Skip to content

RAGFlow is an agent-native database: a Retrieval-Augmented Generation (RAG) engine that grew into context infrastructure.

18.2 KB
Raw Blame
//
//  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 models

import (
	"bytes"
	"context"
	"encoding/json"
	"fmt"
	"io"
	"net/http"
	"ragflow/internal/common"
	"strings"
)

// VolcEngine implements ModelDriver for VolcEngine
type VolcEngine struct {
	baseModel BaseModel
}

// NewVolcEngine creates a new VolcEngine model instance
func NewVolcEngine(baseURL map[string]string, urlSuffix URLSuffix) *VolcEngine {
	return &VolcEngine{
		baseModel: BaseModel{
			BaseURL:    baseURL,
			URLSuffix:  urlSuffix,
			httpClient: common.GetSSRFHTTPClient(),
		},
	}
}

func (v *VolcEngine) NewInstance(baseURL map[string]string) ModelDriver {
	return NewVolcEngine(baseURL, v.baseModel.URLSuffix)
}

func (v *VolcEngine) Name() string {
	return "volcengine"
}

// getAPIKey extracts the actual API key from VolcEngine's stored format.
// VolcEngine stores the api_key as a JSON string like:
//
//	{"ark_api_key": "...", "endpoint_id": "..."}
//
// Falls back to the raw key string if JSON parsing fails.
func (v *VolcEngine) getAPIKey(apiConfig *APIConfig) string {
	if apiConfig == nil || apiConfig.ApiKey == nil {
		return ""
	}
	key := strings.TrimSpace(*apiConfig.ApiKey)
	var volcConfig struct {
		ArkAPIKey string `json:"ark_api_key"`
	}
	if err := json.Unmarshal([]byte(key), &volcConfig); err == nil && volcConfig.ArkAPIKey != "" {
		return volcConfig.ArkAPIKey
	}
	return key
}

// effectiveAPIConfig returns an APIConfig with the extracted API key
// suitable for use with doRequest/doStreamRequest.
func (v *VolcEngine) effectiveAPIConfig(apiConfig *APIConfig) *APIConfig {
	if apiConfig == nil {
		return nil
	}
	key := v.getAPIKey(apiConfig)
	return &APIConfig{
		ApiKey:  &key,
		Region:  apiConfig.Region,
		BaseURL: apiConfig.BaseURL,
	}
}

// ChatWithMessages sends multiple messages with roles and returns response
func (v *VolcEngine) ChatWithMessages(ctx context.Context, modelName string, messages []Message, apiConfig *APIConfig, chatModelConfig *ChatConfig, modelUsage *common.ModelUsage) (*ChatResponse, error) {
	if err := v.baseModel.APIConfigCheck(apiConfig); err != nil {
		return nil, err
	}

	if len(messages) == 0 {
		return nil, fmt.Errorf("messages is empty")
	}

	url, err := v.baseModel.EndpointURL(apiConfig, v.baseModel.URLSuffix.Chat)
	if err != nil {
		return nil, err
	}

	// Build request body
	reqBody := buildRequestBody(chatModelConfig, modelName, messages, false)

	if chatModelConfig != nil {
		if chatModelConfig.Thinking != nil {
			if *chatModelConfig.Thinking {
				var thinkingFlag string
				effort := "medium"
				if chatModelConfig.Effort != nil {
					effort = *chatModelConfig.Effort
				}
				switch effort {
				case "none", "minimal":
					thinkingFlag = "disabled"
					reqBody["reasoning_effort"] = "minimal"
				case "low":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "low"
				case "medium":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "medium"
				case "auto", "default":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "medium"
				case "high":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "high"
				case "xhigh":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "xhigh"
				default:
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = effort
				}
				reqBody["thinking"] = map[string]interface{}{
					"type": thinkingFlag,
				}
			} else {
				reqBody["thinking"] = map[string]interface{}{
					"type": "disabled",
				}
			}
		}

