Update app.py

app.py

@@ -154,13 +154,42 @@ ESCALATION_QUESTIONS = [
 ]
 
 def analyze_single_message(text, thresholds, motif_flags):
+    motif_hits, matched_phrases = detect_motifs(text)
+    sentiment = {"label": "undermining"}  # fallback in case sentiment fails
+    try:
+        input_ids = sentiment_tokenizer(f"emotion: {text}", return_tensors="pt").input_ids
+        with torch.no_grad():
+            outputs = sentiment_model.generate(input_ids)
+        emotion = sentiment_tokenizer.decode(outputs[0], skip_special_tokens=True).strip().lower()
+        sentiment = {
+            "label": EMOTION_TO_SENTIMENT.get(emotion, "undermining"),
+            "emotion": emotion
+        }
+    except:
+        sentiment["emotion"] = "unknown"
+
+    sentiment_score = 0.5 if sentiment["label"] == "undermining" else 0.0
+    contradiction_flag = detect_contradiction(text)
+    motifs = [phrase for _, phrase in matched_phrases]
+
     inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
     with torch.no_grad():
         outputs = model(**inputs)
     scores = torch.sigmoid(outputs.logits.squeeze(0)).numpy()
-    threshold_labels = [label for label, score in zip(LABELS, scores) if score > thresholds[label]]
+
+    labels = [label for label, score in zip(LABELS, scores) if score > thresholds[label]]
     top_patterns = sorted([(label, score) for label, score in zip(LABELS, scores)], key=lambda x: x[1], reverse=True)[:2]
-    return (np.mean([score for _, score in top_patterns]) * 100, threshold_labels, top_patterns, 0.0, {"label": "supportive"})
+    pattern_labels = [label for label, _ in top_patterns]
+
+    darvo_score = calculate_darvo_score(pattern_labels, 0.0, sentiment_score, motifs, contradiction_flag)
+
+    return (
+        np.mean([score for _, score in top_patterns]) * 100,
+        labels,
+        top_patterns,
+        darvo_score,
+        sentiment
+    )
 
 # --- Composite Analysis with Escalation Quiz ---
 def analyze_composite(msg1, msg2, msg3, *answers_and_none):