Update app.py

app.py

@@ -1,12 +1,12 @@
-: import gradio as gr
+import gradio as gr
 import re
 from transformers import pipeline as hf_pipeline
 
-# Load SST model (temporary baseline)
+# Load SST model
 sst_classifier = hf_pipeline(
     "text-classification",
     model="distilbert-base-uncased-finetuned-sst-2-english",
-    top_k=None,
+    top_k=1,
     truncation=True
 )
 
@@ -18,7 +18,7 @@ emotion_pipeline = hf_pipeline(
     truncation=True
 )
 
-# Lexicon enhancement preprocessing
+# Lexicon rules
 negations = {"not", "never", "no", "none", "nobody", "nothing", "neither", "nowhere", "hardly", "scarcely", "barely"}
 amplifiers = {"very", "really", "extremely", "so", "totally", "completely", "absolutely", "utterly", "super"}
 softeners = {"slightly", "somewhat", "a bit", "a little", "mildly", "fairly", "kind of"}
@@ -49,14 +49,12 @@ def preprocess_sentiment_text(text):
 
     return " ".join(modified)
 
-# Emotion mapping
 def get_emotion_profile(text):
     emotions = emotion_pipeline(text)
     if isinstance(emotions, list) and isinstance(emotions[0], list):
         emotions = emotions[0]
     return {e['label'].lower(): round(e['score'], 3) for e in emotions}
 
-# Tone tagging logic
 def get_emotional_tone_tag(emotions, sentiment, patterns, abuse_score=0):
     sadness = emotions.get("sadness", 0)
     joy = emotions.get("joy", 0)
@@ -82,19 +80,22 @@ def get_emotional_tone_tag(emotions, sentiment, patterns, abuse_score=0):
 
     return None
 
-# Main function
 def analyze_message(text):
     preprocessed = preprocess_sentiment_text(text)
-    sst_output = sst_classifier(preprocessed)
-    sentiment = sst_output[0]
-    sentiment_label = "supportive" if sentiment["label"] == "POSITIVE" else "undermining"
-    sentiment_score = round(sentiment["score"] * 100, 2)
 
+    # Run sentiment classification
+    sst_result = sst_classifier(preprocessed)
+    sentiment_output = sst_result[0] if isinstance(sst_result, list) else sst_result
+    sentiment_label = "supportive" if sentiment_output["label"] == "POSITIVE" else "undermining"
+    sentiment_score = round(sentiment_output["score"] * 100, 2)
+
+    # Run emotion model
     emotions = get_emotion_profile(text)
     emotion_summary = "\n".join([f"{k.title()}: {v:.2f}" for k, v in emotions.items()])
 
-    # Temporarily pass empty abuse pattern list until Tether model is added
-    tone_tag = get_emotional_tone_tag(emotions, sentiment_label, patterns=[])
+    # No abuse patterns yet
+    patterns = []
+    tone_tag = get_emotional_tone_tag(emotions, sentiment_label, patterns)
     tone_output = tone_tag if tone_tag else "None detected"
 
     return (
@@ -102,13 +103,13 @@ def analyze_message(text):
         f"🎭 Emotional Profile:\n{emotion_summary}\n\n"
         f"🔍 Tone Tag: {tone_output}"
     )
-# Interface
+
 iface = gr.Interface(
     fn=analyze_message,
     inputs=gr.Textbox(lines=4, placeholder="Paste a message here..."),
     outputs="text",
     title="Tether SST + Emotional Tone Tagger",
-    description="Applies lexicon-enhanced preprocessing, classifies sentiment, profiles emotion, and infers tone tags based on behavior logic."
+    description="Applies lexicon-enhanced preprocessing, classifies sentiment, profiles emotion, and infers tone tags using behavioral logic."
 )
 
 iface.launch()