mirror of
https://gitcode.com/JianFeeeee/HomeAgent.git
synced 2026-09-21 17:38:10 +00:00
feat(ai_image): base_url 设置项支持自定义 OpenAI 兼容网关 v1.1.0
generateOpenAI 原硬编码上游为 https://api.openai.com，无法接入 本机 llmsproxy(ModelRouter) 等标准 OpenAI 兼容网关。新增： - 设置项 base_url（string，默认空）：为空保持官方直连；非空时 上游改为 {base_url}/v1/images/generations（约定不带 /v1 尾缀， 拼接时自动去重避免 /v1/v1） - Plugin.baseURL 字段 Start 时读入；注册 ConfigDef（category ai_image）供 WebUI 配置页展示 部署：plugindev 打包 1.1.0 → pluginmgr overwrite:true 升级安装 config_kept=true → plgreload 热加载。 配置（category ai_image）：api_key=sk-gw-local-0001, base_url=http://127.0.0.1:8081, provider=openai, model=Kwai-Kolors/Kolors, size=1024x1024 验收：agent 实调 ai_image_generate 出图成功——llmsproxy audit 记录 type=image src=siliconflow model=AUTO ok=true（经网关非直连）， 返回临时 S3 URL 下载为有效 PNG (1024x1024)。
This commit is contained in:
JianFeeeee
20
third_party/homeagent-sdk/example/ai_image/plg.json
vendored
Normal file
20
third_party/homeagent-sdk/example/ai_image/plg.json
vendored
Normal file
@ -0,0 +1,20 @@
|
||||
{
|
||||
"name": "ai_image",
|
||||
"name_zh": "AI绘图",
|
||||
"name_en": "AI Image",
|
||||
"version": "1.1.0",
|
||||
"description": "AI 图像生成插件，支持 OpenAI DALL·E / Stable Diffusion",
|
||||
"author": "HomeAgent",
|
||||
"entry": "plugin.so",
|
||||
"tags": [
|
||||
"ai",
|
||||
"image",
|
||||
"draw",
|
||||
"generate"
|
||||
],
|
||||
"targets": "linux/amd64",
|
||||
"outdir": "dist",
|
||||
"bundle": true,
|
||||
"replaces": {},
|
||||
"source_dirs": []
|
||||
}
|
||||
357
third_party/homeagent-sdk/example/ai_image/plugin.go
vendored
Normal file
357
third_party/homeagent-sdk/example/ai_image/plugin.go
vendored
Normal file
@ -0,0 +1,357 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"io"
|
||||
"net/http"
|
||||
"strconv"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
sdk "gitcode.com/JianFeeeee/homeagent-sdk/sdk"
|
||||
)
|
||||
|
||||
type Plugin struct {
|
||||
name string
|
||||
sdk *sdk.PluginSDK
|
||||
client *http.Client
|
||||
apiKey string
|
||||
provider string
|
||||
model string
|
||||
size string
|
||||
baseURL string
|
||||
}
|
||||
|
||||
func NewPluginFactory(name string, config map[string]interface{}) (sdk.Plugin, error) {
|
||||
return &Plugin{name: name}, nil
|
||||
}
|
||||
|
||||
func (p *Plugin) Name() string { return p.name }
|
||||
|
||||
func getSetting[T string | int64 | float64](s sdk.SettingsAPI, key string, def T) T {
|
||||
v, err := s.Get(key)
|
||||
if err != nil || v == nil {
|
||||
return def
|
||||
}
|
||||
switch any(def).(type) {
|
||||
case string:
|
||||
if sv, ok := v.(string); ok {
|
||||
return any(sv).(T)
|
||||
}
|
||||
case int64:
|
||||
switch n := v.(type) {
|
||||
case float64:
|
||||
return any(int64(n)).(T)
|
||||
case int64:
|
||||
return any(n).(T)
|
||||
case string:
|
||||
if i, err := strconv.ParseInt(n, 10, 64); err == nil {
|
||||
return any(i).(T)
|
||||
}
|
||||
}
|
||||
case float64:
|
||||
switch n := v.(type) {
|
||||
case float64:
|
||||
return any(n).(T)
|
||||
case int64:
|
||||
return any(float64(n)).(T)
|
||||
case string:
|
||||
if f, err := strconv.ParseFloat(n, 64); err == nil {
|
||||
return any(f).(T)
|
||||
}
|
||||
}
|
||||
}
|
||||
return def
|
||||
}
|
||||
|
||||
func getArg[T string | int64 | float64](args map[string]interface{}, key string, def T) T {
|
||||
v, ok := args[key]
|
||||
if !ok || v == nil {
|
||||
return def
|
||||
}
|
||||
switch any(def).(type) {
|
||||
case string:
|
