made fixes

app.py

@@ -6,8 +6,138 @@ from datetime import datetime, timedelta
 import requests
 from bs4 import BeautifulSoup
 
-# Previous imports and model loading code remains the same...
-# (Keep all the previous code until the create_interface function)
+# Download required NLTK data
+try:
+    nltk.data.find('tokenizers/punkt')
+except LookupError:
+    nltk.download('punkt')
+
+# Global variables to cache models
+CACHED_MODELS = {}
+
+# Initialize models and tokenizers
+def load_models():
+    global CACHED_MODELS
+    
+    if CACHED_MODELS:
+        return (
+            CACHED_MODELS['generator_tokenizer'],
+            CACHED_MODELS['generator'],
+            CACHED_MODELS['sentiment_analyzer'],
+            CACHED_MODELS['content_checker']
+        )
+    
+    try:
+        # Use a smaller text generation model
+        generator_model = "distilgpt2"  # Smaller than opt-350m
+        generator_tokenizer = AutoTokenizer.from_pretrained(generator_model)
+        generator = AutoModelForCausalLM.from_pretrained(generator_model)
+        
+        # Sentiment analysis
+        sentiment_analyzer = pipeline(
+            "sentiment-analysis",
+            model="finiteautomata/bertweet-base-sentiment-analysis"
+        )
+        
+        # Content safety checker
+        content_checker = pipeline(
+            "text-classification",
+            model="facebook/roberta-hate-speech-dynabench-r4-target"
+        )
+        
+        # Cache the models
+        CACHED_MODELS['generator_tokenizer'] = generator_tokenizer
+        CACHED_MODELS['generator'] = generator
+        CACHED_MODELS['sentiment_analyzer'] = sentiment_analyzer
+        CACHED_MODELS['content_checker'] = content_checker
+        
+        return generator_tokenizer, generator, sentiment_analyzer, content_checker
+    except Exception as e:
+        print(f"Error loading models: {str(e)}")
+        raise
+
+def generate_content(
+    product_name,
+    product_description,
+    target_audience,
+    key_features,
+    unique_benefits,
+    platform,
+    tone,
+    generator_tokenizer,
+    generator,
+    sentiment_analyzer,
+    content_checker
+):
+    char_limit = 280 if platform == "Twitter" else 500
+    
+    prompt = f"""Write a {platform} post:
+Product: {product_name}
+Description: {product_description}
+Target: {target_audience}
+Features: {key_features}
+Benefits: {unique_benefits}
+Tone: {tone}
+
+Write a compelling {platform} post in {tone} tone that highlights the product benefits:"""
+    
+    try:
+        # Generate initial content
+        inputs = generator_tokenizer(prompt, return_tensors="pt", max_length=256, truncation=True)
+        outputs = generator.generate(
+            inputs["input_ids"],
+            max_length=char_limit,
+            num_return_sequences=2,  # Reduced from 3 to 2 for speed
+            temperature=0.7,
+            top_p=0.9,
+            do_sample=True,
+            pad_token_id=generator_tokenizer.eos_token_id
+        )
+        
+        generated_texts = [generator_tokenizer.decode(output, skip_special_tokens=True) for output in outputs]
+        
+        # Filter and analyze content
+        filtered_content = []
+        for text in generated_texts:
+            # Clean up text
+            text = text.replace(prompt, "").strip()
+            if len(text) < 10:  # Skip very short texts
+                continue
+                
+            # Truncate to character limit if needed
+            text = text[:char_limit]
+            
+            # Check sentiment
+            try:
+                sentiment = sentiment_analyzer(text)[0]
+            except:
+                sentiment = {'label': 'unknown', 'score': 0.0}
+            
+            # Check content safety
+            try:
+                safety_check = content_checker(text)[0]
+            except:
+                safety_check = {'label': 'unknown', 'score': 0.0}
+            
+            # More lenient filtering
+            filtered_content.append({
+                'text': text,
+                'sentiment': sentiment['label'],
+                'safety_score': f"{float(safety_check.get('score', 0)):.2f}"
+            })
+        
+        return filtered_content if filtered_content else [{
+            'text': generated_texts[0].replace(prompt, "").strip()[:char_limit],
+            'sentiment': 'not analyzed',
+            'safety_score': 'not analyzed'
+        }]
+    except Exception as e:
+        print(f"Error in content generation: {str(e)}")
+        return [{
+            'text': 'Error generating content. Please try again.',
+            'sentiment': 'error',
+            'safety_score': 'error'
+        }]
 
 def create_interface():
     print("Loading models...")
@@ -63,77 +193,89 @@ def create_interface():
             return f"An error occurred: {str(e)}"
 
