This introduces a Makefile and a graphical user interface (based on streamlit) so users can more easily try out the tool with the sample data.

This release introduces HTTP endpoints to upload files, calculate solutions from scratch and from manually adjusted routes, and download every output files. This will allow us to have a browser-based GUI in the next release.

This release offers some more robustness features and a preview of using HTTP requests to communicate with the routing tool.

Super quick instructions:

Install yq and docker.

sh s_build_docker_dev.sh

docker run --network host dkroutingtool:dev /opt/conda/bin/python src/py/server.py

Then in a different terminal:

curl -X "POST" "localhost:5001/provide_files" -H "accept: application/json" -H "Content-Type: multipart/form-data" -F "files=@local_data/customer_data.xlsx" -F "files=@local_data/config.json" -F "files=@local_data/extra_points.csv"

curl https://localhost:5001/get_solution

curl -o download.zip https://localhost:5001/download

A file representing a small area is attached, edit the url in build_parameters.yml with:
geofabrik-url: https://github.com/datakind/dk-routing/releases/download/datadive_2023/Brooklyn.osm.pbf

Various improvements in terms of viz, manual edits, file management, and having an API beyond a GDrive integration.