modif interpretation results

app.py

@@ -141,6 +141,52 @@ def html_result(badge_text, duration, rating_text, downloads_text, extra_html=""
 </div>
 """.strip()
 
+def normalize_avg_rating_label_fr(label) -> str:
+    """
+    Convertit n'importe quel label avg_rating (EN/FR/variantes) en FR stable.
+    Sorties possibles : "Informations manquantes", "Faible", "Moyen", "Élevé"
+    """
+    if label is None:
+        return "Informations manquantes"
+
+    s = str(label).strip().lower()
+
+    # manquant
+    if "miss" in s or "missing" in s or "none" in s or "no" in s or "nan" in s:
+        return "Informations manquantes"
+    if "info" in s and "manq" in s:
+        return "Informations manquantes"
+
+    # élevé
+    if "high" in s or "élev" in s or "eleve" in s:
+        return "Élevé"
+
+    # moyen
+    if "medium" in s or "moy" in s:
+        return "Moyen"
+
+    # faible
+    if "low" in s or "faibl" in s:
+        return "Faible"
+
+    # fallback (si le modèle renvoie un truc inattendu)
+    return "Informations manquantes"
+
+
+def avg_fr_to_class(avg_fr: str) -> int:
+    """
+    Convertit l'étiquette FR en classe 0..3 pour interpret_results()
+    """
+    s = str(avg_fr).strip().lower()
+    if "manqu" in s:
+        return 0
+    if "faibl" in s:
+        return 1
+    if "moy" in s:
+        return 2
+    if "élev" in s or "eleve" in s:
+        return 3
+    return 0
 
 # =========================
 # INTERPRETATION (COMMUNE)
@@ -449,10 +495,11 @@ def predict_from_freesound_url(url: str):
     if MIN_EFFECT <= duration <= MAX_EFFECT:
         badge = "🔊 Effet sonore (URL → features API)"
         dl_class = int(predict_with_model_fs(xgb_effect_num, sound, xgb_effect_feat_num))
-        avg_text = str(predict_with_model_fs(xgb_effect_avg, sound, xgb_effect_feat_avg, le_effect_avg))
+        avg_text_raw = str(predict_with_model_fs(xgb_effect_avg, sound, xgb_effect_feat_avg, le_effect_avg))
+        avg_text = normalize_avg_rating_label_fr(avg_text_raw)
+        avg_class = avg_fr_to_class(avg_text)
         dl_text = NUM_DOWNLOADS_MAP_FR.get(dl_class, str(dl_class))
 
-        avg_class = avg_label_to_class(avg_text)
         conclusion = interpret_results(avg_class, dl_class)
 
         extra = f"""
@@ -467,10 +514,12 @@ def predict_from_freesound_url(url: str):
     if MIN_MUSIC <= duration <= MAX_MUSIC:
         badge = "🎵 Musique (URL → features API)"
         dl_class = int(predict_with_model_fs(xgb_music_num, sound, xgb_music_feat_num))
-        avg_text = str(predict_with_model_fs(xgb_music_avg, sound, xgb_music_feat_avg, le_music_avg))
+        avg_text_raw = str(predict_with_model_fs(xgb_music_avg, sound, xgb_music_feat_avg, le_music_avg))
+        avg_text = normalize_avg_rating_label_fr(avg_text_raw)
+        avg_class = avg_fr_to_class(avg_text)
+
         dl_text = NUM_DOWNLOADS_MAP_FR.get(dl_class, str(dl_class))
 
-        avg_class = avg_label_to_class(avg_text)
         conclusion = interpret_results(avg_class, dl_class)
 
         extra = f"""
@@ -843,10 +892,10 @@ def predict_from_metadata_url(url: str):
     pred_downloads_text = NUM_DOWNLOADS_MAP.get(pred_num_downloads_val, str(pred_num_downloads_val))
 
     # avg rating (label)
-    pred_avg_rating_label = predict_with_model_meta(model_ar, df_for_model, le=current_le)
+    pred_avg_rating_label_raw = predict_with_model_meta(model_ar, df_for_model, le=current_le)
+    pred_avg_rating_label = normalize_avg_rating_label_fr(pred_avg_rating_label_raw)
+    avg_class = avg_fr_to_class(pred_avg_rating_label)
 
-    # 8) transformer en classes (pour interprétation + cohérence visuelle)
-    avg_class = avg_label_to_class(pred_avg_rating_label)
     dl_class = int(pred_num_downloads_val) if isinstance(pred_num_downloads_val, (int, np.integer)) else 0
 
     rating_display = str(pred_avg_rating_label)