添加chat gemini、chatfire端点、 图片生成 gemini、chatfire 更轻松的AI配置
This commit is contained in:
Connor
277
pkg/image/gemini_image_client.go
Normal file
277
pkg/image/gemini_image_client.go
Normal file
@@ -0,0 +1,277 @@
|
||||
package image
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"encoding/base64"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"io"
|
||||
"net/http"
|
||||
"strings"
|
||||
"time"
|
||||
)
|
||||
|
||||
type GeminiImageClient struct {
|
||||
BaseURL string
|
||||
APIKey string
|
||||
Model string
|
||||
Endpoint string
|
||||
HTTPClient *http.Client
|
||||
}
|
||||
|
||||
type GeminiImageRequest struct {
|
||||
Contents []struct {
|
||||
Parts []GeminiPart `json:"parts"`
|
||||
} `json:"contents"`
|
||||
GenerationConfig struct {
|
||||
ResponseModalities []string `json:"responseModalities"`
|
||||
} `json:"generationConfig"`
|
||||
}
|
||||
|
||||
type GeminiPart struct {
|
||||
Text string `json:"text,omitempty"`
|
||||
InlineData *GeminiInlineData `json:"inlineData,omitempty"`
|
||||
}
|
||||
|
||||
type GeminiInlineData struct {
|
||||
MimeType string `json:"mimeType"`
|
||||
Data string `json:"data"` // base64 编码的图片数据
|
||||
}
|
||||
|
||||
type GeminiImageResponse struct {
|
||||
Candidates []struct {
|
||||
Content struct {
|
||||
Parts []struct {
|
||||
InlineData struct {
|
||||
MimeType string `json:"mimeType"`
|
||||
Data string `json:"data"`
|
||||
} `json:"inlineData,omitempty"`
|
||||
Text string `json:"text,omitempty"`
|
||||
} `json:"parts"`
|
||||
} `json:"content"`
|
||||
} `json:"candidates"`
|
||||
UsageMetadata struct {
|
||||
PromptTokenCount int `json:"promptTokenCount"`
|
||||
CandidatesTokenCount int `json:"candidatesTokenCount"`
|
||||
TotalTokenCount int `json:"totalTokenCount"`
|
||||
} `json:"usageMetadata"`
|
||||
}
|
||||
|
||||
// downloadImageToBase64 下载图片 URL 并转换为 base64
|
||||
func downloadImageToBase64(imageURL string) (string, string, error) {
|
||||
resp, err := http.Get(imageURL)
|
||||
if err != nil {
|
||||
return "", "", fmt.Errorf("download image: %w", err)
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
if resp.StatusCode != http.StatusOK {
|
||||
return "", "", fmt.Errorf("download image failed with status: %d", resp.StatusCode)
|
||||
}
|
||||
|
||||
imageData, err := io.ReadAll(resp.Body)
|
||||
if err != nil {
|
||||
return "", "", fmt.Errorf("read image data: %w", err)
|
||||
}
|
||||
|
||||
// 根据 Content-Type 确定 mimeType
|
||||
mimeType := resp.Header.Get("Content-Type")
|
||||
if mimeType == "" {
|
||||
mimeType = "image/jpeg"
|
||||
}
|
||||
|
||||
base64Data := base64.StdEncoding.EncodeToString(imageData)
|
||||
return base64Data, mimeType, nil
|
||||
}
|
||||
|
||||
func NewGeminiImageClient(baseURL, apiKey, model, endpoint string) *GeminiImageClient {
|
||||
if baseURL == "" {
|
||||
baseURL = "https://generativelanguage.googleapis.com"
|
||||
}
|
||||
if endpoint == "" {
|
||||
endpoint = "/v1beta/models/{model}:generateContent"
|
||||
}
|
||||
if model == "" {
|
||||
model = "gemini-3-pro-image-preview"
|
||||
}
|
||||
return &GeminiImageClient{
|
||||
BaseURL: baseURL,
|
||||
APIKey: apiKey,
|
||||
Model: model,
|
||||
