def execute(self, options, args):
sess = db.session
if options.clear:
self.logger.info("Clearing all weather data!")
sess.query(model.RideWeather).delete()
if options.limit:
self.logger.info("Fetching weather for first {0} rides".format(
options.limit))
else:
self.logger.info("Fetching weather for all rides")
# Find rides that have geo, but no weather
sess.query(model.RideWeather)
q = text("""
select R.id from rides R
join ride_geo G on G.ride_id = R.id
left join ride_weather W on W.ride_id = R.id
where W.ride_id is null
and date(R.start_date) < CURDATE()
and time(R.start_date) != '00:00:00' -- Exclude bad entries.
;
""")
c = wu_api.Client(
api_key=app.config['WUNDERGROUND_API_KEY'],
cache_dir=app.config['WUNDERGROUND_CACHE_DIR'],
pause=7.0, # Max requests 10/minute for developer license
cache_only=options.cache_only)
rows = db.engine.execute(q).fetchall() # @UndefinedVariable
num_rides = len(rows)
for i, r in enumerate(rows):
if options.limit and i > options.limit:
logging.info("Limit ({0}) reached".format(options.limit))
break
ride = sess.query(model.Ride).get(r['id'])
self.logger.info("Processing ride: {0} ({1}/{2})".format(
ride.id, i, num_rides))
try:
start_geo_wkt = db.session.scalar(
ride.geo.start_geo.wkt) # @UndefinedVariable
point = parse_point_wkt(start_geo_wkt)
lon = point.lon
lat = point.lat
# We are doing only lat/lon now instead of us_city, since us_city seems to resolve to regional weather stations
# rather than the closest weather stations ...
# hist = c.history(ride.start_date, us_city=ride.location, lat=lat, lon=lon)
hist = c.history(ride.start_date, lat=lat, lon=lon)
ride_start = ride.start_date.replace(tzinfo=hist.date.tzinfo)
ride_end = ride_start + timedelta(seconds=ride.elapsed_time)
# NOTE: if elapsed_time is significantly more than moving_time then we need to assume
# that the rider wasn't actually riding for this entire time (and maybe just grab temps closest to start of
# ride as opposed to averaging observations during ride.
ride_observations = hist.find_observations_within(
ride_start, ride_end)
start_obs = hist.find_nearest_observation(ride_start)
end_obs = hist.find_nearest_observation(ride_end)
def avg(l):
no_nulls = [e for e in l if e is not None]
if not no_nulls:
return None
return sum(no_nulls) / len(
no_nulls) * 1.0 # to force float
rw = model.RideWeather()
rw.ride_id = ride.id
rw.ride_temp_start = start_obs.temp
rw.ride_temp_end = end_obs.temp
if len(ride_observations) <= 2:
# if we dont' have many observations, bookend the list with the start/end observations
ride_observations = [start_obs
] + ride_observations + [end_obs]
rw.ride_temp_avg = avg([o.temp for o in ride_observations])
rw.ride_windchill_start = start_obs.windchill
rw.ride_windchill_end = end_obs.windchill
rw.ride_windchill_avg = avg(
[o.windchill for o in ride_observations])
rw.ride_precip = sum([
o.precip for o in ride_observations if o.precip is not None
])
rw.ride_rain = any([o.rain for o in ride_observations])
rw.ride_snow = any([o.snow for o in ride_observations])
rw.day_temp_min = hist.min_temp
rw.day_temp_max = hist.max_temp
# ride.weather_fetched = True # (We don't have such an attribute, actually.)
# (We get this from the activity now.)
# ride.timezone = hist.date.tzinfo.zone
sess.add(rw)
sess.flush()
if lat and lon:
try:
sun = Sun(lat=lat, lon=lon)
rw.sunrise = sun.sunrise(ride_start)
rw.sunset = sun.sunset(ride_start)
except:
self.logger.exception(
"Error getting sunrise/sunset for ride {0}".format(
ride))
# But soldier on ...
except:
self.logger.exception(
"Error getting weather data for ride: {0}".format(ride))
# But soldier on ...
sess.commit()
