ConservationMetrics / mapgl-tile-renderer

Headless Node.js Maplibre-GL renderer for generating MBTiles with styled raster tiles.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

mapgl-tile-renderer

Publish to DockerHub

This headless Node.js MapGL renderer can be used to generate styled raster tiles in an MBTiles format. It can work with self-provided tilesets and a stylesheet, or an online API source with optional GeoJSON and OpenStreetMap data overlays.

The motivation to build this utility is to create offline background maps for use in mobile data collection applications, such as Mapeo and ODK Collect / KoboToolbox Collect, or other offline-compatible tools that can work with self-hosted maps like Terrastories. However, it can be helpful for any use case where having self-hosted raster MBTiles is a requirement.

This tool started as an extension of mbgl-renderer, which was built to export single static map images. Our thanks go out to the contributors of that project.

Usage

This tool can be used in the following ways:

  1. Via CLI (npm, Node.js or Docker)
  2. Using a Github template to generate tiles using a manifest.json.
  3. As a task worker service to poll a queue for new requests, and update a database table with the render results.
    • Currently supported: Azure Storage Queue and PostgreSQL.
    • The tool may be extended with RabbitMQ for self-hosting in the future.

Supported online API sources

❗️ To use these services, you are responsible for providing your own API token. In doing so, please carefully consult the terms of service and API limitations for each service.

Note that depending on your bounding box and maximum zoom level, this tool has the capability to send a lot of requests. You can use a utility like the Mapbox offline tile count estimator to ensure that your request will be reasonable, and in the case of any API sources with a freemium API limit, won't end up costing you.

Install via npm

To install the tool using npm, run:

$ npm install -g mapgl-tile-renderer

CLI options

  • -s or --style: Specify the style source. Use "self" for a self-provided style or one of the following for an online source: "bing", "esri", "google", "mapbox", "mapbox-satellite", "planet", "protomaps"

If using a self-provided style (--style self):

  • --stylelocation: Location of your provided map style
  • --stylesources: Directory where any local source files (GeoJSON, XYZ directory, MBTiles) specified in your provided style are located

If using an online style (--style with any online style name):

  • -a or --overlay: (Optional) Provide a GeoJSON object for a feature layer to overlay on top of the online source
  • -k or --apikey: (Optional) API key that may be required for your online source

If using any of the imagery online styles ("bing", "esri", "google", "mapbox-satellite", or "planet"):

  • -O or --openstreetmap: (Optional) Overlay OSM vector data on top of your imagery. Currently mapped: hydrology, roads, and points of interest (with labels). This is a boolean variable; set to "true" if you want to use this.

If your style is mapbox:

  • -m or --mapboxstyle in the format <yourusername>/<styleid>

If your style is planet:

  • -p or --monthyear: The month and year (in YYYY-MM format) of the Planet Monthly Visual Basemap to use

Common options:

  • -b or --bounds: Bounding box in WSEN format, comma separated (required)
  • -Z or --maxzoom: Maximum zoom level (required)
  • -z or --minzoom: Minimum zoom level (optional, 0 if not provided)
  • -r or --ratio: Output pixel ratio (optional, 1 if not provided)
  • -t or --tiletype: Output tile type (jpg, png, or webp) (optional, jpg if not provided)
  • -o or --outputdir: Output directory (optional, "outputs/" if not provided)
  • -f or --filename: Name of the output MBTiles file (optional, "output" if not provided)

CLI example usage

Using a self-provided style:

mapgl-tile-renderer --style self --stylelocation tests/fixtures/alert/style-with-geojson.json --stylesources tests/fixtures/alert/sources --bounds "-79,37,-77,38" -Z 8

From an online source (Bing), with OpenStreetMap data overlaid:

mapgl-tile-renderer --style bing --bounds "-79,37,-77,38" --openstreetmap true -Z 8 --apikey YOUR_API_KEY_HERE

From an online source (Mapbox):

mapgl-tile-renderer --style mapbox --mapboxstyle YOUR_USERNAME/YOUR_MAPBOX_STYLE_ID --apikey YOUR_API_KEY_HERE --bounds "-79,37,-77,38" -Z 8

From an online source (Planet):

mapgl-tile-renderer --style planet --monthyear 2013-12 --apikey YOUR_API_KEY_HERE --bounds "-54,3,-53,4" -Z 8

Online source (Esri) with GeoJSON overlay:

mapgl-tile-renderer --style esri --apikey YOUR_API_KEY_HERE --bounds "-54,3,-53,4" -Z 8 --overlay '{"type": "FeatureCollection", "features": [{"geometry": {"coordinates": [[[-54.25348208981326, 3.140689896338671], [-54.25348208981326, 3.140600064810259], [-54.253841415926914, 3.140600064810259], [-54.25348208981326, 3.140689896338671]]], "geodesic": false, "type": "Polygon"}, "id": "-603946+34961", "properties": {"month": "09", "year": "2023"}, "type": "Feature"}]}'

