spatio 0.2.9

High-performance, embedded spatio-temporal database

Spatio for Python

High-performance spatial database for Python, powered by Rust.

What is it?

Spatio is an embedded database optimized for tracking moving objects. It uses a specialized Hot/Cold State architecture to provide real-time spatial queries and historical trajectory tracking with minimal overhead.

Quick Start

import spatio

# Open an in-memory database
db = spatio.Spatio.memory()

# Track a moving object
pos = spatio.Point(-74.006, 40.712, z=10.0)
db.upsert("drones", "drone_1", pos, {"status": "active"})

# Find objects within 500 meters
nearby = db.query_radius("drones", pos, radius=500, limit=10)
for object_id, point, metadata, distance in nearby:
    print(f"Found {object_id} at {distance:.1f}m")

core API

Spatio (Local Database)

import spatio

# Creation
db = spatio.Spatio.memory()
db = spatio.Spatio.open("data.db")

# Tracking
db.upsert(namespace, object_id, point, metadata=None)
db.get(namespace, object_id)  # Returns CurrentLocation or None
db.delete(namespace, object_id)

# Proximity Queries
# Returns list of (object_id, point, metadata, distance)
db.query_radius(namespace, center_point, radius, limit=100)
db.query_near(namespace, object_id, radius, limit=100)

# K-Nearest Neighbors
db.knn(namespace, center_point, k)
db.knn_near_object(namespace, object_id, k)

# Volume Queries
db.query_bbox(namespace, min_x, min_y, max_x, max_y, limit=100)
db.query_within_cylinder(namespace, center_point, min_z, max_z, radius, limit=100)
db.query_within_bbox_3d(namespace, min_x, min_y, min_z, max_x, max_y, max_z, limit=100)

# Relative Queries
db.query_bbox_near_object(namespace, object_id, width, height, limit=100)
db.query_cylinder_near_object(namespace, object_id, min_z, max_z, radius, limit=100)

# Trajectories
db.insert_trajectory(namespace, object_id, list_of_temporal_points)
db.query_trajectory(namespace, object_id, start_ts, end_ts, limit=100)

Data Types

Point

spatio.Point(lon, lat, z=None)

• point.lon, point.lat, point.alt
• point.distance_to(other_point) -> distance in meters (Haversine)

TemporalPoint

spatio.TemporalPoint(point, timestamp)

• Used for bulk trajectory ingestion.

Performance Tips

1. Namespaces: Use namespaces to logically separate different types of objects (e.g., "delivery_trucks", "warehouses").
2. Object Queries: Use query_near or knn_near_object whenever possible, as they avoid re-passing coordinates and leverage the internal Hot State index directly.

License

MIT - see LICENSE