gradio_analytics.py

# gradio_analytics.py
import gradio as gr
import logging
import json
import sqlite3
import os
from datetime import datetime

logger = logging.getLogger(__name__)

try:
    from app import (
        get_trackio_database_path,
        get_project_statistics_with_nulls,
        get_recent_interactions_with_nulls,
        create_dashboard_html_with_nulls,
        calculate_response_quality,
        refresh_analytics_data_persistent as refresh_analytics_data,
        export_metrics_json_persistent as export_metrics_json,
        export_metrics_csv_persistent as export_metrics_csv,
        load_analytics_state,
        get_global_state_debug_info,
        sync_trackio_with_global_state,
        global_state_manager,
        evaluate_educational_quality_with_tracking,
    )
except ImportError:
    def get_trackio_database_path(project_name):
        return None
    
    def get_project_statistics_with_nulls(cursor, project_name):
        return {
            "total_conversations": None,
            "avg_session_length": None,
            "success_rate": None
        }
    
    def get_recent_interactions_with_nulls(cursor, project_name, limit=10):
        return []
    
    def create_dashboard_html_with_nulls(project_name, project_stats):
        return f"<div>Mock dashboard for {project_name}</div>"
    
    def calculate_response_quality(response):
        return 3.0
    
    def refresh_analytics_data():
        return {}, [], "<div>Mock analytics</div>"
    
    def export_metrics_json():
        gr.Info("Mock JSON export")
    
    def export_metrics_csv():
        gr.Info("Mock CSV export")
    
    def load_analytics_state():
        return {}, [], "<div>Mock analytics state</div>"
    
    def get_global_state_debug_info():
        return {"status": "mock"}
    
    def sync_trackio_with_global_state():
        pass
    
    def evaluate_educational_quality_with_tracking(*args, **kwargs):
        return {"educational_score": 0.5}
    
    class MockStateManager:
        def get_cache_status(self):
            return {"status": "mock"}
        def get_evaluation_summary(self, include_history=False):
            return {"aggregate_metrics": {}, "total_evaluations": {}}
        def clear_all_states(self):
            pass
        def _backup_to_hf_dataset(self):
            pass
    
    global_state_manager = MockStateManager()

def load_custom_css():
    try:
        with open("styles.css", "r", encoding="utf-8") as css_file:
            css_content = css_file.read()
            logger.info(f"CSS loaded successfully for analytics page")
            return css_content
    except FileNotFoundError:
        logger.warning("styles.css file not found for analytics page")
        return ""
    except Exception as e:
        logger.warning(f"Error reading styles.css: {e}")
        return ""

def show_cache_info():
    try:
        from pathlib import Path
        from huggingface_hub import scan_cache_dir
        
        cache_info = scan_cache_dir(cache_dir="/tmp/huggingface")
        
        info_text = f"""
**HuggingFace Cache Status:**

**Total Size:** {cache_info.size_on_disk / (1024**3):.2f} GB
**Number of Repos:** {len(cache_info.repos)}

**Cached Models:**
"""
        
        for repo in cache_info.repos:
            size_gb = repo.size_on_disk / (1024**3)
            info_text += f"""
- **{repo.repo_id}**
  - Size: {size_gb:.2f} GB
  - Type: {repo.repo_type}
  - Revisions: {len(repo.revisions)}
"""
        
        return info_text
        
    except Exception as e:
        return f"Error inspecting cache: {str(e)}"
        
def launch_external_trackio():
    try:
        import subprocess
        result = subprocess.run(
            ["trackio", "show", "--project", "Mimir"],
            capture_output=False,
            text=True
        )
        
        if result.returncode == 0:
            gr.Info("Trackio dashboard launched in browser")
        else:
            gr.Warning("Could not launch trackio dashboard")
            
    except Exception as e:
        logger.error(f"Failed to launch trackio: {e}")
        gr.Warning(f"Failed to launch trackio dashboard: {str(e)}")

def show_cache_status():
    try:
        debug_info = get_global_state_debug_info()
        cache_status = debug_info.get("cache_status", {})
        
        status_text = f"""
**Global State Cache Status:**
- Session ID: {cache_status.get('session_id', 'Unknown')}
- Analytics Cached: {'Yes' if cache_status.get('analytics_cached') else 'No'}
- Conversation Cached: {'Yes' if cache_status.get('conversation_cached') else 'No'}
- Analytics Last Refresh: {cache_status.get('analytics_last_refresh', 'Never')}
- Total Analytics Sessions: {cache_status.get('total_analytics_sessions', 0)}
- Total Conversation Sessions: {cache_status.get('total_conversation_sessions', 0)}