	}

	effectiveConfig := v.effectiveAPIConfig(apiConfig)
	body, err := v.baseModel.doRequest(ctx, url, effectiveConfig, reqBody, nonStreamCallTimeout)
	if err != nil {
		return nil, err
	}

	return HandleNonStreamingResponse(ctx, body, modelUsage, chatModelConfig, OpenAIParserConfig)
}

// ChatStreamlyWithSender sends messages and streams response via sender function (best performance, no channel)
func (v *VolcEngine) ChatStreamlyWithSender(ctx context.Context, modelName string, messages []Message, apiConfig *APIConfig, modelConfig *ChatConfig, modelUsage *common.ModelUsage, sender func(*string, *string) error) error {
	if err := v.baseModel.APIConfigCheck(apiConfig); err != nil {
		return err
	}

	if len(messages) == 0 {
		return fmt.Errorf("messages is empty")
	}

	url, err := v.baseModel.EndpointURL(apiConfig, v.baseModel.URLSuffix.Chat)
	if err != nil {
		return err
	}

	// Build request body with streaming enabled
	reqBody := buildRequestBody(modelConfig, modelName, messages, true)
	reqBody["stream_options"] = map[string]interface{}{"include_usage": true}

	if modelConfig != nil {
		// TODO VolcEngine has `auto` mode
		if modelConfig.Thinking != nil {
			if *modelConfig.Thinking {
				var thinkingFlag string
				effort := "medium"
				if modelConfig.Effort != nil {
					effort = *modelConfig.Effort
				}
				switch effort {
				case "none", "minimal":
					thinkingFlag = "disabled"
					reqBody["reasoning_effort"] = "minimal"
					break
				case "low":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "low"
					break
				case "medium":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "medium"
					break
				case "auto", "default":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "medium"
					break
				case "high":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "high"
					break
				case "xhigh":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "xhigh"
					break
				case "max":
					thinkingFlag = "enabled"
					reqBody["reasoning_effort"] = "max"
					break
				default:
					return fmt.Errorf("invalid effort level")
				}
				reqBody["thinking"] = map[string]interface{}{
					"type": thinkingFlag,
				}
			} else {
				reqBody["thinking"] = map[string]interface{}{
					"type": "disabled",
				}
			}
		}

	}

	effectiveConfig := v.effectiveAPIConfig(apiConfig)
	return v.baseModel.doStreamRequest(ctx, url, effectiveConfig, reqBody, streamCallTimeout, func(body io.ReadCloser) error {
		return HandleStreamingResponse(body, modelUsage, modelConfig, OpenAIParserConfig, sender)
	})
}

type volcengineEmbeddingResponse struct {
	Created int64                   `json:"created"`
	Data    volcengineEmbeddingData `json:"data"`
	ID      string                  `json:"id"`
	Model   string                  `json:"model"`
	Object  string                  `json:"object"`
	Usage   volcengineUsage         `json:"usage"`
}

type volcengineEmbeddingData struct {
	Embedding []float64 `json:"embedding"`
	Object    string    `json:"object"`
}

type volcengineUsage struct {
	PromptTokens        int                            `json:"prompt_tokens"`
	TotalTokens         int                            `json:"total_tokens"`
	PromptTokensDetails *volcenginePromptTokensDetails `json:"prompt_tokens_details,omitempty"`
}

type volcenginePromptTokensDetails struct {
	ImageTokens int `json:"image_tokens"`
	TextTokens  int `json:"text_tokens"`
}

