PGVecto.rs

這個筆記本展示了如何使用與 Postgres 向量資料庫相關的功能 (pgvecto.rs)。

%pip install "pgvecto_rs[sdk]" langchain-community

from typing import List

from langchain_community.document_loaders import TextLoader
from langchain_community.embeddings.fake import FakeEmbeddings
from langchain_community.vectorstores.pgvecto_rs import PGVecto_rs
from langchain_core.documents import Document
from langchain_text_splitters import CharacterTextSplitter

API 參考文檔：TextLoader | FakeEmbeddings | PGVecto_rs | Document | CharacterTextSplitter

loader = TextLoader("../../how_to/state_of_the_union.txt")
documents = loader.load()
text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
docs = text_splitter.split_documents(documents)

embeddings = FakeEmbeddings(size=3)

使用 官方演示 docker 镜像 啟動資料庫。

! docker run --name pgvecto-rs-demo -e POSTGRES_PASSWORD=mysecretpassword -p 5432:5432 -d tensorchord/pgvecto-rs:latest

然後構建資料庫 URL

## PGVecto.rs needs the connection string to the database.
## We will load it from the environment variables.
import os

PORT = os.getenv("DB_PORT", 5432)
HOST = os.getenv("DB_HOST", "localhost")
USER = os.getenv("DB_USER", "postgres")
PASS = os.getenv("DB_PASS", "mysecretpassword")
DB_NAME = os.getenv("DB_NAME", "postgres")

# Run tests with shell:
URL = "postgresql+psycopg://{username}:{password}@{host}:{port}/{db_name}".format(
    port=PORT,
    host=HOST,
    username=USER,
    password=PASS,
    db_name=DB_NAME,
)

最後，從文檔建立 VectorStore

db1 = PGVecto_rs.from_documents(
    documents=docs,
    embedding=embeddings,
    db_url=URL,
    # The table name is f"collection_{collection_name}", so that it should be unique.
    collection_name="state_of_the_union",
)

稍後您可以使用以下方法連接到該表

# Create new empty vectorstore with collection_name.
# Or connect to an existing vectorstore in database if exists.
# Arguments should be the same as when the vectorstore was created.
db1 = PGVecto_rs.from_collection_name(
    embedding=embeddings,
    db_url=URL,
    collection_name="state_of_the_union",
)

請確保使用者有權建立表格。

具有分數的相似度搜尋

使用歐幾里得距離的相似度搜尋 (預設)

query = "What did the president say about Ketanji Brown Jackson"
docs: List[Document] = db1.similarity_search(query, k=4)
for doc in docs:
    print(doc.page_content)
    print("======================")

使用篩選器的相似度搜尋

from pgvecto_rs.sdk.filters import meta_contains

query = "What did the president say about Ketanji Brown Jackson"
docs: List[Document] = db1.similarity_search(
    query, k=4, filter=meta_contains({"source": "../../how_to/state_of_the_union.txt"})
)

for doc in docs:
    print(doc.page_content)
    print("======================")

或者

query = "What did the president say about Ketanji Brown Jackson"
docs: List[Document] = db1.similarity_search(
    query, k=4, filter={"source": "../../how_to/state_of_the_union.txt"}
)

for doc in docs:
    print(doc.page_content)
    print("======================")

相關內容

• 向量儲存 概念指南
• 向量儲存 操作指南