File size: 21,262 Bytes
3bbd581 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 |
import time
from typing import Dict, List, Any, Optional
from dataclasses import dataclass, asdict
import re
# Import visualization dependencies with fallbacks
try:
import networkx as nx
import matplotlib.pyplot as plt
plt.switch_backend('Agg')
import matplotlib
matplotlib.use('Agg')
import warnings
warnings.filterwarnings('ignore', category=UserWarning, module='matplotlib')
except ImportError:
print("Warning: Visualization deps missing. Install with: pip install networkx matplotlib")
nx = None
plt = None
@dataclass
class WorkflowStep:
step_id: str
step_type: str
timestamp: float
content: str
metadata: Dict[str, Any]
duration: Optional[float] = None
status: str = 'pending'
parent_step: Optional[str] = None
details: Optional[Dict[str, Any]] = None
mcp_server: Optional[str] = None # Added to track MCP server
tool_name: Optional[str] = None # Added to track specific tool
class EnhancedWorkflowVisualizer:
def __init__(self):
self.steps: List[WorkflowStep] = []
self.current_step: Optional[WorkflowStep] = None
self.start_time = time.time()
self.step_counter = 0
# MCP server mapping for better display names
self.server_display_names = {
"7860": "Semantic Server",
"7861": "Token Counter",
"7862": "Sentiment Analysis",
"7863": "Health Monitor"
}
def _extract_mcp_server_from_url(self, url_or_content: str) -> Optional[str]:
"""Extract MCP server name from URL or content."""
if not url_or_content:
return None
# Extract port from URL
port_match = re.search(r':(\d{4})', url_or_content)
if port_match:
port = port_match.group(1)
return self.server_display_names.get(port, f"Port {port}")
# Check for server keywords in content
if "semantic" in url_or_content.lower():
return "Semantic Server"
elif "token" in url_or_content.lower():
return "Token Counter"
elif "sentiment" in url_or_content.lower():
return "Sentiment Analysis"
elif "health" in url_or_content.lower():
return "Health Monitor"
return None
def _extract_tool_name(self, content: str) -> Optional[str]:
"""Extract tool name from content."""
# Enhanced tool patterns - prioritize actual function names
function_patterns = [
# Specific MCP tool functions (high priority)
r'\b(sentiment_analysis)\s*\(',
r'\b(count_tokens_openai_gpt4)\s*\(',
r'\b(count_tokens_openai_gpt3)\s*\(',
r'\b(count_tokens_openai_davinci)\s*\(',
r'\b(count_tokens_bert_family)\s*\(',
r'\b(count_tokens_roberta_family)\s*\(',
r'\b(count_tokens_gpt2_family)\s*\(',
r'\b(count_tokens_t5_family)\s*\(',
r'\b(count_tokens_distilbert)\s*\(',
r'\b(semantic_similarity)\s*\(',
r'\b(find_similar_sentences)\s*\(',
r'\b(extract_semantic_keywords)\s*\(',
r'\b(semantic_search_in_text)\s*\(',
r'\b(health_check)\s*\(',
r'\b(server_status)\s*\(',
r'\b(get_server_info)\s*\(',
# Generic patterns (lower priority)
r'(\w*sentiment_analysis\w*)',
r'(\w*semantic_similarity\w*)',
r'(\w*find_similar_sentences\w*)',
r'(\w*extract_semantic_keywords\w*)',
r'(\w*semantic_search_in_text\w*)',
r'(\w*count_tokens_\w+)',
r'(\w*health_check\w*)',
r'(\w*server_status\w*)'
]
# Try high-priority function call patterns first
for pattern in function_patterns:
match = re.search(pattern, content, re.IGNORECASE)
if match:
tool_name = match.group(1)
# Skip common non-tool functions
if tool_name not in ['print', 'len', 'str', 'int', 'float', 'final_answer', 'sse', 'model']:
return tool_name
# Check for execution logs that contain actual function names
if "count_tokens_openai_gpt4" in content:
return "count_tokens_openai_gpt4"
elif "sentiment_analysis" in content:
return "sentiment_analysis"
elif "extract_semantic_keywords" in content:
return "extract_semantic_keywords"
elif "semantic_similarity" in content:
return "semantic_similarity"
elif "find_similar_sentences" in content:
return "find_similar_sentences"
elif "semantic_search_in_text" in content:
return "semantic_search_in_text"
