Automatically generate a map of zoning and planning decisions from municipal minutes pdfs. OpenAI GPT is used to extract addresses, summaries, resolution status, and tags.
-
Play with interactive map here: Interactive Map
-
Easily build a kepler.gl map (demo/sql/to_kepler_table.sql):
- Query a property history by address (demo/find_address_summaries.sql).
-
In your python env:
pip install -r requirements.txt -
Install tesseract dependency
brew install tesseract(mac)- set
POPPLER_PATHto your poppler path in components/processors/pdf2text.py
-
Install duckdb cli
brew install duckdb(mac) -
Install a local copy of chromium driver https://chromedriver.chromium.org/getting-started
-
Set your openai key as
OPENAI_KEYenvvar:export OPENAI_KEY=<your key>
Weekend project to play with
- GPT for structuring unstructured local municipal data
- duckdb
- selenium
There are no tests, this code is provided as is!
If you would like to help write munisource crawlers for other cities, reach out.