Skip to content

pandaxbacon/WatsonWeatherBot

 
 

Repository files navigation

Watson Weather Bot

Build Status

A Kitura project which uses the swift-watson-sdk to create a Slack bot for grabbing current weather information.

Prerequisites:

  1. Install Open source Swift 3.0 - SNAPSHOT 05-03-a, making sure to update your $PATH.

  2. Register for a Bluemix account. We will be using two services from BlueMix for this application:

  • The Natural Language Classifier service interprets the intent behind the queries to the bot, and will return a classification corresponding to the nature of the query.
  • Insights for Weather enables you to integrate real-time or historical data into your applications.
  1. Have a Slack team with admin privileges ready, or create a new Slack team.

Quick Start:

  1. Clone the Watson Weather repository from a directory you'd like to store the bot:

git clone https://github.com/IBM-MIL/WatsonWeatherBot/

  1. Copy the Configuration-Sample.swift file to Sources/Configuration.swift

cp Config/Configuration-Sample.swift Sources/Configuration.swift

  1. If you haven't already, login to BlueMix. To do so, set the API endpoint and then login to your account.
cf api https://api.ng.bluemix.net
cf login

Quick Start with the IBM Cloud Tool for Swift:

  1. Sign in to Bluemix and create Insights for Weather and Natural Language Classifier services

Using the Web Interface, go to the Bluemix Catalog and create an Insights for Weather application called 'weatherinsights'. Next, create a Natural Language Classifier called natural_language_classifier.

  1. Download and install the IBM Cloud Tool (beta)

  2. Create a new IBM Cloud Tool project called 'Watson Weather Bot'

Open ICT and from the Projects view, click on the blue "+" icon in the upper right hand corner. Select Create Kitura Project. Enter the project name "Watson Weather Bot" and click Next.

  1. Add Cloud Runtime

Enter a Cloud Runtime Name, "WatsonWeatherBot". This will generate a Cloud runtime URL on Bluemix. Select the space on Bluemix where you would like to deploy (dev is fine). Click Next.

  1. Choose a location to save your files.

You will see a loading screen that ICT is cloning the Kitura sample project to your computer, deploying the compiling the code to Bluemix, then running the project on the server. We will overwrite what's in the sample project with the code you cloned from the repository next.

  1. Copy all the files that you cloned from the Watson Weather Bot repository and place them in the directory you chose when you created the Cloud Runtime.

The Package.swift and Manifest.yml file should be at the root of the directory after you copy them over. If the OS asks to confirm overwriting the files, select OK.

  1. Deploy the Watson Weather Bot code using the tool.

Deploy the updated files to Bluemix by clicking the deploy icon (it looks like a cloud with an arrowing going into it). This step can take 4-7 minutes to complete since it pulls all the dependencies for Kitura and compiles the app on the server.

Quick Start with Cloud Foundry Command Line Tool:

  1. Install the Cloud Foundry CLI interface.

  2. Create the services the Bot uses: Weather Insights and Natural Language Classifier.

Note that you will receive a warning that the Natural Language Classifier incurs a cost. As of the time this bot was created, a starter level of usage is included at not cost. For more details, see the NLC pricing.

cf create-service weatherinsights Free weatherbot-weather
cf create-service natural_language_classifier standard weatherbot-nlc
  1. Deploy the app from your local environment to BlueMix. There will be a delay of several minutes to install the system dependencies, download app dependencies, and compile the application.

cf push

  1. Get the URL for your app and also credentials such as username and password from both weatherbot-weather and weatherbot-nlc

cf env WatsonWeatherBot

The complete VCAP information will be dumped to the screen. Record somewhere the username and passwords:

 "weatherinsights": [
  {
   "credentials": {
    "host": "twcservice.mybluemix.net",
    "password": "password appears here",
    "port": 443,
    "url": "url appears here",
    "username": "user appears here"
   },
   "label": "weatherinsights",
   "name": "weatherbot-weather",
   "plan": "Free",
   "tags": [
    "big_data",
    "ibm_created",
    "ibm_dedicated_public"
   ]
  }
 ]

Record the URIs that appears for your deployment:

 "name": "WatsonWeatherBot",
 "space_id": "f900a490-dbd5-4053-abe2-c7645e3527eb",
 "space_name": "Swift",
 "uris": [
  "watsonweatherbot-posttympanic-marathon.mybluemix.net"
 ],
 "users": null,
 "version": "dfa2cccf-8990-4dc4-9a0b-876579beae29"
}
  1. Train the Natural Language Classifier.

During this process, you seed the classifier with some strings and corresponding classifications. A training set is provided in Config/weather_question_corpus.csv. Note that it may take several minutes for the training process to complete.

Replace username:password below with the credentials from the natural_language_classifier section of VCAP_SERVICES you recorded in the previous step.

curl -u username:password -F training_data=@Training/weather_question_corpus.csv -F training_metadata="{\"language\":\"en\",\"name\":\"My Classifier\"}" "https://gateway.watsonplatform.net/natural-language-classifier/api/v1/classifiers"
  1. Record the classifier_id. After the training step, it will be returned in a message similar to this:
{
"classifier_id" : "classifier id appears here",
"name" : "My Classifier",
"language" : "en",
"created" : "2016-06-10T15:14:49.322Z",
"url" : "https://gateway.watsonplatform.net/natural-language-classifier/api/v1/classifiers/classifier id",
"status" : "Training",
"status_description" : "The classifier instance is in its training phase, not yet ready to accept classify requests"
}%
  1. Add Slash Command integration to your Slack team.

From https://<TEAM-NAME>.slack.com/apps, search for the Slash Commands integration.

Use the following configuration:

Command: `/weather`
 URL: `URL to WatsonWeatherBot service recorded when calling cf env. Use an https:// prefix`
 Method: `POST`
 Token: this is not settable, record this to add to Configuration.swift
 Autocomplete help text: Select the box
    Description: `Get the weather`
    Usage hint: `What is the current temperature?`
  1. Modify Configuration.swift in Sources directory

Open in your favorite editor Configuration.swift. All of these values need to be set.

public struct Configuration  {
   
   static let staticGeocode = "37.7839,122.4012" You can replace this with any longitude and latitude
   static let slackToken = "replace this with what you got in Step 9"
   static let classifierID = "replace this value with what you got in Step 8"
   static let weatherUsername = "Replace this with what you got in Step 6"
   static let weatherPassword = "Replace this with what you got in Step 6"
   static let naturalLanguageClassifierUsername = "Replace this with what you got in Step 6"
   static let naturalLanguageClassifierPassword = "Replace this with what you got in Step 6"
}
  1. Redeploy your app now with the Credentials set properly

cf push

The deployment of the app usually takes about 4-7 minutes to complete. Once it is finished, you should get a message like:

1 of 1 instances running 

App started
  1. Type in your Slack channel: /weather What are the current weather conditions?

If it worked properly, you should get a similar response:

Today: Sunshine and a few afternoon clouds. High 91F. Winds SE at 5 to 10 mph. 
Tonight Partly cloudy. Low 71F. Winds SSE at 5 to 10 mph. 
The current temperature in San Francisco is 78 F.

About

Watson Weather Bot written in Swift 3 and Kitura

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Swift 95.5%
  • HTML 4.5%