return None
def add_step(self, step_type: str, content: str, metadata: Optional[Dict[str, Any]] = None,
parent_step: Optional[str] = None, details: Optional[Dict[str, Any]] = None,
mcp_server: Optional[str] = None, tool_name: Optional[str] = None) -> str:
step_id = f"{step_type}_{self.step_counter}"
self.step_counter += 1
# Auto-extract MCP server and tool if not provided
if not mcp_server:
mcp_server = self._extract_mcp_server_from_url(content)
if not tool_name:
tool_name = self._extract_tool_name(content)
step = WorkflowStep(
step_id=step_id,
step_type=step_type,
timestamp=time.time(),
content=content,
metadata=metadata or {},
status='running',
parent_step=parent_step,
details=details or {},
mcp_server=mcp_server,
tool_name=tool_name
)
self.steps.append(step)
self.current_step = step
return step_id
def complete_step(self, step_id: str, status: str = 'completed',
additional_metadata: Optional[Dict[str, Any]] = None,
details: Optional[Dict[str, Any]] = None):
for step in self.steps:
if step.step_id == step_id:
step.status = status
step.duration = time.time() - step.timestamp
if additional_metadata and step.metadata is not None:
step.metadata.update(additional_metadata)
if details and step.details is not None:
step.details.update(details)
break
def add_communication_step(self, from_component: str, to_component: str,
message_type: str, content: str,
parent_step: Optional[str] = None) -> str:
"""Add a communication step between components."""
step_type = f"comm_{from_component}_to_{to_component}"
# Extract server info for communication steps
mcp_server = self._extract_mcp_server_from_url(content)
tool_name = self._extract_tool_name(content)
details = {
"from": from_component,
"to": to_component,
"message_type": message_type,
"content_preview": content[:100] + "..." if len(content) > 100 else content
}
return self.add_step(step_type, f"{message_type}: {from_component} → {to_component}",
parent_step=parent_step, details=details,
mcp_server=mcp_server, tool_name=tool_name)
def add_tool_execution_step(self, tool_name: str, mcp_server: str,
input_data: str, parent_step: Optional[str] = None) -> str:
"""Specialized method for tool execution steps."""
content = f"Executing {tool_name} on {mcp_server}"
return self.add_step("tool_execution", content,
parent_step=parent_step,
mcp_server=mcp_server,
tool_name=tool_name,
details={"input_preview": input_data[:50] + "..." if len(input_data) > 50 else input_data})
def generate_graph(self) -> Any:
if nx is None:
return None
G = nx.DiGraph()
# Enhanced color mapping with server-specific colors
color_map = {
'input': '#e3f2fd', # Light blue
'agent_init': '#f3e5f5', # Light purple
'agent_process': '#e8f5e8', # Light green
'comm_agent_to_mcp': '#fff3e0', # Light orange
'comm_mcp_to_server': '#ffebee', # Light red
'comm_server_to_mcp': '#e0f2f1', # Light teal
'comm_mcp_to_agent': '#f9fbe7', # Light lime
'llm_call': '#fce4ec', # Light pink
'tool_execution': '#e1f5fe', # Light cyan
'response': '#f1f8e9', # Light green
'error': '#ffcdd2' # Light red
}
# Add nodes with enhanced labeling
for step in self.steps:
color = color_map.get(step.step_type, '#f5f5f5')
# Create enhanced label with MCP server and tool info
duration_str = f" ({step.duration:.2f}s)" if step.duration else ""
# Build comprehensive label
label_parts = []
# Add step type
step_display = step.step_type.replace('_', ' ').title()
label_parts.append(step_display)
# Add MCP server info
if step.mcp_server:
label_parts.append(f"📡 {step.mcp_server}")
# Add tool name prominently
if step.tool_name:
label_parts.append(f"🔧 {step.tool_name}")
# Add content preview (shortened to make room for server/tool info)
content_preview = step.content[:20] + "..." if len(step.content) > 20 else step.content
if not step.tool_name or step.tool_name.lower() not in content_preview.lower():
label_parts.append(content_preview)
# Add duration
if duration_str:
