geocode-sqlite 0.4.0

Geocode rows from a SQLite table

geocode-sqlite

Geocode rows from a SQLite table

Installation

Install this tool using pip or pipx:

# install inside a virtualenv
pip install geocode-sqlite

# install globally
pipx install geocode-sqlite

Usage

Let's say you have a spreadsheet with addresses in it, and you'd like to map those locations. First, create a SQLite database and insert rows from that spreadsheet using sqlite-utils.

sqlite-utils insert data.db data data.csv --csv

Now, geocode it using OpenStreetMap's Nominatum geocoder.

geocode-sqlite nominatum data.db data \
 --location="{address}, {city}, {state} {zip}" \
 --delay=1 \
 --user-agent="this-is-me"

In the command above, you're using Nominatum, which is free and only asks for a unique user agent (--user-agent).

This will connect to a database (data.db) and read all rows from the table data (skipping any that already have both a latitude and longitude column filled).

You're also telling the geocoder how to extract a location query (--location) from a row of data, using Python's built-in string formatting, and setting a rate limit (--delay) of one request per second.

For each row where geocoding succeeds, latitude and longitude will be populated. If you hit an error, or a rate limit, run the same query and pick up where you left off.

Under the hood, this package uses the excellent geopy library, which is stable and thoroughly road-tested. If you need help understanding a particular geocoder's options, consult geopy's documentation.

Supported Geocoders

The CLI currently supports these geocoders:

• bing
• googlev3
• mapquest (and open-mapquest)
• nominatum

More will be added soon.

Common arguments and options

Each geocoder needs to know where to find the data it's working with. These are the first two arguments:

• database: a path to a SQLite file, which must already exist
• table: the name of a table, in that database, which exists and has data to geocode

From there, we have a set of options passed to every geocoder:

• location: a string format that will be expanded with each row to build a full query, to be geocoded
• delay: a delay between each call (some services require this)
• latitude: latitude column name
• longitude: longitude column name

Each geocoder takes additional, specific arguments beyond these, such as API keys. Again, geopy's documentation is an excellent resource.

Python API

The command line interface aims to support the most common options for each geocoder. For more find-grained control, use the Python API.

As with the CLI, this assumes you already have a SQLite database and a table of location data.

from geocode_sqlite import geocode_table
from geopy.geocoders import Nominatum

# create a geocoder instance, with some extra options
nominatum = Nominatum(user_agent="this-is-me", domain="nominatum.local.dev", scheme="http")

# assuming our database is in the same directory
count = geocode_sqlite("data.db", "data", query_template="{address}, {city}, {state} {zip}")

# when it's done
print(f"Geocoded {count} rows")

Any geopy geocoder can be used with the Python API.

Development

To contribute to this tool, first checkout the code. Then create a new virtual environment:

cd geocode-sqlite
python -m venv .venv
source .venv/bin/activate

Or if you are using pipenv:

pipenv shell

Now install the dependencies and tests:

pip install -e '.[test]'

To run the tests:

pytest

Please remember that this library is mainly glue code between other well-tested projects, specifically: click, geopy and sqlite-utils. Tests should focus on making sure those parts fit together correctly. We can assume the parts themselves already work.

To that end, there is a test geocoder included: geocode_sqlite.testing.DummyGeocoder. That geocoder works with an included dataset of In-N-Out Burger locations provided by AllThePlaces. It works like a normal GeoPy geocoder, except it will only return results for In-N-Out locations using the included database.