pgvector-pascal

title: pgvector-pascal slug: pgvector-pascal brand: pgvector category: sdks-libraries tags:

• sdk
• pgvector
• vector-store source_url: https://github.com/pgvector/pgvector-pascal images:
• https://opengraph.githubassets.com/1/pgvector/pgvector-pascal

Overview

pgvector-pascal provides Pascal examples and bindings for using the pgvector PostgreSQL extension, enabling vector similarity search from Pascal applications. It currently supports the SQLDB database library from Free Pascal.

Features

• Pascal bindings for pgvector

  • Demonstrates how to use PostgreSQL’s vector type from Pascal code
  • Integrates with the pgvector extension for vector similarity search

• SQLDB support

  • Works with Free Pascal’s SQLDB library
  • Uses TSQLQuery (or equivalent) to execute SQL with vector types and parameters

• Extension setup

  • Example code to enable the pgvector extension:
    CREATE EXTENSION IF NOT EXISTS vector
  • Shows how to run this via Query.SQL.Text and Query.ExecSQL

• Schema definition with vector columns

  • Example table creation with a vector column:
    CREATE TABLE items (id bigserial PRIMARY KEY, embedding vector(3))

• Inserting vectors from Pascal

  • Parameterized insert statements using Pascal string parameters cast to vector type
  • Demonstrates inserting multiple vector rows in one statement:
    INSERT INTO items (embedding) VALUES ((:embedding1)::vector), ((:embedding2)::vector)

• Nearest neighbor search

  • Example query ordering by vector distance:
    SELECT * FROM items ORDER BY embedding <-> (:embedding)::vector LIMIT 5
  • Shows how to bind query parameters and iterate over result rows in Pascal

• Approximate vector indexing

  • Examples of creating HNSW index:
    CREATE INDEX ON items USING hnsw (embedding vector_l2_ops)
  • Examples of creating IVFFlat index with configuration:
    CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 100)

• Multiple distance metrics

  • Notes on choosing operator classes for different similarity metrics:
    • vector_l2_ops for Euclidean (L2) distance
    • vector_ip_ops for inner product
    • vector_cosine_ops for cosine distance

• Runnable example program

  • Repository includes a complete example (example.pp) showing end-to-end usage:
    extension creation, table creation, inserts, search, and iteration over results

• Development and testing setup

  • Example commands to clone and run:
    • git clone https://github.com/pgvector/pgvector-pascal.git
    • cd pgvector-pascal
    • createdb pgvector_pascal_test
    • fpc example.pp and ./example
  • Instructions for specifying the path to libpq by symlinking libpq.dylib when needed

Usage

• Install and enable pgvector in PostgreSQL.
• Use SQLDB in a Free Pascal application to:
  • Enable the extension (CREATE EXTENSION IF NOT EXISTS vector).
  • Create tables with vector(n) columns.
  • Insert vectors as strings cast to vector type via parameters.
  • Run similarity queries with <-> and limit results.
  • Add HNSW or IVFFlat indexes for approximate nearest neighbor search.

Pricing

• Not specified in the provided content (GitHub-hosted example/binding, typically open source).