Endpoint: endpoint,
|
||||
HTTPClient: &http.Client{
|
||||
Timeout: 10 * time.Minute,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
func (c *GeminiImageClient) GenerateImage(prompt string, opts ...ImageOption) (*ImageResult, error) {
|
||||
options := &ImageOptions{
|
||||
Size: "1024x1024",
|
||||
Quality: "standard",
|
||||
}
|
||||
|
||||
for _, opt := range opts {
|
||||
opt(options)
|
||||
}
|
||||
|
||||
model := c.Model
|
||||
if options.Model != "" {
|
||||
model = options.Model
|
||||
}
|
||||
|
||||
promptText := prompt
|
||||
if options.NegativePrompt != "" {
|
||||
promptText += fmt.Sprintf("\n\nNegative prompt: %s", options.NegativePrompt)
|
||||
}
|
||||
if options.Size != "" {
|
||||
promptText += fmt.Sprintf("\n\nImage size: %s", options.Size)
|
||||
}
|
||||
|
||||
// 构建请求的 parts，支持参考图
|
||||
parts := []GeminiPart{}
|
||||
|
||||
// 如果有参考图，先添加参考图
|
||||
if len(options.ReferenceImages) > 0 {
|
||||
for _, refImg := range options.ReferenceImages {
|
||||
var base64Data string
|
||||
var mimeType string
|
||||
var err error
|
||||
|
||||
// 检查是否是 HTTP/HTTPS URL
|
||||
if strings.HasPrefix(refImg, "http://") || strings.HasPrefix(refImg, "https://") {
|
||||
// 下载图片并转换为 base64
|
||||
base64Data, mimeType, err = downloadImageToBase64(refImg)
|
||||
if err != nil {
|
||||
continue
|
||||
}
|
||||
} else if strings.HasPrefix(refImg, "data:") {
|
||||
// 如果是 data URI 格式，需要解析
|
||||
// 格式: data:image/jpeg;base64,xxxxx
|
||||
mimeType = "image/jpeg"
|
||||
parts := []byte(refImg)
|
||||
for i := 0; i < len(parts); i++ {
|
||||
if parts[i] == ',' {
|
||||
base64Data = refImg[i+1:]
|
||||
// 提取 mime type
|
||||
if i > 11 {
|
||||
mimeTypeEnd := i
|
||||
for j := 5; j < i; j++ {
|
||||
if parts[j] == ';' {
|
||||
mimeTypeEnd = j
|
||||
break
|
||||
}
|
||||
}
|
||||
mimeType = refImg[5:mimeTypeEnd]
|
||||
}
|
||||
break
|
||||
}
|
||||
}
|
||||
} else {
|
||||
// 假设已经是 base64 编码
|
||||
base64Data = refImg
|
||||
mimeType = "image/jpeg"
|
||||
}
|
||||
|
||||
if base64Data != "" {
|
||||
parts = append(parts, GeminiPart{
|
||||
InlineData: &GeminiInlineData{
|
||||
MimeType: mimeType,
|
||||
Data: base64Data,
|
||||
},
|
||||
})
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// 添加文本提示词
|
||||
parts = append(parts, GeminiPart{
|
||||
Text: promptText,
|
||||
})
|
||||
|
||||
reqBody := GeminiImageRequest{
|
||||
Contents: []struct {
|
||||
Parts []GeminiPart `json:"parts"`
|
||||
}{
|
||||
{
|
||||
Parts: parts,
|
||||
},
|
||||
},
|
||||
GenerationConfig: struct {
|
||||
ResponseModalities []string `json:"responseModalities"`
|
||||
}{
|
||||
ResponseModalities: []string{"IMAGE"},
|
||||
},
|
||||
}
|
||||
|
||||
jsonData, err := json.Marshal(reqBody)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("marshal request: %w", err)
|
||||
}
|
||||
|
||||
endpoint := c.BaseURL + c.Endpoint
|
||||
endpoint = replaceModelPlaceholder(endpoint, model)
|
||||
url := fmt.Sprintf("%s?key=%s", endpoint, c.APIKey)
|
||||
|
||||
req, err := http.NewRequest("POST", url, bytes.NewBuffer(jsonData))
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("create request: %w", err)
|
||||
}
|
||||
