BytesRouter

主题

LlamaIndex 文档版本 跟随上游 源快照 d8e6865b 更新于 2026-10-18

阿里云 MySQL，也称为 ApsaraDB RDS for MySQL。ApsaraDB RDS for MySQL 是基于 MySQL 源代码分支的在线数据库服务，可提供高性能。ApsaraDB RDS for MySQL 是一种经过验证的解决方案，曾在双 11 期间处理大量并发流量。ApsaraDB RDS for MySQL 提供白名单配置、备份与恢复、透明数据加密（TDE）、数据迁移以及实例、账户和数据库管理等基本功能。更多信息，请参见 RDS MySQL 功能概述。

要运行此 notebook，你需要在云中运行一个 ApsaraDB RDS MySQL 实例，创建一个账户并创建所需的数据库。你可以参考此链接。

在本 notebook 中，我们需要在你的 ApsaraDB RDS MySQL 实例中创建名为 llama_index_test 和 llama_index_meta_test 的数据库。

设置

如果你在 colab 上打开此 Notebook，可能需要确保已安装 llama-index：

!pip install llama-index
%pip install llama-index-vector-stores-alibabacloud-mysql
# choose dashscope as embedding and llm model, your can also use default openai or other model to test
%pip install llama-index-embeddings-dashscope
%pip install llama-index-llms-dashscope

配置 dashscope 嵌入和 llm 模型，你也可以使用默认的 openai 或其他模型进行测试。如果你选择使用 dashscope 模型，可以在此处获取 API 密钥，并在以下代码中进行设置：

!export DASHSCOPE_API_KEY="your_api_key"

下载示例数据

!mkdir -p 'data/paul_graham/'
!wget 'https://raw.githubusercontent.com/run-llama/llama_index/main/docs/examples/data/paul_graham/paul_graham_essay.txt' -O 'data/paul_graham/paul_graham_essay.txt'

使用阿里云 MySQL 的 RAG 演示

简单查询

为简单查询加载数据

import os

DASHSCOPE_API_KEY = os.environ.get("DASHSCOPE_API_KEY")

from llama_index.core import (
    VectorStoreIndex,
    SimpleDirectoryReader,
    StorageContext,
)
from llama_index.vector_stores.alibabacloud_mysql.base import (
    AlibabaCloudMySQLVectorStore,
)

# set Embbeding model
from llama_index.core import Settings
from llama_index.embeddings.dashscope import DashScopeEmbedding

Settings.embed_model = DashScopeEmbedding(api_key=DASHSCOPE_API_KEY)

# config llm model
from llama_index.llms.dashscope import DashScope, DashScopeGenerationModels

dashscope_llm = DashScope(
    model_name=DashScopeGenerationModels.QWEN_MAX, api_key=DASHSCOPE_API_KEY
)

documents = SimpleDirectoryReader("data/paul_graham/").load_data()
print(f"Total documents: {len(documents)}")
print(f"First document, id: {documents[0].doc_id}")
print(f"First document, hash: {documents[0].hash}")
print(
    "First document, text"
    f" ({len(documents[0].text)} characters):\n{'='*20}\n{documents[0].text[:360]} ..."
)

print(
    """
#################
# simple generate vector
#################"""
)
client = AlibabaCloudMySQLVectorStore.from_params(
    host="rm-***.mysql.***.rds.aliyuncs.com",
    port=3306,
    user="user",
    password="password",
    database="llama_index_test",
    distance_method="COSINE",
)
storage_context = StorageContext.from_defaults(vector_store=client)
VectorStoreIndex.from_documents(
    documents, storage_context=storage_context, show_progress=True
)

使用 AlibabaCloudMySQL 进行搜索测试查询

import os

DASHSCOPE_API_KEY = os.environ.get("DASHSCOPE_API_KEY")

from llama_index.core import VectorStoreIndex
from llama_index.vector_stores.alibabacloud_mysql.base import (
    AlibabaCloudMySQLVectorStore,
)

