File size: 21,927 Bytes
824bf31
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
"""
Module: agents.ayrton_senna
Codinome: Ayrton Senna - Navegador das Rotas Perfeitas
Description: Semantic router for directing queries to appropriate agents with precision and speed
Author: Anderson H. Silva
Date: 2025-01-24
License: Proprietary - All rights reserved
"""

import re
from typing import Any, Dict, List, Optional, Tuple

from pydantic import BaseModel, Field as PydanticField

from src.core import AgentStatus, get_logger
from src.core.exceptions import AgentError, ValidationError
from .deodoro import (
    AgentContext,
    AgentMessage,
    AgentResponse,
    BaseAgent,
)


class RoutingRule(BaseModel):
    """Rule for routing queries to agents."""
    
    name: str = PydanticField(..., description="Rule name")
    patterns: List[str] = PydanticField(..., description="Regex patterns to match")
    keywords: List[str] = PydanticField(default_factory=list, description="Keywords to match")
    target_agent: str = PydanticField(..., description="Target agent name")
    action: str = PydanticField(..., description="Action to perform")
    priority: int = PydanticField(default=5, description="Rule priority (1-10)")
    confidence_threshold: float = PydanticField(default=0.7, description="Confidence threshold")
    metadata: Dict[str, Any] = PydanticField(default_factory=dict, description="Additional metadata")


class RoutingDecision(BaseModel):
    """Result of routing decision."""
    
    target_agent: str = PydanticField(..., description="Selected agent")
    action: str = PydanticField(..., description="Action to perform")
    confidence: float = PydanticField(..., description="Confidence in decision")
    rule_used: str = PydanticField(..., description="Rule that matched")
    parameters: Dict[str, Any] = PydanticField(default_factory=dict, description="Parameters for agent")
    fallback_agents: List[str] = PydanticField(default_factory=list, description="Fallback agents")


class SemanticRouter(BaseAgent):
    """
    Semantic router that analyzes queries and routes them to appropriate agents.
    
    The router uses:
    - Rule-based routing with regex patterns and keywords
    - Semantic similarity for complex queries
    - Intent detection for conversational flows
    - Fallback strategies for ambiguous cases
    """
    
    def __init__(
        self,
        llm_service: Any,
        embedding_service: Optional[Any] = None,
        confidence_threshold: float = 0.7,
        **kwargs: Any
    ) -> None:
        """
        Initialize semantic router.
        
        Args:
            llm_service: LLM service for intent detection
            embedding_service: Embedding service for semantic similarity
            confidence_threshold: Minimum confidence for routing decisions
            **kwargs: Additional arguments
        """
        super().__init__(
            name="SemanticRouter",
            description="Routes queries to appropriate agents based on semantic analysis",
            capabilities=[
                "route_query",
                "detect_intent",
                "analyze_query_type",
                "suggest_agents",
                "validate_routing",
            ],
            **kwargs
        )
        
        self.llm_service = llm_service
        self.embedding_service = embedding_service
        self.confidence_threshold = confidence_threshold
        self.routing_rules: List[RoutingRule] = []
        self.agent_capabilities: Dict[str, List[str]] = {}
        
        self._initialize_default_rules()
        
        self.logger.info(
            "semantic_router_initialized",
            confidence_threshold=confidence_threshold,
            rules_count=len(self.routing_rules),
        )
    
    async def initialize(self) -> None:
        """Initialize semantic router."""
        self.logger.info("semantic_router_initializing")
        
        # Initialize services
        if hasattr(self.llm_service, 'initialize'):
            await self.llm_service.initialize()
        
        if self.embedding_service and hasattr(self.embedding_service, 'initialize'):
            await self.embedding_service.initialize()
        
        self.status = AgentStatus.IDLE
        self.logger.info("semantic_router_initialized")
    
    async def shutdown(self) -> None:
        """Shutdown semantic router."""
        self.logger.info("semantic_router_shutting_down")
        
        if hasattr(self.llm_service, 'shutdown'):
            await self.llm_service.shutdown()
        
        if self.embedding_service and hasattr(self.embedding_service, 'shutdown'):
            await self.embedding_service.shutdown()
        
        self.logger.info("semantic_router_shutdown_complete")
    
    async def process(
        self,
        message: AgentMessage,
        context: AgentContext,
    ) -> AgentResponse:
        """
        Process routing requests.
        