Docker

To run the tool with Docker, run:

docker run -it --rm -v "$(pwd)":/app/outputs communityfirst/mapgl-tile-renderer --style "mapbox" --bounds "-79,37,-77,38" -Z 8 --mapboxstyle YOUR_USERNAME/YOUR_MAPBOX_STYLE_ID --apikey YOUR_API_KEY_HERE

This automatically pulls the latest image from Docker hub. The docker run command is used to execute the mapgl-tile-renderer tool with a set of options that define how the map tiles will be rendered and saved. Here's a breakdown of the command and its variables:

  • -it: This option ensures that the Docker container runs in interactive mode, allowing you to interact with the command-line interface.
  • --rm: This option automatically removes the container when it exits, which helps to clean up and save disk space.
  • -v "$(pwd)":/app/outputs: This mounts the current working directory ($(pwd)) to the /app/outputs directory inside the container, allowing the container to write the output files to your local file system.
  • communityfirst/mapgl-tile-renderer: This is the name of the Docker image that contains the mapgl-tile-renderer tool. Make sure to replace the placeholder values with your actual information before running the command.

To run locally first build the Docker image:

docker build -t mapgl-tile-renderer .

Then run:

docker run -it --rm -v "$(pwd)":/app/outputs/ mapgl-tile-renderer --style "mapbox" --bounds "-79,37,-77,38" -Z 8 --mapboxstyle YOUR_USERNAME/YOUR_MAPBOX_STYLE_ID --apikey YOUR_API_KEY_HERE

Task worker listening to message queue

mapgl-tile-renderer can be configured as a task worker that listens for messages from a queue service, as submitted via map-packer. Upon retrieval of a message, the tool will initiate rendering, store the file on the container (ideally a volume mount location), and update a PostgreSQL table with the render results.

To set up mapgl-tile-renderer as a task worker, deploy the Docker image and provide the following environmental variables:

  • QUEUE_NAME: the name of your message queue where map-packer will submit messages.
  • QUEUE_CONNECTION_STRING: (for Azure usage) the storage account connection string.
  • DB_CONNECTION_STRING: connection string of your Postgres DB.
  • DB_TABLE: the name of your Postgres DB table.

For Azure Storage Queue (and other queue services in the future), mapgl-tile-renderer expects a message with a JSON body, composed of the various input options. Example:

{
  "style": "bing",
  "apiKey": "bing-api-key",
  "bounds": "-79,37,-77,38",
  "minZoom": 0,
  "maxZoom": 8,
  "outputFilename": "bing"
  "outputDir": "/maps"
}

Note that outputDir likely needs to be a volume mount directory on your mapgl-tile-renderer container, so that it can be accessed by map-packer or other tools for sharing and downloading.

For more information on the complete flow wherein mapgl-tile-renderer is deployed as a task worker, see the map-packer documentation.

Running with Github Actions

To use the Github Actions workflow defined in .github/workflows/gen-tiles-input.yml for generating tiles, follow these steps:

  1. Fork the repository containing the workflow file.
  2. Navigate to your forked repository on Github.
  3. Click on the "Actions" tab.
  4. You will see a list of available workflows. Click on the "Generate Tiles from Input" workflow.
  5. Click on the "Run workflow" dropdown button.
  6. Fill in the required and optional input fields. See CLI options above for the list of fields.
  7. After providing the necessary information in the input fields, click on the "Run workflow" button to initiate the tile generation process.
  8. The workflow will execute the steps defined in gen-tiles-input.yml and upon completion, the generated tiles will be available as artifacts in the workflow run.

Make sure to review the gen-tiles-input.yml file to understand the inputs and outputs of the workflow.

Inspect the outputs

Three easy ways to examine and inspect the MBTiles:

  1. Upload them to a Felt map.
  2. Use the mbview tool to view them in the browser.
  3. Load them in QGIS.

Formats other than MBTiles

In the future, we may decide to extend this tool to support creating raster tiles in a different format, such as PMTiles. However, for the time being, you can use tools like go-pmtiles to convert the MBTiles outputs generated by this tool.

For developers

Mapgl-tile-renderer uses Maplibre-GL Native to render tiles, Sharp to save them as an image, and Mapbox's mbtiles Node package to compile them into an MBTiles database (which is a SQLite file).

Node installation requirements

Node version: 18.17.0 to 20.x.x.

(Sharp requires 18.17.0 at minimum, and MapLibre Native is currently only supported on stable releases of Node, 20 being the latest)

Tests

To run tests and view coverage, run:

npm run test

To run tests that require a Mapbox or Planet access token, create a .env.test file and add MAPBOX_TOKEN, PLANET_TOKEN, and PROTOMAPS_TOKEN vars with your own token. (If not provided, tests requiring these will be skipped.)

About

Headless Node.js Maplibre-GL renderer for generating MBTiles with styled raster tiles.

License:MIT License


Languages

Language:JavaScript 94.8%Language:HCL 3.4%Language:Dockerfile 1.1%Language:Shell 0.7%