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
|
||||
resp, err := c.HTTPClient.Do(req)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("send request: %w", err)
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
body, err := io.ReadAll(resp.Body)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("read response: %w", err)
|
||||
}
|
||||
|
||||
if resp.StatusCode != http.StatusOK {
|
||||
bodyStr := string(body)
|
||||
if len(bodyStr) > 1000 {
|
||||
bodyStr = fmt.Sprintf("%s ... %s", bodyStr[:500], bodyStr[len(bodyStr)-500:])
|
||||
}
|
||||
return nil, fmt.Errorf("API error (status %d): %s", resp.StatusCode, bodyStr)
|
||||
}
|
||||
|
||||
var result GeminiImageResponse
|
||||
if err := json.Unmarshal(body, &result); err != nil {
|
||||
return nil, fmt.Errorf("parse response: %w", err)
|
||||
}
|
||||
|
||||
if len(result.Candidates) == 0 || len(result.Candidates[0].Content.Parts) == 0 {
|
||||
return nil, fmt.Errorf("no image generated in response")
|
||||
}
|
||||
|
||||
base64Data := result.Candidates[0].Content.Parts[0].InlineData.Data
|
||||
if base64Data == "" {
|
||||
return nil, fmt.Errorf("no base64 image data in response")
|
||||
}
|
||||
|
||||
dataURI := fmt.Sprintf("data:image/jpeg;base64,%s", base64Data)
|
||||
|
||||
return &ImageResult{
|
||||
Status: "completed",
|
||||
ImageURL: dataURI,
|
||||
Completed: true,
|
||||
Width: 1024,
|
||||
Height: 1024,
|
||||
}, nil
|
||||
}
|
||||
|
||||
func (c *GeminiImageClient) GetTaskStatus(taskID string) (*ImageResult, error) {
|
||||
return nil, fmt.Errorf("not supported for Gemini (synchronous generation)")
|
||||
}
|
||||
|
||||
func replaceModelPlaceholder(endpoint, model string) string {
|
||||
result := endpoint
|
||||
if bytes.Contains([]byte(result), []byte("{model}")) {
|
||||
result = string(bytes.ReplaceAll([]byte(result), []byte("{model}"), []byte(model)))
|
||||
}
|
||||
return result
|
||||
}
|
||||
@@ -1,14 +1,5 @@
|
||||
package image
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"io"
|
||||
"net/http"
|
||||
"time"
|
||||
)
|
||||
|
||||
type ImageClient interface {
|
||||
GenerateImage(prompt string, opts ...ImageOption) (*ImageResult, error)
|
||||
GetTaskStatus(taskID string) (*ImageResult, error)
|
||||
@@ -100,285 +91,3 @@ func WithReferenceImages(images []string) ImageOption {
|
||||
o.ReferenceImages = images
|
||||
}
|
||||
}
|
||||
|
||||
type OpenAIImageClient struct {
|
||||
BaseURL string
|
||||
APIKey string
|
||||
Model string
|
||||
HTTPClient *http.Client
|
||||
}
|
||||
|
||||
type DALLERequest struct {
|
||||
Model string `json:"model"`
|
||||
Prompt string `json:"prompt"`
|
||||
Size string `json:"size,omitempty"`
|
||||
Quality string `json:"quality,omitempty"`
|
||||
N int `json:"n"`
|
||||
Image []string `json:"image,omitempty"` // 参考图片URL列表
|
||||
}
|
||||
|
||||
type DALLEResponse struct {
|
||||
Created int64 `json:"created"`
|
||||
Data []struct {
|
||||
URL string `json:"url"`
|
||||
RevisedPrompt string `json:"revised_prompt,omitempty"`
|
||||
} `json:"data"`
|
||||
}
|
||||
|
||||
func NewOpenAIImageClient(baseURL, apiKey, model string) *OpenAIImageClient {
|
||||
return &OpenAIImageClient{
|
||||
BaseURL: baseURL,