        Args:
            message: Message to process
            context: Agent context
            
        Returns:
            Agent response with routing decision
        """
        action = message.action
        payload = message.payload
        
        self.logger.info(
            "semantic_router_processing",
            action=action,
            context_id=context.investigation_id,
        )
        
        try:
            if action == "route_query":
                result = await self._route_query(payload, context)
            elif action == "detect_intent":
                result = await self._detect_intent(payload, context)
            elif action == "analyze_query_type":
                result = await self._analyze_query_type(payload, context)
            elif action == "suggest_agents":
                result = await self._suggest_agents(payload, context)
            elif action == "validate_routing":
                result = await self._validate_routing(payload, context)
            else:
                raise AgentError(
                    f"Unknown action: {action}",
                    details={"action": action, "available_actions": self.capabilities}
                )
            
            return AgentResponse(
                agent_name=self.name,
                status=AgentStatus.COMPLETED,
                result=result,
                metadata={"action": action, "context_id": context.investigation_id},
            )
            
        except Exception as e:
            self.logger.error(
                "semantic_router_processing_failed",
                action=action,
                error=str(e),
                context_id=context.investigation_id,
            )
            
            return AgentResponse(
                agent_name=self.name,
                status=AgentStatus.ERROR,
                error=str(e),
                metadata={"action": action, "context_id": context.investigation_id},
            )
    
    def register_agent_capabilities(
        self,
        agent_name: str,
        capabilities: List[str],
    ) -> None:
        """
        Register agent capabilities for routing decisions.
        
        Args:
            agent_name: Name of the agent
            capabilities: List of capabilities
        """
        self.agent_capabilities[agent_name] = capabilities
        self.logger.info(
            "agent_capabilities_registered",
            agent_name=agent_name,
            capabilities=capabilities,
        )
    
    def add_routing_rule(self, rule: RoutingRule) -> None:
        """
        Add a custom routing rule.
        
        Args:
            rule: Routing rule to add
        """
        self.routing_rules.append(rule)
        # Sort by priority (higher priority first)
        self.routing_rules.sort(key=lambda r: r.priority, reverse=True)
        
        self.logger.info(
            "routing_rule_added",
            rule_name=rule.name,
            target_agent=rule.target_agent,
            priority=rule.priority,
        )
    
    async def route_query(
        self,
        query: str,
        context: AgentContext,
        user_preferences: Optional[Dict[str, Any]] = None,
    ) -> RoutingDecision:
        """
        Route a query to the most appropriate agent.
        
        Args:
            query: Query to route
            context: Agent context
            user_preferences: Optional user preferences
            
        Returns:
            Routing decision
        """
        self.logger.info(
            "routing_query",
            query=query[:100],  # Log first 100 chars
            context_id=context.investigation_id,
        )
        
        # Step 1: Rule-based routing
        rule_decision = await self._apply_routing_rules(query, context)
        
        if rule_decision and rule_decision.confidence >= self.confidence_threshold:
            self.logger.info(
                "rule_based_routing_success",
                target_agent=rule_decision.target_agent,
                confidence=rule_decision.confidence,
                rule=rule_decision.rule_used,
            )
            return rule_decision
        
        # Step 2: Semantic routing using LLM
        semantic_decision = await self._semantic_routing(query, context)
        
        if semantic_decision and semantic_decision.confidence >= self.confidence_threshold:
            self.logger.info(
                "semantic_routing_success",
                target_agent=semantic_decision.target_agent,
                confidence=semantic_decision.confidence,
            )
            return semantic_decision
        
        # Step 3: Fallback to master agent
        fallback_decision = RoutingDecision(
            target_agent="MasterAgent",
            action="investigate",
            confidence=0.5,
            rule_used="fallback",
            parameters={"query": query},
            fallback_agents=["InvestigatorAgent", "AnalystAgent"],
        )
        
        self.logger.warning(
            "routing_fallback_used",
            query=query[:50],
            confidence=fallback_decision.confidence,
        )
        
        return fallback_decision
    
    async def _route_query(
        self,
        payload: Dict[str, Any],
        context: AgentContext,
    ) -> RoutingDecision:
        """Route query based on payload."""
        query = payload.get("query", "")
        if not query:
            raise ValidationError("Query is required for routing")
        
        user_preferences = payload.get("user_preferences")
        return await self.route_query(query, context, user_preferences)
    
    async def _apply_routing_rules(
        self,
        query: str,
        context: AgentContext,
    ) -> Optional[RoutingDecision]:
        """Apply rule-based routing."""
        query_lower = query.lower()
        
        for rule in self.routing_rules:
            confidence = 0.0
            
            # Check regex patterns
            pattern_matches = 0
            for pattern in rule.patterns:
                if re.search(pattern, query_lower, re.IGNORECASE):
                    pattern_matches += 1
            
            if rule.patterns:
                confidence += (pattern_matches / len(rule.patterns)) * 0.6
            
