "test_user_memory"
|
||||
|
||||
print(f"\n使用 thread_id: {thread_id[:8]}...")
|
||||
print(f"使用 user_id: {user_id}")
|
||||
|
||||
# 第一轮对话
|
||||
print("\n[第一轮] 发送消息: '我叫张三，今年28岁'")
|
||||
result1 = await agent_service.process_message(
|
||||
"我叫张三，今年28岁", thread_id, "local", user_id
|
||||
)
|
||||
print(f"回复: {result1['reply'][:100]}...")
|
||||
|
||||
# 第二轮对话 - 测试记忆
|
||||
print("\n[第二轮] 发送消息: '我叫什么名字？今年多大？'")
|
||||
result2 = await agent_service.process_message(
|
||||
"我叫什么名字？今年多大？", thread_id, "local", user_id
|
||||
)
|
||||
print(f"回复: {result2['reply']}")
|
||||
|
||||
# 验证记忆是否存在
|
||||
if "张三" in result2['reply'] or "28" in result2['reply']:
|
||||
print("\n✅ 短期记忆测试通过！")
|
||||
return True
|
||||
else:
|
||||
print("\n❌ 短期记忆测试失败！")
|
||||
return False
|
||||
|
||||
async def test_tool_calling(agent_service):
|
||||
"""测试工具调用（RAG 搜索）"""
|
||||
await print_section("测试 2: 工具调用（Tool Calling）")
|
||||
|
||||
thread_id = str(uuid.uuid4())
|
||||
user_id = "test_user_tools"
|
||||
|
||||
print(f"\n使用 thread_id: {thread_id[:8]}...")
|
||||
print(f"使用 user_id: {user_id}")
|
||||
|
||||
# 发送需要 RAG 搜索的问题
|
||||
print("\n发送消息: '请告诉我，黄双银在魔王大陆的故事？'")
|
||||
result = await agent_service.process_message(
|
||||
"请告诉我，黄双银在魔王大陆的故事？", thread_id, "local", user_id
|
||||
)
|
||||
print(f"回复: {result['reply'][:200]}...")
|
||||
|
||||
# 检查是否调用了 RAG 工具（回复中会有黄双银相关内容）
|
||||
if "黄双银" in result['reply']:
|
||||
print("\n✅ 工具调用测试通过！")
|
||||
return True
|
||||
else:
|
||||
print("\n⚠️ 工具调用测试结果不确定，需要手动验证")
|
||||
return None
|
||||
|
||||
async def test_streaming(agent_service):
|
||||
"""测试流式对话"""
|
||||
await print_section("测试 3: 流式对话（Streaming）")
|
||||
|
||||
thread_id = str(uuid.uuid4())
|
||||
user_id = "test_user_stream"
|
||||
|
||||
print(f"\n使用 thread_id: {thread_id[:8]}...")
|
||||
print(f"使用 user_id: {user_id}")
|
||||
|
||||
print("\n发送消息: '用100字介绍一下AI人工智能' (流式)...")
|
||||
print("流式输出: ", end="", flush=True)
|
||||
|
||||
full_reply = ""
|
||||
chunk_count = 0
|
||||
|
||||
try:
|
||||
async for chunk in agent_service.process_message_stream(
|
||||
"用100字介绍一下AI人工智能", thread_id, "local", user_id
|
||||
):
|
||||
chunk_count += 1
|
||||
if chunk.get("type") == "llm_token":
|
||||
token = chunk.get("token", "")
|
||||
print(token, end="", flush=True)
|
||||
full_reply += token
|
||||
elif chunk.get("type") == "state_update":
|
||||
pass # 状态更新不显示
|
||||
|
||||
print(f"\n\n共收到 {chunk_count} 个 chunk")
|
||||
print(f"完整回复长度: {len(full_reply)} 字")
|
||||
|
||||
if chunk_count > 0 and len(full_reply) > 10:
|
||||
print("\n✅ 流式对话测试通过！")
|
||||
return True
|
||||
else:
|
||||
print("\n❌ 流式对话测试失败！")
|
||||
return False
|
||||
|
||||
except Exception as e:
|
||||
print(f"\n❌ 流式对话异常: {e}")
|
||||
return False
|
||||
|
||||
async def test_history_service(agent_service, history_service):
|
||||
"""测试历史查询服务"""
|
||||
await print_section("测试 4: 历史查询服务（History Service）")
|
||||
|
||||
user_id = "test_user_history"
|
||||
|
||||
# 先创建几个对话
|
||||
print(f"\n为 user_id={user_id} 创建测试对话...")
