Converting int to float with known precision

October 20, 2022

I’ve exported my location data from Google but one thing they have in the resulting json is that both latitudes and longitudes are stored as integers. For example:

-37785037   # latitude
144939405  # longitude

Since I know where I roughly was at that time it’s clear that the actual coords are -37.785.. and 144.939…, i.e. somewhere in northern suburbs of Melbourne, AU.

Also, by examining the data I established that in all cases the values have a 7-digit precision. So reading this json I’m converting coordinates to integers using this very ugly construct: first I convert int to string, then take integer part, add a dot, add floating part and then convert all to float:

loc = {}

for i in data['locations']:
    try:
        t = dt.datetime.strptime(i['timestamp'], "%Y-%m-%dT%H:%M:%S.%fZ")
    except ValueError:
        t = dt.datetime.strptime(i['timestamp'], "%Y-%m-%dT%H:%M:%SZ")
        
    lat = float(f"{str(i['latitudeE7'])[:-7]}.{str(i['latitudeE7'])[-7:]}")
    lon = float(f"{str(i['longitudeE7'])[:-7]}.{str(i['longitudeE7'])[-7:]}")
    loc[t] = {
        'latitude': lat,
        'longitude': lon
            }

This works, but looks hella ugly.

Is there a cleaner way of converting an int to float when you know the precision of the resulting float?