# set Embbeding model
from llama_index.core import Settings
from llama_index.embeddings.dashscope import DashScopeEmbedding

embed_model = DashScopeEmbedding(api_key=DASHSCOPE_API_KEY)
# Global Settings
Settings.embed_model = embed_model

# config llm model
from llama_index.llms.dashscope import DashScope, DashScopeGenerationModels

dashscope_llm = DashScope(
    model_name=DashScopeGenerationModels.QWEN_MAX, api_key=DASHSCOPE_API_KEY
)

print(
    """
#################
# Basic Querying including Search Test
#################
"""
)
client = AlibabaCloudMySQLVectorStore.from_params(
    host="rm-***.mysql.eu-west-1.rds.***.com",
    port=3306,
    user="user",
    password="password",
    database="llama_index_test",
    distance_method="COSINE",
)
index = VectorStoreIndex.from_vector_store(
    vector_store=client, embed_model=embed_model
)

QUESTION = "What did the author do growing up?"
# Set Retriever
vector_retriever = index.as_retriever()
# search
source_nodes = vector_retriever.retrieve(QUESTION)
# check source_nodes
print(f"Question: {QUESTION}")
for node in source_nodes:
    print(f"---------------------------------------------")
    print("Search Test")
    print(f"---------------------------------------------")
    print(f"Score: {node.score:.3f}")
    print(node.get_content())
    print(f"---------------------------------------------")

# run query
query_engine = index.as_query_engine(llm=dashscope_llm)
res = query_engine.query(QUESTION)
print(f"Answer: {res.response}")
print(f"---------------------------------------------\n\n")

元数据过滤

为元数据过滤加载数据

import os

DASHSCOPE_API_KEY = os.environ.get("DASHSCOPE_API_KEY")

from llama_index.core import (
    VectorStoreIndex,
    SimpleDirectoryReader,
    StorageContext,
)
from llama_index.vector_stores.alibabacloud_mysql.base import (
    AlibabaCloudMySQLVectorStore,
)

# set Embbeding model
from llama_index.core import Settings
from llama_index.embeddings.dashscope import DashScopeEmbedding

Settings.embed_model = DashScopeEmbedding(api_key=DASHSCOPE_API_KEY)

# config llm model
from llama_index.llms.dashscope import DashScope, DashScopeGenerationModels

dashscope_llm = DashScope(
    model_name=DashScopeGenerationModels.QWEN_MAX, api_key=DASHSCOPE_API_KEY
)

documents = SimpleDirectoryReader("data/paul_graham/").load_data()
print(f"Total documents: {len(documents)}")
print(f"First document, id: {documents[0].doc_id}")
print(f"First document, hash: {documents[0].hash}")
print(
    "First document, text"
    f" ({len(documents[0].text)} characters):\n{'='*20}\n{documents[0].text[:360]} ..."
)

print(
    """
#################
# generate vector with some metadata for Metadata Filtering
#################"""
)
client = AlibabaCloudMySQLVectorStore.from_params(
    host="rm-***.mysql.***.rds.aliyuncs.com",
    port=3306,
    user="user",
    password="password",
    database="llama_index_meta_test",
    distance_method="COSINE",
)
storage_context = StorageContext.from_defaults(vector_store=client)
index = VectorStoreIndex.from_documents(
    documents, storage_context=storage_context, show_progress=True
)

from llama_index.core import Document
import regex as re

# Split the text into paragraphs.
text_chunks = documents[0].text.split("\n\n")

# Create a document for each footnote
footnotes = [
    Document(
        text=chunk,
        id=documents[0].doc_id,
        metadata={
            "is_footnote": bool(re.search(r"^\s*\[\d+\]\s*", chunk)),
            "mark_id": i,
        },
    )
    for i, chunk in enumerate(text_chunks)
    if bool(re.search(r"^\s*\[\d+\]\s*", chunk))
]

