Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func NewEntityExtractor ¶
func NewEntityExtractor(extractor retrieval.EntityExtractor, logger logging.Logger) *entityExtractor
NewEntityExtractor creates a new entity extraction step with logger.
Types ¶
type ChunkStep ¶
type ChunkStep struct {
// contains filtered or unexported fields
}
ChunkStep 语义分块步骤（用于索引阶段）
func NewChunkStep ¶
func NewChunkStep(chunker dataprep.SemanticChunker) *ChunkStep
NewChunkStep 创建分块步骤
type EmbedStep ¶
type EmbedStep struct {
// contains filtered or unexported fields
}
EmbedStep 向量化步骤
func NewEmbedStep ¶
func NewEmbedStep(embedder embedding.Provider, metrics abstraction.Metrics) *EmbedStep
NewEmbedStep 创建向量化步骤（支持指标收集）
type FileDiscoveryStep ¶
type FileDiscoveryStep struct{}
FileDiscoveryStep 文件发现与验证步骤
func NewFileDiscoveryStep ¶
func NewFileDiscoveryStep() *FileDiscoveryStep
NewFileDiscoveryStep 创建文件发现步骤
type MultimodalEmbeddingStep ¶
type MultimodalEmbeddingStep struct {
// contains filtered or unexported fields
}
MultimodalEmbeddingStep encodes the query text (and optionally an image) using a MultimodalEmbedder and writes the resulting vectors into the pipeline state so that subsequent VectorSearchStep and ImageSearchStep can use them independently.
Vectors are stored in AgenticMetadata.Custom:
- "query_vector" []float32 – text embedding of state.Query.Text
- "image_vector" []float32 – image embedding (only when image_data is present)
To pass an image for cross-modal search, callers must set:
state.Agentic.Custom["image_data"] = []byte{...}
before executing the pipeline.
func NewMultimodalEmbeddingStep ¶
func NewMultimodalEmbeddingStep(embedder abstraction.MultimodalEmbedder, logger logging.Logger) *MultimodalEmbeddingStep
NewMultimodalEmbeddingStep creates a new multimodal embedding step.
func (*MultimodalEmbeddingStep) Execute ¶
func (s *MultimodalEmbeddingStep) Execute(ctx context.Context, state *entity.PipelineState) error
Execute encodes the query text and, if present, the image data.
func (*MultimodalEmbeddingStep) Name ¶
func (s *MultimodalEmbeddingStep) Name() string
type ParseStep ¶
type ParseStep struct {
// contains filtered or unexported fields
}
ParseStep 流式解析步骤（支持多解析器）
func NewParseStep ¶
NewParseStep 创建解析步骤（支持多个解析器）
type StoreStep ¶
type StoreStep struct {
// contains filtered or unexported fields
}
StoreStep 存储步骤
func NewStoreStep ¶
func NewStoreStep(vectorStore abstraction.VectorStore, metrics abstraction.Metrics) *StoreStep
NewStoreStep 创建存储步骤（支持指标收集）
Source Files
¶
Directories
¶
| Path | Synopsis |
|---|---|
|
Package agentic provides agentic orchestration steps for autonomous RAG workflows.
|
Package agentic provides agentic orchestration steps for autonomous RAG workflows. |
|
Package post_retrieval provides steps that process and optimize retrieval results.
|
Package post_retrieval provides steps that process and optimize retrieval results. |
|
Package pre_retrieval provides query enhancement steps that occur before retrieval.
|
Package pre_retrieval provides query enhancement steps that occur before retrieval. |
|
Package retrieval provides retrieval strategy steps that execute different search algorithms.
|
Package retrieval provides retrieval strategy steps that execute different search algorithms. |