Next.js Endpoint: /api/get-cities?query={query}
This is an example of how to use the OpenAI chat completion API to provide suggested city names to a frontend using a Next.js API route. Firebase is used to store all searches from OpenAI. Care is taken to throttle the number of requests to OpenAI, with a 10-second interval that can be adjusted.

• Model used: gpt-3.5-turbo
• Range of tokens used per request: 150 to 200
• Query must be at least 3 letters long and no more than 20 letters long
• Queries are limited to a maximum of 2 words per request

Start your application

Please fill in your OpenAI and Firebase credentials in the .env.local file

FIREBASE_apiKey=
FIREBASE_authDomain=
FIREBASE_projectId=
FIREBASE_messagingSenderId=
FIREBASE_appId=
FIREBASE_measurementId=

OPENAI_API_KEY=

To start the application, run the following commands:

npm i
npm run dev

Adjust the number of suggestions and interval of requests in the index.js file:

const maxSuggestions = 5;
const openAIInterval = 10 * 1000; // call OpenAI API once per 10 seconds

Note for learners

1. To learn how to implement the chatGPT feature, please refer to the openai-city-search.js file.
2. Only one API route is created: get-cities.js. In this file, I used logic to call the OpenAI API only when necessary.

Test cases

Endpoint: /api/get-cities?query={query}

1. /api/get-cities?query=chennai

{
  "matchType": "Exact match",
  "cities": [
    "Chennai in India"
  ]
}

1. /api/get-cities?query=new

{
  "matchType": "Partial match",
  "cities": [
    "New Delhi in India",
    "New York City in United States",
    "New South Wales in Australia",
    "New Orleans in United States",
    "Newcastle upon Tyne in United Kingdom"
  ]
}

1. /api/get-cities?query=sponge+bob

{
  "matchType": "No match",
  "cities": []
}