IndexError in geocodio-library-python When Geocoding

[rway314]

Certain addresses cause the geocodio-library-python library (v0.7.0) to throw an IndexError exception. See attached simplified example.

I have used the library for years with no issues. About once a month I geocode a batch of about 14000 addresses (breaking them into <10000 address chunks). Several months to a year ago it stopped working. It seemed to want a newer version of Python, so I upgraded to 3.11.15. Once I upgraded certain data sets caused IndexError exceptions when geocoding (in client.py:geocode).

The addresses may be bad, but (1) they used to work, (2) there has to be a better way to report bad addresses, and (3) I know of no way to programmatically determine which addresses the library doesn't like. It simply throws IndexError on the entire batch (and charges me for the geocoding).

Things I've tried:

• Upgrading from version 0.7.0 to 1.0.0 of the library
• Filtering out addresses the library doesn't like, but it's time consuming (and rings up a big geocoding bill) to isolate all the addresses and I could not feasibly find them all.

GeocodioApiIndexError.py

[MiniCodeMonkey]

Thanks for the detailed report and especially for the minimal repro — that made this easy to confirm.

This is a bug we already fixed. When a batch element came back with zero matches (which is what UPR 94501, APO, AE, 09305 does), the parser indexed results[0] on an empty list and blew up the whole batch. It was fixed in 3ae19ff and shipped in v1.1.0, so upgrading past the 1.0.0 you tested will resolve it:

pip install --upgrade geocodio-library-python

Running your script against current code gives:

formatted_address='1600 Pennsylvania Ave'  matched=True   query='1600 Pennsylvania Ave, Washington, DC'
formatted_address=''                       matched=False  query='UPR 94501, APO, AE, 09305'

Two things worth calling out, since they cover points (2) and (3) in your report:

• Unmatched addresses stay in the result list in their original position, so response.results stays index-aligned with the list you submitted. No more guessing which row dropped out.

• Each result now has a matched boolean, so you can pull out the failures programmatically instead of bisecting batches:

  batch = client.geocode(addresses)
  failed = [r.query for r in batch.results if not r.matched]

Closing this out as fixed — but please reopen if you still hit an IndexError on 1.4.0 and I'll dig in further.