# Insert the footnotes into the index
for f in footnotes:
    index.insert(f)

使用 AlibabaCloudMySQL 进行搜索测试查询

import os

DASHSCOPE_API_KEY = os.environ.get("DASHSCOPE_API_KEY")

from llama_index.core import VectorStoreIndex
from llama_index.vector_stores.alibabacloud_mysql.base import (
    AlibabaCloudMySQLVectorStore,
)

# set Embbeding model
from llama_index.core import Settings
from llama_index.embeddings.dashscope import DashScopeEmbedding

embed_model = DashScopeEmbedding(api_key=DASHSCOPE_API_KEY)
# Global Settings
Settings.embed_model = embed_model

# config llm model
from llama_index.llms.dashscope import DashScope, DashScopeGenerationModels

dashscope_llm = DashScope(
    model_name=DashScopeGenerationModels.QWEN_MAX, api_key=DASHSCOPE_API_KEY
)

print(
    """
#################
# Querying with Metadata Filtering including Search Test
#################
"""
)
client = AlibabaCloudMySQLVectorStore.from_params(
    host="rm-***.mysql.***.rds.aliyuncs.com",
    port=3306,
    user="user",
    password="password",
    database="llama_index_meta_test",
    distance_method="COSINE",
)
index = VectorStoreIndex.from_vector_store(
    vector_store=client, embed_model=embed_model
)

from llama_index.core.vector_stores import (
    MetadataFilters,
    MetadataFilter,
    FilterOperator,
    FilterCondition,
)

QUESTION = "What did the author about space aliens and lisp?"
print(f"---------------------------------------------")
print(f"Question: {QUESTION}")
filters = MetadataFilters(
    filters=[
        MetadataFilter(
            key="is_footnote", value="true", operator=FilterOperator.EQ
        ),
        MetadataFilter(key="mark_id", value=0, operator=FilterOperator.GTE),
    ],
    condition=FilterCondition.AND,
)
print(f"---------------------------------------------")
for i in range(len(filters.filters)):
    print(f"Filter[{i}]: {filters.filters[i]}")
print(f"Filter Condition: {filters.condition}")
print(f"---------------------------------------------")
retriever = index.as_retriever(
    filters=filters,
)
result = retriever.retrieve(QUESTION)
for node in result:
    print("Search Test")
    print(f"---------------------------------------------")
    print(f"Score: {node.score:.3f}")
    print(node.get_content())
    print(f"---------------------------------------------")

# Create a query engine that only searches certain footnotes.
footnote_query_engine = index.as_query_engine(
    filters=filters,
    llm=dashscope_llm,
)

res = footnote_query_engine.query(QUESTION)
print(f"Answer: {res.response}")
print(f"---------------------------------------------\n\n")

友情链接

  • Agents.md面向 AI 编码助手的仓库说明文件规范
  • LangChainLangChain 是一个开放的Agent平台
  • LangSmithAgent 工程平台：可观测与评估、部署、Fleet、Engine 等
  • EmbabelEmbabel 是一个JVM平台Agent框架
  • SpringAISpringAI 是一个基于 Spring 的 AI 框架
  • MCPModel Context Protocol，连接 AI 应用与外部数据/工具的开放标准
  • LangChain4jJava 生态的 LLM 应用开发框架
  • LlamaIndex把数据接入 LLM 的框架与文档平台
  • LlamaParse智能体式 OCR 文档解析：把 PDF、扫描件变成 LLM 可用的文本与结构化数据
  • LiteParse本地优先、轻量的文档解析库，不需要 API key
  • CrewAI多智能体编排框架：以 Crew/Flow 组织自主协作的 AI Agent
  • MAFMicrosoft Agent Framework，企业级 .NET/Python 智能体框架

BytesRouter 是社区翻译站点，所有译文仅供参考，一切以官方英文原文为准。

© 2026 BytesRouter · 译文采用 CC BY 4.0 许可