apr_visualization.py

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.dates as mdates
import plotly.graph_objects as go
import plotly.express as px
from plotly.subplots import make_subplots
import random
from datetime import datetime, timedelta
import requests
import sys
import json
from typing import List, Dict, Any

# Global variable to store the data for reuse
global_df = None

# Configuration
API_BASE_URL = "https://afmdb.autonolas.tech"

def get_agent_type_by_name(type_name: str) -> Dict[str, Any]:
    """Get agent type by name"""
    response = requests.get(f"{API_BASE_URL}/api/agent-types/name/{type_name}")
    if response.status_code == 404:
        print(f"Error: Agent type '{type_name}' not found")
        return None
    response.raise_for_status()
    return response.json()

def get_attribute_definition_by_name(attr_name: str) -> Dict[str, Any]:
    """Get attribute definition by name"""
    response = requests.get(f"{API_BASE_URL}/api/attributes/name/{attr_name}")
    if response.status_code == 404:
        print(f"Error: Attribute definition '{attr_name}' not found")
        return None
    response.raise_for_status()
    return response.json()

def get_agents_by_type(type_id: int) -> List[Dict[str, Any]]:
    """Get all agents of a specific type"""
    response = requests.get(f"{API_BASE_URL}/api/agent-types/{type_id}/agents/")
    if response.status_code == 404:
        print(f"No agents found for type ID {type_id}")
        return []
    response.raise_for_status()
    return response.json()

def get_attribute_values_by_type_and_attr(agents: List[Dict[str, Any]], attr_def_id: int) -> List[Dict[str, Any]]:
    """Get all attribute values for a specific attribute definition across all agents of a given list"""
    all_attributes = []
    
    # For each agent, get their attributes and filter for the one we want
    for agent in agents:
        agent_id = agent["agent_id"]
        
        # Call the /api/agents/{agent_id}/attributes/ endpoint
        response = requests.get(f"{API_BASE_URL}/api/agents/{agent_id}/attributes/", params={"limit": 1000})
        if response.status_code == 404:
            print(f"No attributes found for agent ID {agent_id}")
            continue
        
        try:
            response.raise_for_status()
            agent_attrs = response.json()
            
            # Filter for the specific attribute definition ID
            filtered_attrs = [attr for attr in agent_attrs if attr.get("attr_def_id") == attr_def_id]
            all_attributes.extend(filtered_attrs)
        except requests.exceptions.RequestException as e:
            print(f"Error fetching attributes for agent ID {agent_id}: {e}")
    
    return all_attributes

def get_agent_name(agent_id: int, agents: List[Dict[str, Any]]) -> str:
    """Get agent name from agent ID"""
    for agent in agents:
        if agent["agent_id"] == agent_id:
            return agent["agent_name"]
    return "Unknown"

def extract_apr_value(attr: Dict[str, Any]) -> Dict[str, Any]:
    """Extract APR value and timestamp from JSON value"""
    try:
        # The APR value is stored in the json_value field
        if attr["json_value"] is None:
            return {"apr": None, "timestamp": None, "agent_id": attr["agent_id"], "is_dummy": False}
        
        # If json_value is a string, parse it
        if isinstance(attr["json_value"], str):
            json_data = json.loads(attr["json_value"])
        else:
            json_data = attr["json_value"]
        
        apr = json_data.get("apr")
        timestamp = json_data.get("timestamp")
        
        # Convert timestamp to datetime if it exists
        timestamp_dt = None
        if timestamp:
            timestamp_dt = datetime.fromtimestamp(timestamp)
            
        return {"apr": apr, "timestamp": timestamp_dt, "agent_id": attr["agent_id"], "is_dummy": False}
    except (json.JSONDecodeError, KeyError, TypeError) as e:
        print(f"Error parsing JSON value: {e}")
        return {"apr": None, "timestamp": None, "agent_id": attr["agent_id"], "is_dummy": False}

def fetch_apr_data_from_db():
    """
    Fetch APR data from database using the API.
    """
    global global_df
    
    try:
        # Step 1: Find the Modius agent type
        modius_type = get_agent_type_by_name("Modius")
        if not modius_type:
            print("Modius agent type not found, using placeholder data")
            global_df = pd.DataFrame([])
            return global_df
        
        type_id = modius_type["type_id"]
        
