langchain-alibabacloud-mysql 1.0.1

An integration package connecting AlibabaCloud RDS MySQL (with Vector support) and LangChain

🦜️🔗 LangChain AlibabaCloud MySQL

This package provides LangChain integration with AlibabaCloud RDS MySQL's native vector search capabilities.

Requirements

• Python 3.10+
• AlibabaCloud RDS MySQL 8.0.36+ with Vector support
• rds_release_date >= 20251031

Installation

pip install -U langchain-alibabacloud-mysql

Optional Dependencies

For DashScope embeddings (recommended for Alibaba Cloud):

pip install langchain-community dashscope

Quick Start

Using DashScope Embeddings (Recommended)

from langchain_alibabacloud_mysql import AlibabaCloudMySQL
from langchain_community.embeddings import DashScopeEmbeddings

embeddings = DashScopeEmbeddings(
    model="text-embedding-v4",
    dashscope_api_key="your-dashscope-api-key",
)

vectorstore = AlibabaCloudMySQL(
    host="your-rds-host.mysql.rds.aliyuncs.com",
    port=3306,
    user="your-user",
    password="your-password",
    database="your-database",
    embedding=embeddings,
    table_name="langchain_vectors",
    distance_strategy="cosine",
    hnsw_m=6,
)

# Add texts
vectorstore.add_texts(["Hello world", "LangChain is great"])

# Search
results = vectorstore.similarity_search("Hello", k=2)

Using Other Embedding Models

Any LangChain-compatible embedding model can be used:

from langchain_openai import OpenAIEmbeddings

embeddings = OpenAIEmbeddings()
# ... rest remains the same

Features

• Native VECTOR data type for efficient storage
• HNSW vector index for fast approximate nearest neighbor search
• Cosine and Euclidean distance metrics
• MMR search for diverse results
• Metadata filtering support
• Batch operations for efficient data loading

API Reference

AlibabaCloudMySQL

Main vector store class.

Constructor Parameters:

Parameter	Type	Default	Description
host	str	-	MySQL host
port	int	-	MySQL port
user	str	-	Username
password	str	-	Password
database	str	-	Database name
embedding	Embeddings	-	Embedding model
table_name	str	"langchain_vectors"	Table name
distance_strategy	str	"cosine"	"cosine" or "euclidean"
hnsw_m	int	6	HNSW M parameter

Methods:

• add_texts(texts, metadatas, ids) - Add texts to the store
• similarity_search(query, k, filter) - Search for similar documents
• similarity_search_with_score(query, k, filter) - Search with similarity scores
• max_marginal_relevance_search(query, k, fetch_k, lambda_mult, filter) - MMR search
• delete(ids) - Delete vectors by IDs
• get_by_ids(ids) - Get documents by IDs
• count() - Count total vectors
• clear() - Clear all vectors
• drop_table() - Drop the vector table
• from_texts(texts, embedding, ...) - Create from texts
• from_documents(documents, embedding, ...) - Create from documents

Demo Tests

See tests/demo_tests/ for comprehensive examples:

• RAG-agent.py - RAG agent with retrieval tools
• filter-query.py - Metadata filtering examples
• semantic-search.py - Basic similarity search

License

MIT License