|
||||
APIKey: apiKey,
|
||||
Model: model,
|
||||
HTTPClient: &http.Client{
|
||||
Timeout: 10 * time.Minute,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
func (c *OpenAIImageClient) GenerateImage(prompt string, opts ...ImageOption) (*ImageResult, error) {
|
||||
options := &ImageOptions{
|
||||
Size: "1920x1920",
|
||||
Quality: "standard",
|
||||
}
|
||||
|
||||
for _, opt := range opts {
|
||||
opt(options)
|
||||
}
|
||||
|
||||
model := c.Model
|
||||
if options.Model != "" {
|
||||
model = options.Model
|
||||
}
|
||||
|
||||
reqBody := DALLERequest{
|
||||
Model: model,
|
||||
Prompt: prompt,
|
||||
Size: options.Size,
|
||||
Quality: options.Quality,
|
||||
N: 1,
|
||||
Image: options.ReferenceImages,
|
||||
}
|
||||
|
||||
jsonData, err := json.Marshal(reqBody)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("marshal request: %w", err)
|
||||
}
|
||||
|
||||
endpoint := c.BaseURL + "/v1/images/generations"
|
||||
req, err := http.NewRequest("POST", endpoint, bytes.NewBuffer(jsonData))
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("create request: %w", err)
|
||||
}
|
||||
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
req.Header.Set("Authorization", "Bearer "+c.APIKey)
|
||||
|
||||
resp, err := c.HTTPClient.Do(req)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("send request: %w", err)
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
body, err := io.ReadAll(resp.Body)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("read response: %w", err)
|
||||
}
|
||||
|
||||
if resp.StatusCode != http.StatusOK {
|
||||
return nil, fmt.Errorf("API error (status %d): %s", resp.StatusCode, string(body))
|
||||
}
|
||||
|
||||
// 打印原始响应以便调试
|
||||
fmt.Printf("OpenAI API Response: %s\n", string(body))
|
||||
|
||||
var result DALLEResponse
|
||||
if err := json.Unmarshal(body, &result); err != nil {
|
||||
return nil, fmt.Errorf("parse response: %w, body: %s", err, string(body))
|
||||
}
|
||||
|
||||
if len(result.Data) == 0 {
|
||||
return nil, fmt.Errorf("no image generated, response: %s", string(body))
|
||||
}
|
||||
|
||||
return &ImageResult{
|
||||
Status: "completed",
|
||||
ImageURL: result.Data[0].URL,
|
||||
Completed: true,
|
||||
}, nil
|
||||
}
|
||||
|
||||
func (c *OpenAIImageClient) GetTaskStatus(taskID string) (*ImageResult, error) {
|
||||
return nil, fmt.Errorf("not supported for OpenAI/DALL-E")
|
||||
}
|
||||
|
||||
type StableDiffusionClient struct {
|
||||
BaseURL string
|
||||
APIKey string
|
||||
Model string
|
||||
HTTPClient *http.Client
|
||||
}
|
||||
|
||||
type SDRequest struct {
|
||||
Prompt string `json:"prompt"`
|
||||
NegativePrompt string `json:"negative_prompt,omitempty"`
|
||||
Model string `json:"model,omitempty"`
|
||||
Width int `json:"width,omitempty"`
|
||||
Height int `json:"height,omitempty"`
|
||||
Steps int `json:"steps,omitempty"`
|
||||
CfgScale float64 `json:"cfg_scale,omitempty"`
|
||||
Seed int64 `json:"seed,omitempty"`
|
||||
Samples int `json:"samples"`
|
||||
Image []string `json:"image,omitempty"` // 参考图片URL列表
|
||||
}
|
||||
|
||||
type SDResponse struct {
|
||||
Status string `json:"status"`
|
||||
