Files
homeagent-sdk/example/ai_image/plugin.go
JianFeeeee b6a6168e65 chore(module)!: module 路径随托管地迁移 gitcode → github
SDK 仓已迁到 GitHub 作为权威源,go module 路径同步:
`gitcode.com/JianFeeeee/homeagent-sdk` → `github.com/JianFeeeee/homeagentsdk`

## 范围

74 个文件:71 个 Go 源码 + go.mod/go.sum + hmapdev 模板 + CI 与打包脚本。

主仓(HomeAgent)侧对应改动是 6d38a164。本仓这些文件当时**不在工作树上**
(见下),所以那次的批量替换扫不到它们——这里是补偿。

## 为何这批文件当时不在工作树上(事故与恢复)

10-08 07:58 主仓做 checkout/cherry-pick 序列时,把本仓工作树的 20 个文件
误清理了(`example/*/plugin.go` + `plg.json`,10 个示例)。

根因是**结构误判**:本仓是「一目录两仓库」结构下主仓 `third_party/` 里的
**独立 git 仓库**,而主仓 .gitignore 忽略了 `example/`,于是它在主仓
`git status` 里不可见 —— 看着像「不受跟踪的构建产物」,实际是本仓受跟踪的源码。

已全部恢复(`git checkout -- <D 文件>`,只恢复删除、不动迁移改动),
20 个文件与 HEAD 逐字节一致。

**破坏的连锁后果**:恢复出来的 `plugin.go` 仍写着旧 module 路径,
因为迁移时它们已不在工作树上。所以这次补偿是必须的,不是可选优化。

## 保留 gitcode 引用

以下位置的 `gitcode` **有意保留**,不要"顺手清掉":
- README.md / docs 里「二进制实际分发地址目前是 gitcode」的说明
- mkdocs.yml 的 gitcode 镜像链接(国内直连更快)
- `tools/hmapdev/templates.go` 里「SDK 已从 gitcode 迁到 GitHub」的注释

gitcode 仍是**只读镜像**(源码同步;release 附件暂时仍在那边)。

## 验证

- 本仓根 module:`go build ./...` + `go test ./...` 全通过
- 全部 20 个 example:`go vet` 通过(补偿前 10 个因旧 module 路径失败)
- example/qq:`go test` 通过,且能从源码 build 出可用 plugin 产物
2026-10-09 07:25:25 +08:00

435 lines
13 KiB
Go
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

package main
import (
"bytes"
"encoding/json"
"fmt"
"io"
"log"
"net/http"
"os"
"path/filepath"
"strconv"
"strings"
"time"
sdk "github.com/JianFeeeee/homeagentsdk/sdk"
)
type Plugin struct {
name string
sdk *sdk.PluginSDK
client *http.Client
apiKey string
provider string
model string
size string
baseURL string
dataDir string // <data>/ai_images:生成本地图片存放目录
}
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", "")), "/")
// 生图本地存放目录:插件专属数据目录(SDK DataDir API,内核保证存在)。
if p.sdk != nil {
if dd := s.Settings().DataDir(); dd != "" {
p.dataDir = dd
}
}
if p.dataDir == "" {
// 旧版内核无 DataDir API 时退到 /tmp
p.dataDir = filepath.Join(os.TempDir(), "homeagent_ai_images")
}
os.MkdirAll(p.dataDir, 0755)
tp := p.name + "_"
s.RegisterTool(tp+"generate", sdk.ToolDef{
Name: tp + "generate", Description: "Generate image from text prompt using AI. Downloads the result locally and returns a local file path (permanent, no expiry). To show the user, send it via output_send with type=image and payload=the returned path.",
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",
}
log.Printf("[ai_image] endpoint=%s baseURL=%q model=%q", endpoint, p.baseURL, model)
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
}
// 下载到本地 data 目录,返回本地文件路径(而非临时 S3 URL):
// - S3 临时 URL 约 1 小时过期,且对无浏览器 UA 的客户端拒绝访问
// - 本地路径可经 webui /files/ 永久下发给所有客户端(含 API key 客户端)
localPaths := make([]string, len(urls))
var errs []string
for i, u := range urls {
path, err := p.downloadImage(u, fmt.Sprintf("ai_%s_%d", model, time.Now().UnixNano()))
if err != nil {
errs = append(errs, fmt.Sprintf("第%d张下载失败: %v", i+1, err))
continue
}
localPaths[i] = path
}
content := fmt.Sprintf("Generated %d image(s) with model %s:", len(urls), model)
for _, pth := range localPaths {
if pth != "" {
content += "\n" + pth
}
}
if len(errs) > 0 {
content += "\n\n" + strings.Join(errs, "\n")
}
content += "\n\n已将图片保存到本地(不会过期)。如需展示请用 output_send__webui(payload=本地路径, type=image)。"
return map[string]interface{}{
"content": content,
"images": localPaths,
"prompt": prompt,
"model": model,
"local_paths": localPaths,
}, nil
}
// downloadImage 把生图返回的临时 URL 下载为本地文件,返回本地路径。
// 带浏览器 UA 以规避图床对无 UA 客户端的拦截。
func (p *Plugin) downloadImage(url, baseName string) (string, error) {
dl := &http.Client{Timeout: 60 * time.Second}
req, err := http.NewRequest("GET", url, nil)
if err != nil {
return "", err
}
req.Header.Set("User-Agent", "Mozilla/5.0 (compatible; HomeAgent/1.0)")
resp, err := dl.Do(req)
if err != nil {
return "", err
}
defer resp.Body.Close()
if resp.StatusCode != http.StatusOK {
b, _ := io.ReadAll(resp.Body)
return "", fmt.Errorf("HTTP %d: %s", resp.StatusCode, strings.TrimSpace(string(b))[:200])
}
data, err := io.ReadAll(resp.Body)
if err != nil {
return "", err
}
ext := ".png"
if ct := resp.Header.Get("Content-Type"); strings.Contains(ct, "jpeg") || strings.Contains(ct, "jpg") {
ext = ".jpg"
} else if strings.Contains(ct, "webp") {
ext = ".webp"
}
path := filepath.Join(p.dataDir, baseName+ext)
if err := os.WriteFile(path, data, 0644); err != nil {
return "", err
}
return path, 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\n\n图片已保存到本地,如需展示请用 output_send(type=image)。", len(urls), strings.Join(urls, "\n")),
"images": urls,
"prompt": prompt,
"model": model,
}, nil
}