**Analytics Data Status:**
- Has Analytics Data: {'Yes' if cache_status.get('analytics_has_data') else 'No'}
- Conversation Length: {cache_status.get('conversation_length', 0)} messages
- Chat History Length: {cache_status.get('chat_history_length', 0)} messages

*Last Updated: {datetime.now().strftime('%H:%M:%S')}*
        """
        
        gr.Info("Cache status updated - check the Status panel")
        return status_text
        
    except Exception as e:
        error_text = f"Error getting cache status: {str(e)}"
        gr.Warning(error_text)
        return error_text

def manual_backup_to_hf():
    try:
        global_state_manager._backup_to_hf_dataset()
        gr.Info("Manual backup to HF dataset completed successfully")
        return f"Backup completed at {datetime.now().strftime('%H:%M:%S')}"
    except Exception as e:
        gr.Warning(f"Backup failed: {str(e)}")
        return f"Backup failed: {str(e)}"

def get_persistence_status():
    try:
        status_info = {
            "SQLite DB": "Active" if os.path.exists(global_state_manager._db_path) else "Not Found",
            "HF Dataset": global_state_manager.dataset_repo,
            "Last HF Backup": global_state_manager._last_hf_backup.strftime('%Y-%m-%d %H:%M:%S'),
            "DB Path": global_state_manager._db_path,
            "Backup Interval": f"{global_state_manager._hf_backup_interval}s"
        }
        return status_info
    except Exception as e:
        return {"error": str(e)}

def clear_all_global_states():
    try:
        global_state_manager.clear_all_states()
        gr.Info("All global states cleared successfully")
        
        empty_stats = {
            "total_conversations": None,
            "avg_session_length": None,
            "success_rate": None,
            "model_type": "Cleared",
            "last_updated": datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        }
        
        empty_html = """
        <div style="text-align: center; padding: 40px; border: 2px dashed #ccc; border-radius: 8px; background: #f8f9fa;">
            <h3>States Cleared</h3>
            <p>All global states have been cleared.</p>
            <p>Click "Refresh Data" to reload analytics.</p>
        </div>
        """
        
        return empty_stats, [], empty_html
        
    except Exception as e:
        gr.Warning(f"Failed to clear states: {str(e)}")
        return load_analytics_state()

def show_evaluation_metrics():
    try:
        eval_summary = global_state_manager.get_evaluation_summary(include_history=True)
        
        metrics_data = [
            ["Educational Quality", f"{eval_summary['aggregate_metrics']['avg_educational_quality']:.3f}"],
            ["User Satisfaction", f"{eval_summary['aggregate_metrics']['user_satisfaction_rate']:.3f}"]
        ]
        
        recent_evaluations = []
        if 'history' in eval_summary:
            for eval_item in eval_summary['history']['recent_educational_scores'][-5:]:
                recent_evaluations.append([
                    eval_item['timestamp'][:16],
                    f"{eval_item['educational_score']:.3f}",
                    f"{eval_item['semantic_quality']:.3f}",
                    f"{eval_item['response_time']:.3f}s"
                ])
        
        return eval_summary, metrics_data, recent_evaluations
        
    except Exception as e:
        logger.error(f"Error getting evaluation metrics: {e}")
        return {}, [], []

def sync_and_refresh_all():
    try:
        sync_trackio_with_global_state()
        project_stats, recent_interactions, dashboard_html = refresh_analytics_data()
        eval_summary, metrics_data, recent_evaluations = show_evaluation_metrics()
        
        gr.Info("All data synced and refreshed successfully")
        
        return project_stats, recent_interactions, dashboard_html, eval_summary, metrics_data, recent_evaluations
        
    except Exception as e:
        logger.error(f"Sync and refresh failed: {e}")
        gr.Warning(f"Sync failed: {str(e)}")
        return load_analytics_state() + ({}, [], [])

with gr.Blocks() as demo:
    custom_css = load_custom_css()
    if custom_css:
        gr.HTML(f'<style>{custom_css}</style>')
    
    gr.HTML('<div class="analytics-title"><h2>Mimir Analytics Dashboard</h2></div>')
    
    gr.Markdown("Monitor educational AI performance and effectiveness metrics with persistent state management.")
    
