Adapt Engineering Take-home

How to run

yarn install

yarn run --silent start

What it does

index.ts has a main function which takes an "input", supplied in the original take-home assignment spec.

[{
    "carrier": "MOCK_INDEMNITY",
    "customerId": "a0dfjw9a"
},{
  "carrier": "PLACEHOLDER_CARRIER",
    "customerId": "f02dkl4e"
}]

The script fetches a website associated with the given carrier and returns a json response containing data found for the given customer id.

Organization

There is a scrapers directory which exposes various scraping strategies in the form of a module. Each strategy is specific to a particular carrier. The strategy contains all the logic for scraping data from that carrier.

Each scraper module exposes a 'scrapeCustomer' function. Each scraper module could conceivably expose many other methods to scrape data related to that carrier.

Each scraper strategy is free to use its own methods for scraping data.

I separated "how" to fetch from the individual scrapers. some scrapers may use the same library; others my have radically different ways of retrieving data. Maybe insurance policy updates are available via a simple api call; maybe they get sent to an email address which must be checked? At a very high, hand-wavy level, this basic structure should accommodate such challenges.

Similarly, the parsing library and methodology is up to the individual scraper. Maybe cheerio is not the right tool, maybe the data is not HTML; maybe it's JSON or a PDF.

Constraints

Most of my time was eaten up figuring out how to parse the PLACEHOLDER_CARRIER carrier url, and re-familiarizing myself with cheerio/jquery.

Typescript

Due to time considerations, I did not attempt to specify a robust type language for the final data format. Some of what I would have liked to do is merely implied by the format of the response from the scrapeCustomer function, that is, something along the lines of

type Agency = {
    name: string
    code: string
}
type Agent = {
    name: string
    producerCode: string
    agency: Agency
}
type Policy = {
  id: string
  premium: number
  status: string
  effectiveDate: Date
  terminationDate: Date
  lastPaymentDate: Date
}
type Customer = {
    name: string
    id: string
    email: string
    address: string
    agent: Agent
    policies: [Policy]
}

Exception Handling

Additionally, and probably more grievously, I did little to no exception or error handling, and there are plenty of places to do it. Namely:

url retrieval/fetch how to recover if there are network errors, incorrect urls, etc.
exceptions in parsing the fetched documents websites change all the time, and a production version of this would likely encounter exceptions regularly as sites are updated, edge cases emerge, fields are added or removed. My parsing code is very naive.
also no account for missing data again, I am relying on this being a time-limited, constrained toy demo

Omitted fields

The PLACEHOLDER_CARRIER carrier site had some additional fields for the policies that the MOCK_INDEMNITY site did not. I opted not to include these. Again, mostly a consideration of time and hemming and hawing on how I would handle it on a typescript level.

scelerat/adapt-api-scraper