This is a repository for an interview challenge question at Openhouse.ai. The demo is currently hosted as a live Heroku app for your convenience @ https://openhouse-log-api.herokuapp.com/logs/ I recommend playing around with the query parameters in the browser to test the filtering functionality, and postman for sending batch requests for actions and other calls.

This application uses the Java Spring Boot framework as a RESTful API, MongoDB hosted on MongoDB Atlas for log storage, and Heroku as a quick and easy cloud provider for our API.

The REST API and it's available calls are described below.

This API route fetches all the logs if no query parameters are passed, otherwise the query filter parameters can be used in any combination to refine the log search results. Additionally, lists of users can be passed in as comma seperated values (?users=userIdA,userIdB) and lists of types can be used in the same fashion (?types=CLICK,NAVIGATE).

GET /logs/

curl -i -H 'Accept: application/json' https://openhouse-log-api.herokuapp.com/logs/?users=ABC123XYZ,XYZ123ABC&types=CLICK

• users: [Array of User Id's] - This parameter will filter out the results to only contain logs from the specified user array (Optional)
• types: [Array of Type's] - This parameter will filter out the results to only contain logs of the matching type (Optional)
• start: Date (YYYY-MM-DD) - This parameter will filter the results to contain logs AFTER the specified start date (Optional)
• end: Date (YYYY-MM-DD) - This parameter will filter the results to contain logs BEFORE the specified start date (Optional)

[
  {
    "id": "625f2b756d0820fdd7b9e3d3",
    "userId": "ABC123XYZ",
    "sessionId": "XYZ456ABC",
    "actions": [
      {
        "time": "2018-10-18T21:37:28-06:00",
        "type": "CLICK",
        "properties": {
          "locationX": 52,
          "locationY": 11
        }
      }
    ]
  },
  {
    "id": "625f5ab26d0820fdd7b9e3d5",
    "userId": "XYZ123ABC",
    "sessionId": "789XYZ123",
    "actions": [
      {
        "time": "2018-10-20T21:37:28-06:00",
        "type": "CLICK",
        "properties": {
          "locationX": 120,
          "locationY": 40
        }
      }
    ]
  }
]

This example link calls the api between a date range of 7 days, this can be used by other applications who wish to have better granularity over date ranges.

GET https://openhouse-log-api.herokuapp.com/logs/?start=2018-10-18&end=2018-10-22 [Browser Link]

[
  {
    "id": "625f2b756d0820fdd7b9e3d3",
    "userId": "ABC123XYZ",
    "sessionId": "XYZ456ABC",
    "actions": [
      {
        "time": "2018-10-18T21:37:28-06:00",
        "type": "CLICK",
        "properties": {
          "locationX": 52,
          "locationY": 11
        }
      },
      {
        "time": "2018-10-18T21:37:30-06:00",
        "type": "VIEW",
        "properties": {
          "viewedId": "FDJKLHSLD"
        }
      },
      {
        "time": "2018-10-18T21:37:30-06:00",
        "type": "NAVIGATE",
        "properties": {
          "pageFrom": "communities",
          "pageTo": "inventory"
        }
      }
    ]
  },
  {
    "id": "625f5ab26d0820fdd7b9e3d5",
    "userId": "XYZ123ABC",
    "sessionId": "789XYZ123",
    "actions": [
      {
        "time": "2018-10-20T21:37:28-06:00",
        "type": "CLICK",
        "properties": {
          "locationX": 120,
          "locationY": 40
        }
      },
      {
        "time": "2018-10-21T21:37:30-06:00",
        "type": "VIEW",
        "properties": {
          "viewedId": "ABC123"
        }
      }
    ]
  }
] 
  

This endpoint is for adding a single action to a logs list of actions (See Batch endpoint below for batch action uploads). If the UserId and SessionId do not exist, a new Log will be automatically created. Otherwise, if the UserId and SessionId exist in a log the activity is added to the logs activities. The basic JSON request body looks something like this:

