azavea/tilertwo

Publish static vector tile sets to S3 from GeoJSON with a single command

Tiler Two

This repository contains a Docker image that can be used to take a GeoJson file or directory of line-delimited GeoJson and generate a static vector tile set from it, using almost any combination of available Tippecanoe options.

If you're not sure about which of the extensive Tippecanoe options to try first, a good default for most datasets is:

--drop-densest-as-needed -zg --hilbert

First, ensure Docker and Docker Compose 3.0+ are installed. Then build the container with:

docker-compose build

For complete documentation, run docker-compose run --rm tiler-two --help

Examples

Only generate an mbtiles file using relative paths

The mbtiles file will be placed in the root of the mounted data volume.

docker-compose run --rm -v $(pwd)/sample-data:/data tiler-two \
    file:///data/ne_110m_admin_0_countries.geojson \
    file:///data/tiles/ne_110m_admin_0_countries \
    --no-clean \
    --skip-export \
    --tmp /data \
    --tippecanoe-opts "-l ne_110m_admin0 --drop-densest-as-needed -z14"

Export tiles to S3 from a local geojson file

AWS_PROFILE='<profile>' docker-compose run --rm \
    -v $(pwd)/sample-data:/data tiler-two \
    file:///data/ne_110m_admin_0_countries.geojson \
    s3://bucket/path/to/tiles/ne_110m_admin_0_countries \
    --tippecanoe-opts "--drop-densest-as-needed -zg --coalesce --reorder --hilbert"

Environment Variables

AWS_PROFILE

If you're exporting tiles to S3, set this and mount your ~/.aws directory to /home/tilertwo/.aws in the container.

If you run the container with the included docker-compose script, the aws directory mount is handled for you.

TIPPECANOE_VERSION

Defaults to 1.34.3. Update this and rebuild the image if you'd prefer to install a different version of Tippecanoe.

tilertwo's Issues

Publish container image to quay.io/azavea account

In order to save users the step of having to build the image locally.

Not pressing, as build times aren't long, but a nice to have at some point.

Add AWS S3 copy command to README

Users who have generated a local XYZ tileset might want to just publish their local tileset to S3 without rerunning Tippecanoe or mbutil.

We should include either the command to run in the README or add a new option that allows users to skip Tippecanoe+mbutil.

Its easy to miss required magiks on the S3 copy command and end up with an unusable tileset in S3. The command is:

aws s3 cp \
  ./path/to/local/tile/dir \
  s3://bucket/key/prefix/ \
  --recursive \
  --acl public-read \
  --content-encoding gzip

Support generating multiple vector tile sets in one pass

Sometimes we have multiple geojson files and want to generate multiple mbtiles with different tippecanoe options, concatenate them with tile-join and then export tiles with mb-util.

For now, users can just start a container and run all of their commands manually within the container while a data volume is mounted, but this doesn't really help reuse or work for automated pipelines.

Develop a proposal for how to handle this use case and integrate it into the tilertwo.py script.