Update app.py

app.py

@@ -9,21 +9,18 @@ from nltk.tokenize import word_tokenize
 from keras.models import load_model
 import nltk
 import cloudpickle
-import easyocr
 
-# Download required NLTK data (only needed once)
+# Download required NLTK data
 nltk.download('stopwords')
 nltk.download('punkt')
 
-# Load the pre-trained model and tokenizer
+# Load the pre-trained model
 model = load_model('Sarcasmmodel.h5')
+
 with open('tokenizer.pkl', 'rb') as file:
     tokenizer_obj = cloudpickle.load(file)
 
-# Initialize EasyOCR Reader once
-ocr_reader = easyocr.Reader(['en'])
-
-# Text cleaning function
+# Function to clean the text
 def clean_text(text):
     text = text.lower()
     text = re.sub(r"http\S+|www\S+|https\S+", '', text, flags=re.MULTILINE)
@@ -32,7 +29,7 @@ def clean_text(text):
     text = re.sub(r'\d+', '', text)
     return text
 
-# Tokenize and remove stopwords
+# Function to tokenize and clean the text data
 def CleanTokenize(df):
     head_lines = []
     lines = df["headline"].values.tolist()
@@ -45,34 +42,35 @@ def CleanTokenize(df):
         head_lines.append(words)
     return head_lines
 
-# Predict sarcasm with confidence
+# Function to predict sarcasm
 def predict_sarcasm(text, max_length=25):
     x_final = pd.DataFrame({"headline": [text]})
     test_lines = CleanTokenize(x_final)
     test_sequences = tokenizer_obj.texts_to_sequences(test_lines)
     test_review_pad = pad_sequences(test_sequences, maxlen=max_length, padding='post')
     pred = model.predict(test_review_pad)
-    confidence = pred[0][0] * 100  # percentage
-
+    confidence = pred[0][0] * 100  # Convert to percentage
     result = "It's a sarcasm!" if confidence >= 50 else "It's not a sarcasm."
     return f"**Result:** {result}\n**Confidence:** {confidence:.2f}%"
 
-# OCR + Sarcasm prediction pipeline
-def ocr_sarcasm_detection(image):
-    # Extract text from image with OCR
-    extracted_text = " ".join(ocr_reader.readtext(image, detail=0))
-    if not extracted_text.strip():
-        return "No text detected in the image."
-    return predict_sarcasm(extracted_text)
+# Gradio interface
+def gradio_interface(text):
+    return predict_sarcasm(text)
 
-# Gradio interface takes only image input; no text input or recommendations
+# Create the Gradio app
 iface = gr.Interface(
-    fn=ocr_sarcasm_detection,
-    inputs=gr.Image(type="filepath", label="Upload Image with Text"),
-    outputs=gr.Textbox(label="Sarcasm Detection Result"),
-    title="OCR-based Sarcasm Detection 🤖",
-    description="Upload an image containing text (e.g., meme or screenshot). The app extracts text via OCR and predicts sarcasm.",
+    fn=gradio_interface,
+    inputs=gr.Textbox(lines=2, placeholder="Type something sarcastic...", label="Input Text"),
+    outputs=gr.Textbox(label="Prediction"),
+    title="🤖 Sarcasm Detection",
+    description="This app detects whether a given text is sarcastic or not.",
+    examples=[
+        ["Oh great, another Monday morning!"],
+        ["I just love spending hours in traffic."],
+        ["This is the best day of my life!"]
+    ],
     theme="default"
 )
 
-iface.launch()
+# Launch the app
+iface.launch()