Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
This section is empty.
Types ¶
type AgenticSearchResponse ¶
type AgenticSearchResponse struct {
SearchResponse
// SubQueries used in parallel retrieval
SubQueries []string
// CRAGEvaluation result
CRAGEvaluation *retrieval.CRAGEvaluation
// RAGEvaluation result
RAGEvaluation *retrieval.RAGEScores
// ToolExecutions tracks any external tool calls
ToolExecutions []string
}
AgenticSearchResponse extends SearchResponse with Agentic RAG metadata.
type AgenticSearchService ¶
type AgenticSearchService struct {
// contains filtered or unexported fields
}
AgenticSearchService orchestrates the complete Agentic RAG search workflow.
This is a thin orchestration layer that coordinates multiple domain interfaces to complete the full Agentic RAG use case.
Business logic lives in infra/* implementations, not here.
func NewAgenticSearchService ¶
func NewAgenticSearchService( ic IntentClassifier, d QueryDecomposer, ce CRAGEvaluator, r Retriever, g Generator, re RAGEvaluator, ) *AgenticSearchService
NewAgenticSearchService creates a new Agentic search service.
func (*AgenticSearchService) Search ¶
func (s *AgenticSearchService) Search(ctx context.Context, req *SearchRequest) (*AgenticSearchResponse, error)
Search executes the complete Agentic RAG workflow.
Workflow: 1. Intent Classification → Determine if RAG is needed 2. Query Decomposition → Break into sub-queries (if complex) 3. Parallel Retrieval → Retrieve chunks for all sub-queries 4. CRAG Evaluation → Assess retrieval quality 5. Answer Generation → Generate response from chunks 6. RAG Evaluation → Assess answer quality
Parameters: - ctx: The context for cancellation and timeouts - req: The search request containing query and configuration
Returns: - *AgenticSearchResponse: Complete response with metadata - error: Any error encountered during the workflow
type CRAGEvaluator ¶
type CRAGEvaluator interface {
Evaluate(ctx context.Context, query *entity.Query, chunks []*entity.Chunk) (*retrieval.CRAGEvaluation, error)
}
CRAGEvaluator defines the interface for retrieval quality evaluation.
type ChatRequest ¶
type ChatRequest struct {
// Message is the user's chat message
Message string
// UserID identifies the user
UserID string
// SessionID tracks the conversation session
SessionID string
// History contains previous conversation messages
History []string
}
ChatRequest is the DTO for chat use case input.
type ChatResponse ¶
type ChatResponse struct {
// Message is the chatbot's response
Message string
// SessionID tracks the conversation session
SessionID string
}
ChatResponse is the DTO for chat use case output.
type Generator ¶
type Generator interface {
Generate(ctx context.Context, query *entity.Query, chunks []*entity.Chunk) (*retrieval.GenerationResult, error)
}
Generator defines the interface for answer generation. The return type matches retrieval.Generator.Generate which returns *retrieval.GenerationResult.
type IndexRequest ¶
type IndexRequest struct {
// Documents to index
Documents []*entity.Document
// Collection is the target collection name
Collection string
// BatchSize for bulk indexing (default: 100)
BatchSize int
}
IndexRequest is the DTO for indexing use case input.
type IndexResponse ¶
type IndexResponse struct {
// TotalDocuments indexed successfully
TotalDocuments int
// FailedDocuments count
FailedDocuments int
// Errors encountered during indexing
Errors []string
}
IndexResponse is the DTO for indexing use case output.
type IntentClassifier ¶
type IntentClassifier interface {
Classify(ctx context.Context, query *entity.Query) (*retrieval.IntentResult, error)
}
IntentClassifier defines the interface for intent classification.
type QueryDecomposer ¶
type QueryDecomposer interface {
Decompose(ctx context.Context, query *entity.Query) (*retrieval.DecompositionResult, error)
}
QueryDecomposer defines the interface for query decomposition.
type RAGEvaluator ¶
type RAGEvaluator interface {
Evaluate(ctx context.Context, query, answer, context string) (*retrieval.RAGEScores, error)
}
RAGEvaluator defines the interface for answer quality evaluation.
type Retriever ¶
type Retriever interface {
Retrieve(ctx context.Context, queries []string, topK int) ([]*retrieval.RetrievalResult, error)
}
Retriever defines the interface for document retrieval. The signature aligns with retrieval.Retriever and infra/service.retriever.
type SearchRequest ¶
type SearchRequest struct {
// Query is the user's search query
Query *entity.Query
// TopK is the number of chunks to retrieve (default: 5)
TopK int
// UserID identifies the user making the request
UserID string
// SessionID tracks the conversation session
SessionID string
}
SearchRequest is the DTO for search use case input.
type SearchResponse ¶
type SearchResponse struct {
// Answer is the generated response
Answer string
// Chunks contains the retrieved chunks used in the answer
Chunks []string
// Score is the relevance score (0-1)
Score float32
// Intent is the classified intent
Intent retrieval.IntentType
// SourceDocuments lists the source document IDs
SourceDocuments []string
}
SearchResponse is the DTO for search use case output.
Source Files