|
||||
|
||||
thread_ids = []
|
||||
for i in range(3):
|
||||
thread_id = str(uuid.uuid4())
|
||||
thread_ids.append(thread_id)
|
||||
|
||||
await agent_service.process_message(
|
||||
f"这是第 {i+1} 个测试对话", thread_id, "local", user_id
|
||||
)
|
||||
print(f" 创建线程 {i+1}: {thread_id[:8]}...")
|
||||
|
||||
# 1. 测试获取用户线程列表
|
||||
print("\n[4.1] 测试获取用户线程列表...")
|
||||
threads = await history_service.get_user_threads(user_id, limit=10)
|
||||
print(f" 找到 {len(threads)} 个线程")
|
||||
|
||||
if len(threads) >= 3:
|
||||
print(" ✅ 线程列表查询通过")
|
||||
else:
|
||||
print(" ⚠️ 线程数量少于预期")
|
||||
|
||||
# 2. 测试获取单个线程的消息历史
|
||||
if thread_ids:
|
||||
test_thread_id = thread_ids[0]
|
||||
print(f"\n[4.2] 测试获取线程消息历史 (thread_id={test_thread_id[:8]}...)")
|
||||
messages = await history_service.get_thread_messages(test_thread_id)
|
||||
print(f" 找到 {len(messages)} 条消息")
|
||||
|
||||
if len(messages) >= 2: # 至少有一问一答
|
||||
print(" ✅ 消息历史查询通过")
|
||||
else:
|
||||
print(" ⚠️ 消息数量少于预期")
|
||||
|
||||
# 3. 测试获取线程摘要
|
||||
print(f"\n[4.3] 测试获取线程摘要...")
|
||||
summary = await history_service.get_thread_summary(test_thread_id)
|
||||
print(f" 摘要: {summary.get('summary', '')[:50]}...")
|
||||
print(f" 消息数: {summary.get('message_count', 0)}")
|
||||
|
||||
if summary.get('message_count', 0) > 0:
|
||||
print(" ✅ 线程摘要查询通过")
|
||||
else:
|
||||
print(" ⚠️ 摘要查询结果不确定")
|
||||
|
||||
return len(threads) >= 3
|
||||
|
||||
async def test_long_term_memory(agent_service):
|
||||
"""测试长期记忆（mem0）"""
|
||||
await print_section("测试 5: 长期记忆（Long-term Memory - mem0）")
|
||||
|
||||
thread_id1 = str(uuid.uuid4())
|
||||
thread_id2 = str(uuid.uuid4()) # 不同的线程
|
||||
user_id = "test_user_longterm"
|
||||
|
||||
print(f"\n使用 user_id: {user_id}")
|
||||
print(f"线程 1: {thread_id1[:8]}...")
|
||||
print(f"线程 2: {thread_id2[:8]}...")
|
||||
|
||||
# 在第一个线程中保存信息
|
||||
print("\n[线程 1] 发送消息: '记住，我的宠物名字叫小白，是一只猫'")
|
||||
result1 = await agent_service.process_message(
|
||||
"记住，我的宠物名字叫小白，是一只猫", thread_id1, "local", user_id
|
||||
)
|
||||
print(f"回复: {result1['reply'][:100]}...")
|
||||
|
||||
# 等待一下，让 mem0 保存
|
||||
await asyncio.sleep(1)
|
||||
|
||||
# 在第二个线程中询问（不同的 thread_id）
|
||||
print("\n[线程 2] 发送消息: '我的宠物叫什么名字？是什么动物？'")
|
||||
result2 = await agent_service.process_message(
|
||||
"我的宠物叫什么名字？是什么动物？", thread_id2, "local", user_id
|
||||
)
|
||||
print(f"回复: {result2['reply']}")
|
||||
|
||||
# 验证长期记忆
|
||||
if "小白" in result2['reply'] or "猫" in result2['reply']:
|
||||
print("\n✅ 长期记忆测试通过！")
|
||||
return True
|
||||
else:
|
||||
print("\n⚠️ 长期记忆可能未启用，或需要手动验证")
|
||||
return None
|
||||
|
||||
async def main():
|
||||
"""主测试函数"""
|
||||
print("\n" + "=" * 70)
|
||||
print(" 后端完整功能测试")
|
||||
print("=" * 70)
|
||||
|
||||
results = {}
|
||||
|
||||
try:
|
||||
# 创建数据库连接和服务
|
||||
print("\n正在初始化数据库连接...")
