Update app.py

app.py

@@ -1,111 +1,91 @@
-# app.py - very simple Flask emotion detector
-# English (joblib) + Sinhala (joblib) + Tamil (transformers)
-
-from flask import Flask, render_template, request
+import gradio as gr
 import joblib
 from huggingface_hub import hf_hub_download
 from transformers import pipeline
 
-app = Flask(__name__)
-
 # ────────────────────────────────────────────────
-# English model (already joblib)
+# 1. Load Models & Components
 # ────────────────────────────────────────────────
-print("Loading English model...")
-en_vectorizer = joblib.load(hf_hub_download(
-    "E-motionAssistant/Englsih_Trained_Model_LR",
-    "tfidf_vectorizer.joblib"
-))
-en_classifier = joblib.load(hf_hub_download(
-    "E-motionAssistant/Englsih_Trained_Model_LR",
-    "logreg_model.joblib"
-))
-en_label_encoder = joblib.load(hf_hub_download(
-    "E-motionAssistant/Englsih_Trained_Model_LR",
-    "label_encoder.joblib"
-))
+print("Loading Models...")
 
-# ────────────────────────────────────────────────
-# Sinhala model – now using joblib (LR + TF-IDF)
-# Replace REPO_SINHALA with your actual Sinhala LR model repository
-# ────────────────────────────────────────────────
-print("Loading Sinhala LR model...")
-REPO_SINHALA = "E-motionAssistant/Sinhala_Text_Emotion_Model_LR"   # ← CHANGE THIS to your real repo name
+# English (LR)
+en_vectorizer = joblib.load(hf_hub_download("E-motionAssistant/Englsih_Trained_Model_LR", "tfidf_vectorizer.joblib"))
+en_classifier = joblib.load(hf_hub_download("E-motionAssistant/Englsih_Trained_Model_LR", "logreg_model.joblib"))
+en_label_encoder = joblib.load(hf_hub_download("E-motionAssistant/Englsih_Trained_Model_LR", "label_encoder.joblib"))
 
-si_vectorizer = joblib.load(hf_hub_download(
-    "E-motionAssistant/Sinhala_Text_Emotion_Model_LR",
-    "tfidf_vectorizer.joblib"
-))
-si_classifier = joblib.load(hf_hub_download(
-    "E-motionAssistant/Sinhala_Text_Emotion_Model_LR",
-    "logreg_model.joblib"
-))
-si_label_encoder = joblib.load(hf_hub_download(
-    "E-motionAssistant/Sinhala_Text_Emotion_Model_LR",
-    "label_encoder.joblib"
-))
+# Sinhala (LR)
+si_vectorizer = joblib.load(hf_hub_download("E-motionAssistant/Sinhala_Text_Emotion_Model_LR", "tfidf_vectorizer.joblib"))
+si_classifier = joblib.load(hf_hub_download("E-motionAssistant/Sinhala_Text_Emotion_Model_LR", "logreg_model.joblib"))
+si_label_encoder = joblib.load(hf_hub_download("E-motionAssistant/Sinhala_Text_Emotion_Model_LR", "label_encoder.joblib"))
 
-# ────────────────────────────────────────────────
-# Tamil model (still transformers)
-# ────────────────────────────────────────────────
-print("Loading Tamil model...")
-tamil_pipe = pipeline(
-    "text-classification",
-    model="E-motionAssistant/Tamil_Emotion_Recognition_Model"
-)
+# Tamil (Transformers)
+tamil_pipe = pipeline("text-classification", model="E-motionAssistant/Tamil_Emotion_Recognition_Model")
 
 print("All models loaded successfully!")
 
 # ────────────────────────────────────────────────
-# Routes
+# 2. Prediction Functions
 # ────────────────────────────────────────────────
-@app.route("/", methods=["GET", "POST"])
-def home():
-    english_result = ""
-    sinhala_result = ""
-    tamil_result = ""
-
-    if request.method == "POST":
-        action = request.form.get("action")
 
-        # English – joblib
-        if action == "predict_english":
-            text = request.form.get("english_text", "").strip()
-            if text:
-                X = en_vectorizer.transform([text])
-                pred = en_classifier.predict(X)[0]
-                emotion = en_label_encoder.inverse_transform([pred])[0]
-                english_result = f"Emotion: {emotion}"
-            else:
-                english_result = "Please write something"
+def predict_emotion(text, language):
+    if not text.strip():
+        return "Please enter some text."
+    
+    try:
+        if language == "English":
+            X = en_vectorizer.transform([text])
+            pred = en_classifier.predict(X)[0]
+            return en_label_encoder.inverse_transform([pred])[0]
+            
+        elif language == "Sinhala":
+            X = si_vectorizer.transform([text])
+            pred = si_classifier.predict(X)[0]
+            return si_label_encoder.inverse_transform([pred])[0]
+            
+        elif language == "Tamil":
+            res = tamil_pipe(text)[0]
+            return f"{res['label']} (Confidence: {res['score']:.3f})"
+            
+    except Exception as e:
+        return f"Error: {str(e)}"
 
-        # Sinhala – now joblib (same pattern as English)
-        elif action == "predict_sinhala":
-            text = request.form.get("sinhala_text", "").strip()
-            if text:
-                X = si_vectorizer.transform([text])
-                pred = si_classifier.predict(X)[0]
-                emotion = si_label_encoder.inverse_transform([pred])[0]
-                sinhala_result = f"Emotion: {emotion}"
-            else:
-                sinhala_result = "Please write something"
+# ────────────────────────────────────────────────
+# 3. Gradio Interface
+# ────────────────────────────────────────────────
 
-        # Tamil – still transformers / pipeline
-        elif action == "predict_tamil":
-            text = request.form.get("tamil_text", "").strip()
-            if text:
-                res = tamil_pipe(text)[0]
-                tamil_result = f"Emotion: {res['label']} ({res['score']:.3f})"
-            else:
-                tamil_result = "Please write something"
+with gr.Blocks(theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 🌍 Multi-Language Emotion Detector")
+    gr.Markdown("Select a language and enter your text to analyze the emotion.")
+    
+    with gr.Row():
+        with gr.Column():
+            input_text = gr.Textbox(label="Input Text", placeholder="Type here...", lines=3)
+            lang_dropdown = gr.Dropdown(
+                choices=["English", "Sinhala", "Tamil"], 
+                value="English", 
+                label="Select Language"
+            )
+            submit_btn = gr.Button("Analyze Emotion", variant="primary")
+        
+        with gr.Column():
+            output_label = gr.Label(label="Predicted Emotion")
 
-    return render_template(
-        "index.html",
-        english_result=english_result,
-        sinhala_result=sinhala_result,
-        tamil_result=tamil_result
+    # This defines the API logic automatically
+    submit_btn.click(
+        fn=predict_emotion, 
+        inputs=[input_text, lang_dropdown], 
+        outputs=output_label,
+        api_name="predict" # This creates the /predict API endpoint
     )
 
+    gr.Examples(
+        examples=[
+            ["I am very happy today!", "English"],
+            ["මම අද ගොඩක් සතුටින්", "Sinhala"],
+            ["நான் இன்று மிகவும் மகிழ்ச்சியாக இருக்கிறேன்", "Tamil"]
+        ],
+        inputs=[input_text, lang_dropdown]
+    )
 
 if __name__ == "__main__":
-    app.run(host="0.0.0.0", port=7860, debug=False)
+    demo.launch()