TaskID string `json:"task_id,omitempty"`
|
||||
Output []struct {
|
||||
URL string `json:"url"`
|
||||
} `json:"output,omitempty"`
|
||||
Error string `json:"error,omitempty"`
|
||||
}
|
||||
|
||||
func NewStableDiffusionClient(baseURL, apiKey, model string) *StableDiffusionClient {
|
||||
return &StableDiffusionClient{
|
||||
BaseURL: baseURL,
|
||||
APIKey: apiKey,
|
||||
Model: model,
|
||||
HTTPClient: &http.Client{
|
||||
Timeout: 10 * time.Minute,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
func (c *StableDiffusionClient) GenerateImage(prompt string, opts ...ImageOption) (*ImageResult, error) {
|
||||
options := &ImageOptions{
|
||||
Width: 1024,
|
||||
Height: 1024,
|
||||
Steps: 30,
|
||||
CfgScale: 7.5,
|
||||
}
|
||||
|
||||
for _, opt := range opts {
|
||||
opt(options)
|
||||
}
|
||||
|
||||
model := c.Model
|
||||
if options.Model != "" {
|
||||
model = options.Model
|
||||
}
|
||||
|
||||
reqBody := SDRequest{
|
||||
Prompt: prompt,
|
||||
NegativePrompt: options.NegativePrompt,
|
||||
Model: model,
|
||||
Width: options.Width,
|
||||
Height: options.Height,
|
||||
Steps: options.Steps,
|
||||
CfgScale: options.CfgScale,
|
||||
Seed: options.Seed,
|
||||
Samples: 1,
|
||||
Image: options.ReferenceImages,
|
||||
}
|
||||
|
||||
jsonData, err := json.Marshal(reqBody)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("marshal request: %w", err)
|
||||
}
|
||||
|
||||
endpoint := c.BaseURL + "/v1/images/generations"
|
||||
req, err := http.NewRequest("POST", endpoint, bytes.NewBuffer(jsonData))
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("create request: %w", err)
|
||||
}
|
||||
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
req.Header.Set("Authorization", "Bearer "+c.APIKey)
|
||||
|
||||
resp, err := c.HTTPClient.Do(req)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("send request: %w", err)
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
body, err := io.ReadAll(resp.Body)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("read response: %w", err)
|
||||
}
|
||||
|
||||
if resp.StatusCode != http.StatusOK {
|
||||
return nil, fmt.Errorf("API error (status %d): %s", resp.StatusCode, string(body))
|
||||
}
|
||||
|
||||
var result SDResponse
|
||||
if err := json.Unmarshal(body, &result); err != nil {
|
||||
return nil, fmt.Errorf("parse response: %w", err)
|
||||
}
|
||||
|
||||
if result.Error != "" {
|
||||
return nil, fmt.Errorf("SD error: %s", result.Error)
|
||||
}
|
||||
|
||||
if result.Status == "processing" {
|
||||
return &ImageResult{
|
||||
TaskID: result.TaskID,
|
||||
Status: "processing",
|
||||
Completed: false,
|
||||
}, nil
|
||||
}
|
||||
|
||||
if len(result.Output) == 0 {
|
||||
return nil, fmt.Errorf("no image generated")
|
||||
}
|
||||
|
||||
return &ImageResult{
|
||||
Status: "completed",
|
||||
ImageURL: result.Output[0].URL,
|
||||
Width: options.Width,
|
||||
Height: options.Height,
|
||||
Completed: true,
|
||||
}, nil
|
||||
}
|
||||
|
||||
func (c *StableDiffusionClient) GetTaskStatus(taskID string) (*ImageResult, error) {
|
||||
endpoint := c.BaseURL + "/v1/images/status/" + taskID
|
||||