||||
if s, ok := v.(string); ok {
|
||||
return any(s).(T)
|
||||
}
|
||||
case int64:
|
||||
switch n := v.(type) {
|
||||
case float64:
|
||||
return any(int64(n)).(T)
|
||||
case int64:
|
||||
return any(n).(T)
|
||||
case string:
|
||||
if i, err := strconv.ParseInt(n, 10, 64); err == nil {
|
||||
return any(i).(T)
|
||||
}
|
||||
}
|
||||
case float64:
|
||||
switch n := v.(type) {
|
||||
case float64:
|
||||
return any(n).(T)
|
||||
case int64:
|
||||
return any(float64(n)).(T)
|
||||
case string:
|
||||
if f, err := strconv.ParseFloat(n, 64); err == nil {
|
||||
return any(f).(T)
|
||||
}
|
||||
}
|
||||
}
|
||||
return def
|
||||
}
|
||||
|
||||
func (p *Plugin) Start(s *sdk.PluginSDK) error {
|
||||
s.SetAutoRestart(true)
|
||||
p.sdk = s
|
||||
p.client = &http.Client{Timeout: 120 * time.Second}
|
||||
|
||||
s.Settings().RegisterDef(sdk.ConfigDef{
|
||||
Key: "api_key", Default: "", Type: "string",
|
||||
DisplayName: "API Key", Description: "OpenAI / Stable Diffusion API Key",
|
||||
Category: "ai_image", Secret: true,
|
||||
})
|
||||
s.Settings().RegisterDef(sdk.ConfigDef{
|
||||
Key: "base_url", Default: "", Type: "string",
|
||||
DisplayName: "Base URL", Description: "自定义 OpenAI 兼容网关地址（不带 /v1 尾缀，如 http://127.0.0.1:8081）；为空走官方 https://api.openai.com",
|
||||
Category: "ai_image",
|
||||
})
|
||||
s.Settings().RegisterDef(sdk.ConfigDef{
|
||||
Key: "provider", Default: "openai", Type: "string",
|
||||
DisplayName: "Provider", Description: "Image generation provider: openai / stability",
|
||||
Category: "ai_image",
|
||||
})
|
||||
s.Settings().RegisterDef(sdk.ConfigDef{
|
||||
Key: "model", Default: "dall-e-3", Type: "string",
|
||||
DisplayName: "Model", Description: "Model name (dall-e-3, sd-xl, etc.)",
|
||||
Category: "ai_image",
|
||||
})
|
||||
s.Settings().RegisterDef(sdk.ConfigDef{
|
||||
Key: "size", Default: "1024x1024", Type: "string",
|
||||
DisplayName: "Size", Description: "Default image size (1024x1024, 1024x1792, 1792x1024)",
|
||||
Category: "ai_image",
|
||||
})
|
||||
|
||||
p.apiKey = getSetting(s.Settings(), "api_key", "")
|
||||
p.provider = getSetting(s.Settings(), "provider", "openai")
|
||||
p.model = getSetting(s.Settings(), "model", "dall-e-3")
|
||||
p.size = getSetting(s.Settings(), "size", "1024x1024")
|
||||
p.baseURL = strings.TrimRight(strings.TrimSpace(getSetting(s.Settings(), "base_url", "")), "/")
|
||||
|
||||
tp := p.name + "_"
|
||||
s.RegisterTool(tp+"generate", sdk.ToolDef{
|
||||
Name: tp + "generate", Description: "Generate image from text prompt using AI. Returns image URL.",
|
||||
Parameters: map[string]interface{}{
|
||||
"type": "object",
|
||||
"properties": map[string]interface{}{
|
||||
"prompt": map[string]interface{}{"type": "string", "description": "Text description of the image to generate"},
|
||||
"size": map[string]interface{}{"type": "string", "description": "Image size (1024x1024, 1024x1792, 1792x1024), default from config"},
|
||||
"model": map[string]interface{}{"type": "string", "description": "Model override (dall-e-3, dall-e-2)"},
|
||||
"n": map[string]interface{}{"type": "integer", "description": "Number of images to generate (1-10), default 1"},
|
||||
},
|
||||
"required": []string{"prompt"},
|
||||
},
|
||||
}, p.handleGenerate)
|
||||
|
||||
fmt.Printf("[%s] started (provider=%s, model=%s)\n", p.name, p.provider, p.model)
|
||||
return nil
|
||||
}
|
||||
|
||||