label_parts.append(duration_str)
label = "\n".join(label_parts)
G.add_node(step.step_id,
label=label,
color=color,
step_type=step.step_type,
status=step.status,
mcp_server=step.mcp_server,
tool_name=step.tool_name)
# Add edges based on parent relationships and chronological order
for i, step in enumerate(self.steps):
if step.parent_step:
# Add edge from parent step
G.add_edge(step.parent_step, step.step_id, edge_type='parent')
elif i > 0:
# Add chronological edge to previous step
G.add_edge(self.steps[i-1].step_id, step.step_id, edge_type='sequence')
return G
def create_matplotlib_visualization(self) -> str:
if nx is None or plt is None:
return ""
G = self.generate_graph()
if not G or len(G.nodes()) == 0:
return ""
# Create larger figure to accommodate enhanced labels
fig, ax = plt.subplots(figsize=(20, 12))
# Use hierarchical layout if possible
try:
pos = nx.spring_layout(G, k=3, iterations=150, seed=42)
except:
pos = nx.circular_layout(G)
# Prepare node visualization with server-aware coloring
node_colors = []
node_labels = {}
node_sizes = []
for node_id in G.nodes():
step = next(s for s in self.steps if s.step_id == node_id)
# Enhanced color coding based on status and server
if step.status == 'error':
color = '#ff5252'
elif step.status == 'completed':
# Server-specific color coding
if step.mcp_server == "Semantic Server":
base_color = '#4caf50' # Green for semantic
elif step.mcp_server == "Token Counter":
base_color = '#2196f3' # Blue for token counting
elif step.mcp_server == "Sentiment Analysis":
base_color = '#ff9800' # Orange for sentiment
elif step.mcp_server == "Health Monitor":
base_color = '#9c27b0' # Purple for health
else:
# Default colors by step type
base_colors = {
'input': '#4caf50',
'agent_init': '#9c27b0',
'agent_process': '#2e7d32',
'comm_agent_to_mcp': '#ff9800',
'comm_mcp_to_server': '#f44336',
'comm_server_to_mcp': '#009688',
'comm_mcp_to_agent': '#8bc34a',
'llm_call': '#e91e63',
'tool_execution': '#03a9f4',
'response': '#4caf50'
}
base_color = base_colors.get(step.step_type, '#607d8b')
color = base_color
else:
color = '#bdbdbd'
node_colors.append(color)
# Create enhanced node labels
label_parts = []
# Step type
step_display = step.step_type.replace('_', ' ').title()
label_parts.append(f"**{step_display}**")
# MCP Server (prominent)
if step.mcp_server:
label_parts.append(f"📡 {step.mcp_server}")
# Tool name (most prominent)
if step.tool_name:
label_parts.append(f"🔧 **{step.tool_name}**")
# Duration
if step.duration:
label_parts.append(f"⏱️ {step.duration:.2f}s")
node_labels[node_id] = "\n".join(label_parts)
# Size based on importance - larger for tool executions
if step.step_type == 'tool_execution':
node_sizes.append(5000)
elif step.step_type in ['input', 'response']:
node_sizes.append(4000)
elif 'comm_' in step.step_type:
node_sizes.append(2500)
else:
node_sizes.append(3000)
# Draw the graph
nx.draw(G, pos,
node_color=node_colors,
node_size=node_sizes,
font_size=9,
font_weight='bold',
arrows=True,
arrowsize=20,
edge_color='#666666',
alpha=0.9,
ax=ax,
arrowstyle='->')
# Draw enhanced labels
nx.draw_networkx_labels(G, pos, node_labels, font_size=8, ax=ax)
# Add title and formatting
ax.set_title("MCP Agent Workflow: Server & Tool Execution Flow",
fontsize=20, pad=25, fontweight='bold')
ax.axis('off')
# Enhanced legend with server info
legend_elements = [
plt.Rectangle((0,0),1,1, facecolor='#4caf50', label='Semantic Server'),
plt.Rectangle((0,0),1,1, facecolor='#2196f3', label='Token Counter Server'),
plt.Rectangle((0,0),1,1, facecolor='#ff9800', label='Sentiment Analysis Server'),
plt.Rectangle((0,0),1,1, facecolor='#9c27b0', label='Health Monitor Server'),
plt.Rectangle((0,0),1,1, facecolor='#e91e63', label='LLM Calls'),
plt.Rectangle((0,0),1,1, facecolor='#607d8b', label='Agent Processing'),
]
ax.legend(handles=legend_elements, loc='upper left', bbox_to_anchor=(0, 1))
fig.set_constrained_layout(True)