{
  "time":"2018-10-24T21:37:28-06:00"
  "type":"CLICK"
  "properties":{
    "locationX": 690,
    "locationY": 105
  }
}

POST /logs/[sessionId]/[userId]

curl -X POST https://openhouse-log-api.herokuapp.com/logs/000ABC123/321CBA000 -H 'Content-Type: application/json' -d '{"time":"2018-10-25T21:37:28-06:00","type":"CLICK","properties":{"locationX": 1001,"locationY": 578}}'

{
  "id":"625f97bea2d4765be423f776",
  "userId":"000ABC123",
  "sessionId":"321CBA000",
  "actions":[{
      "time":"2018-10-25T21:37:28-06:00",
      "type":"CLICK",
      "properties":{
        "locationX":1001,
        "locationY":578
       }
     },
     {
      "time":"2018-10-25T21:37:28-06:00",
      "type":"CLICK",
      "properties":{
        "locationX":1920,
        "locationY":1080
       }
   }]
}

This endpoint is for sending large actions in batchs to be added to a Log. Like the singular addition endpoint, a new Log will be created if the SessionId and UserId do not already exist. The JSON body has a structure like:

{
  "actions": [
    {
      "time":"2018-10-24T21:37:28-06:00"
      "type":"CLICK"
      "properties":{
        "locationX": 690,
        "locationY": 105
      }
    },
    {
      "More": "Actions..."
    }
   ]
}

POST /logs/[sessionId]/[userId]/batch

curl -X POST https://openhouse-log-api.herokuapp.com/logs/000ABC123/321CBA000/batch -H 'Content-Type: application/json' -d '{"actions":[{"time":"2018-10-24T21:37:28-06:00","type":"CLICK","properties":{"locationX": 1001,"locationY": 578}}, {"time":"2018-10-29T21:37:28-06:00","type":"CLICK","properties":{"locationX": 456,"locationY": 123}}]}'

{
  "id":"625f97bea2d4765be423f776",
  "userId":"000ABC123",
  "sessionId":"321CBA000",
  "actions":[{
      "time":"2018-10-25T21:37:28-06:00",
      "type":"CLICK",
      "properties":{
        "locationX":1001,
        "locationY":578
       }
     },
     {
      "time":"2018-10-25T21:37:28-06:00",
      "type":"CLICK",
      "properties":{
        "locationX":1920,
        "locationY":1080
       }
   }]
}

• Provide your comments on how you would make this solution cloud-scalable?
  • I would first start by adding a load balancer to the Heroku app to help deal with the surge of requests. This way we can be sure the app will be able to handle any additional request as the load balancer will scale the application out horizontally until demands are met.
  • Some of the filtering logic could be refined down to use MongoDB's aggregate function to build the query, instead of manual post processing of embedded documents.
• Is your code well tested?
  • Manually yes, but no automated JUnit tests or anything have been made due to time constraints (this was my first foray into Spring Boot API's)
• Is the service performant for the use cases provided?
  • There does seem to be a cold start on the app if it hasn't been called recently, but due to the 5 min interval for batch calls this should keep the application alive between calls so cold start is only a factor on the first request!
• Is the data stored efficiently?
  • Yes, it is layed out in the MongoDB Container in the exact same way the sample log is layed out.
  • I chose MongoDB to make the data a little more straightforward to work with, specifically the array of actions in each log document as the schema needed some flexibility.
• Do you provide error handling and appropriate REST status codes?
  • Somewhat, http status codes return as expected, but error handling could be greatly improved (Again time constraints)
• Is the code deployable?
  • Yes! it's live at https://openhouse-log-api.herokuapp.com/logs/

• I have removed the MongoDB connection string from the repo, if you would like to run it locally you can either create your own MongoDB instance to connect to, or email me and I can give you a temporary user and password for connecting to my existing MongoDB!