func (p *Plugin) Stop() error {
|
||||
fmt.Printf("[%s] stopped\n", p.name)
|
||||
return nil
|
||||
}
|
||||
|
||||
type openAIReq struct {
|
||||
Model string `json:"model"`
|
||||
Prompt string `json:"prompt"`
|
||||
N int `json:"n"`
|
||||
Size string `json:"size"`
|
||||
ResponseFormat string `json:"response_format"`
|
||||
}
|
||||
|
||||
type openAIResp struct {
|
||||
Created int64 `json:"created"`
|
||||
Data []struct {
|
||||
RevisedPrompt string `json:"revised_prompt"`
|
||||
URL string `json:"url"`
|
||||
} `json:"data"`
|
||||
Error *struct {
|
||||
Message string `json:"message"`
|
||||
Type string `json:"type"`
|
||||
} `json:"error"`
|
||||
}
|
||||
|
||||
func (p *Plugin) handleGenerate(args map[string]interface{}) (interface{}, error) {
|
||||
prompt := getArg(args, "prompt", "")
|
||||
if prompt == "" {
|
||||
return map[string]interface{}{"isError": true, "content": "prompt is required"}, nil
|
||||
}
|
||||
|
||||
key := getSetting(p.sdk.Settings(), "api_key", p.apiKey)
|
||||
if key == "" {
|
||||
return map[string]interface{}{"isError": true, "content": "API key not configured. Set plugin.ai_image.api_key via CLI."}, nil
|
||||
}
|
||||
|
||||
provider := getSetting(p.sdk.Settings(), "provider", p.provider)
|
||||
model := getArg(args, "model", getSetting(p.sdk.Settings(), "model", p.model))
|
||||
size := getArg(args, "size", getSetting(p.sdk.Settings(), "size", p.size))
|
||||
n := getArg(args, "n", int64(1))
|
||||
if n < 1 {
|
||||
n = 1
|
||||
}
|
||||
if n > 10 {
|
||||
n = 10
|
||||
}
|
||||
|
||||
switch provider {
|
||||
case "openai":
|
||||
return p.generateOpenAI(prompt, model, size, int(n), key)
|
||||
case "stability":
|
||||
return p.generateStability(prompt, model, size, int(n), key)
|
||||
default:
|
||||
return map[string]interface{}{"isError": true, "content": "Unknown provider: " + provider + ". Supported: openai, stability"}, nil
|
||||
}
|
||||
}
|
||||
|
||||
func (p *Plugin) generateOpenAI(prompt, model, size string, n int, apiKey string) (interface{}, error) {
|
||||
// 上游地址：base_url 非空时走自定义网关（如本机 llmsproxy），约定不带 /v1 尾缀；
|
||||
// 为空保持官方直连。兼容误配了 /v1 尾缀的情况（去重）。
|
||||
endpoint := "https://api.openai.com/v1/images/generations"
|
||||
if p.baseURL != "" {
|
||||
base := strings.TrimSuffix(p.baseURL, "/v1")
|
||||
endpoint = base + "/v1/images/generations"
|
||||
}
|
||||
|
||||
body := openAIReq{
|
||||
Model: model,
|
||||
Prompt: prompt,
|
||||
N: n,
|
||||
Size: size,
|
||||
ResponseFormat: "url",
|
||||
}
|
||||
|
||||
b, _ := json.Marshal(body)
|
||||
req, _ := http.NewRequest("POST", endpoint, bytes.NewReader(b))
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
req.Header.Set("Authorization", "Bearer "+apiKey)
|
||||
|
||||
resp, err := p.client.Do(req)
|
||||
if err != nil {
|
||||
return map[string]interface{}{"isError": true, "content": "Request failed: " + err.Error()}, nil
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
respBody, _ := io.ReadAll(resp.Body)
|
||||
var result openAIResp
|
||||
if err := json.Unmarshal(respBody, &result); err != nil {
|
||||
return map[string]interface{}{"isError": true, "content": "Failed to parse response: " + err.Error()}, nil
|
||||
}
|
||||
|
||||
if result.Error != nil {
|
||||
return map[string]interface{}{"isError": true, "content": "API error: " + result.Error.Message}, nil
|
||||
}
|
||||
|
||||
if len(result.Data) == 0 {
|
||||
return map[string]interface{}{"isError": true, "content": "No images returned"}, nil
