docker images for the kartotherian project
They use a mixed architecture of Kartotherian and openmaptiles You need a docker version > 18.06
Use these commmands to download, import data and create tiles jobs for Luxembourg:
(Optional) First, delete all related containers and volumes (from an older import):
If sudo required, use "-E"!!
./exec.py cleanDownload, import and start the tiles generation (of Luxembourg by default):
./exec.py load-db # it runs the build command as a dependency on 'load-db'Once all tiles are generated, the map is visible on http://localhost:8585! (If not, take a look at docker ps and see what the port of the image qwantresearch/erdapfel is.)
If you want to see the list of all available commands, use -h or --help option on exec.py.
If you want to update the generation process, you need to edit Qwant/openmaptiles then update the openmaptiles submodule.
To launch kartotherian just do:
./exec.py kartotherian
(you might need sudo permissions depending on your setup)
to download a pbf and load data in postgres and generate tiles you need:
./exec.py load-db
Note that you can specify the PBF you want to give by using the --osm-file option.
The different way to configure the import can be seen in this readme.
Note: the first import might be quite long are some additional data will be downloaded (cf load_db)
If you want to use already downloaded data (especially usefull for a quicker dev cycle), you can use a mounted docker volume.
The file local-compose.yml gives an example of how to bind a named docker volume (the file uses a ./data directory but you can change it if you want).
To use a locally mounted volume add the local-compose.yml with the -f docker-compose option.
For an easier dev experience, you can use the docker-compose additional file local-compose.yml that forward ports, use a locally ./data mounted volume (to avoid some unnecessary download) and run a front end to view the tiles.
For example with this setup you can also provide an already downloaded pbf (it needs to be in the ./data volume) with --osm-file:
./exec.py --osm-file /data/input/luxembourg-latest.osm.pbf load-dbNote: even if the local directoy in ./data the osm file path is "/data/input/" because it's the directory path inside the container.
You can also specify a download url:
./exec.py --osm-file https://download.geofabrik.de/europe/luxembourg-latest.osm.pbf load-dbThe tiles generation is also handle by the load_db container.
To only generate 1 tile, you can set --tiles-coords [[x, y, z]]. x, y, and z are based on the Slippy Map Tile name system and you can use Geofabrik's tool to generate these for a specific location.
The different ways to configure the tiles generation can be seen in the default configuration file.
If you have forwarded the port, you can check the tile generation at http://localhost:16534/jobs and check a vector tile based map on http://localhost:8585
During the initial creation of the PG database, state and configuration files is initialized in the update_tiles_data volume from the .pbf metadata.
To launch the tiles update, run the update-tiles task (defined in load_db tasks):
./exec.py update-tiles
During this task:
The tile server architecture can be seen at QwantMaps
Normally you shouldn't need to change anything in the subfolders: everything is handled inside the docker files through exec.py. However, if you're interested in what these sub-folders are used for, go take a look to their README.md file.
The SQL and imposm mapping generation is quite straigthforward (cf. generate-sql and generate-imposm3 in the documentation).
The data_tm2source_*.xml generation is a bit more complex. We use generate-tm2source to generate a Carto project .yml file. This file is transformed to a Mapnik .xml project using kosmtik.
See more details on https://github.com/Qwant/openmaptiles