mirror of
https://github.com/infiniflow/ragflow.git
synced 2026-08-05 15:20:30 +08:00
4e78f1f440
Port Python rag/advanced_rag agentic search to Go: ES-backed dataset-nav service, agentic-search harness, and agent tools. Includes agentic-search port plan and self-review docs.
81 lines
2.6 KiB
Go
81 lines
2.6 KiB
Go
//
|
|
// Copyright 2026 The InfiniFlow Authors. All Rights Reserved.
|
|
//
|
|
// Licensed under the Apache License, Version 2.0 (the "License");
|
|
// you may not use this file except in compliance with the License.
|
|
// You may obtain a copy of the License at
|
|
//
|
|
// http://www.apache.org/licenses/LICENSE-2.0
|
|
//
|
|
// Unless required by applicable law or agreed to in writing, software
|
|
// distributed under the License is distributed on an "AS IS" BASIS,
|
|
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
// See the License for the specific language governing permissions and
|
|
// limitations under the License.
|
|
//
|
|
|
|
package service
|
|
|
|
import (
|
|
"context"
|
|
"fmt"
|
|
"strings"
|
|
)
|
|
|
|
// NavEmbedder is the production implementation of nlp.NavEmbedder. It resolves
|
|
// the tenant's embedding model on each call and returns float32 vectors (the
|
|
// dataset-nav index stores q_<dim>_vec as float). It lives in the service
|
|
// package (not nlp) so it can import model_service without an import cycle.
|
|
type NavEmbedder struct {
|
|
modelSvc *ModelProviderService
|
|
// embdModelName is the composite embedding model name (e.g.
|
|
// "embedding_model@..." ). Empty falls back to resolving the tenant default.
|
|
embdModelName string
|
|
}
|
|
|
|
// NewNavEmbedder builds the production embedder used by NavService.
|
|
func NewNavEmbedder(modelSvc *ModelProviderService, embdModelName string) *NavEmbedder {
|
|
return &NavEmbedder{modelSvc: modelSvc, embdModelName: embdModelName}
|
|
}
|
|
|
|
// Encode embeds texts for the tenant and returns float32 vectors.
|
|
func (e *NavEmbedder) Encode(ctx context.Context, tenantID string, texts []string) ([][]float32, error) {
|
|
if e.modelSvc == nil {
|
|
return nil, fmt.Errorf("datasetnav: embedding model service not initialized")
|
|
}
|
|
name := e.embdModelName
|
|
if name == "" {
|
|
name = tenantID // composite name falls back to tenant default resolution
|
|
}
|
|
model, err := e.modelSvc.GetEmbeddingModel(ctx, tenantID, name)
|
|
if err != nil {
|
|
return nil, fmt.Errorf("datasetnav: resolve embedding model for tenant %s: %w", tenantID, err)
|
|
}
|
|
nonEmpty := make([]string, 0, len(texts))
|
|
for _, t := range texts {
|
|
if strings.TrimSpace(t) != "" {
|
|
nonEmpty = append(nonEmpty, t)
|
|
}
|
|
}
|
|
if len(nonEmpty) == 0 {
|
|
return nil, nil
|
|
}
|
|
embeds, err := model.ModelDriver.Embed(ctx, model.ModelName, nonEmpty, model.APIConfig, nil, nil)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
out := make([][]float32, 0, len(embeds))
|
|
for _, e := range embeds {
|
|
out = append(out, toF32(e.Embedding))
|
|
}
|
|
return out, nil
|
|
}
|
|
|
|
func toF32(v []float64) []float32 {
|
|
out := make([]float32, len(v))
|
|
for i, x := range v {
|
|
out[i] = float32(x)
|
|
}
|
|
return out
|
|
}
|