rapidgeo 0.2.4

Fast geographic and planar distance calculations with Python bindings

rapidgeo

Geographic calculations for Python: distance calculations, polyline encoding/decoding, line simplification, and curve similarity measures.

Installation

pip install rapidgeo          # Base package
pip install rapidgeo[numpy]   # With NumPy support

Quick Start

from rapidgeo import LngLat
from rapidgeo.distance.geo import haversine

# Create coordinates (longitude, latitude) - multiple formats supported
sf = LngLat.new_deg(-122.4194, 37.7749)   # LngLat object
nyc = (-74.0060, 40.7128)                  # Tuple (lng, lat)
la = [-118.2437, 34.0522]                  # List [lng, lat]

# Calculate distance
distance = haversine(sf, nyc)
print(f"Distance: {distance / 1000:.1f} km")  # ~4,130 km

What it does

Distance Calculations:

• Haversine: Spherical Earth approximation, good for most uses
• Vincenty: Ellipsoidal Earth model for higher precision
• Euclidean: Flat plane calculations
• Batch operations for multiple points
• NumPy array support for vectorized operations

Polyline Encoding/Decoding:

• Google Polyline Algorithm implementation
• Compress coordinate sequences to text strings
• Configurable precision levels

Line Simplification:

• Douglas-Peucker algorithm
• Reduce point count while preserving shape
• Configurable distance tolerance

Curve Similarity:

• Fréchet distance: Compare trajectories (point order matters)
• Hausdorff distance: Compare shapes (point order doesn't matter)
• Useful for GPS track analysis and route comparison

Coordinate System

All coordinates use longitude, latitude ordering (lng, lat):

# Correct
point = LngLat.new_deg(-122.4194, 37.7749)  # lng first, lat second

# Common mistake
# point = LngLat.new_deg(37.7749, -122.4194)  # lat, lng - WRONG

Usage Examples

Compare GPS Tracks:

from rapidgeo.similarity.frechet import discrete_frechet

# Two similar walking routes
route_a = [LngLat.new_deg(-122.419, 37.775), LngLat.new_deg(-122.418, 37.776)]
route_b = [LngLat.new_deg(-122.419, 37.775), LngLat.new_deg(-122.417, 37.777)]

similarity = discrete_frechet(route_a, route_b)
print(f"Routes differ by {similarity:.1f} meters")

Simplify GPS Tracks:

from rapidgeo.simplify import douglas_peucker

# Reduce GPS track complexity
detailed_track = [LngLat.new_deg(-122.4, 37.7), LngLat.new_deg(-122.39, 37.71), ...]
simplified = douglas_peucker(detailed_track, tolerance=10.0)  # 10m tolerance
print(f"Reduced from {len(detailed_track)} to {len(simplified)} points")

Encode/Decode Polylines:

from rapidgeo.polyline import encode, decode

# Compress GPS data - supports multiple coordinate formats
points = [
    LngLat.new_deg(-122.4, 37.7),  # LngLat objects
    (-122.3, 37.8),                 # Tuples
    [-122.2, 37.9],                 # Lists
]
encoded = encode(points, precision=5)
decoded = decode(encoded, precision=5)

NumPy Batch Operations:

import numpy as np
from rapidgeo.numpy import pairwise_haversine, path_length_haversine

# High-performance vectorized operations
lng = np.array([-122.4, -122.3, -122.2])
lat = np.array([37.7, 37.8, 37.9])

# Calculate distances between consecutive points
distances = pairwise_haversine(lng, lat)

# Total path length
total_distance = path_length_haversine(lng, lat)
print(f"Total: {total_distance / 1000:.1f} km")

License

Licensed under either of Apache License, Version 2.0 or MIT License at your option.