Fix: Add robust lang detection

app.py

@@ -100,15 +100,42 @@ def _predict_np2_compat(self, text, k=1, threshold=0.0, on_unicode_error='strict
 lid_model.predict = types.MethodType(_predict_np2_compat, lid_model)
 
 ### Check if lang is english #####################################################
-def is_eng(review: str):
-    lang_preds = lid_model.predict(review, k=3)
+def is_eng(text: str, k: int = 3, threshold: float = 0.1):
+    
+    out = lid_model.predict(text, k=k)
 
-    print(lang_preds)
+    # Normalisieren auf zwei Listen: labels[], probs[]
+    labels, probs = [], []
+
+    # Fall A: (labels, probs)
+    if isinstance(out, tuple) and len(out) == 2:
+        labels, probs = out
+
+    # Fall B: [(prob, '__label__xx'), ...]
+    elif (
+        isinstance(out, (list, tuple))
+        and len(out) > 0
+        and isinstance(out[0], (list, tuple))
+        and len(out[0]) == 2
+        and isinstance(out[0][1], str)
+    ):
+        probs, labels = zip(*out)            # entpacken
+        labels, probs = list(labels), list(probs)
+
+    # Fall C: ['__label__en', '__label__de', ...]  (ohne Probs)
+    elif isinstance(out, (list, tuple)) and (len(out) == 0 or isinstance(out[0], str)):
+        labels = list(out)
+        probs = [1.0] * len(labels)          # Dummy-Prob, falls nicht geliefert
+
+    else:
+        # Unbekanntes Format
+        return True, 0.0
+
+    if "__label__en" in labels:
+        i = labels.index("__label__en")
+        p = float(probs[i])
+        return (p >= threshold), p
 
-    for prob, label in lang_preds:
-        if label == "__label__en":
-            return True, float(prob)
-    
     return False, 0.0
 
 ### Do actual prediction #########################################################