Hier Baut Berlin

A tool to visualize and inform citizens of Berlin about governmental decisions.

You want to help? Awesome. Scroll through the issues, open a new one, or just send a short notice to mail@hierbautberlin.de. We are happy about every person who wants to help.

Development setup

HierBautBerlin uses Elixir and Phoenix. Information on how to install Elixir can be found here.

As database it uses PostgreSQL with the PostGIS extension. If you don't want to install PostGIS locally, you can run it in a container:

podman run -d --name hierbautberlin-db -e POSTGRES_PASSWORD=postgres -p 5434:5432 docker.io/postgis/postgis:17-3.5
export DATABASE_PORT=5434

The importers need qpdf, pdftotext (poppler-utils) and dumppdf.py (pip install pdfminer.six). Assets are built with esbuild and dart-sass (installed by mix), JavaScript dependencies are installed with npm (Node.js 22+). Common tasks are run with just (just lists all recipes).

After you installed everything, the setup is as follows:

just update
just setup
just start

The app runs on http://localhost:4000, sent emails show up in http://localhost:4000/dev/mailbox.

Before you contribute code, please make sure to read the CONTRIBUTING.md

How to run the test suite

just check

You can also run the ExUnit tests in watch mode with:

just run-tests

Import geo objects

The text analysis (finding streets, addresses and places in news) and the map need the streets, street numbers and places of Berlin. They are imported from OpenStreetMap (streets, addresses, parks) and from gdi.berlin.de (LOR planning areas, schools):

just import-geo

It downloads the OSM extract of Berlin (~100 MB) and needs ogr2ogr (Debian: gdal-bin). A full import takes about a minute. Running it again updates the data; ids and links to news items are kept. In the Docker container use bin/import_geo_objects.sh --release.

After the address matching or the geo data changed, the links of existing news can be updated. Without --apply it only shows what would change:

just reanalyze BERLIN_AMTSBLATT
just reanalyze BERLIN_PRESSE --since 2026-01-01 --apply

The list next to the map is sorted by relevance (see lib/hierbautberlin/geo_data/relevance.ex). After changing those rules, calculate it again for all news items with mix geo.relevance.

Working with a production dump

just restore-dump dump.sql hierbautberlin_prod   # creates the database
export DATABASE_NAME=hierbautberlin_prod
mix ecto.migrate
mix geo.relevance
just import-geo
just start

Copy the production file storage into file_storage/ to see the Amtsblatt PDFs.

Deployment

The app runs as Docker image built from the Dockerfile (e.g. with Coolify). The container runs the web server and the importers, so run only one instance. The migrations run on every start.

It needs Postgres with PostGIS and these environment variables:

Variable
DATABASE_URL	required, e.g. ecto://user:password@host/database
SECRET_KEY_BASE	required, mix phx.gen.secret
PHX_HOST	the domain, default hierbautberlin.de
MAILGUN_API, MAILGUN_DOMAIN	sending emails
BUGSNAG_API	optional, error reporting
PLAUSIBLE_SCRIPT_ID	optional, empty disables the visitor statistics
POOL_SIZE, PORT	optional, default 10 and 4000

The Amtsblatt PDFs, the import folder and the OSM download are stored in /app/data, mount a volume there.

Funding

This project is funded by the German Federal Ministry of Education and Research and is part of the 9th batch of the prototype fund.