|
||||
}
|
||||
|
||||
urls := make([]string, len(result.Data))
|
||||
for i, d := range result.Data {
|
||||
urls[i] = d.URL
|
||||
}
|
||||
|
||||
return map[string]interface{}{
|
||||
"content": fmt.Sprintf("Generated %d image(s) with model %s:\n%s", len(urls), model, strings.Join(urls, "\n")),
|
||||
"images": urls,
|
||||
"prompt": prompt,
|
||||
"model": model,
|
||||
}, nil
|
||||
}
|
||||
|
||||
type stabilityReq struct {
|
||||
TextPrompts []stabilityPrompt `json:"text_prompts"`
|
||||
Width int `json:"width"`
|
||||
Height int `json:"height"`
|
||||
Samples int `json:"samples"`
|
||||
}
|
||||
|
||||
type stabilityPrompt struct {
|
||||
Text string `json:"text"`
|
||||
Weight float64 `json:"weight,omitempty"`
|
||||
}
|
||||
|
||||
type stabilityArtifact struct {
|
||||
Base64 string `json:"base64"`
|
||||
Seed int `json:"seed"`
|
||||
}
|
||||
|
||||
type stabilityResp struct {
|
||||
Artifacts []stabilityArtifact `json:"artifacts"`
|
||||
Message string `json:"message,omitempty"`
|
||||
}
|
||||
|
||||
func (p *Plugin) generateStability(prompt, model, size string, n int, apiKey string) (interface{}, error) {
|
||||
width, height := 1024, 1024
|
||||
if parts := strings.Split(size, "x"); len(parts) == 2 {
|
||||
if w, err := strconv.Atoi(parts[0]); err == nil {
|
||||
width = w
|
||||
}
|
||||
if h, err := strconv.Atoi(parts[1]); err == nil {
|
||||
height = h
|
||||
}
|
||||
}
|
||||
|
||||
body := stabilityReq{
|
||||
TextPrompts: []stabilityPrompt{{Text: prompt, Weight: 1.0}},
|
||||
Width: width,
|
||||
Height: height,
|
||||
Samples: n,
|
||||
}
|
||||
|
||||
apiURL := "https://api.stability.ai/v1/generation/stable-diffusion-xl-1024-v1-0/text-to-image"
|
||||
|
||||
b, _ := json.Marshal(body)
|
||||
req, _ := http.NewRequest("POST", apiURL, bytes.NewReader(b))
|
||||
req.Header.Set("Content-Type", "application/json")
|
||||
req.Header.Set("Authorization", "Bearer "+apiKey)
|
||||
req.Header.Set("Accept", "application/json")
|
||||
|
||||
resp, err := p.client.Do(req)
|
||||
if err != nil {
|
||||
return map[string]interface{}{"isError": true, "content": "Request failed: " + err.Error()}, nil
|
||||
}
|
||||
defer resp.Body.Close()
|
||||
|
||||
if resp.StatusCode != 200 {
|
||||
respBody, _ := io.ReadAll(resp.Body)
|
||||
return map[string]interface{}{"isError": true, "content": fmt.Sprintf("API error (status %d): %s", resp.StatusCode, string(respBody))}, nil
|
||||
}
|
||||
|
||||
respBody, _ := io.ReadAll(resp.Body)
|
||||
var result stabilityResp
|
||||
if err := json.Unmarshal(respBody, &result); err != nil {
|
||||
return map[string]interface{}{"isError": true, "content": "Failed to parse response: " + err.Error()}, nil
|
||||
}
|
||||
|
||||
if len(result.Artifacts) == 0 {
|
||||
msg := result.Message
|
||||
if msg == "" {
|
||||
msg = "No images returned"
|
||||
}
|
||||
return map[string]interface{}{"isError": true, "content": msg}, nil
|
||||
}
|
||||
|
||||
urls := make([]string, len(result.Artifacts))
|
||||
for i, a := range result.Artifacts {
|
||||
urls[i] = "data:image/png;base64," + a.Base64
|
||||
}
|
||||
|
||||
return map[string]interface{}{
|
||||
"content": fmt.Sprintf("Generated %d image(s) via Stability AI:\n%s", len(urls), strings.Join(urls, "\n")),
|
||||
"images": urls,
|
||||
"prompt": prompt,
|
||||
"model": model,
|
||||
}, nil
|
||||
}
|
||||
Reference in New Issue
Block a user