            # Check keywords
            keyword_matches = 0
            for keyword in rule.keywords:
                if keyword.lower() in query_lower:
                    keyword_matches += 1
            
            if rule.keywords:
                confidence += (keyword_matches / len(rule.keywords)) * 0.4
            
            # Apply rule priority weight
            confidence = min(confidence * (rule.priority / 10), 1.0)
            
            if confidence >= rule.confidence_threshold:
                return RoutingDecision(
                    target_agent=rule.target_agent,
                    action=rule.action,
                    confidence=confidence,
                    rule_used=rule.name,
                    parameters={"query": query, **rule.metadata},
                )
        
        return None
    
    async def _semantic_routing(
        self,
        query: str,
        context: AgentContext,
    ) -> Optional[RoutingDecision]:
        """Use LLM for semantic routing."""
        try:
            routing_prompt = self._create_routing_prompt(query)
            
            response = await self.llm_service.generate(
                prompt=routing_prompt,
                context=context,
            )
            
            # Parse LLM response
            decision = self._parse_routing_response(response, query)
            return decision
            
        except Exception as e:
            self.logger.error(
                "semantic_routing_failed",
                query=query[:50],
                error=str(e),
            )
            return None
    
    async def _detect_intent(
        self,
        payload: Dict[str, Any],
        context: AgentContext,
    ) -> Dict[str, Any]:
        """Detect intent from query."""
        query = payload.get("query", "")
        
        # Simple intent detection based on patterns
        intents = {
            "investigation": ["investigar", "analisar", "verificar", "buscar"],
            "explanation": ["explicar", "entender", "como", "por que"],
            "comparison": ["comparar", "diferença", "melhor", "versus"],
            "trend_analysis": ["tendência", "evolução", "histórico", "ao longo"],
            "anomaly_detection": ["suspeito", "anômalo", "irregular", "estranho"],
        }
        
        query_lower = query.lower()
        detected_intents = []
        
        for intent, keywords in intents.items():
            confidence = sum(1 for keyword in keywords if keyword in query_lower)
            if confidence > 0:
                detected_intents.append({
                    "intent": intent,
                    "confidence": min(confidence / len(keywords), 1.0),
                })
        
        # Sort by confidence
        detected_intents.sort(key=lambda x: x["confidence"], reverse=True)
        
        return {
            "query": query,
            "intents": detected_intents,
            "primary_intent": detected_intents[0]["intent"] if detected_intents else "unknown",
        }
    
    async def _analyze_query_type(
        self,
        payload: Dict[str, Any],
        context: AgentContext,
    ) -> Dict[str, Any]:
        """Analyze query type and complexity."""
        query = payload.get("query", "")
        
        # Simple query analysis
        analysis = {
            "length": len(query),
            "word_count": len(query.split()),
            "has_numbers": bool(re.search(r'\d', query)),
            "has_dates": bool(re.search(r'\d{4}|\d{2}/\d{2}', query)),
            "has_organizations": bool(re.search(r'ministério|prefeitura|secretaria', query, re.IGNORECASE)),
            "complexity": "simple",
        }
        
        # Determine complexity
        if analysis["word_count"] > 20 or "e" in query.lower():
            analysis["complexity"] = "complex"
        elif analysis["word_count"] > 10:
            analysis["complexity"] = "medium"
        
        return analysis
    
    async def _suggest_agents(
        self,
        payload: Dict[str, Any],
        context: AgentContext,
    ) -> List[Dict[str, Any]]:
        """Suggest possible agents for a query."""
        query = payload.get("query", "")
        
        suggestions = []
        
        # Analyze query and match with agent capabilities
        for agent_name, capabilities in self.agent_capabilities.items():
            score = 0.0
            reasons = []
            
            query_lower = query.lower()
            
            # Score based on capabilities
            if "investigar" in query_lower and "investigate" in capabilities:
                score += 0.8
                reasons.append("Query requires investigation")
            
            if "analisar" in query_lower and "analyze" in capabilities:
                score += 0.7
                reasons.append("Query requires analysis")
            
            if "relatório" in query_lower and "report" in capabilities:
                score += 0.6
                reasons.append("Query mentions reports")
            
            if score > 0:
                suggestions.append({
                    "agent_name": agent_name,
                    "score": score,
                    "reasons": reasons,
                    "capabilities": capabilities,
                })
        
