Hotel Search API with AI

Experimental project.

Goal

• Small modular frontend with React, Tailwind and TypeScript
• Small backend with Express, TypeScript and AI

Details

• Frontend
• Backend

How to use

• Start client
• Start backend
• Open browser
• Enter an idea
• Get results

Note

Mental model

• Render client
  • Enter prompt
    • Backend get suggestions for autocompletion
    • Send search text
      • AI parse it
        • Handle error if occures
          • Render error
        • Auth to Hotel Search API
          • Handle error
          • Get Location ID
            • Handle error
            • Get coordinates based on the location ID
              • Handle error
              • Build search payload from AI results and retrieved API results
                • Handle errors
                • Trigger hotel search
                  • Render results

Disclaimer

• The code is experimental
• The backend uses a demo API endpiont which is inconsistent
• No requirements for erorr handling strategy
• No requirements for logging
• No requirements for technologies, other than JavaScript, React and Express
• The client use Axios for HTTP endpoints
• The server use Axios and Node's Fetch for HTTP endpoints

Known issues

• Loading animation is not the most perfect
• No dark mode
• Hotel search API might not respond properly, discovered responses:
  • Empty, no payload
  • Timeout (after 8 seconds)
  • Timeout (after 3 seconds)
  • Empty list
  • Text, Two new line character
  • Text, Two separated hotel list but as text, the two list is just two JSON array as text, near each other
  • JSON object with 3 different hotel list (with redundant elements)
  • JSON object with 2 different hotel list (no redundant element)
  • JSON object with 1 hotel list
• Experimental AI data parsing sometime corrupts the JSON structure which stops the process (mitigated)