Shortbread-Mapnik – Vector Map Styles for Mapnik

Installation

• Create vector tiles using the Shortbread schema.
• Convert the .mbtiles file returned by Tilemaker into plain z/x/y.pbf structure using mbutil.
• Add geometry types to the metadata.json file of the vector tile tileset using the add_geometry_types.py script.
• Build map styles for zoom levels 0 to 14 using cd shortbread-mapnik && make. Some Makefile variables are used, There are sensible defaults. CARTO:: defined the the path to the carto binary. The default works if you ran npm install -g carto. Use CARTO=./node_modules/.bin/carto if you ran npm install carto in the root of the project TILEDIR:: Path where MVTs are loaded from. Default is vectortiles in the root of the project e.g. make CARTO=/usr/local/bin/carto TILEDIR=/bigdisk/vtiles all

Technical Notes

The map style uses the OGR input plugin of Mapnik and GDAL's MVT driver to use vector tiles as data source. This leads to a few oddities compared to established Mapnik styles for raster tile rendering:

• Due to a bug/feature in OGR that needs sorting out, we currently need one map style per input zoom level, i.e. one XML file for zoom 0, one for zoom 1, … one for zoom 13 and one for zoom 14+. The Tirex queue manager has a feature that allows one to present this collection of different styles as one map style to the outside again but in the long run OGR needs to be modified to not default to the highest available vector tile zoom level for any rendering request.
• The file parameter of the data source in the Mapnik style has to point to the directory of the zoom level. GDAL expects a metadata.json file the parent directory of the zoom level. That file specifies the layer and their geometry types. Without the information about the geometry, GDAL will do a sequential read on all vector tiles on that zoom level in order to detect the geometry type for each layer.
• Vector tiles have to contain the features in the order they should be rendered on the map because the OGR data source does not support sorting.

Usage with Kosmtik

Kosmtik does not support OGR data sources. As a workaround, you can load the data in the same structure into a PostgreSQL database.

Set up a database:

# Create user as PostgreSQL superuser (not needed if you have some knowledge about PostgreSQL and know how to create a database):
sudo -u postgres createuser --superuser $USER
createdb -E utf8 -O $USER vtiles
psql -d vtiles -c "CREATE EXTENSION postgis;"
psql -d vtiles -c "CREATE EXTENSION hstore;"

Load into database using Osm2pgsql and its Flex output:

osm2pgsql --output flex --style osm2pgsql/shortbread.lua -d vtiles --multi-geometry --merc your_osm_file.osm.pbf

Clone Git repository of Tilemaker configuration and use its script to download shape files with vector data for low zoom levels and oceans:

git clone https://github.com/geofabrik/shortbread-tilemaker.git
cd shortbread-tilemaker
bash get-shapefiles.sh

Shape files will be located in the data subdirectory in the Git repository. Import them into the database now. Replace SHAPE_DIRECTORY by the path to the directory where the shape files are located.

bash load_shapefiles.sh -i SHAPE_DIRECTORY -d vtiles

Create .mml file for Kosmtik based on the template:

python3 prepare_mml.py --source vector --database vtiles shortbread_mapnik.mml > shortbread_mapnik_kosmtik.mml

Now you can start Kosmtik. Use the file created by prepare_mml.py as .mml file for Kosmtik. If you want to edit the .mml file, edit the template an re-run prepare_mml.py.

See also

• Tilemaker
• Tilemaker source code
• Shortbread schema for Tilemaker
• Shortbread documentation

Authors

This set of configuration files has been created for Geofabrik by Christine Karch and Amanda McCann before it was put on Github. Further contributors may be visible in the Git history.

License and Copyright

You are free to use this map style under the same license that applies to the tilemaker software, the FTWPL license.