Update app.py

app.py

@@ -1,7 +1,9 @@
 import gradio as gr
 import torch
 from transformers import pipeline as hf_pipeline, AutoModelForSequenceClassification, AutoTokenizer
-
+from PIL import Image
+import pytesseract
+import io
 
 # ——— 1) Emotion Pipeline ————————————————————————————————————————————————
 emotion_pipeline = hf_pipeline(
@@ -12,11 +14,7 @@ emotion_pipeline = hf_pipeline(
 )
 
 def get_emotion_profile(text):
-    """
-    Returns a dict of { emotion_label: score } for the input text.
-    """
     results = emotion_pipeline(text)
-    # some pipelines return [[…]]
     if isinstance(results, list) and isinstance(results[0], list):
         results = results[0]
     return {r["label"].lower(): round(r["score"], 3) for r in results}
@@ -39,6 +37,7 @@ THRESHOLDS = {
     "projection":      0.09, "recovery phase":           0.33, "threat":    0.15
 }
 
+
 # ——— 3) Emotional-tone tagging (no abuse_score / DARVO) —————————————————————————————
 def get_emotional_tone_tag(emotion_profile, patterns):
     anger   = emotion_profile.get("anger",   0)
@@ -60,7 +59,7 @@ def get_emotional_tone_tag(emotion_profile, patterns):
     if (anger + disgust) > 0.5 and any(p in patterns for p in ["insults", "control", "threat"]):
         return "confrontational"
 
-    # 4) Manipulative: neutral high + classic manipulation patterns
+    # 4) Manipulative: neutral high + manipulation patterns
     if neutral > 0.4 and any(p in patterns for p in ["gaslighting", "dismissiveness", "projection", "guilt tripping", "blame shifting"]):
         return "manipulative"
 
@@ -79,38 +78,38 @@ def get_emotional_tone_tag(emotion_profile, patterns):
     return None
 
 
-# ——— 3) Single-message analysis ——————————————————————————————————————————————
+# ——— 4) Single-message analysis —————————————————————————————————————————————
 def analyze_message(text):
-    """
-    Runs emotion profiling, and the abuse-pattern classifier.
-    Returns a dict with:
-      - emotion_profile: { emotion: score }
-      - active_patterns: [ labels above their threshold ]
-    """
-    
+    # 1) emotion profiling
     emotion_profile = get_emotion_profile(text)
 
-    # get raw model scores
+    # 2) abuse-pattern classification
     toks = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
     with torch.no_grad():
         logits = model(**toks).logits.squeeze(0)
     scores = torch.sigmoid(logits).cpu().numpy()
 
-    # pick up all labels whose score >= threshold
-    active = [lab for lab, sc in zip(LABELS, scores) if sc >= THRESHOLDS[lab]]
+    # 3) identify active patterns
+    active_patterns = [
+        label for label, prob in zip(LABELS, scores)
+        if prob >= THRESHOLDS[label]
+    ]
 
+    # 4) tone tagging
     tone_tag = get_emotional_tone_tag(emotion_profile, active_patterns)
+
     return {
-      "emotion_profile": emotion_profile,
-      "active_patterns": active_patterns,
-      "tone_tag": tone_tag     # ← make sure this line is present
-}
+        "emotion_profile": emotion_profile,
+        "active_patterns": active_patterns,
+        "tone_tag": tone_tag
+    }
 
 
+# ——— 5) Composite wrapper (handles .txt or image + text boxes) ——————————————————————
 def analyze_composite(uploaded_file, *texts):
     outputs = []
 
-    # 1) Uploaded file block
+    # file upload
     if uploaded_file is not None:
         raw = uploaded_file.read()
         name = uploaded_file.name.lower()
@@ -126,21 +125,21 @@ def analyze_composite(uploaded_file, *texts):
         r = analyze_message(content)
         outputs.append(
             "── Uploaded File ──\n"
-            f"Emotion Profile   : {r['emotion_profile']}\n"
-            f"Active Patterns   : {r['active_patterns']}\n"
-            f"Emotional Tone : {r['tone_tag']}\n"
+            f"Emotion Profile : {r['emotion_profile']}\n"
+            f"Active Patterns : {r['active_patterns']}\n"
+            f"Emotional Tone  : {r['tone_tag']}\n"
         )
 
-    # 2) Text-box blocks
+    # text inputs
     for idx, txt in enumerate(texts, start=1):
         if not txt:
             continue
         r = analyze_message(txt)
         outputs.append(
             f"── Message {idx} ──\n"
-            f"Emotion Profile   : {r['emotion_profile']}\n"
-            f"Active Patterns   : {r['active_patterns']}\n"
-            f"Emotional Tone : {r['tone_tag']}\n"
+            f"Emotion Profile : {r['emotion_profile']}\n"
+            f"Active Patterns : {r['active_patterns']}\n"
+            f"Emotional Tone  : {r['tone_tag']}\n"
         )
 
     if not outputs:
@@ -148,19 +147,17 @@ def analyze_composite(uploaded_file, *texts):
 
     return "\n".join(outputs)
 
+
 # ——— 6) Gradio interface ————————————————————————————————————————————————
 message_inputs = [gr.Textbox(label=f"Message {i+1}") for i in range(3)]
 
 iface = gr.Interface(
     fn=analyze_composite,
-    inputs=[ 
-        gr.File(file_types=[".txt", ".png", ".jpg", ".jpeg"], 
-                label="Upload text or image") 
-    ] + message_inputs,
+    inputs=[gr.File(file_types=[".txt", ".png", ".jpg", ".jpeg"], label="Upload text or image")] + message_inputs,
     outputs=gr.Textbox(label="Analysis"),
-    title="Tether Analyzer (with Tone Tags)",
-    description="Extracts motifs, emotions, patterns—and now an emotional tone tag—no abuse score or DARVO."
+    title="Tether Analyzer (streamlined)",
+    description="Emotion profiling, pattern tags, and tone tagging—no motif detection, no abuse score or DARVO."
 )
 
 if __name__ == "__main__":
-    iface.launch()
+    iface.launch()