req, err := http.NewRequest("GET", endpoint, nil)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("create request: %w", err)
|
||||
}
|
||||
|
||||
req.Header.Set("Authorization", "Bearer "+c.APIKey)
|
||||
|
||||
resp, err := c.HTTPClient.Do(req)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("send request: %w", err)
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
body, err := io.ReadAll(resp.Body)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("read response: %w", err)
|
||||
}
|
||||
|
||||
var result SDResponse
|
||||
if err := json.Unmarshal(body, &result); err != nil {
|
||||
return nil, fmt.Errorf("parse response: %w", err)
|
||||
}
|
||||
|
||||
imageResult := &ImageResult{
|
||||
TaskID: taskID,
|
||||
Status: result.Status,
|
||||
Completed: result.Status == "completed",
|
||||
}
|
||||
|
||||
if result.Error != "" {
|
||||
imageResult.Error = result.Error
|
||||
}
|
||||
|
||||
if len(result.Output) > 0 {
|
||||
imageResult.ImageURL = result.Output[0].URL
|
||||
}
|
||||
|
||||
return imageResult, nil
|
||||
}
|
||||
|
||||
128
pkg/image/openai_image_client.go
Normal file
128
pkg/image/openai_image_client.go
Normal file
@@ -0,0 +1,128 @@
|
||||
package image
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"io"
|
||||
"net/http"
|
||||
"time"
|
||||
)
|
||||
|
||||
type OpenAIImageClient struct {
|
||||
BaseURL string
|
||||
APIKey string
|
||||
Model string
|
||||
Endpoint string
|
||||
HTTPClient *http.Client
|
||||
}
|
||||
|
||||
type DALLERequest struct {
|
||||
Model string `json:"model"`
|
||||
Prompt string `json:"prompt"`
|
||||
Size string `json:"size,omitempty"`
|
||||
Quality string `json:"quality,omitempty"`
|
||||
N int `json:"n"`
|
||||
Image []string `json:"image,omitempty"`
|
||||
}
|
||||
|
||||
type DALLEResponse struct {
|
||||
Created int64 `json:"created"`
|
||||
Data []struct {
|
||||
URL string `json:"url"`
|
||||
RevisedPrompt string `json:"revised_prompt,omitempty"`
|
||||
} `json:"data"`
|
||||
}
|
||||
|
||||
func NewOpenAIImageClient(baseURL, apiKey, model, endpoint string) *OpenAIImageClient {
|
||||
if endpoint == "" {
|
||||
endpoint = "/v1/images/generations"
|
||||
}
|
||||
return &OpenAIImageClient{
|
||||
BaseURL: baseURL,
|
||||
APIKey: apiKey,
|
||||
Model: model,
|
||||
Endpoint: endpoint,
|
||||
HTTPClient: &http.Client{
|
||||
Timeout: 10 * time.Minute,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
func (c *OpenAIImageClient) GenerateImage(prompt string, opts ...ImageOption) (*ImageResult, error) {
|
||||
options := &ImageOptions{
|
||||
Size: "1920x1920",
|
||||
Quality: "standard",
|
||||
}
|
||||
|
||||
for _, opt := range opts {
|
||||
opt(options)
|
||||
}
|
||||
|
||||
model := c.Model
|
||||
if options.Model != "" {
|
||||
model = options.Model
|
||||
}
|
||||
|
||||
reqBody := DALLERequest{
|
||||
Model: model,
|
||||
Prompt: prompt,
|
||||
Size: options.Size,
|
||||
Quality: options.Quality,
|
||||
N: 1,
|
||||
Image: options.ReferenceImages,
|
||||
}
|
||||
|
||||
jsonData, err := json.Marshal(reqBody)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("marshal request: %w", err)
|
||||
}
|
||||
|
||||
url := c.BaseURL + c.Endpoint
|
||||
