Create `service` endpoint

[GabrielEValenzuela]

Description

Develop an API endpoint within the FastAPI library that receives a JSON payload containing a query, processes it with the ML model, and returns the probability score based on the model’s similarity detection within a knowledge graph. The endpoint should validate input, handle exceptions, has query limit, and format the response as a JSON object with the prediction result.

User Stories

• As an API Client, I want to send a JSON query to the service and receive a probability score, so I can determine if two subgraphs are similar.
• As a Developer, I want the endpoint to be well-documented and robust, so it handles a variety of valid inputs and gracefully manages errors.

---

Details

• Objective: Create an endpoint /service that takes a JSON query, invokes the ML module for similarity detection, and returns a probability score indicating the likelihood of similarity between subgraphs.
• Requirements:
  • Endpoint: Define a POST endpoint /service under src/app/api/routers/service.py.
  • Input Validation: Ensure the JSON payload includes required fields for the ML model (e.g., subgraph1 and subgraph2 data).
  • Output: Return a JSON response with the probability score, formatted as:

    {
    "probability": <float>
    }

  • Error Handling:
    • If input validation fails, return a 400 Bad Request with a relevant error message.
    • If the ML module fails, return a 500 Internal Server Error.
    • If the query limit has been reached, return a 429 To Many Request

---

Example Usage and Response

• Request (JSON):

  {
  "input": [{...}]
  }

• Response (JSON):

  {
  "probability": 0.85,
  }

• Error Response (JSON):

  {
  "error": "Invalid input: missing subgraph data."
  }

---

Mockup of Endpoint Implementation

from fastapi import APIRouter, HTTPException
from pydantic import BaseModel
from src.ml.{{ module_name }}.modeling.predict import predict_similarity

router = APIRouter()

class SimilarityRequest(BaseModel):
    subgraph: list(dict)

@router.post("/service")
async def predict_similarity_endpoint(request: SimilarityRequest):
    try:
        # Logic for valid API key and rate limit

        # Invoke ML model prediction
        probability = predict_similarity(request.subgraph)

        # Return JSON response with the probability score
        return {
            "probability": probability,
        }
    except ValueError as e:
        # Handle invalid input
        raise HTTPException(status_code=400, detail=str(e))
    except Exception as e:
        # Handle general errors
        raise HTTPException(status_code=500, detail="Internal server error.")

---

Edge Cases

• Invalid JSON Format: Return a 400 Bad Request if the JSON payload is malformed.
• Missing Required Fields: Return a 400 Bad Request if subgraph is missing.
• ML Model Errors: Return a 500 Internal Server Error if there’s an issue with the ML model, such as a missing model file or processing error.

---