        # Step 2: Find the APR attribute definition
        apr_attr_def = get_attribute_definition_by_name("APR")
        if not apr_attr_def:
            print("APR attribute definition not found, using placeholder data")
            global_df = pd.DataFrame([])
            return global_df
            
        attr_def_id = apr_attr_def["attr_def_id"]
        
        # Step 3: Get all agents of type Modius
        modius_agents = get_agents_by_type(type_id)
        if not modius_agents:
            print("No agents of type 'Modius' found")
            global_df = pd.DataFrame([])
            return global_df
        
        # Step 4: Fetch all APR values for Modius agents
        apr_attributes = get_attribute_values_by_type_and_attr(modius_agents, attr_def_id)
        if not apr_attributes:
            print("No APR values found for 'Modius' agents")
            global_df = pd.DataFrame([])
            return global_df
        
        # Step 5: Extract APR data
        apr_data_list = []
        for attr in apr_attributes:
            apr_data = extract_apr_value(attr)
            if apr_data["apr"] is not None and apr_data["timestamp"] is not None:
                # Get agent name
                agent_name = get_agent_name(attr["agent_id"], modius_agents)
                # Add agent name to the data
                apr_data["agent_name"] = agent_name
                # Add is_dummy flag (all real data)
                apr_data["is_dummy"] = False
                
                # Mark negative values as "Performance" metrics
                if apr_data["apr"] < 0:
                    apr_data["metric_type"] = "Performance"
                else:
                    apr_data["metric_type"] = "APR"
                    
                apr_data_list.append(apr_data)
        
        # Convert list of dictionaries to DataFrame
        if not apr_data_list:
            print("No valid APR data extracted")
            global_df = pd.DataFrame([])
            return global_df
        
        global_df = pd.DataFrame(apr_data_list)
        return global_df
        
    except requests.exceptions.RequestException as e:
        print(f"API request error: {e}")
        global_df = pd.DataFrame([])
        return global_df
    except Exception as e:
        print(f"Error fetching APR data: {e}")
        global_df = pd.DataFrame([])
        return global_df

def generate_apr_visualizations():
    """Generate APR visualizations with real data only (no dummy data)"""
    global global_df
    
    # Fetch data from database
    df = fetch_apr_data_from_db()
    
    # If we got no data at all, return placeholder figures
    if df.empty:
        print("No APR data available. Using fallback visualization.")
        # Create empty visualizations with a message using Plotly
        fig = go.Figure()
        fig.add_annotation(
            x=0.5, y=0.5,
            text="No APR data available",
            font=dict(size=20),
            showarrow=False
        )
        fig.update_layout(
            xaxis=dict(showgrid=False, zeroline=False, showticklabels=False),
            yaxis=dict(showgrid=False, zeroline=False, showticklabels=False)
        )
        
        # Save as static files for reference
        fig.write_html("modius_apr_per_agent_graph.html")
        fig.write_image("modius_apr_per_agent_graph.png")
        fig.write_html("modius_apr_combined_graph.html")
        fig.write_image("modius_apr_combined_graph.png")
        
        csv_file = None
        return fig, fig, csv_file
    
    # No longer generating dummy data
    # Set global_df for access by other functions
    global_df = df
    
    # Save to CSV before creating visualizations
    csv_file = save_to_csv(df)
    
    # Create per-agent time series graph (returns figure object)
    per_agent_fig = create_time_series_graph_per_agent(df)
    
    # Create combined time series graph (returns figure object)
    combined_fig = create_combined_time_series_graph(df)
    
    return per_agent_fig, combined_fig, csv_file

def create_time_series_graph_per_agent(df):
    """Create a time series graph for each agent using Plotly"""
    # Get unique agents
    unique_agents = df['agent_id'].unique()
    
    if len(unique_agents) == 0:
        print("No agent data to plot")
        fig = go.Figure()
        fig.add_annotation(
            text="No agent data available",
            x=0.5, y=0.5,
            showarrow=False, font=dict(size=20)
        )
        return fig
    
