Hey there! That 100k apartment coordinate problem definitely sounds like a headache—sending individual Tiles Query requests for each would be a total nightmare in terms of latency, cost, and hitting rate limits. Let’s break down some practical, scalable solutions tailored to your needs:

First, Clarify Your Core Need #

Before diving into solutions, let’s pin down exactly what you’re trying to achieve:

• If you only need to highlight the apartment locations (points): You don’t need to query MapBox’s building features at all! MapBox Flutter SDK handles large datasets efficiently with vector layers.
  • Convert all your (lat, lng) coordinates into a GeoJSON FeatureCollection of Point features.
  • Add a CircleLayer or SymbolLayer to your map style, pointing to this GeoJSON source. MapBox’s rendering engine is optimized for tens of thousands of points, so this will perform smoothly.
• If you need to highlight the entire building polygon that each apartment belongs to: This requires matching each coordinate to MapBox’s building features, and we need smarter ways than 100k individual requests.

Scalable Solutions for Building Feature Matching #

1. Pre-Batch Local Processing (Most Recommended) #

This approach avoids any API calls altogether by processing data offline once, then reusing the results:

• Step 1: Download Building Vector Tiles
  Grab the MapBox vector tiles for your target city at a zoom level where buildings are clearly represented (usually z=16 or z=17). You can fetch tiles via MapBox’s Tile API, or use tools to bulk download tiles for the city’s bounding box.
• Step 2: Local Spatial Matching
  Use a Dart spatial processing library like turf_dart to perform "point-in-polygon" checks:
  • Parse each downloaded vector tile to extract building polygons.
  • For every apartment coordinate, check which building polygon contains it. Store the matching building’s ID or geometry data in a local database (like Hive or SQLite).
• Step 3: Render Highlighted Buildings
  When loading your map, pull the pre-matched building geometries from your local database, add them as a GeoJSON source, and use a FillLayer to style them with a transparent highlight (e.g., fill-color: rgba(255, 0, 0, 0.3)).

2. Batch API Requests (For Real-Time Matching) #

If you can’t pre-process data, MapBox’s Batch API lets you bundle multiple Tiles Query requests into one, drastically reducing the number of calls:

• How it works: Instead of sending one request per coordinate, bundle up to 50 Tiles Query requests into a single POST request to the batch endpoint.
• Request Structure Example:

  {
    "requests": [
      {
        "method": "GET",
        "path": "/v4/mapbox.mapbox-streets-v8/tilequery/{lng1},{lat1}.json?radius=10&layers=building&access_token=YOUR_TOKEN"
      },
      {
        "method": "GET",
        "path": "/v4/mapbox.mapbox-streets-v8/tilequery/{lng2},{lat2}.json?radius=10&layers=building&access_token=YOUR_TOKEN"
      }
    ]
  }
  

• Calculate Requests: For 100k coordinates, this would only require ~2000 requests (50 per batch), which is manageable within rate limits and much faster than 100k individual calls.
• Flutter Implementation: Use a HTTP client like dio to send batch requests, parse the responses to extract matching building features, then render them on the map.

3. Reverse Geocoding Workaround (For Lower Precision Needs) #

If absolute precision isn’t critical, you can use batch reverse geocoding to get addresses for your coordinates, then use those addresses to fetch building polygons:

• Use MapBox’s Batch Reverse Geocoding API to get addresses for all coordinates in batches.
• For each address, use the Geocoding API to fetch the corresponding building polygon.
• Note: This is less accurate than direct point-in-polygon matching, as addresses might map to larger parcels rather than exact buildings.

Flutter Optimization Tips #

• Lazy Loading: Instead of loading all 100k points/buildings at once, only load those within the current map viewport. Use MapBox’s onCameraIdle callback to trigger loading when the map stops moving.
• Memory Management: Use efficient data structures and avoid holding all geometries in memory at once. Stream data from your local database as needed.

内容来源于stack exchange