fmt.Printf("[OpenAI Image] Request URL: %s\n", url)
|
||||
fmt.Printf("[OpenAI Image] Request Body: %s\n", string(jsonData))
|
||||
|
||||
req, err := http.NewRequest("POST", url, bytes.NewBuffer(jsonData))
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("create request: %w", err)
|
||||
}
|
||||
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
req.Header.Set("Authorization", "Bearer "+c.APIKey)
|
||||
|
||||
resp, err := c.HTTPClient.Do(req)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("send request: %w", err)
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
body, err := io.ReadAll(resp.Body)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("read response: %w", err)
|
||||
}
|
||||
|
||||
if resp.StatusCode != http.StatusOK {
|
||||
return nil, fmt.Errorf("API error (status %d): %s", resp.StatusCode, string(body))
|
||||
}
|
||||
|
||||
fmt.Printf("OpenAI API Response: %s\n", string(body))
|
||||
|
||||
var result DALLEResponse
|
||||
if err := json.Unmarshal(body, &result); err != nil {
|
||||
return nil, fmt.Errorf("parse response: %w, body: %s", err, string(body))
|
||||
}
|
||||
|
||||
if len(result.Data) == 0 {
|
||||
return nil, fmt.Errorf("no image generated, response: %s", string(body))
|
||||
}
|
||||
|
||||
return &ImageResult{
|
||||
Status: "completed",
|
||||
ImageURL: result.Data[0].URL,
|
||||
Completed: true,
|
||||
}, nil
|
||||
}
|
||||
|
||||
func (c *OpenAIImageClient) GetTaskStatus(taskID string) (*ImageResult, error) {
|
||||
return nil, fmt.Errorf("not supported for OpenAI/DALL-E")
|
||||
}
|
||||
158
pkg/image/volcengine_image_client.go
Normal file
158
pkg/image/volcengine_image_client.go
Normal file
@@ -0,0 +1,158 @@
|
||||
package image
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"io"
|
||||
"net/http"
|
||||
"time"
|
||||
)
|
||||
|
||||
type VolcEngineImageClient struct {
|
||||
BaseURL string
|
||||
APIKey string
|
||||
Model string
|
||||
Endpoint string
|
||||
QueryEndpoint string
|
||||
HTTPClient *http.Client
|
||||
}
|
||||
|
||||
type VolcEngineImageRequest struct {
|
||||
Model string `json:"model"`
|
||||
Prompt string `json:"prompt"`
|
||||
Image []string `json:"image,omitempty"`
|
||||
SequentialImageGeneration string `json:"sequential_image_generation,omitempty"`
|
||||
Size string `json:"size,omitempty"`
|
||||
Watermark bool `json:"watermark,omitempty"`
|
||||
}
|
||||
|
||||
type VolcEngineImageResponse struct {
|
||||
Model string `json:"model"`
|
||||
Created int64 `json:"created"`
|
||||
Data []struct {
|
||||
URL string `json:"url"`
|
||||
Size string `json:"size"`
|
||||
} `json:"data"`
|
||||
Usage struct {
|
||||
GeneratedImages int `json:"generated_images"`
|
||||
OutputTokens int `json:"output_tokens"`
|
||||
TotalTokens int `json:"total_tokens"`
|
||||
} `json:"usage"`
|
||||
Error interface{} `json:"error,omitempty"`
|
||||
}
|
||||
|
||||
func NewVolcEngineImageClient(baseURL, apiKey, model, endpoint, queryEndpoint string) *VolcEngineImageClient {
|
||||
if endpoint == "" {
|
||||
endpoint = "/api/v3/images/generations"
|
||||
}
|
||||
if queryEndpoint == "" {
|
||||
queryEndpoint = endpoint
|
||||
}
|
||||
return &VolcEngineImageClient{
|
