nativ_maps

Amazon Location Service v2 in pure Dart, shaped like Google Maps.

All 44 operations —Places, Routes, Maps, geofencing and device tracking— behind a typed client, with a spending cap, retries and no Flutter dependency at all.

import 'package:nativ_maps/nativ_maps.dart';

final maps = NativMaps(
  region: 'us-east-1',
  credentials: const ApiKeyCredentials('your-key'),
  language: 'es',
  budget: Budget(maxUnits: 500),
);

final results = await maps.places.searchText(queryText: 'gasolinera');
final route = await maps.routes.calculateRoutes(
  origin: here,
  destination: results.places.first.position!,
);

Looking for the map?

This package does not draw. The widget lives in nativ_maps_flutter, which re-exports all of this: installing just that one gets you all forty-four operations plus the map.

This one lives apart so it can be tested without an emulator and used in a command-line tool or a Dart server.

The 44 operations

Places · 7	Routes · 5	Maps · 5
`autocomplete`	`calculateRoutes`	`styleDescriptorUrl`
`searchText`	`calculateRouteMatrix`	`staticMap`
`reverseGeocode`	`calculateIsolines`	`tileUrlTemplate`
`getPlace`	`snapToRoads`	`glyphsUrlTemplate`
`geocode`	`optimizeWaypoints`	`spritesUrl`
`searchNearby`
`suggest`

The two in bold do not exist in Google Maps.

Geofencing · 12	Device tracking · 15
`putGeofence` · `batchPutGeofence`	`batchUpdateDevicePosition`
`getGeofence` · `listGeofences`	`getDevicePosition`
`batchDeleteGeofence`	`batchGetDevicePosition`
`batchEvaluateGeofences`	`getDevicePositionHistory`
`forecastGeofenceEvents`	`listDevicePositions`
5 for managing collections	`verifyDevicePosition`
	9 for management and linking

forecastGeofenceEvents warns before a device enters or leaves a zone. verifyDevicePosition detects spoofed locations. Neither has an equivalent in Google Maps.

⚠️ Geofencing and tracking belong to the previous generation of Amazon Location: you have to create a resource and they do not accept an API key —SigV4 credentials are needed—. The package checks this before sending, with a message that says what to do, instead of letting it arrive as a 403 identical to every other one.

The calculation layer

Pure Dart, no requests: what you need on top of the service for a taxi, bidding or tracking app to actually work.

Class	What it solves
`TripRecorder`	measures the trip without GPS noise inflating it
`Tariff`	taximeter with an auditable breakdown, time bands and waiting time
`RouteTracker`	ETA per maneuver and off-route detection, no calls
`RideAuction` · `BidAdvisor`	inDrive-style auction and real profitability
`PriceAdvisor`	suggested price with the reason behind each factor
`TariffCalibration`	fits the tariff to the prices in your city
`DispatchPlanner`	the closest driver in time, not in a straight line
`TelemetryAnalyzer`	harsh acceleration, harsh braking, cornering and speeding

A parked car with ±20 m of uncertainty racks up more than 5 km per hour if you add up the distances between readings. TripRecorder racks up less than 50 m. That difference is what the passenger ends up paying.

Spending control

Amazon Location bills per request, and some requests cost more than one unit:

Operation	What it looks like	What it costs
`calculateIsolines` with 5 thresholds	1 request	5 units
`calculateRouteMatrix` of 10×10	1 request	100 units

Budget counts the real units and cuts off before sending. The service's hard limits are also checked beforehand, with the math spelled out in the error message.

Documentation

Recipes — 40 copy-paste examples
For AI agents
Coverage

License

MIT