|
||||
async with AsyncPostgresSaver.from_conn_string(DB_URI) as checkpointer:
|
||||
await checkpointer.setup()
|
||||
print("✅ 数据库连接成功")
|
||||
|
||||
# 创建服务实例
|
||||
print("\n正在初始化 Agent 服务...")
|
||||
agent_service = AIAgentService(checkpointer)
|
||||
await agent_service.initialize()
|
||||
print("✅ Agent 服务初始化成功")
|
||||
|
||||
history_service = ThreadHistoryService(checkpointer)
|
||||
print("✅ 历史服务初始化成功")
|
||||
|
||||
print(f"\n可用模型: {list(agent_service.graphs.keys())}")
|
||||
|
||||
# 运行测试
|
||||
results["短期记忆"] = await test_short_term_memory(agent_service)
|
||||
await asyncio.sleep(1)
|
||||
|
||||
results["工具调用"] = await test_tool_calling(agent_service)
|
||||
await asyncio.sleep(1)
|
||||
|
||||
results["流式对话"] = await test_streaming(agent_service)
|
||||
await asyncio.sleep(1)
|
||||
|
||||
results["历史查询"] = await test_history_service(agent_service, history_service)
|
||||
await asyncio.sleep(1)
|
||||
|
||||
results["长期记忆"] = await test_long_term_memory(agent_service)
|
||||
await asyncio.sleep(1)
|
||||
|
||||
# 打印总结
|
||||
await print_section("测试总结")
|
||||
print("\n测试结果:")
|
||||
print("-" * 40)
|
||||
|
||||
pass_count = 0
|
||||
fail_count = 0
|
||||
skip_count = 0
|
||||
|
||||
for test_name, result in results.items():
|
||||
if result is True:
|
||||
status = "✅ 通过"
|
||||
pass_count += 1
|
||||
elif result is False:
|
||||
status = "❌ 失败"
|
||||
fail_count += 1
|
||||
else:
|
||||
status = "⚠️ 待验证"
|
||||
skip_count += 1
|
||||
print(f" {test_name:12s}: {status}")
|
||||
|
||||
print("-" * 40)
|
||||
print(f"总计: {len(results)} 个测试")
|
||||
print(f"通过: {pass_count}, 失败: {fail_count}, 待验证: {skip_count}")
|
||||
|
||||
if fail_count == 0:
|
||||
print("\n🎉 所有核心测试通过！")
|
||||
else:
|
||||
print(f"\n⚠️ 有 {fail_count} 个测试失败")
|
||||
|
||||
except Exception as e:
|
||||
error(f"\n❌ 测试运行异常: {e}")
|
||||
import traceback
|
||||
traceback.print_exc()
|
||||
return 1
|
||||
|
||||
return 0 if fail_count == 0 else 1
|
||||
|
||||
if __name__ == "__main__":
|
||||
exit_code = asyncio.run(main())
|
||||
sys.exit(exit_code)
|
||||
66
tools/test/test_dqrant.py
Normal file
66
tools/test/test_dqrant.py
Normal file
@@ -0,0 +1,66 @@
|
||||
"""检查 Qdrant 中存储的向量质量。"""
|
||||
|
||||
import os
|
||||
import sys
|
||||
import numpy as np
|
||||
from dotenv import load_dotenv
|
||||
from qdrant_client import QdrantClient
|
||||
|
||||
# 添加项目根目录和 backend 目录到 Python 路径
|
||||
project_root = os.path.join(os.path.dirname(__file__), "..")