        # Sort by score
        suggestions.sort(key=lambda x: x["score"], reverse=True)
        
        return suggestions
    
    async def _validate_routing(
        self,
        payload: Dict[str, Any],
        context: AgentContext,
    ) -> Dict[str, Any]:
        """Validate a routing decision."""
        decision_data = payload.get("decision", {})
        
        target_agent = decision_data.get("target_agent")
        action = decision_data.get("action")
        
        validation = {
            "valid": True,
            "warnings": [],
            "errors": [],
        }
        
        # Check if agent exists
        if target_agent not in self.agent_capabilities:
            validation["valid"] = False
            validation["errors"].append(f"Agent {target_agent} not registered")
        
        # Check if agent supports the action
        elif action not in self.agent_capabilities.get(target_agent, []):
            validation["warnings"].append(f"Agent {target_agent} may not support action {action}")
        
        return validation
    
    def _initialize_default_rules(self) -> None:
        """Initialize default routing rules."""
        rules = [
            # Investigation rules
            RoutingRule(
                name="investigation_query",
                patterns=[r"investigar|verificar|analisar.*gasto"],
                keywords=["investigar", "verificar", "analisar", "suspeito"],
                target_agent="MasterAgent",
                action="investigate",
                priority=9,
            ),
            
            # Anomaly detection rules
            RoutingRule(
                name="anomaly_detection",
                patterns=[r"suspeito|anômalo|irregular|estranho"],
                keywords=["suspeito", "anômalo", "irregular", "superfaturamento"],
                target_agent="InvestigatorAgent",
                action="detect_anomalies",
                priority=8,
            ),
            
            # Pattern analysis rules
            RoutingRule(
                name="pattern_analysis",
                patterns=[r"padrão|tendência|evolução"],
                keywords=["padrão", "tendência", "evolução", "histórico"],
                target_agent="AnalystAgent",
                action="analyze_patterns",
                priority=7,
            ),
            
            # Report generation rules
            RoutingRule(
                name="report_generation",
                patterns=[r"relatório|resumo|gerar.*relatório"],
                keywords=["relatório", "resumo", "documento"],
                target_agent="ReporterAgent",
                action="generate_report",
                priority=6,
            ),
            
            # Memory/context rules
            RoutingRule(
                name="memory_query",
                patterns=[r"lembrar|anterior|histórico.*investigação"],
                keywords=["lembrar", "anterior", "histórico"],
                target_agent="ContextMemoryAgent",
                action="retrieve_episodic",
                priority=5,
            ),
        ]
        
        for rule in rules:
            self.routing_rules.append(rule)
        
        # Sort by priority
        self.routing_rules.sort(key=lambda r: r.priority, reverse=True)
    
    def _create_routing_prompt(self, query: str) -> str:
        """Create prompt for LLM-based routing."""
        agents_info = []
        for agent_name, capabilities in self.agent_capabilities.items():
            agents_info.append(f"- {agent_name}: {', '.join(capabilities)}")
        
        agents_text = "\n".join(agents_info) if agents_info else "- MasterAgent: investigate, coordinate"
        
        return f"""
        Analise a seguinte consulta e determine qual agente é mais adequado para processá-la:
        
        Consulta: "{query}"
        
        Agentes disponíveis:
        {agents_text}
        
        Responda em formato JSON:
        {{
            "target_agent": "nome_do_agente",
            "action": "ação_a_executar", 
            "confidence": 0.8,
            "reasoning": "explicação da escolha"
        }}
        """
    
    def _parse_routing_response(
        self,
        response: str,
        query: str,
    ) -> Optional[RoutingDecision]:
        """Parse LLM routing response."""
        try:
            import json
            
            # Extract JSON from response
            json_start = response.find('{')
            json_end = response.rfind('}') + 1
            
            if json_start >= 0 and json_end > json_start:
                json_str = response[json_start:json_end]
                data = json.loads(json_str)
                
                return RoutingDecision(
                    target_agent=data.get("target_agent", "MasterAgent"),
                    action=data.get("action", "investigate"),
                    confidence=data.get("confidence", 0.5),
                    rule_used="llm_semantic",
                    parameters={"query": query, "reasoning": data.get("reasoning", "")},
                )
            
        except Exception as e:
            self.logger.error(
                "routing_response_parse_failed",
                response=response[:100],
                error=str(e),
            )
        
        return None