Is this repo redundant/outdated?

[easherma]

Does the libpostal library effectively replace the function of this python app (assuming some sort of basic wrapper, Flask or otherwise)?

https://github.com/openvenues/libpostal/blob/parser-data/scripts/geodata/names/deduping.py

[albarrentine]

Outdated, yes, but not entirely redundant. This app uses libpostal's Python-only, English-only predecessor, address_normalizer, which is deprecated and has been replaced by libpostal/pypostal. Swapping it out for pypostal would take a tiny amount of refactoring but not much.

There's been some discussion around re-architecting this for use in the Pelias import process. The Pelias team mostly uses NodeJS, and since libpostal has Node bindings, it might make more sense on their side to rewrite this in Node and embed calls to libpostal directly rather than making batches of calls over HTTP.

The class mentioned in libpostal's geodata package is used for deduping place names across data sets (Quattroshapes/WOE names vs. OSM) rather than addresses.

[easherma]

Julian and I chatted about this and I pointed him towards that node library for libpostal. However, for my current use case I'm working on a data-pipeline for bulk geocoding locally and that will be written in python (using luigi), so I might be interested in a refactor of this to use pypostal.