qail-qdrant

QAIL driver for Qdrant vector database.

⚠️ BETA - This crate is under active development. API may change.

Overview

AST-native Rust driver for Qdrant vector search over gRPC.

The driver uses:

• direct protobuf wire encoding (no tonic-generated request structs)
• HTTP/2 transport with reconnect + timeout handling
• QAIL AST integration for search filters

This crate does not use SQL. "SQL bytes vs SQL strings" terminology only applies to PostgreSQL crates (qail-core + qail-pg).

Features

• 🔍 Vector similarity search with filters
• 📦 Upsert points with payload metadata
• 🗑️ Delete points by ID
• 📁 Collection management (create, delete)
• 🚀 Zero-copy protobuf encoding for gRPC

Quick Start

use qail_qdrant::{Distance, Point, QdrantDriver};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    // Connect to Qdrant gRPC (default: 6334)
    let mut driver = QdrantDriver::connect("localhost", 6334).await?;

    // Create collection
    driver
        .create_collection("products", 384, Distance::Cosine, false)
        .await?;

    // Upsert points
    let point = Point::new("p1", vec![0.1, 0.2, 0.3]).with_payload("name", "iPhone 15");
    driver.upsert("products", &[point], false).await?;

    // Search
    let embedding = vec![0.1, 0.2, 0.3];
    let results = driver.search("products", &embedding, 10, None).await?;

    Ok(())
}

Requirements

• Qdrant server running with gRPC enabled (default gRPC port: 6334)
• Rust 2024 edition

License

Apache-2.0