    # Create a subplot figure for each agent
    fig = make_subplots(rows=len(unique_agents), cols=1, 
                       subplot_titles=[f"Agent: {df[df['agent_id'] == agent_id]['agent_name'].iloc[0]}" 
                                      for agent_id in unique_agents],
                       vertical_spacing=0.1)
    
    # Plot data for each agent
    for i, agent_id in enumerate(unique_agents):
        agent_data = df[df['agent_id'] == agent_id].copy()
        agent_name = agent_data['agent_name'].iloc[0]
        row = i + 1
        
        # Add zero line to separate APR and Performance
        fig.add_shape(
            type="line", line=dict(dash="solid", width=1.5, color="black"),
            y0=0, y1=0, x0=agent_data['timestamp'].min(), x1=agent_data['timestamp'].max(),
            row=row, col=1
        )
        
        # Add background colors
        fig.add_shape(
            type="rect", fillcolor="rgba(230, 243, 255, 0.3)", line=dict(width=0),
            y0=0, y1=1000, x0=agent_data['timestamp'].min(), x1=agent_data['timestamp'].max(),
            row=row, col=1, layer="below"
        )
        fig.add_shape(
            type="rect", fillcolor="rgba(255, 230, 230, 0.3)", line=dict(width=0),
            y0=-1000, y1=0, x0=agent_data['timestamp'].min(), x1=agent_data['timestamp'].max(),
            row=row, col=1, layer="below"
        )
        
        # Create separate dataframes for different data types
        apr_data = agent_data[agent_data['metric_type'] == 'APR']
        perf_data = agent_data[agent_data['metric_type'] == 'Performance']
        
        # Sort all data by timestamp for the line plots
        combined_agent_data = agent_data.sort_values('timestamp')
        
        # Add main line connecting all points
        fig.add_trace(
            go.Scatter(
                x=combined_agent_data['timestamp'], 
                y=combined_agent_data['apr'],
                mode='lines',
                line=dict(color='purple', width=2),
                name=f'{agent_name}',
                legendgroup=agent_name,
                showlegend=(i == 0),  # Only show in legend once
                hovertemplate='Time: %{x}<br>Value: %{y:.2f}<extra></extra>'
            ),
            row=row, col=1
        )
        
        # Add scatter points for APR values
        if not apr_data.empty:
            fig.add_trace(
                go.Scatter(
                    x=apr_data['timestamp'], 
                    y=apr_data['apr'],
                    mode='markers',
                    marker=dict(color='blue', size=10, symbol='circle'),
                    name='APR',
                    legendgroup='APR',
                    showlegend=(i == 0),
                    hovertemplate='Time: %{x}<br>APR: %{y:.2f}<extra></extra>'
                ),
                row=row, col=1
            )
        
        # Add scatter points for Performance values
        if not perf_data.empty:
            fig.add_trace(
                go.Scatter(
                    x=perf_data['timestamp'], 
                    y=perf_data['apr'],
                    mode='markers',
                    marker=dict(color='red', size=10, symbol='square'),
                    name='Performance',
                    legendgroup='Performance',
                    showlegend=(i == 0),
                    hovertemplate='Time: %{x}<br>Performance: %{y:.2f}<extra></extra>'
                ),
                row=row, col=1
            )
        
        # Update axes
        fig.update_xaxes(title_text="Time", row=row, col=1)
        fig.update_yaxes(title_text="Value", row=row, col=1, gridcolor='rgba(0,0,0,0.1)')
    
    # Update layout
    fig.update_layout(
        height=400 * len(unique_agents),
        width=1000,
        title_text="APR and Performance Values per Agent",
        template="plotly_white",
        legend=dict(
            orientation="h",
            yanchor="bottom",
            y=1.02,
            xanchor="right",
            x=1
        ),
        margin=dict(r=20, l=20, t=30, b=20),
        hovermode="closest"
    )
    
    # Save the figure (still useful for reference)
    graph_file = "modius_apr_per_agent_graph.html"
    fig.write_html(graph_file, include_plotlyjs='cdn', full_html=False)
    