|
||||
backend_dir = os.path.join(project_root, "backend")
|
||||
sys.path.insert(0, project_root)
|
||||
sys.path.insert(0, backend_dir)
|
||||
load_dotenv()
|
||||
|
||||
from rag_core import LlamaCppEmbedder
|
||||
|
||||
QDRANT_URL = os.getenv("QDRANT_URL", "http://127.0.0.1:6333")
|
||||
QDRANT_API_KEY = os.getenv("QDRANT_API_KEY")
|
||||
COLLECTION_NAME = "rag_documents"
|
||||
|
||||
client = QdrantClient(url=QDRANT_URL, api_key=QDRANT_API_KEY)
|
||||
embedder = LlamaCppEmbedder()
|
||||
|
||||
# 获取样本
|
||||
points, _ = client.scroll(
|
||||
collection_name=COLLECTION_NAME,
|
||||
limit=1,
|
||||
with_vectors=True,
|
||||
with_payload=True,
|
||||
)
|
||||
|
||||
if not points:
|
||||
print(f"集合 '{COLLECTION_NAME}' 为空")
|
||||
exit()
|
||||
|
||||
sample = points[0]
|
||||
raw_vec = sample.vector
|
||||
if isinstance(raw_vec, dict):
|
||||
stored_vec = list(raw_vec.values())[0]
|
||||
elif isinstance(raw_vec, list):
|
||||
stored_vec = raw_vec
|
||||
else:
|
||||
stored_vec = []
|
||||
|
||||
stored_payload = sample.payload or {}
|
||||
stored_text = str(stored_payload.get("page_content", ""))[:200]
|
||||
|
||||
print(f"内容预览:\n{stored_text}...\n")
|
||||
print(f"向量维度: {len(stored_vec)}") # type: ignore
|
||||
print(f"前5个值: {stored_vec[:5]}") # type: ignore
|
||||
print(f"是否全零: {all(v == 0.0 for v in stored_vec)}") # type: ignore
|
||||
|
||||
# 重新编码对比
|
||||
if stored_text:
|
||||
new_vec = embedder.embed_query(stored_text)
|
||||
similarity = np.dot(stored_vec, new_vec) / (np.linalg.norm(stored_vec) * np.linalg.norm(new_vec)) # type: ignore
|
||||
print(f"\n重新编码前5个值: {new_vec[:5]}")
|
||||
print(f"余弦相似度: {similarity:.4f}")
|
||||
|
||||
if similarity < 0.8:
|
||||
print("\n⚠️ 相似度过低，建议删除集合并重建索引")
|
||||
else:
|
||||
print("\n✅ 向量一致")
|
||||
else:
|
||||
print("\n⚠️ 样本无文本内容")
|
||||
69
tools/test/test_frontend.py
Normal file
69
tools/test/test_frontend.py
Normal file
@@ -0,0 +1,69 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
前端快速测试脚本
|
||||
验证前端导入是否正常工作
|
||||
"""
|
||||
|
||||
import sys
|
||||
import os
|
||||
|
||||
# 添加必要的路径
|
||||
project_root = os.path.dirname(os.path.abspath(__file__))
|
||||
frontend_src = os.path.join(project_root, "frontend", "src")
|
||||
backend_dir = os.path.join(project_root, "backend")
|
||||
|
||||
sys.path.insert(0, project_root)
|
||||
sys.path.insert(0, frontend_src)
|
||||
sys.path.insert(0, backend_dir)
|
||||
|
||||
print("=" * 60)
|
||||
print("前端导入测试")
|
||||
print("=" * 60)
|
||||
|
||||
# 测试 1: 直接导入前端模块
|
||||
print("\n[测试 1] 直接导入前端模块...")
|
||||
try:
|
||||
from frontend.src.frontend_main import main
|
||||
print("✅ frontend_main 导入成功")
|
||||
except Exception as e:
|
||||
print(f"❌ 导入失败: {e}")
|
||||
sys.exit(1)
|
||||
|
||||
# 测试 2: 导入配置
|
||||
print("\n[测试 2] 导入配置...")
|
||||
try:
|
||||
from config import config
|
||||
print(f"✅ config 导入成功: page_title={config.page_title}")
|
||||
except Exception as e:
|
||||
print(f"❌ 导入失败: {e}")
|
||||
|
||||
# 测试 3: 导入状态管理
|
||||
print("\n[测试 3] 导入状态管理...")
|
||||
try:
|
||||
from state import AppState
|
||||
print("✅ AppState 导入成功")
|
||||
except Exception as e:
|
||||
print(f"❌ 导入失败: {e}")
|
||||
|
||||
# 测试 4: 导入 API 客户端
|
||||
print("\n[测试 4] 导入 API 客户端...")
|
||||
try:
|
||||
from api_client import api_client
|
||||
print("✅ api_client 导入成功")
|
||||
except Exception as e:
|
||||
print(f"❌ 导入失败: {e}")
|
||||
|
||||
# 测试 5: 导入组件
|
||||
print("\n[测试 5] 导入组件...")