    with gr.Tabs():
        with gr.TabItem("Traditional Analytics"):
            with gr.Row():
                with gr.Column(scale=1):
                    gr.Markdown("## Controls")
                    refresh_btn = gr.Button("Refresh Data", variant="primary")
                    sync_all_btn = gr.Button("Sync & Refresh All", variant="primary")
                    
                    with gr.Row():
                        export_json_btn = gr.Button("Export JSON", variant="secondary", size="sm")
                        export_csv_btn = gr.Button("Export CSV", variant="secondary", size="sm")
                    
                    launch_trackio_btn = gr.Button("Launch Trackio Dashboard", variant="secondary")
                    
                    gr.Markdown("### State Management")
                    with gr.Row():
                        cache_status_btn = gr.Button("Cache Status", size="sm")
                        clear_states_btn = gr.Button("Clear All States", size="sm", variant="stop")
                    
                    with gr.Group():
                        gr.Markdown("### Project Information")
                        project_info = gr.JSON(
                            value={
                                "total_conversations": None,
                                "avg_session_length": None,
                                "success_rate": None,
                                "model_type": None
                            },
                            label="Project Stats"
                        )
                    
                    with gr.Group():
                        gr.Markdown("### System Status")
                        status_panel = gr.Markdown(
                            "Click 'Cache Status' to view global state information.",
                            label="Status Information"
                        )
                
                with gr.Column(scale=2):
                    gr.Markdown("## Key Metrics Dashboard")
                    trackio_iframe = gr.HTML(
                        value="""
                        <div style="text-align: center; padding: 40px; border: 2px dashed #ccc; border-radius: 8px; background: #f8f9fa;">
                            <h3>Trackio Dashboard</h3>
                            <p>Analytics data will appear here after conversations.</p>
                            <p>Data is automatically cached and persists across page navigation.</p>
                            <p>To launch trackio dashboard separately, run:</p>
                            <code style="background: #e9ecef; padding: 4px 8px; border-radius: 4px;">trackio show --project "Mimir"</code>
                        </div>
                        """,
                        label="Dashboard"
                    )
            
            with gr.Row():
                with gr.Column():
                    gr.Markdown("## Recent Interactions")
                    gr.Markdown("*Data persists when switching between Chatbot and Analytics pages*")
                    recent_metrics = gr.Dataframe(
                        headers=["Timestamp", "Response Time", "Prompt Mode", "Tools Used", "Quality Score", "Adapter"],
                        datatype=["str", "number", "str", "bool", "number", "str"],
                        row_count=10,
                        col_count=6,
                        interactive=False,
                        label="Latest Sessions",
                        value=[],
                        show_label=True
                    )
        
        with gr.TabItem("ML Performance"):
            gr.Markdown("## Agent-Based Performance & Global State Metrics")
            
            with gr.Row():
                with gr.Column(scale=1):
                    eval_metrics_btn = gr.Button("Get Evaluation Metrics", variant="primary")
                    
                    with gr.Group():
                        gr.Markdown("### Model Cache Status")
                        cache_status_display = gr.JSON(
                            value={},
                            label="Cache Information"
                        )
                
                with gr.Column(scale=2):
                    gr.Markdown("### Aggregate Performance Metrics")
                    eval_metrics_table = gr.Dataframe(
                        headers=["Metric", "Score"],
                        datatype=["str", "str"],
                        label="Model Performance",
                        value=[]
                    )
                    
                    eval_summary_display = gr.JSON(
                        value={},
                        label="Detailed Evaluation Summary"
                    )
            
            with gr.Row():
                with gr.Column():
                    gr.Markdown("### Recent Quality Evaluations")
                    recent_evaluations_table = gr.Dataframe(
                        headers=["Timestamp", "Educational Score", "Semantic Quality", "Response Time"],
                        datatype=["str", "str", "str", "str"],
                        label="Recent Evaluations",
                        value=[]
                    )
        
        with gr.TabItem("System Status"):
            gr.Markdown("## Global State Manager & System Diagnostics")
            
            with gr.Row():
                with gr.Column():
                    gr.Markdown("### Global State Cache")
                    cache_details = gr.Markdown("Click 'Show Cache Status' to view detailed information.")
                    