// Embed embeds a list of texts into embeddings
func (v *VolcEngine) Embed(ctx context.Context, modelName *string, request EmbedRequest, apiConfig *APIConfig, embeddingConfig *EmbeddingConfig, modelUsage *common.ModelUsage) ([]EmbeddingData, error) {
	if err := v.baseModel.APIConfigCheck(apiConfig); err != nil {
		return nil, err
	}

	if len(request.Texts) == 0 {
		return []EmbeddingData{}, nil
	}
	if modelName == nil || *modelName == "" {
		return nil, fmt.Errorf("model name is required")
	}

	url, err := v.baseModel.EndpointURL(apiConfig, v.baseModel.URLSuffix.Embedding)
	if err != nil {
		return nil, err
	}

	var embeddings []EmbeddingData
	var totalUsage TokenUsage
	var requestID string

	for i, text := range request.Texts {

		reqBody := map[string]interface{}{
			"model":           *modelName,
			"encoding_format": "float",
			"input": []map[string]interface{}{
				{
					"type": "text",
					"text": text,
				},
			},
		}
		if embeddingConfig != nil && embeddingConfig.Dimension > 0 {
			reqBody["dimensions"] = embeddingConfig.Dimension
		}

		jsonData, err := json.Marshal(reqBody)
		if err != nil {
			return nil, fmt.Errorf(
				"failed to marshal request: %w",
				err,
			)
		}
		// Run each per-text request in its own scope so the context's
		// deadline is cancelled at the end of every iteration instead of
		// piling up deferred cancels until the whole batch finishes.
		parsed, err := func() (volcengineEmbeddingResponse, error) {
			var parsed volcengineEmbeddingResponse

			ctx, cancel := context.WithTimeout(ctx, nonStreamCallTimeout)
			defer cancel()

			req, err := http.NewRequestWithContext(ctx, "POST", url, bytes.NewBuffer(jsonData))
			if err != nil {
				return parsed, fmt.Errorf("failed to create request: %w", err)
			}

			req.Header.Set("Content-Type", "application/json")
			req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", v.getAPIKey(apiConfig)))

			resp, err := v.baseModel.httpClient.Do(req)
			if err != nil {
				return parsed, fmt.Errorf("failed to send request: %w", err)
			}
			defer resp.Body.Close()

			body, err := io.ReadAll(resp.Body)
			if err != nil {
				return parsed, fmt.Errorf("failed to read response: %w", err)
			}

			if resp.StatusCode != http.StatusOK {
				return parsed, fmt.Errorf("API request failed with status %d: %s", resp.StatusCode, string(body))
			}

			if err = json.Unmarshal(body, &parsed); err != nil {
				return parsed, fmt.Errorf("failed to parse response: %w", err)
			}
			return parsed, nil
		}()
		if err != nil {
			return nil, err
		}
		requestID = parsed.ID
		totalUsage.PromptTokens += parsed.Usage.PromptTokens
		totalUsage.TotalTokens += parsed.Usage.TotalTokens

		var embeddingData EmbeddingData
		embeddingData.Index = i
		embeddingData.Embedding = parsed.Data.Embedding
		embeddings = append(embeddings, embeddingData)
	}
	recordResponseUsage(modelUsage, requestID, &totalUsage, "embedding")

	return embeddings, nil
}

// Rerank calculates similarity scores between query and documents
func (v *VolcEngine) Rerank(ctx context.Context, modelName *string, request RerankRequest, apiConfig *APIConfig, rerankConfig *RerankConfig, modelUsage *common.ModelUsage) (*RerankResponse, error) {
	return nil, fmt.Errorf("%s, Rerank not implemented", v.Name())
}

// TranscribeAudio transcribe audio
func (v *VolcEngine) TranscribeAudio(ctx context.Context, modelName *string, file *string, apiConfig *APIConfig, asrConfig *ASRConfig, modelUsage *common.ModelUsage) (*ASRResponse, error) {
	return nil, fmt.Errorf("%s: no such method: TranscribeAudio", v.Name())
}

func (v *VolcEngine) TranscribeAudioWithSender(ctx context.Context, modelName *string, file *string, apiConfig *APIConfig, asrConfig *ASRConfig, modelUsage *common.ModelUsage, sender func(*string, *string) error) error {
	return fmt.Errorf("%s: no such method: TranscribeAudioWithSender", v.Name())
}