||||
BaseURL: baseURL,
|
||||
APIKey: apiKey,
|
||||
Model: model,
|
||||
Endpoint: endpoint,
|
||||
QueryEndpoint: queryEndpoint,
|
||||
HTTPClient: &http.Client{
|
||||
Timeout: 10 * time.Minute,
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
func (c *VolcEngineImageClient) GenerateImage(prompt string, opts ...ImageOption) (*ImageResult, error) {
|
||||
options := &ImageOptions{
|
||||
Size: "1024x1024",
|
||||
Quality: "standard",
|
||||
}
|
||||
|
||||
for _, opt := range opts {
|
||||
opt(options)
|
||||
}
|
||||
|
||||
model := c.Model
|
||||
if options.Model != "" {
|
||||
model = options.Model
|
||||
}
|
||||
|
||||
promptText := prompt
|
||||
if options.NegativePrompt != "" {
|
||||
promptText += fmt.Sprintf(". Negative: %s", options.NegativePrompt)
|
||||
}
|
||||
|
||||
size := options.Size
|
||||
if size == "" {
|
||||
if model == "doubao-seedream-4-5-251128" {
|
||||
size = "2K"
|
||||
} else {
|
||||
size = "1K"
|
||||
}
|
||||
}
|
||||
|
||||
reqBody := VolcEngineImageRequest{
|
||||
Model: model,
|
||||
Prompt: promptText,
|
||||
Image: options.ReferenceImages,
|
||||
SequentialImageGeneration: "disabled",
|
||||
Size: size,
|
||||
Watermark: false,
|
||||
}
|
||||
|
||||
jsonData, err := json.Marshal(reqBody)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("marshal request: %w", err)
|
||||
}
|
||||
|
||||
url := c.BaseURL + c.Endpoint
|
||||
fmt.Printf("[VolcEngine Image] Request URL: %s\n", url)
|
||||
fmt.Printf("[VolcEngine Image] Request Body: %s\n", string(jsonData))
|
||||
|
||||
req, err := http.NewRequest("POST", url, bytes.NewBuffer(jsonData))
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("create request: %w", err)
|
||||
}
|
||||
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
req.Header.Set("Authorization", "Bearer "+c.APIKey)
|
||||
|
||||
resp, err := c.HTTPClient.Do(req)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("send request: %w", err)
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
body, err := io.ReadAll(resp.Body)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("read response: %w", err)
|
||||
}
|
||||
|
||||
fmt.Printf("VolcEngine Image API Response: %s\n", string(body))
|
||||
|
||||
if resp.StatusCode != http.StatusOK && resp.StatusCode != http.StatusCreated {
|
||||
return nil, fmt.Errorf("API error (status %d): %s", resp.StatusCode, string(body))
|
||||
}
|
||||
|
||||
var result VolcEngineImageResponse
|
||||
if err := json.Unmarshal(body, &result); err != nil {
|
||||
return nil, fmt.Errorf("parse response: %w", err)
|
||||
}
|
||||
|
||||
if result.Error != nil {
|
||||
return nil, fmt.Errorf("volcengine error: %v", result.Error)
|
||||
}
|
||||
|
||||
if len(result.Data) == 0 {
|
||||
return nil, fmt.Errorf("no image generated")
|
||||
}
|
||||
|
||||
return &ImageResult{
|
||||
Status: "completed",
|
||||
ImageURL: result.Data[0].URL,
|
||||
Completed: true,
|
||||
}, nil
|
||||
}
|
||||
|
||||
func (c *VolcEngineImageClient) GetTaskStatus(taskID string) (*ImageResult, error) {
|
||||
return nil, fmt.Errorf("not supported for VolcEngine Seedream (synchronous generation)")
|
||||
}
|
||||
Reference in New Issue
Block a user