Dark Sky Wrapper

This is a wrapper for the Dark Sky (formerly forecast.io) API. It allows you to get the weather for any location, now, in the past, or future.

The Basic Use section covers enough to get you going. I suggest also reading the source if you want to know more about how to use the wrapper or what its doing (it's very simple).

Installation

You should use pip to install python-forecastio.

• To install pip install python-forecastio
• To remove pip uninstall python-forecastio

Simple!

Requirements

• You need an API key to use it (https://darksky.net/dev/). Don't worry a key is free.

Basic Use

Although you don't need to know anything about the Dark Sky API to use this module, their docs are available at https://darksky.net/dev/.

To use the wrapper:

import forecastio

api_key = "YOUR API KEY"
lat = -31.967819
lng = 115.87718

forecast = forecastio.load_forecast(api_key, lat, lng)
...

The load_forecast() method has a few optional parameters. Providing your API key, a latitude and longitude are the only required parameters.

Use the forecast.DataBlockType() eg. currently(), daily(), hourly(), minutely() methods to load the data you are after.

These methods return a DataBlock. Except currently() which returns a DataPoint.

byHour = forecast.hourly()
print byHour.summary
print byHour.icon

The .data attributes for each DataBlock is a list of DataPoint objects. This is where all the good data is :)

for hourlyData in byHour.data:
        print hourlyData.temperature

Advanced

function forecastio.load_forecast(key, latitude, longitude)

This makes an API request and returns a Forecast object (see below).

Parameters:

• key - Your API key from https://darksky.net/dev/.
• latitude - The latitude of the location for the forecast
• longitude - The longitude of the location for the forecast
• time - (optional) A datetime object for the forecast either in the past or future - see How Timezones Work below for the details on how timezones are handled in this library.
• units - (optional) A string of the preferred units of measurement, "auto" is the default. "us","ca","uk","si" are also available. See the API Docs (https://darksky.net/dev/docs/forecast) for exactly what each unit means.
• lazy - (optional) Defaults to false. If true the function will request the json data as it is needed. Results in more requests, but maybe a faster response time.
• callback - (optional) Pass a function to be used as a callback. If used, load_forecast() will use an asynchronous HTTP call and will not return the forecast object directly, instead it will be passed to the callback function. Make sure it can accept it.

---

function forecastio.manual(url)

This function allows manual creation of the URL for the Dark Sky API request. This method won't be required often but can be used to take advantage of new or beta features of the API which this wrapper does not support yet. Returns a Forecast object (see below).

Parameters:

• url - The URL which the wrapper will attempt build a forecast from.
• callback - (optional) Pass a function to be used as a callback. If used, an asynchronous HTTP call will be used and forecastio.manual will not return the forecast object directly, instead it will be passed to the callback function. Make sure it can accept it.

---

class forecastio.models.Forecast

The Forecast object, it contains both weather data and the HTTP response from Dark Sky

Attributes

• response

  • The Response object returned from requests request.get() method. See https://requests.readthedocs.org/en/latest/api/#requests.Response

• http_headers

  • A dictionary of response headers. 'X-Forecast-API-Calls' might be of interest, it contains the number of API calls made by the given API key for today.

• json

  • A dictionary containing the json data returned from the API call.

Methods

• currently()

  • Returns a ForecastioDataPoint object

• minutely()

  • Returns a ForecastioDataBlock object

• hourly()

  • Returns a ForecastioDataBlock object

• daily()

  • Returns a ForecastioDataBlock object

• update()

  • Refreshes the forecast data by making a new request.

---

class forecastio.models.ForecastioDataBlock

Contains data about a forecast over time.

Attributes (descriptions taken from the darksky.net website)

• summary

  • A human-readable text summary of this data block.

• icon

  • A machine-readable text summary of this data block.

• data

  • An array of ForecastioDataPoint objects (see below), ordered by time, which together describe the weather conditions at the requested location over time.

---

class forecastio.models.ForecastioDataPoint

Contains data about a forecast at a particular time.

Data points have many attributes, but not all of them are always available. Some commonly used ones are:

Attributes (descriptions taken from the darksky.net website)

• summary - A human-readable text summary of this data block.
• icon - A machine-readable text summary of this data block.
• time - The time at which this data point occurs.
• temperature - (not defined on daily data points): A numerical value representing the temperature at the given time.
• precipProbability - A numerical value between 0 and 1 (inclusive) representing the probability of precipitation occurring at the given time.

For a full list of ForecastioDataPoint attributes and attribute descriptions, take a look at the Dark Sky data point documentation (https://darksky.net/dev/docs/response#data-point)

---

How Timezones Work

Requests with a naive datetime (no time zone specified) will correspond to the supplied time in the requesting location. If a timezone aware datetime object is supplied, the supplied time will be in the associated timezone.

Returned times eg the time parameter on the currently DataPoint are always in UTC time even if making a request with a timezone. If you want to manually convert to the locations local time, you can use the offset and timezone attributes of the forecast object.

Typically, would would want to do something like this:

# Amsterdam
lat  = 52.370235
lng  = 4.903549
current_time = datetime(2015, 2, 27, 6, 0, 0)
forecast = forecastio.load_forecast(api_key, lat, lng, time=current_time)

Be caerful, things can get confusing when doing something like the below. Given that I'm looking up the weather in Amsterdam (+2) while I'm in Perth, Australia (+8).

# Amsterdam
lat  = 52.370235
lng  = 4.903549

current_time = datetime.datetime.now()

forecast = forecastio.load_forecast(api_key, lat, lng, time=current_time)

The result is actually a request for the weather in the future in Amsterdam (by 6 hours). In addition, since all returned times are in UTC, it will report a time two hours behind the local time in Amsterdam.

If you're doing lots of queries in the past/future in different locations, the best approach is to consistently use UTC time. Keep in mind datetime.datetime.utcnow() is still a naive datetime. To use proper timezone aware datetime objects you will need to use a library like pytz