                    show_cache_btn = gr.Button("Show Cache Status", variant="primary")
                    refresh_cache_btn = gr.Button("Refresh Cache Info", variant="secondary")
                    
                    gr.Markdown("### Persistence Controls")
                    backup_btn = gr.Button("Manual Backup to HF Dataset", variant="primary")
                    backup_status = gr.Textbox(label="Backup Status", value="No recent backup", interactive=False)
                
                with gr.Column():
                    gr.Markdown("### System Actions")
                    sync_trackio_btn = gr.Button("Sync to Database", variant="secondary")
                    clear_all_btn = gr.Button("Clear All Global States", variant="stop")
                    
                    gr.Markdown("### Persistence Status")
                    persistence_info = gr.JSON(
                        value={},
                        label="Persistence Information"
                    )
                    
                    gr.Markdown("### Performance Monitor")
                    perf_info = gr.JSON(
                        value={},
                        label="Performance Information"
                    )
            
            # NEW: HuggingFace Cache Viewer Section
            with gr.Row():
                with gr.Column():
                    gr.Markdown("### 🗂️ HuggingFace Model Cache")
                    gr.Markdown("*View cached models and disk usage*")
                    
                    cache_viewer_btn = gr.Button("Inspect Model Cache", variant="primary", size="lg")
                    
                    with gr.Row():
                        clear_cache_btn = gr.Button("Clear Cache (⚠️ Dangerous)", variant="stop", size="sm")
                        refresh_models_btn = gr.Button("Re-download Models", variant="secondary", size="sm")
                    
                    cache_info_display = gr.Markdown(
                        "Click **Inspect Model Cache** to view detailed cache information.",
                        label="Cache Details"
                    )
        
    demo.load(
        load_analytics_state,
        inputs=None,
        outputs=[project_info, recent_metrics, trackio_iframe],
        show_progress="hidden"
    )
    
    demo.load(
        fn=lambda: global_state_manager.get_cache_status(),
        inputs=None,
        outputs=[cache_status_display],
        show_progress="hidden"
    )
    
    demo.load(
        fn=get_persistence_status,
        inputs=None,
        outputs=[persistence_info],
        show_progress="hidden"
    )
    
    refresh_btn.click(
        fn=refresh_analytics_data,
        inputs=[],
        outputs=[project_info, recent_metrics, trackio_iframe],
        show_progress="full"
    )
    
    sync_all_btn.click(
        fn=sync_and_refresh_all,
        inputs=[],
        outputs=[project_info, recent_metrics, trackio_iframe, eval_summary_display, eval_metrics_table, recent_evaluations_table],
        show_progress="full"
    )
    
    export_json_btn.click(
        fn=export_metrics_json,
        inputs=[],
        outputs=[],
        show_progress="full"
    )
    
    export_csv_btn.click(
        fn=export_metrics_csv,
        inputs=[],
        outputs=[],
        show_progress="full"
    )
    
    launch_trackio_btn.click(
        fn=launch_external_trackio,
        inputs=[],
        outputs=[],
        show_progress="full"
    )
    
    cache_status_btn.click(
        fn=show_cache_status,
        inputs=[],
        outputs=[status_panel],
        show_progress="full"
    )
    
    clear_states_btn.click(
        fn=clear_all_global_states,
        inputs=[],
        outputs=[project_info, recent_metrics, trackio_iframe],
        show_progress="full"
    )
    
    eval_metrics_btn.click(
        fn=show_evaluation_metrics,
        inputs=[],
        outputs=[eval_summary_display, eval_metrics_table, recent_evaluations_table],
        show_progress="full"
    )
    
    show_cache_btn.click(
        fn=show_cache_status,
        inputs=[],
        outputs=[cache_details],
        show_progress="full"
    )
    
    refresh_cache_btn.click(
        fn=lambda: global_state_manager.get_cache_status(),
        inputs=[],
        outputs=[perf_info],
        show_progress="full"
    )
    
    backup_btn.click(
        fn=manual_backup_to_hf,
        inputs=[],
        outputs=[backup_status],
        show_progress="full"
    )
    
    sync_trackio_btn.click(
        fn=sync_trackio_with_global_state,
        inputs=[],
        outputs=[],
        show_progress="full"
    )
    
    clear_all_btn.click(
        fn=clear_all_global_states,
        inputs=[],
        outputs=[project_info, recent_metrics, trackio_iframe],
        show_progress="full"
    )

if __name__ == "__main__":
    logger.info("Running analytics dashboard standalone with global state management")
    demo.launch(server_name="0.0.0.0", server_port=7861)