ChatGPT wrote most of the code for me. Created it in about three hours, where the time is combined with creating the DB, and testing it with the rest of the code for a different project.
I needed to look up about 200.000 coordinated to addresses from a government data dump, to be able to do filtering based on zip-codes. Found that it was a bit to expensive to use Google for reverse geocoding for a hobby project.
Norway address data in OpenStreetMap gets updated monthly from a government provided source (https://wiki.openstreetmap.org/wiki/Import/Catalogue/Address...). So it's uncharacteristic complete, normalized, has all tags and the address positions might have matched those from the (Norway) government data dump in your project. I'm not saying you got lucky but it was kind of the ideal scenario of matching data together.
The coordinates are placed by people using a map (bymelding.no), and not house addresses. If it was for coordinates for addresses that could speed it up a lot. And thanks for the link.
Based on the title ("Self Hosted OpenStreetMap using only Sqlite3 and Python") I was expecting something more, like tiles or maybe even geocoding. I think perhaps saying it is specifically only for reverse geocoding and also only tested for single country extracts might be good.
I needed to look up about 200.000 coordinated to addresses from a government data dump, to be able to do filtering based on zip-codes. Found that it was a bit to expensive to use Google for reverse geocoding for a hobby project.
Hope this could help others save some $.