// AudioSpeech convert text to audio
func (v *VolcEngine) AudioSpeech(ctx context.Context, modelName *string, audioContent *string, apiConfig *APIConfig, ttsConfig *TTSConfig, modelUsage *common.ModelUsage) (*TTSResponse, error) {
	return nil, fmt.Errorf("%s: no such method: AudioSpeech", v.Name())
}

func (v *VolcEngine) AudioSpeechWithSender(ctx context.Context, modelName *string, audioContent *string, apiConfig *APIConfig, ttsConfig *TTSConfig, modelUsage *common.ModelUsage, sender func(*string, *string) error) error {
	return fmt.Errorf("%s: no such method: AudioSpeechWithSender", v.Name())
}

// OCRFile OCR file
func (v *VolcEngine) OCRFile(ctx context.Context, modelName *string, content []byte, url *string, apiConfig *APIConfig, ocrConfig *OCRConfig, modelUsage *common.ModelUsage) (*OCRFileResponse, error) {
	return nil, fmt.Errorf("%s: no such method: OCRFile", v.Name())
}

// ParseFile parse file
func (v *VolcEngine) ParseFile(ctx context.Context, modelName *string, content []byte, url *string, apiConfig *APIConfig, parseFileConfig *ParseFileConfig, modelUsage *common.ModelUsage) (*ParseFileResponse, error) {
	return nil, fmt.Errorf("%s: no such method: ParseFile", v.Name())
}

// volcengineModelData is a VolcEngine-specific model list item with
// extra fields for model type inference.
type volcengineModelData struct {
	ID         string   `json:"id"`
	Object     string   `json:"object"`
	OwnedBy    string   `json:"owned_by"`
	Status     string   `json:"status"`
	Domain     string   `json:"domain"`
	TaskType   []string `json:"task_type"`
	Modalities struct {
		InputModalities  []string `json:"input_modalities"`
		OutputModalities []string `json:"output_modalities"`
	} `json:"modalities"`
}

// volcengineModelList is the VolcEngine-specific model list response.
type volcengineModelList struct {
	Object string                `json:"object"`
	Models []volcengineModelData `json:"data"`
}

// inferVolcengineModelTypes infers model types from a VolcEngine model's
// domain, task_type, and modalities fields.
func inferVolcengineModelTypes(m volcengineModelData) []string {
	var types []string
	outputModalities := m.Modalities.OutputModalities
	inputModalities := m.Modalities.InputModalities

	if m.Domain == "Embedding" {
		types = append(types, "embedding")
		return types
	}

	taskSet := make(map[string]bool, len(m.TaskType))
	for _, t := range m.TaskType {
		taskSet[t] = true
	}
	if taskSet["TextEmbedding"] || taskSet["ImageEmbedding"] {
		types = append(types, "embedding")
		return types
	}

	for _, out := range outputModalities {
		switch out {
		case "text":
			types = append(types, "chat")
		case "embeddings":
			types = append(types, "embedding")
		case "audio":
			types = append(types, "tts")
		}
	}
	for _, in := range inputModalities {
		switch in {
		case "image":
			for _, out := range outputModalities {
				if out == "text" {
					types = append(types, "image2text")
				}
			}
		case "audio":
			for _, out := range outputModalities {
				if out == "text" {
					types = append(types, "speech2text")
				}
			}
		}
	}

	if len(types) == 0 {
		types = append(types, "chat")
	}
	return types
}

func (v *VolcEngine) ListModels(ctx context.Context, apiConfig *APIConfig) ([]ListModelResponse, error) {
	if err := v.baseModel.APIConfigCheck(apiConfig); err != nil {
		return nil, err
	}

	modelsSuffix := strings.Trim(strings.TrimSpace(v.baseModel.URLSuffix.Models), "/")
	if modelsSuffix == "" {
		return nil, fmt.Errorf("volcengine: models URL suffix is not configured")
	}
	url, err := v.baseModel.EndpointURL(apiConfig, modelsSuffix)
	if err != nil {
		return nil, err
	}

	ctx, cancel := context.WithTimeout(ctx, nonStreamCallTimeout)
	defer cancel()

	req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
	if err != nil {
		return nil, fmt.Errorf("failed to create request: %w", err)
	}