    # Also save as image for compatibility
    img_file = "modius_apr_per_agent_graph.png"
    fig.write_image(img_file)
    
    print(f"Per-agent graph saved to {graph_file} and {img_file}")
    
    # Return the figure object for direct use in Gradio
    return fig

def create_combined_time_series_graph(df):
    """Create a combined time series graph for all agents using Plotly"""
    if len(df) == 0:
        print("No data to plot combined graph")
        fig = go.Figure()
        fig.add_annotation(
            text="No data available",
            x=0.5, y=0.5,
            showarrow=False, font=dict(size=20)
        )
        return fig
    
    # Create Plotly figure
    fig = go.Figure()
    
    # Get unique agents
    unique_agents = df['agent_id'].unique()
    
    # Define a color scale for different agents
    colors = px.colors.qualitative.Plotly[:len(unique_agents)]
    
    # Add background shapes for APR and Performance regions
    min_time = df['timestamp'].min()
    max_time = df['timestamp'].max()
    
    # Add shape for APR region (above zero)
    fig.add_shape(
        type="rect",
        fillcolor="rgba(230, 243, 255, 0.3)",
        line=dict(width=0),
        y0=0, y1=1000,
        x0=min_time, x1=max_time,
        layer="below"
    )
    
    # Add shape for Performance region (below zero)
    fig.add_shape(
        type="rect",
        fillcolor="rgba(255, 230, 230, 0.3)",
        line=dict(width=0),
        y0=-1000, y1=0,
        x0=min_time, x1=max_time,
        layer="below"
    )
    
    # Add zero line
    fig.add_shape(
        type="line",
        line=dict(dash="solid", width=1.5, color="black"),
        y0=0, y1=0,
        x0=min_time, x1=max_time
    )
    
    # Add data for each agent
    for i, agent_id in enumerate(unique_agents):
        agent_data = df[df['agent_id'] == agent_id].copy()
        agent_name = agent_data['agent_name'].iloc[0]
        color = colors[i % len(colors)]
        
        # Sort the data by timestamp
        agent_data = agent_data.sort_values('timestamp')
        
        # Add the combined line for both APR and Performance
        fig.add_trace(
            go.Scatter(
                x=agent_data['timestamp'],
                y=agent_data['apr'],
                mode='lines',
                line=dict(color=color, width=2),
                name=f'{agent_name}',
                legendgroup=agent_name,
                hovertemplate='Time: %{x}<br>Value: %{y:.2f}<br>Agent: ' + agent_name + '<extra></extra>'
            )
        )
        
        # Add scatter points for APR values
        apr_data = agent_data[agent_data['metric_type'] == 'APR']
        if not apr_data.empty:
            fig.add_trace(
                go.Scatter(
                    x=apr_data['timestamp'],
                    y=apr_data['apr'],
                    mode='markers',
                    marker=dict(color=color, symbol='circle', size=8),
                    name=f'{agent_name} APR',
                    legendgroup=agent_name,
                    showlegend=False,
                    hovertemplate='Time: %{x}<br>APR: %{y:.2f}<br>Agent: ' + agent_name + '<extra></extra>'
                )
            )
        
        # Add scatter points for Performance values
        perf_data = agent_data[agent_data['metric_type'] == 'Performance']
        if not perf_data.empty:
            fig.add_trace(
                go.Scatter(
                    x=perf_data['timestamp'],
                    y=perf_data['apr'],
                    mode='markers',
                    marker=dict(color=color, symbol='square', size=8),
                    name=f'{agent_name} Perf',
                    legendgroup=agent_name,
                    showlegend=False,
                    hovertemplate='Time: %{x}<br>Performance: %{y:.2f}<br>Agent: ' + agent_name + '<extra></extra>'
                )
            )
    
    # Update layout
    fig.update_layout(
        title="APR and Performance Values for All Agents",
        xaxis_title="Time",
        yaxis_title="Value",
        template="plotly_white",
        height=600,
        width=1000,
        legend=dict(
            orientation="h",
            yanchor="bottom",
            y=1.02,
            xanchor="right",
            x=1,
            groupclick="toggleitem"
        ),
        margin=dict(r=20, l=20, t=30, b=20),
        hovermode="closest"
    )
    