|
||||
try:
|
||||
from components.sidebar import render_sidebar
|
||||
from components.chat_area import render_chat_area
|
||||
from components.info_panel import render_info_panel
|
||||
print("✅ 所有组件导入成功")
|
||||
except Exception as e:
|
||||
print(f"❌ 导入失败: {e}")
|
||||
|
||||
print("\n" + "=" * 60)
|
||||
print("🎉 所有前端导入测试通过！")
|
||||
print("=" * 60)
|
||||
print("\n现在可以使用 ./scripts/start.sh both 启动完整服务")
|
||||
142
tools/test/test_rag.py
Normal file
142
tools/test/test_rag.py
Normal file
@@ -0,0 +1,142 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
RAG 系统使用示例（重构版）
|
||||
|
||||
演示：
|
||||
1. 使用 IndexBuilder 获取父子块检索器
|
||||
2. 创建固定流程的 RAGPipeline（多路改写 → RRF融合 → 重排序 → 返回父文档）
|
||||
3. 将流水线封装为 LangChain 工具，供 Agent 调用
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
import sys
|
||||
import os
|
||||
|
||||
from dotenv import load_dotenv
|
||||
|
||||
# 加载环境变量（Qdrant URL、PostgreSQL 连接等）
|
||||
load_dotenv()
|
||||
|
||||
# 添加项目根目录到路径
|
||||
sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
|
||||
from pydantic import SecretStr
|
||||
from langchain_openai import ChatOpenAI
|
||||
from rag_indexer.index_builder import IndexBuilderConfig
|
||||
from rag_indexer.splitters import SplitterType
|
||||
from backend.app.rag.pipeline import RAGPipeline
|
||||
from backend.app.rag.tools import create_rag_tool_sync
|
||||
from backend.rag_core.retriever_factory import create_parent_retriever
|
||||
|
||||
def create_llm():
|
||||
"""创建本地 vLLM 服务 LLM"""
|
||||
vllm_base_url = os.getenv(
|
||||
"VLLM_BASE_URL",
|
||||
"http://127.0.0.1:8081/v1"
|
||||
)
|
||||
|
||||
return ChatOpenAI(
|
||||
base_url=vllm_base_url,
|
||||
api_key=SecretStr(os.getenv("LLAMACPP_API_KEY", "token-abc123")),
|
||||
model="gemma-4-E2B-it",
|
||||
timeout=60.0, # 请求超时时间（秒）
|
||||
max_retries=2, # 失败后自动重试次数
|
||||
streaming=True, # 确保开启流式输出
|
||||
)
|
||||
|
||||
async def demonstrate_full_pipeline():
|
||||
"""
|
||||
完整流水线演示：
|
||||
- 从 IndexBuilder 获取 ParentDocumentRetriever
|
||||
- 创建 RAGPipeline
|
||||
- 执行检索并打印结果
|
||||
"""
|
||||
print("=" * 60)
|
||||
print("演示：固定流程 RAG 检索（多路改写 + RRF + 重排序 + 父文档）")
|
||||
print("=" * 60)
|
||||
|
||||
retriever = create_parent_retriever(collection_name="rag_documents", search_k=5)
|
||||
|
||||
if retriever is None:
|
||||
print("错误：检索器未初始化，请确保索引已构建。")
|
||||
return
|
||||
|
||||
# 3. 创建 LLM 用于查询改写
|
||||
llm = create_llm()
|
||||
|
||||
# 4. 创建 RAGPipeline（固定流程）
|
||||
pipeline = RAGPipeline(
|
||||
retriever=retriever,
|
||||
llm=llm,
|
||||
num_queries=3, # 生成 3 个查询变体
|
||||
rerank_top_n=5, # 最终返回 5 个父文档
|
||||
)
|
||||
|
||||
# 5. 执行检索
|
||||
query = "打虎英雄是谁？"
|
||||
print(f"\n查询: {query}")
|
||||
print("-" * 40)
|
||||
|
||||
try:
|
||||
documents = await pipeline.aretrieve(query)
|
||||
print(f"返回 {len(documents)} 个父文档\n")
|
||||
|
||||
# 打印结果预览
|
||||
for i, doc in enumerate(documents, 1):
|
||||
content_preview = doc.page_content.replace("\n", " ")[:150]
|
||||
source = doc.metadata.get("source", "未知来源")
|
||||
print(f"{i}. 【来源：{source}】")
|
||||
