landlord-tracker

Running this

This should work on Linux, Mac, and Windows.

• install docker

• if you're on Windows, run git config --global core.autocrlf false

• clone this repo

• unzip the data package zipfile into the ./data directory

cd data
unzip /path/to/data.zip

• start services required for the build:

# docker build needs buildkit enabled
export DOCKER_BUILDKIT=1

# copy sample .env file
cp .env-sample .env

# starts the database and other services:
docker compose up --build -d

• to do a full run of the data build (loading and transforming the data):

# run the entire build
# on windows, add "winpty" to the front
docker compose run elt build-all

You'll see a "Finished build" message when it's finished. You can connect to the postgres database on localhost:5433 using a SQL client (like DBeaver) or use psql:

docker exec -it landlord-tracker-db-1 psql -U postgres

• to run the address report (replace 123 MAIN ST with a real address of an apartment complex):

docker compose run elt python3 -m engels.reports.report_address "123 MAIN ST"

• to stop the stack, run docker compose down. This stops the postgres database, but the data should be preserved and available agaion the next time you run compose up.

Doing Development

note: on windows, add "winpty" to the front of docker commands to get interactivity and avoid output buffering

Dev Process

To (re)run only the load or transforms as you are doing development:

# re-runs only the initial load of database tables
docker compose run elt load

# re-runs only the transforms
docker compose run elt transform

# you can also run bash in the container and run other things
docker compose run elt bash

Documentation

After running the transform step, you can view the updated dbt-generated HTML documentation at http://localhost:8000 to get a thousand foot view of the pipelines.

Naming Conventions

For tables:

pre_ - data preprocessed from raw files

raw_ - raw aka source tables, imported into the database directly from extract files

int_ - intermediate tables; these only exist to help create other tables.

stg_ - staged data: these are cleaned versions of raw tables, suitable for querying.

ent_ - derived entity tables built from staging

vld_ - for validation