    # Update axes
    fig.update_xaxes(showgrid=True, gridwidth=1, gridcolor='rgba(0,0,0,0.1)')
    fig.update_yaxes(showgrid=True, gridwidth=1, gridcolor='rgba(0,0,0,0.1)')
    
    # Save the figure (still useful for reference)
    graph_file = "modius_apr_combined_graph.html"
    fig.write_html(graph_file, include_plotlyjs='cdn', full_html=False)
    
    # Also save as image for compatibility
    img_file = "modius_apr_combined_graph.png"
    fig.write_image(img_file)
    
    print(f"Combined graph saved to {graph_file} and {img_file}")
    
    # Return the figure object for direct use in Gradio
    return fig

def save_to_csv(df):
    """Save the APR data DataFrame to a CSV file and return the file path"""
    if df.empty:
        print("No APR data to save to CSV")
        return None
    
    # Define the CSV file path
    csv_file = "modius_apr_values.csv"
    
    # Save to CSV
    df.to_csv(csv_file, index=False)
    print(f"APR data saved to {csv_file}")
    
    # Also generate a statistics CSV file
    stats_df = generate_statistics_from_data(df)
    stats_csv = "modius_apr_statistics.csv"
    stats_df.to_csv(stats_csv, index=False)
    print(f"Statistics saved to {stats_csv}")
    
    return csv_file

def generate_statistics_from_data(df):
    """Generate statistics from the APR data"""
    if df.empty:
        return pd.DataFrame()
    
    # Get unique agents
    unique_agents = df['agent_id'].unique()
    stats_list = []
    
    # Generate per-agent statistics
    for agent_id in unique_agents:
        agent_data = df[df['agent_id'] == agent_id]
        agent_name = agent_data['agent_name'].iloc[0]
        
        # APR statistics
        apr_data = agent_data[agent_data['metric_type'] == 'APR']
        real_apr = apr_data[apr_data['is_dummy'] == False]
        
        # Performance statistics
        perf_data = agent_data[agent_data['metric_type'] == 'Performance']
        real_perf = perf_data[perf_data['is_dummy'] == False]
        
        stats = {
            'agent_id': agent_id,
            'agent_name': agent_name,
            'total_points': len(agent_data),
            'apr_points': len(apr_data),
            'performance_points': len(perf_data),
            'real_apr_points': len(real_apr),
            'real_performance_points': len(real_perf),
            'avg_apr': apr_data['apr'].mean() if not apr_data.empty else None,
            'avg_performance': perf_data['apr'].mean() if not perf_data.empty else None,
            'max_apr': apr_data['apr'].max() if not apr_data.empty else None,
            'min_apr': apr_data['apr'].min() if not apr_data.empty else None,
            'latest_timestamp': agent_data['timestamp'].max().strftime('%Y-%m-%d %H:%M:%S') if not agent_data.empty else None
        }
        stats_list.append(stats)
    
    # Generate overall statistics
    apr_only = df[df['metric_type'] == 'APR']
    perf_only = df[df['metric_type'] == 'Performance']
    
    overall_stats = {
        'agent_id': 'ALL',
        'agent_name': 'All Agents',
        'total_points': len(df),
        'apr_points': len(apr_only),
        'performance_points': len(perf_only),
        'real_apr_points': len(apr_only[apr_only['is_dummy'] == False]),
        'real_performance_points': len(perf_only[perf_only['is_dummy'] == False]),
        'avg_apr': apr_only['apr'].mean() if not apr_only.empty else None,
        'avg_performance': perf_only['apr'].mean() if not perf_only.empty else None,
        'max_apr': apr_only['apr'].max() if not apr_only.empty else None,
        'min_apr': apr_only['apr'].min() if not apr_only.empty else None,
        'latest_timestamp': df['timestamp'].max().strftime('%Y-%m-%d %H:%M:%S') if not df.empty else None
    }
    stats_list.append(overall_stats)
    
    return pd.DataFrame(stats_list)