print(f" {content_preview}...\n")
|
||||
|
||||
# 可选：格式化完整上下文
|
||||
# context = pipeline.format_context(documents)
|
||||
# print(context)
|
||||
|
||||
except Exception as e:
|
||||
print(f"检索失败: {e}")
|
||||
import traceback
|
||||
traceback.print_exc()
|
||||
|
||||
async def demonstrate_tool_creation():
|
||||
"""
|
||||
演示创建 RAG 工具（供 Agent 使用）
|
||||
"""
|
||||
print("\n" + "=" * 60)
|
||||
print("演示：创建 RAG 工具（供 LangGraph Agent 调用）")
|
||||
print("=" * 60)
|
||||
|
||||
# 1. 获取检索器（同上）
|
||||
config = IndexBuilderConfig(
|
||||
collection_name="rag_documents",
|
||||
splitter_type=SplitterType.PARENT_CHILD,
|
||||
)
|
||||
retriever = create_parent_retriever(collection_name="rag_documents", search_k=5)
|
||||
|
||||
# 2. 创建 LLM
|
||||
llm = create_llm()
|
||||
|
||||
# 3. 创建工具
|
||||
rag_tool = create_rag_tool_sync(
|
||||
retriever=retriever,
|
||||
llm=llm,
|
||||
num_queries=3,
|
||||
rerank_top_n=5,
|
||||
collection_name="rag_documents",
|
||||
)
|
||||
|
||||
print(f"工具名称: {rag_tool.name}")
|
||||
print(f"工具描述: {rag_tool.description[:100]}...")
|
||||
|
||||
# 4. 模拟 Agent 调用工具
|
||||
query = "请告诉我 打虎英雄是谁？"
|
||||
print(f"\n模拟调用: {query}")
|
||||
print("-" * 40)
|
||||
|
||||
result = await rag_tool.ainvoke({"query": query})
|
||||
print(result[:800] + "..." if len(result) > 800 else result)
|
||||
|
||||
async def main():
|
||||
await demonstrate_full_pipeline()
|
||||
await demonstrate_tool_creation()
|
||||
|
||||
if __name__ == "__main__":
|
||||
asyncio.run(main())
|
||||
49
tools/test/test_rag_indexer_result.py
Normal file
49
tools/test/test_rag_indexer_result.py
Normal file
@@ -0,0 +1,49 @@
|
||||
#!/usr/bin/env python3
|
||||
"""
|
||||
测试重构后的 IndexBuilder 和 RAGRetriever
|
||||
"""
|
||||
|
||||
import asyncio
|
||||
import os
|
||||
import sys
|
||||
|
||||
# 添加项目根目录到 Python 路径
|
||||
project_root = os.path.join(os.path.dirname(__file__), "..")
|
||||
sys.path.insert(0, project_root)
|
||||
|
||||
from rag_indexer.index_builder import IndexBuilder
|
||||
from rag_indexer.splitters import SplitterType
|
||||
|
||||
async def test_index_builder():
|
||||
"""测试索引构建功能"""
|
||||
print("测试索引构建功能...")
|
||||
|
||||
# 创建 IndexBuilder 实例
|
||||
builder = IndexBuilder(
|
||||
collection_name="rag_documents",
|
||||
splitter_type=SplitterType.PARENT_CHILD,
|
||||
parent_chunk_size=1000,
|
||||
child_chunk_size=200
|
||||
)
|
||||
|
||||
# 测试文档路径
|
||||
test_file = os.path.join(os.path.dirname(__file__), "..", "data", "user_docs", "doublestory.txt")
|
||||
|
||||
if os.path.exists(test_file):
|
||||
# 构建索引
|
||||
print(f"正在为文件 {test_file} 构建索引...")
|
||||
processed = await builder.build_from_file(test_file)
|
||||
print(f"索引构建完成，处理了 {processed} 个文档")
|
||||
|
||||
# 获取集合信息
|
||||
info = builder.get_collection_info()
|
||||
print(f"集合信息: {info}")
|
||||
else:
|
||||
print(f"测试文件不存在: {test_file}")
|
||||
|
||||
# 关闭资源
|
||||
builder.close()
|
||||
print("\n测试完成")
|
||||
|
||||
if __name__ == "__main__":
|
||||
asyncio.run(test_index_builder())
|
||||
Reference in New Issue
Block a user