Clean FastAPI evaluation endpoints

main.py

@@ -1,60 +1,133 @@
 from fastapi import FastAPI, HTTPException
 from pydantic import BaseModel
 from typing import List
-from src.database import init_db, save_evaluation
 import traceback
-from src.aggregator import evaluate_all, evaluate_generation_only
+
+from src.database import (
+    init_db,
+    save_evaluation,
+    get_all_evaluations
+)
+
+from src.aggregator import (
+    evaluate_all,
+    evaluate_generation_only
+)
+
+
+# =========================================================
+# FastAPI Application
+# =========================================================
 
 app = FastAPI(
     title="LLM Evaluation & Hallucination Detection Framework",
     version="2.0.0"
 )
 
+
+# =========================================================
+# Request Models
+# =========================================================
+
 class EvalRequest(BaseModel):
     question: str
     retrieved_contexts: List[str]
     llm_response: str
 
-class EvalResponse(BaseModel):
-    final_verdict: str
-    retrieval_evaluation: dict
-    generation_evaluation: dict
 
 class LLMOnlyEvalRequest(BaseModel):
     question: str
     context: str
     llm_response: str
 
+
+# =========================================================
+# Response Model for Full RAG Evaluation
+# =========================================================
+
+class EvalResponse(BaseModel):
+    final_verdict: str
+    retrieval_evaluation: dict
+    generation_evaluation: dict
+
+
+# =========================================================
+# Full RAG Evaluation
+# =========================================================
+
 @app.post("/evaluate", response_model=EvalResponse)
 def evaluate(request: EvalRequest):
+
+    # -----------------------------
+    # Validate input
+    # -----------------------------
+
     if not request.question.strip():
-        raise HTTPException(status_code=400, detail="Question cannot be empty")
+        raise HTTPException(
+            status_code=400,
+            detail="Question cannot be empty"
+        )
+
     if not request.retrieved_contexts:
-        raise HTTPException(status_code=400, detail="Retrieved contexts cannot be empty")
+        raise HTTPException(
+            status_code=400,
+            detail="Retrieved contexts cannot be empty"
+        )
+
     if not request.llm_response.strip():
-        raise HTTPException(status_code=400, detail="LLM response cannot be empty")
+        raise HTTPException(
+            status_code=400,
+            detail="LLM response cannot be empty"
+        )
 
     try:
+
+        # -----------------------------
+        # Full RAG evaluation
+        # -----------------------------
+
         result = evaluate_all(
             question=request.question,
             retrieved_contexts=request.retrieved_contexts,
             llm_response=request.llm_response
         )
+
+        # -----------------------------
+        # Save evaluation
+        # -----------------------------
+
         save_evaluation(
-            context=" ".join(request.retrieved_contexts),
+            context=" ".join(
+                request.retrieved_contexts
+            ),
             question=request.question,
             llm_response=request.llm_response,
             result=result
         )
+
         return result
+
     except Exception as e:
-        raise HTTPException(status_code=500, detail=str(e) + "\n" + traceback.format_exc())
 
-init_db()
+        raise HTTPException(
+            status_code=500,
+            detail=str(e)
+            + "\n"
+            + traceback.format_exc()
+        )
+
+
+# =========================================================
+# LLM-ONLY Evaluation
+# =========================================================
 
 @app.post("/evaluate-llm")
 def evaluate_llm(request: LLMOnlyEvalRequest):
 
+    # -----------------------------
+    # Validate input
+    # -----------------------------
+
     if not request.question.strip():
         raise HTTPException(
             status_code=400,
@@ -75,12 +148,20 @@ def evaluate_llm(request: LLMOnlyEvalRequest):
 
     try:
 
+        # -----------------------------
+        # Generation-only evaluation
+        # -----------------------------
+
         result = evaluate_generation_only(
             question=request.question,
             context=request.context,
             llm_response=request.llm_response
         )
 
+        # -----------------------------
+        # Save evaluation
+        # -----------------------------
+
         save_evaluation(
             context=request.context,
             question=request.question,
@@ -94,46 +175,37 @@ def evaluate_llm(request: LLMOnlyEvalRequest):
 
         raise HTTPException(
             status_code=500,
-            detail=str(e) + "\n" + traceback.format_exc()
+            detail=str(e)
+            + "\n"
+            + traceback.format_exc()
         )
 
 
+# =========================================================
+# Home
+# =========================================================
+
 @app.get("/")
 def home():
-    return {"message": "LLM Evaluation Framework v2.0 is running"}
 
-@app.post("/evaluate", response_model=EvalResponse)
-def evaluate(request: EvalRequest):
-
-    if not request.question.strip():
-        raise HTTPException(status_code=400, detail="Question cannot be empty")
-    if not request.retrieved_contexts:
-        raise HTTPException(status_code=400, detail="Retrieved contexts cannot be empty")
-    if not request.llm_response.strip():
-        raise HTTPException(status_code=400, detail="LLM response cannot be empty")
-
-    result = evaluate_all(
-        question=request.question,
-        retrieved_contexts=request.retrieved_contexts,
-        llm_response=request.llm_response
-    )
-
-    save_evaluation(
-        context=" ".join(request.retrieved_contexts),
-        question=request.question,
-        llm_response=request.llm_response,
-        result=result
-    )
+    return {
+        "message": "LLM Evaluation Framework v2.0 is running"
+    }
 
-    return result
 
-from src.database import get_all_evaluations
+# =========================================================
+# History
+# =========================================================
 
 @app.get("/history")
 def history():
+
     rows = get_all_evaluations()
+
     results = []
+
     for row in rows:
+
         results.append({
             "id": row[0],
             "question": row[2],
@@ -141,4 +213,15 @@ def history():
             "final_verdict": row[4],
             "created_at": row[11]
         })
-    return {"total": len(results), "evaluations": results}
+
+    return {
+        "total": len(results),
+        "evaluations": results
+    }
+
+
+# =========================================================
+# Initialize Database
+# =========================================================
+
+init_db()