	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", v.getAPIKey(apiConfig)))

	resp, err := v.baseModel.httpClient.Do(req)
	if err != nil {
		return nil, fmt.Errorf("failed to send request: %w", err)
	}
	defer resp.Body.Close()

	body, err := io.ReadAll(resp.Body)
	if err != nil {
		return nil, fmt.Errorf("failed to read response: %w", err)
	}

	if resp.StatusCode != http.StatusOK {
		return nil, fmt.Errorf("VolcEngine models API error: %s, body: %s", resp.Status, string(body))
	}

	// Parse VolcEngine-specific response with extra fields
	var volcList volcengineModelList
	if err = json.Unmarshal(body, &volcList); err != nil {
		return nil, fmt.Errorf("failed to parse response: %w", err)
	}

	// Convert to standard ModelList for enrichment from static config
	var standardModels []ModelListItem
	volcModelInfo := make(map[string]volcengineModelData)
	for _, m := range volcList.Models {
		if m.Status == "Shutdown" {
			continue
		}
		volcModelInfo[m.ID] = m
		standardModels = append(standardModels, ModelListItem{
			ID:      m.ID,
			Object:  m.Object,
			OwnedBy: m.OwnedBy,
		})
	}

	if len(standardModels) == 0 {
		return []ListModelResponse{}, nil
	}

	enrichedModels := ParseListModel(ModelList{
		Object: volcList.Object,
		Models: standardModels,
	})

	// Override model_types with VolcEngine-specific inference
	for i := range enrichedModels {
		if vm, ok := volcModelInfo[enrichedModels[i].Name]; ok {
			enrichedModels[i].ModelTypes = inferVolcengineModelTypes(vm)
		}
	}

	return enrichedModels, nil
}

func (v *VolcEngine) Balance(ctx context.Context, apiConfig *APIConfig) (map[string]interface{}, error) {
	return nil, fmt.Errorf("%s: no such method: Balance", v.Name())
}

func (v *VolcEngine) CheckConnection(ctx context.Context, apiConfig *APIConfig) error {
	if err := v.baseModel.APIConfigCheck(apiConfig); err != nil {
		return err
	}

	url, err := v.baseModel.EndpointURL(apiConfig, v.baseModel.URLSuffix.Files)
	if err != nil {
		return err
	}

	ctx, cancel := context.WithTimeout(ctx, nonStreamCallTimeout)
	defer cancel()

	req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
	if err != nil {
		return fmt.Errorf("failed to create request: %w", err)
	}

	req.Header.Set("Content-Type", "application/json")
	req.Header.Set("Authorization", fmt.Sprintf("Bearer %s", v.getAPIKey(apiConfig)))

	resp, err := v.baseModel.httpClient.Do(req)
	if err != nil {
		return fmt.Errorf("failed to send request: %w", err)
	}
	defer resp.Body.Close()

	body, err := io.ReadAll(resp.Body)
	if err != nil {
		return fmt.Errorf("failed to read response: %w", err)
	}

	if resp.StatusCode != http.StatusOK {
		return fmt.Errorf("API request failed with status %d: %s", resp.StatusCode, string(body))
	}

	return nil
}

func (v *VolcEngine) ListTasks(ctx context.Context, apiConfig *APIConfig) ([]ListTaskStatus, error) {
	return nil, fmt.Errorf("%s: no such method: ListTasks", v.Name())
}

func (v *VolcEngine) ShowTask(ctx context.Context, taskID string, apiConfig *APIConfig) (*TaskResponse, error) {
	return nil, fmt.Errorf("%s: no such method: ShowTask", v.Name())
}

func (v *VolcEngine) Decide(ctx context.Context, modelName *string, request *DecideRequest, apiConfig *APIConfig, modelUsage *common.ModelUsage) (*DecideResponse, error) {
	return nil, fmt.Errorf("%s: no such method: Decide", v.Name())
}