mirror of
https://gitcode.com/JianFeeeee/homeagent-sdk.git
synced 2026-09-20 00:48:12 +00:00
345 lines
9.7 KiB
Go
345 lines
9.7 KiB
Go
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
|
|
}
|
|
|
|
func NewPlugin(name string, config map[string]interface{}) (sdk.Plugin, error) {
|
|
return &Plugin{name: name}, nil
|
|
}
|
|
|
|
func (p *Plugin) Name() string { return p.name }
|
|
|
|
func readCfg[T string | int64 | float64](s sdk.SettingsAPI, key string, fallback T) T {
|
|
v, err := s.Get(key)
|
|
if err == nil && v != nil {
|
|
if sv, ok := v.(string); ok && sv != "" {
|
|
switch any(fallback).(type) {
|
|
case string:
|
|
return any(sv).(T)
|
|
case int64:
|
|
if n, err := strconv.ParseInt(sv, 10, 64); err == nil {
|
|
return any(n).(T)
|
|
}
|
|
case float64:
|
|
if n, err := strconv.ParseFloat(sv, 64); err == nil {
|
|
return any(n).(T)
|
|
}
|
|
}
|
|
}
|
|
}
|
|
v2, err2 := s.GetCore("plugin." + "ai_image" + "." + key)
|
|
if err2 == nil && v2 != nil {
|
|
if sv, ok := v2.(string); ok && sv != "" {
|
|
switch any(fallback).(type) {
|
|
case string:
|
|
return any(sv).(T)
|
|
case int64:
|
|
if n, err := strconv.ParseInt(sv, 10, 64); err == nil {
|
|
return any(n).(T)
|
|
}
|
|
case float64:
|
|
if n, err := strconv.ParseFloat(sv, 64); err == nil {
|
|
return any(n).(T)
|
|
}
|
|
}
|
|
}
|
|
}
|
|
return fallback
|
|
}
|
|
|
|
func readArg[T string | int64 | float64](args map[string]interface{}, key string, fallback T) T {
|
|
v, ok := args[key]
|
|
if !ok || v == nil {
|
|
return fallback
|
|
}
|
|
switch any(fallback).(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 fallback
|
|
}
|
|
|
|
func (p *Plugin) Start(s *sdk.PluginSDK) error {
|
|
p.sdk = s
|
|
|
|
p.apiKey = readCfg(s.Settings(), "api_key", "")
|
|
p.provider = readCfg(s.Settings(), "provider", "openai")
|
|
p.model = readCfg(s.Settings(), "model", "dall-e-3")
|
|
p.size = readCfg(s.Settings(), "size", "1024x1024")
|
|
|
|
p.client = &http.Client{Timeout: 120 * time.Second}
|
|
|
|
s.Settings().RegisterDef(sdk.ConfigDef{
|
|
Key: "plugin.ai_image.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: "plugin.ai_image.provider", Default: "openai", Type: "string",
|
|
DisplayName: "Provider", Description: "Image generation provider: openai / stability",
|
|
Category: "ai_image",
|
|
})
|
|
s.Settings().RegisterDef(sdk.ConfigDef{
|
|
Key: "plugin.ai_image.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: "plugin.ai_image.size", Default: "1024x1024", Type: "string",
|
|
DisplayName: "Size", Description: "Default image size (1024x1024, 1024x1792, 1792x1024)",
|
|
Category: "ai_image",
|
|
})
|
|
|
|
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 := readArg(args, "prompt", "")
|
|
if prompt == "" {
|
|
return map[string]interface{}{"isError": true, "content": "prompt is required"}, nil
|
|
}
|
|
|
|
p.apiKey = readCfg(p.sdk.Settings(), "api_key", p.apiKey)
|
|
if p.apiKey == "" {
|
|
return map[string]interface{}{"isError": true, "content": "API key not configured. Set plugin.ai_image.api_key via CLI."}, nil
|
|
}
|
|
|
|
provider := readCfg(p.sdk.Settings(), "provider", p.provider)
|
|
model := readArg(args, "model", readCfg(p.sdk.Settings(), "model", p.model))
|
|
size := readArg(args, "size", readCfg(p.sdk.Settings(), "size", p.size))
|
|
n := readArg(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))
|
|
case "stability":
|
|
return p.generateStability(prompt, model, size, int(n))
|
|
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) (interface{}, error) {
|
|
body := openAIReq{
|
|
Model: model,
|
|
Prompt: prompt,
|
|
N: n,
|
|
Size: size,
|
|
ResponseFormat: "url",
|
|
}
|
|
|
|
b, _ := json.Marshal(body)
|
|
req, _ := http.NewRequest("POST", "https://api.openai.com/v1/images/generations", bytes.NewReader(b))
|
|
req.Header.Set("Content-Type", "application/json")
|
|
req.Header.Set("Authorization", "Bearer "+p.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) (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 "+p.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
|
|
}
|