Fuzzy Weather

Powered by the Dark Sky API!

This library will retrieve weather data using the Dark Sky API and then provide a text description of the weather for a given day using a "fuzzy" representation suitable for use in a voice interface system. Think of this as a virtual meteorologist.

Note that you will need a Dark Sky API key to use this module!

Install and Basic Use

Install with npm using:

npm install --save fuzzy-weather

Then you can use the module like so:

let weather = require('fuzzy-weather')({
    apiKey: 'abcdefg1234567890',  // your Dark Sky API key!
    location: {
        lat: 38.9649734,
        lng: -77.0207249
    }
});

weather()  // defaults to weather forecast for today
    .then(function(data) {
        console.log(data);
    })
    .catch(console.error);

Options

There are three required pieces of information when you initialize the module, then some other ones you probably want to set:

apiKey: String,  // REQUIRED
location: {
    lat: Number, // REQUIRED
    lng: Number  // REQUIRED
}

That said, you really want to set the avgTemps for the area in question as well. This is what will determine the library's decision about whether it is unseasonably warm or cold in the requested location. This will default to the average temps in Washington, DC (the author's home town), which may be VERY different than the location requested. You can get historical weather data in a variety of places, for example the Open Weather Map project.

avgTemps: [                        // defaults to temps in Washington, DC
    { high: Number, low: Number }, // Jan
    ...,                           // Feb-Nov
    { high: Number, low: Number }  // Dec
]

(Yes, yes... I'd love for this module to get that data from an API service based on the lat and lng, but I can't find a free one that provides this information. Maybe I'll set up a scraper for that some time in the future.)

There are a few other options that define when the library will report that it is humid, windy, or sunny/cloudy:

dewPointBreak: Number,  // The dewpoint (temp) at which the air becomes nasty (humid) [defaults to 69]
humidityBreak: Number,  // The percent humidity (0-1) at which the air becomes nasty [defaults to 0.7]
windBreak: Number,      // The max wind velocity (in mph) at which you consider it "significant" [defaults to 15]
cloudBreak: Number      // The percent cloud coverage (0-1) at which you consider it to be "mostly cloudy" [defaults to 0.8]

Forecast for Requested Date

In addition to the options above, every time you call the module you may pass in the date you want the forecast for. If you leave this blank, then it will use the current date (as defined by the server).

Get today's forecast:

let weather = require('fuzzy-weather')({ /* options... */ });
weather().then(function() { ... });

Get forecast based on string date (must be within 7 days of current date):

let weather = require('fuzzy-weather')({ /* options... */ });
weather('11/30/2016').then(function() { ... });

Get forecast for timestamp (must be within 7 days of current date):

let weather = require('fuzzy-weather')({ /* options... */ });
weather(1480492800000).then(function() { ... });

Response Data

The primary function of this library (see usage above) will always return a Promise which you must then attach handlers to. A rejected Promise will always be fulfilled with an Error object, and a resolved Promise will always provide the same data structure:

{
    date: Date,
    currently: Object | null,
    hourByHour: Object | null,
    dailySummary: Object
}

The date above will be a JavaScript Date object representing the date this forecast is for. The other three sections will always have the same substructure:

{
    forecast: String,
    data: Object,
    conditions: Object
}

That said, they will not always be present. The dailySummary block is the only section that will always be provided. The currently section is only provided if the requested date is the current day. The hourByHour section is only provided if the requested date is within 48 hours (Dark Sky only provides hourly data for the next 48 hours).

The data within those three blocks will contain:

• forecast: This is really the reason you're here. This will be a string that represents the forecast for the requested day. It should be suitable for a voice interface (like Alexa or Google Home).
• data: The data block within these will be the data as provided by the Dark Sky API. You should review the developer documentation on Dark Sky's website for more information.
• conditions: A hash of "condition":"readable text" pairs. For example, it might contain { "heat": "it'll be a scorcher tomorrow" }, but only if the requested date was tomorrow and the forecast data calls for above normal temperatures. Other conditions might include: "wind", "cold", "rain", "snow", "humidity", etc.

Notes and Such

API Key

You will need a Dark Sky API key to use this module. Head over to their developer documentation and sign up. You can then get your API key from the account page.

Limits

Note that Dark Sky does impose some API limits - as of this update, that limit was 1,000 API calls per day for the free tier. If you require more than that, simply update your account with a credit card and they will bill you $0.0001 per call beyond that.

Current Data Retrieved

Currently this module only supports temperature and rain data in the forecast. The rest of the data provided by Dark Sky will come in soon... I just need to get the rain module solid first, and ensure everything is working in the main module.

Forecast Only

Additionally, note that this library is all about forecasting, thus you can only get fuzzy weather data for dates in the future (including the current day) up to 7 days (the Dark Sky limit). You can't request the weather for a date in the past.

Issues? Want to help?

If you have a problem you want addressed, please open an issue on GitHub. If you want to help, please see the CONTRIBUTING documentation, I would be more than happy to see other contributors on this package!

Author and License

This module was written by Jordan Kasper (@jakerella) and is licensed under the MIT license. Feel free to use it any way you want, just be nice.