How do you use pgvector with SQLAlchemy and Python ORMs?

# pip install pgvector sqlalchemy psycopg2-binary

from sqlalchemy import create_engine, Column, Integer, Text, select
from sqlalchemy.orm import declarative_base, Session
from pgvector.sqlalchemy import Vector

engine = create_engine("postgresql+psycopg2://user:pass@localhost/mydb")
Base = declarative_base()

# Define the model with a vector column
class Document(Base):
    __tablename__ = "documents"
    id        = Column(Integer, primary_key=True)
    content   = Column(Text)
    embedding = Column(Vector(1536))  # pgvector Vector type

Base.metadata.create_all(engine)

# Insert a document
with Session(engine) as session:
    doc = Document(content="Python tutorial", embedding=[0.1, 0.2, ...])
    session.add(doc)
    session.commit()

# Query: find 5 nearest neighbours using cosine distance
from pgvector.sqlalchemy import cosine_distance

query_vector = [0.1, 0.2, ...]
with Session(engine) as session:
    results = session.scalars(
        select(Document)
        .order_by(cosine_distance(Document.embedding, query_vector))
        .limit(5)
    ).all()
    for doc in results:
        print(doc.content)

# Create HNSW index via SQLAlchemy:
from sqlalchemy import Index, text
Index(
    "ix_documents_embedding_hnsw",
    Document.embedding,
    postgresql_using="hnsw",
    postgresql_ops={"embedding": "vector_cosine_ops"},
    postgresql_with={"m": 16, "ef_construction": 64},
)