import os
import gradio as gr
import google.generativeai as genai
from PIL import Image
from dotenv import load_dotenv

# 1. Setup & Configuration
# ---------------------------------------------------------
load_dotenv()

# Fetch API Key from environment (HF Secrets or .env file)
API_KEY = os.getenv("GEMINI_API_KEY")

if not API_KEY:
    # Fallback for local testing if env var is missing
    print("⚠️ WARNING: GEMINI_API_KEY not found. Please set it in your environment.")

# Configure Gemini
genai.configure(api_key=API_KEY)

# Use Gemini 1.5 Flash for speed and multimodal capabilities
MODEL_NAME = "gemini-1.5-flash"
generation_config = {
    "temperature": 0.4,  # Lower temperature for more analytical/precise reasoning
    "top_p": 0.95,
    "top_k": 64,
    "max_output_tokens": 8192,
}

model = genai.GenerativeModel(
    model_name=MODEL_NAME,
    generation_config=generation_config,
)


# 2. The "Signal Ghost" Logic (Bayesian Analysis)
# ---------------------------------------------------------
def signal_ghost_analysis(image, audio_path):
    """
    Core function to process the multimodal inputs via Gemini.
    """
    if not API_KEY:
        return "❌ Error: API Key is missing. Please configure GEMINI_API_KEY in Settings."
    
    if image is None or audio_path is None:
        return "⚠️ Please upload both a PCB Image and an Audio Signal file to begin the diagnostic."

    print("👻 Signal Ghost: Processing inputs...")

    # Upload audio file to Gemini File API (required for audio processing)
    # Note: In a production app, manage file deletion/lifecycle. 
    # For Hackathon demo, simple upload is fine.
    try:
        audio_file = genai.upload_file(path=audio_path)
    except Exception as e:
        return f"❌ Error processing audio: {str(e)}"

    # The "Ghost" System Prompt
    # We instruct the model to act as a Bayesian Diagnostic Engine.
    prompt = """
    You are 'Signal Ghost', an advanced AI diagnostic ecosystem for Electrical Engineering. 
    
    **YOUR MISSION:**
    Perform a cross-domain analysis of the provided PCB IMAGE and the AUDIO SIGNAL (EMI noise/interference).
    Use Bayesian reasoning: Update your confidence in a fault based on the combined evidence of visual defects and auditory signal signatures.

    **INPUTS:**
    1. Visual: A circuit board image (look for burnt components, bad traces, grounding issues).
    2. Audio: A recording of electrical noise/interference (listen for 50/60Hz hum, high-pitch switching whine, arcing pops).

    **OUTPUT FORMAT:**
    You must generate a response with two distinct sections separated by a horizontal rule.

    ---
    
    ### 🏗️ SECTION 1: ENGINEER TECHNICAL REPORT (The "Raw Data")
    * **Tone:** Professional, precise, metric-driven (ISO/IEEE standard).
    * **Bayesian Confidence:** "Fault Probability: X%" (Explain the priors: Visual + Audio).
    * **Signal Analysis:** Estimate frequencies (e.g., "120Hz Ripple"), dB estimates, and wave shapes.
    * **Visual Analysis:** Component coordinates (e.g., "Top left capacitor"), specific damage type.
    * **Action Plan:** Bullet points of technical fixes (e.g., "Replace C4 with low-ESR variant", "Add ferrite bead").

    ---

    ### 🎓 SECTION 2: STUDENT SUMMARY (The "Translation")
    * **Tone:** Socratic, encouraging, simplified.
    * **The Analogy:** Explain the technical fault using a real-world analogy (e.g., plumbing, traffic, music).
    * **Why it happened:** Simple cause-and-effect.
    * **Learning Moment:** One key concept to remember (e.g., "Why do capacitors explode?").
    """

    # Generate Content
    try:
        response = model.generate_content([prompt, image, audio_file])
        return response.text
    except Exception as e:
        return f"❌ Analysis Failed: {str(e)}"


# 3. User Interface (Gradio)
# ---------------------------------------------------------

# Custom CSS for that "Cyber/Ghost" Aesthetic
ghost_css = """
body { background-color: #0b0f19; color: #e0e0e0; }
gradio-app { background-color: #0b0f19 !important; }
.header-text { text-align: center; font-family: 'Courier New', monospace; color: #00f3ff; text-shadow: 0px 0px 10px #00f3ff; }
.sub-text { text-align: center; color: #8892b0; font-size: 1.1em; }
button.primary { background: linear-gradient(90deg, #00f3ff 0%, #0077ff 100%); border: none; color: #000; font-weight: bold; transition: all 0.3s ease; }
button.primary:hover { box-shadow: 0px 0px 15px #00f3ff; transform: scale(1.02); }
.panel { border: 1px solid #1f2937; background-color: #111827; border-radius: 8px; }
"""

with gr.Blocks(theme=gr.themes.Soft(), css=ghost_css, title="Signal Ghost") as app:
    
    # --- Header ---
    with gr.Row():
        with gr.Column():
            gr.Markdown("# 👻 SIGNAL GHOST", elem_classes=["header-text"])
            gr.Markdown("### Multimodal ECE Diagnostic Ecosystem", elem_classes=["sub-text"])
            gr.Markdown("Upload a PCB Image and an Audio Recording of the circuit noise to detect EMI faults.", elem_classes=["sub-text"])

    # --- Input Section ---
    with gr.Row(equal_height=True):
        with gr.Column(scale=1, min_width=300):
            gr.Markdown("### 👁️ Visual Input (PCB)")
            img_input = gr.Image(type="pil", label="Upload Circuit Board Image", sources=["upload", "clipboard"])
        
        with gr.Column(scale=1, min_width=300):
            gr.Markdown("### 👂 Auditory Input (Signal Noise)")
            audio_input = gr.Audio(type="filepath", label="Upload Noise Signal (.wav, .mp3)", sources=["upload", "microphone"])

    # --- Action Button ---
    with gr.Row():
        analyze_btn = gr.Button("🔍 INITIATE DIAGNOSTIC SEQUENCE", variant="primary", size="lg")

    # --- Output Section ---
    with gr.Row():
        with gr.Column():
            gr.Markdown("### 📊 Diagnostic Output")
            output_text = gr.Markdown(value="Waiting for signal inputs...", label="Analysis Report")

    # --- Footer ---
    with gr.Row():
        gr.Markdown("---")
        gr.Markdown("*Powered by Gemini 1.5 Flash • Bayesian Inference Engine v1.0*")

    # --- Event Binding ---
    analyze_btn.click(
        fn=signal_ghost_analysis,
        inputs=[img_input, audio_input],
        outputs=output_text
    )

# 4. Launch
# ---------------------------------------------------------
if __name__ == "__main__":
    app.launch(server_name="0.0.0.0", server_port=7860)