Update app.py

app.py

@@ -34,26 +34,38 @@ def _build_vehicle_recognition_markdown(vision_results: dict[str, Any]) -> str:
     confidence = float(vision_results.get("confidence", 0.0) or 0.0)
     method = vision_results.get("method", "fallback")
 
-    confidence_warning = ""
-    if confidence < 0.5 and predicted_class != "Unknown":
-        confidence_warning = "\n⚠️ **Low confidence**: please verify the result manually."
-
-    method_note = ""
+    # Case A: Image analyzed by vision model
     if method == "local_transfer_model":
-        method_note = "\nThe classifier can only predict one of the trained vehicle brands."
-    elif method == "fallback":
-        method_note = "\nNo trained model found. Please train the model first."
+        confidence_warning = ""
+        if confidence < 0.5:
+            confidence_warning = "\n\nThe image model is uncertain. Please verify the result manually."
+
+        return f"""
+### 1. Vehicle recognition – Computer Vision
+
+Detected brand/model group: **{predicted_class}**
+
+The image analysis provides a coarse vehicle brand/model prediction based on deep learning. It is not exact technical vehicle identification.{confidence_warning}
+""".strip()
+
+    # Case B: Manual input without image
     elif method == "manual_input":
-        method_note = "\nCar brand/model entered manually without image analysis."
+        return f"""
+### 1. Vehicle recognition – Manual Input
 
-    return f"""
-### Vehicle recognition
+Vehicle used for the estimate: **{predicted_class}**
 
-- Predicted vehicle brand: **{predicted_class}**
-- Confidence: **{confidence:.2f}**
-- Method: **{method}**
+The price estimate is based on your manual vehicle input and structured used-car listing data.
+""".strip()
+
+    # Case C: Unknown or fallback
+    else:
+        return f"""
+### 1. Vehicle recognition
 
-The image classifier provides a coarse vehicle brand/model prediction, not exact technical vehicle identification.{confidence_warning}{method_note}
+Vehicle: **Unknown**
+
+Please upload a car image or enter a known make/model to get a price estimate.
 """.strip()
 
 
@@ -63,40 +75,44 @@ def _build_price_estimate_markdown(price_prediction: dict[str, Any]) -> str:
     upper_bound = format_currency_chf(price_prediction.get("upper_bound"))
 
     return f"""
-### Price estimate
+### 2. Price estimate – ML Numeric Data
 
 - Estimated market price: **{estimated_price}**
-- Price range: **{lower_bound} - {upper_bound}**
+- Expected price range: **{lower_bound} – {upper_bound}**
 
-The estimate is based on structured used-car listing data.
+This estimate is based on structured used-car listing data.
 """.strip()
 
 
-def _build_budget_assessment_markdown(recommendation: dict[str, Any]) -> str:
-    return f"""
-### Budget assessment
-
-- Assessment: **{recommendation.get('price_budget_assessment', 'n/a')}**
-- Reason: {recommendation.get('price_budget_reason', 'n/a')}
-""".strip()
-
-
-def _build_financing_markdown(recommendation: dict[str, Any]) -> str:
+def _build_purchase_assessment_markdown(recommendation: dict[str, Any]) -> str:
     rough_months_needed = recommendation.get("rough_months_needed")
     financing_gap = recommendation.get("financing_gap")
 
     months_text = "n/a" if rough_months_needed in {None, "", 0} else f"{float(rough_months_needed):.1f} months"
     gap_text = "n/a" if financing_gap in {None, ""} else format_currency_chf(financing_gap)
 
+    budget_assessment = recommendation.get('price_budget_assessment', 'n/a')
+    budget_reason = recommendation.get('price_budget_reason', '')
+    financing_orientation = recommendation.get('financing_orientation', 'n/a')
+    financing_reason = recommendation.get('financing_reason', '')
+    explanation = recommendation.get('full_explanation', 'n/a')
+
     return f"""
-### Simple financing orientation
+### 3. Purchase assessment – NLP Explanation
+
+**Budget assessment:** {budget_assessment}
+{budget_reason}
+
+**Simple financing orientation:** {financing_orientation}
+- Financing gap: {gap_text}
+- Rough months needed: {months_text}
+{financing_reason}
 
-- Orientation: **{recommendation.get('financing_orientation', 'n/a')}**
-- Financing gap: **{gap_text}**
-- Rough months needed: **{months_text}**
-- Note: {recommendation.get('financing_reason', 'n/a')}
+**Assessment:**
+{explanation}
 
-No interest rates, leasing offers, or true financing products are simulated here.
+---
+**Disclaimer:** This output is a non-binding orientation only. It does not replace a technical vehicle inspection, financial advice or an insurance quote.
 """.strip()
 
 
@@ -147,12 +163,9 @@ def _run_advisor(
     # Allow either image OR manual make_model input, but require at least one
     if image is None and not (make_model and make_model.strip()):
         return (
-            "Please upload a car image OR enter a known make/model.",
-            "No price estimate available yet.",
-            "No budget assessment available yet.",
-            "No financing orientation available yet.",
-            "Please provide car details to start the analysis.",
-            "No disclaimer available yet.",
+            "Please upload a car image OR enter a known make/model to start the analysis.",
+            "",
+            "",
             {},
         )
 
@@ -197,16 +210,13 @@ def _run_advisor(
     return (
         _build_vehicle_recognition_markdown(vision_results),
         _build_price_estimate_markdown(price_prediction),
-        _build_budget_assessment_markdown(recommendation),
-        _build_financing_markdown(recommendation),
-        recommendation.get("full_explanation", "n/a"),
-        recommendation.get("disclaimer", "n/a"),
+        _build_purchase_assessment_markdown(recommendation),
         _build_debug_payload(vision_results, price_prediction, recommendation),
     )
 
 
 def _clear_outputs():
-    return "", "", "", "", "", "", {}
+    return "", "", "", {}
 
 
 def build_interface() -> gr.Blocks:
@@ -244,10 +254,7 @@ Image upload -> analyze_car_image() -> predicted_class -> predict_price() -> est
             with gr.Column(scale=1):
                 vehicle_recognition = gr.Markdown(label="Vehicle recognition")
                 price_estimate = gr.Markdown(label="Price estimate")
-                budget_assessment = gr.Markdown(label="Budget assessment")
-                financing_orientation = gr.Markdown(label="Simple financing orientation")
-                ai_description = gr.Markdown(label="AI-generated description")
-                disclaimer = gr.Markdown(label="Disclaimer")
+                purchase_assessment = gr.Markdown(label="Purchase assessment")
 
                 with gr.Accordion("Debug details", open=False):
                     debug_output = gr.JSON(label="Raw pipeline output")
@@ -258,10 +265,7 @@ Image upload -> analyze_car_image() -> predicted_class -> predict_price() -> est
             outputs=[
                 vehicle_recognition,
                 price_estimate,
-                budget_assessment,
-                financing_orientation,
-                ai_description,
-                disclaimer,
+                purchase_assessment,
                 debug_output,
             ],
         )
@@ -272,10 +276,7 @@ Image upload -> analyze_car_image() -> predicted_class -> predict_price() -> est
             outputs=[
                 vehicle_recognition,
                 price_estimate,
-                budget_assessment,
-                financing_orientation,
-                ai_description,
-                disclaimer,
+                purchase_assessment,
                 debug_output,
             ],
         )