import gradio as gr
from utils.ner_helpers import is_llm_model
from typing import Dict, List, Any
from tasks.sentiment_analysis import sentiment_analysis

def sentiment_ui():
    """Sentiment analysis UI component"""
    
    # Define models
    SENTIMENT_MODELS = [
        "gemini-2.0-flash"  # Only allow gemini-2.0-flash for now
        # "gpt-4",
        # "claude-2",
        # "distilbert-base-uncased-finetuned-sst-2-english",
        # "finiteautomata/bertweet-base-sentiment-analysis"
    ]
    DEFAULT_MODEL = "gemini-2.0-flash"
    
    def analyze_sentiment(text, model, custom_instructions):
        """Process text for sentiment analysis"""
        if not text.strip():
            return "No text provided"
            
        use_llm = is_llm_model(model)
        result = sentiment_analysis(
            text=text, 
            model=model, 
            custom_instructions=custom_instructions,
            use_llm=use_llm
        )
        
        # Try to normalize the result
        result = result.lower().strip()
        if "positive" in result:
            return "Positive"
        elif "negative" in result:
            return "Negative"
        elif "neutral" in result:
            return "Neutral"
        else:
            # Return as is for other results
            return result
    
    # UI Components
    with gr.Row():
        with gr.Column():
            input_text = gr.Textbox(
                label="Input Text",
                lines=6,
                placeholder="Enter text to analyze sentiment...",
                elem_id="sentiment-input-text"
            )
            gr.Examples(
                examples=[
                    ["I am very satisfied with the customer service of this company."],
                    ["The product did not meet my expectations and I am disappointed."]
                ],
                inputs=[input_text],
                label="Examples"
            )
            model = gr.Dropdown(
                SENTIMENT_MODELS,
                value=DEFAULT_MODEL,
                label="Model",
                interactive=True,
                elem_id="sentiment-model-dropdown"
            )
            custom_instructions = gr.Textbox(
                label="Custom Instructions (optional)",
                lines=2,
                placeholder="Add any custom instructions for the model...",
                elem_id="sentiment-custom-instructions"
            )

            btn = gr.Button("Analyze Sentiment", variant="primary", elem_id="sentiment-analyze-btn")
        
        with gr.Column():
            output = gr.Textbox(
                label="Sentiment Analysis",
                elem_id="sentiment-output"
            )
            
            # with gr.Accordion("About Sentiment Analysis", open=False):
            #     gr.Markdown("""
            #     ## Sentiment Analysis
                
            #     Sentiment analysis identifies the emotional tone behind text. The model analyzes your input text and classifies it as:
                
            #     - **Positive**: Text expresses positive emotions, approval, or optimism
            #     - **Negative**: Text expresses negative emotions, criticism, or pessimism
            #     - **Neutral**: Text is factual or does not express strong sentiment
                
            #     ### Model Types
                
            #     - **LLM Models** (Gemini, GPT, Claude): Provide sophisticated analysis with better understanding of context
            #     - **Traditional Models**: Specialized models trained specifically for sentiment analysis tasks
                
            #     Use the advanced options to customize how the model analyzes your text.
            #     """)
    
    # Event handlers
    btn.click(
        analyze_sentiment, 
        inputs=[input_text, model, custom_instructions], 
        outputs=output
    )
    
    return None