     # Create input components
-    product_name = gr.Textbox(label="Product Name", placeholder="Enter product name")
-    product_description = gr.Textbox(label="Product Description", lines=3, placeholder="Brief description of your product")
-    target_audience = gr.Textbox(label="Target Audience", placeholder="Who is this product for?")
-    key_features = gr.Textbox(label="Key Features", lines=2, placeholder="Main features of your product")
-    unique_benefits = gr.Textbox(label="Unique Benefits", lines=2, placeholder="What makes your product special?")
-    platform = gr.Radio(
-        choices=["Twitter", "Instagram"],
-        label="Platform",
-        value="Twitter"
-    )
-    tone = gr.Textbox(label="Tone", placeholder="e.g., professional, casual, friendly")
-    
-    # Output component
-    output = gr.Textbox(label="Generated Content", lines=10)
+    with gr.Blocks() as demo:
+        gr.Markdown("# Ethimar - AI Marketing Content Generator")
+        gr.Markdown("Generate ethical marketing content with AI-powered insights.\n⏳ Note: First generation might take 3-5 minutes due to model loading. Subsequent generations will be faster!")
+        
+        with gr.Row():
+            fill_button = gr.Button(
+                "Fill the form with sample data",
+                variant="primary",
+                scale=1,
+                size="sm"
+            )
 
-    # Create the interface with custom layout
-    iface = gr.Interface(
-        fn=process_input,
-        inputs=[
-            product_name,
-            product_description,
-            target_audience,
-            key_features,
-            unique_benefits,
-            platform,
-            tone
-        ],
-        outputs=output,
-        title="Ethimar - AI Marketing Content Generator",
-        description="""Generate ethical marketing content with AI-powered insights.
-        ⏳ Note: First generation might take 3-5 minutes due to model loading.
-        Subsequent generations will be faster!""",
-        theme="default",
-        examples=[
-            [
-                "EcoBottle",
-                "Sustainable water bottle made from recycled ocean plastic",
-                "Environmentally conscious young professionals",
-                "100% recycled materials, Insulated design, Leak-proof",
-                "Helps clean oceans, Keeps drinks cold for 24 hours",
-                "Twitter",
-                "professional"
-            ]
-        ]
-    )
+        with gr.Column():
+            product_name = gr.Textbox(label="Product Name", placeholder="Enter product name")
+            product_description = gr.Textbox(label="Product Description", lines=3, placeholder="Brief description of your product")
+            target_audience = gr.Textbox(label="Target Audience", placeholder="Who is this product for?")
+            key_features = gr.Textbox(label="Key Features", lines=2, placeholder="Main features of your product")
+            unique_benefits = gr.Textbox(label="Unique Benefits", lines=2, placeholder="What makes your product special?")
+            platform = gr.Radio(
+                choices=["Twitter", "Instagram"],
+                label="Platform",
+                value="Twitter"
+            )
+            tone = gr.Textbox(label="Tone", placeholder="e.g., professional, casual, friendly")
+            
+            submit_button = gr.Button("Generate Content", variant="primary")
+            output = gr.Textbox(label="Generated Content", lines=10)
 
-    # Add the sample data button with custom styling
-    fill_button = gr.Button(
-        "Fill the form with sample data",
-        variant="primary",
-        scale=1,
-        size="sm"
-    )
-    
-    # Connect the button to the fill_sample_data function
-    fill_button.click(
-        fn=fill_sample_data,
-        outputs=[
-            product_name,
-            product_description,
-            target_audience,
-            key_features,
-            unique_benefits,
-            platform,
-            tone
-        ]
-    )
+        # Connect the buttons to functions
+        fill_button.click(
+            fn=fill_sample_data,
+            outputs=[
+                product_name,
+                product_description,
+                target_audience,
+                key_features,
+                unique_benefits,
+                platform,
+                tone
+            ]
+        )
+        
+        submit_button.click(
+            fn=process_input,
+            inputs=[
+                product_name,
+                product_description,
+                target_audience,
+                key_features,
+                unique_benefits,
+                platform,
+                tone
+            ],
+            outputs=output
+        )
+        
+        # Add examples
+        gr.Examples(
+            examples=[
+                [
+                    "EcoBottle",
+                    "Sustainable water bottle made from recycled ocean plastic",
+                    "Environmentally conscious young professionals",
+                    "100% recycled materials, Insulated design, Leak-proof",
+                    "Helps clean oceans, Keeps drinks cold for 24 hours",
+                    "Twitter",
+                    "professional"
+                ]
+            ],
+            inputs=[
+                product_name,
+                product_description,
+                target_audience,
+                key_features,
+                unique_benefits,
+                platform,
+                tone
+            ]
+        )
 
-    return iface
+    return demo
 
 # Launch the app
 if __name__ == "__main__":
-    iface = create_interface()
-    iface.launch()
+    demo = create_interface()
+    demo.launch()