# Save to temporary file
import tempfile
temp_file = tempfile.NamedTemporaryFile(delete=False, suffix='.png')
plt.savefig(temp_file.name, format='png', dpi=300, bbox_inches='tight')
plt.close(fig)
return temp_file.name
def get_workflow_summary(self) -> Dict[str, Any]:
total_duration = time.time() - self.start_time
# Count steps by type and server
step_counts = {}
server_usage = {}
tool_usage = {}
communication_steps = []
processing_steps = []
for step in self.steps:
step_counts[step.step_type] = step_counts.get(step.step_type, 0) + 1
# Track server usage
if step.mcp_server:
server_usage[step.mcp_server] = server_usage.get(step.mcp_server, 0) + 1
# Track tool usage
if step.tool_name:
tool_usage[step.tool_name] = tool_usage.get(step.tool_name, 0) + 1
if 'comm_' in step.step_type:
communication_steps.append({
'step_id': step.step_id,
'from': step.details.get('from', 'unknown') if step.details else 'unknown',
'to': step.details.get('to', 'unknown') if step.details else 'unknown',
'message_type': step.details.get('message_type', 'unknown') if step.details else 'unknown',
'mcp_server': step.mcp_server,
'tool_name': step.tool_name,
'duration': step.duration,
'status': step.status
})
else:
processing_steps.append({
'step_id': step.step_id,
'type': step.step_type,
'content': step.content[:50] + "..." if len(step.content) > 50 else step.content,
'mcp_server': step.mcp_server,
'tool_name': step.tool_name,
'duration': step.duration,
'status': step.status
})
# Calculate timing statistics
completed_steps = [s for s in self.steps if s.duration is not None]
avg_duration = (sum(s.duration or 0 for s in completed_steps) / len(completed_steps)) if completed_steps else 0
return {
'total_steps': len(self.steps),
'total_duration': round(total_duration, 3),
'average_step_duration': round(avg_duration, 3),
'step_counts': step_counts,
'server_usage': server_usage, # New: server usage stats
'tool_usage': tool_usage, # New: tool usage stats
'communication_flow': communication_steps,
'processing_steps': processing_steps,
'status': 'completed' if all(s.status in ['completed', 'error'] for s in self.steps) else 'running',
'error_count': sum(1 for s in self.steps if s.status == 'error'),
'success_rate': round((sum(1 for s in self.steps if s.status == 'completed') / len(self.steps)) * 100, 1) if self.steps else 0,
'detailed_steps': [asdict(s) for s in self.steps]
}
# Global instance
workflow_visualizer = EnhancedWorkflowVisualizer()
# Enhanced helper functions
def track_workflow_step(step_type: str, content: str, metadata: Optional[Dict[str, Any]] = None,
parent_step: Optional[str] = None, mcp_server: Optional[str] = None,
tool_name: Optional[str] = None) -> str:
return workflow_visualizer.add_step(step_type, content, metadata, parent_step,
mcp_server=mcp_server, tool_name=tool_name)
def track_communication(from_component: str, to_component: str, message_type: str,
content: str, parent_step: Optional[str] = None) -> str:
return workflow_visualizer.add_communication_step(from_component, to_component,
message_type, content, parent_step)
def track_tool_execution(tool_name: str, mcp_server: str, input_data: str,
parent_step: Optional[str] = None) -> str:
"""New helper for tracking tool executions with clear server/tool info."""
return workflow_visualizer.add_tool_execution_step(tool_name, mcp_server, input_data, parent_step)
def complete_workflow_step(step_id: str, status: str = 'completed',
metadata: Optional[Dict[str, Any]] = None,
details: Optional[Dict[str, Any]] = None):
workflow_visualizer.complete_step(step_id, status, metadata, details)
def get_workflow_visualization() -> str:
return workflow_visualizer.create_matplotlib_visualization()
def get_workflow_summary() -> Dict[str, Any]:
return workflow_visualizer.get_workflow_summary()
def reset_workflow():
global workflow_visualizer
workflow_visualizer = EnhancedWorkflowVisualizer() |