mirror of
https://gitcode.com/JianFeeeee/homeagent-sdk.git
synced 2026-10-15 12:15:11 +00:00
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
145 lines
4.3 KiB
Go
145 lines
4.3 KiB
Go
package main
|
||
|
||
import (
|
||
"encoding/base64"
|
||
"encoding/json"
|
||
"fmt"
|
||
"io"
|
||
"log"
|
||
"net/http"
|
||
"os"
|
||
"os/exec"
|
||
"path/filepath"
|
||
"strings"
|
||
"time"
|
||
|
||
"github.com/JianFeeeee/homeagentsdk/sdk"
|
||
)
|
||
|
||
type Plugin struct {
|
||
name string
|
||
sdk *sdk.PluginSDK
|
||
}
|
||
|
||
func (p *Plugin) Name() string { return p.name }
|
||
|
||
func (p *Plugin) Start(s *sdk.PluginSDK) error {
|
||
s.SetAutoRestart(true)
|
||
p.sdk = s
|
||
tp := p.name + "_"
|
||
|
||
s.RegisterTool(tp+"ocr_image", sdk.ToolDef{
|
||
Name: tp + "ocr_image",
|
||
Description: "对图片进行OCR文字识别,支持中文和英文。可传入图片URL或base64编码。返回识别出的文本内容。",
|
||
Parameters: map[string]interface{}{
|
||
"type": "object",
|
||
"properties": map[string]interface{}{
|
||
"image_url": map[string]interface{}{"type": "string", "description": "图片的HTTP/HTTPS URL,与 image_data 二选一"},
|
||
"image_data": map[string]interface{}{"type": "string", "description": "图片的base64编码数据(不含 data:image/... 前缀),与 image_url 二选一"},
|
||
"language": map[string]interface{}{"type": "string", "description": "识别语言,默认 chi_sim+eng(中文简体+英文),可选 chi_sim / eng / chi_sim+eng"},
|
||
},
|
||
},
|
||
Cleaner: func(output string) string {
|
||
var r struct{ Text string }
|
||
if json.Unmarshal([]byte(output), &r) == nil && r.Text != "" {
|
||
return r.Text
|
||
}
|
||
var r2 struct{ Content string }
|
||
if json.Unmarshal([]byte(output), &r2) == nil && r2.Content != "" {
|
||
return r2.Content
|
||
}
|
||
return output
|
||
},
|
||
}, p.handleOcrImage)
|
||
|
||
log.Printf("[%s] plugin started", p.name)
|
||
return nil
|
||
}
|
||
|
||
func (p *Plugin) Stop() error {
|
||
return nil
|
||
}
|
||
|
||
func (p *Plugin) handleOcrImage(args map[string]interface{}) (interface{}, error) {
|
||
imageURL, _ := args["image_url"].(string)
|
||
imageData, _ := args["image_data"].(string)
|
||
language, _ := args["language"].(string)
|
||
|
||
if imageURL == "" && imageData == "" {
|
||
return map[string]interface{}{"error": "请提供 image_url 或 image_data"}, nil
|
||
}
|
||
|
||
tmpDir, err := os.MkdirTemp("", "ocr-*")
|
||
if err != nil {
|
||
return map[string]interface{}{"error": fmt.Sprintf("创建临时目录失败: %v", err)}, nil
|
||
}
|
||
defer os.RemoveAll(tmpDir)
|
||
|
||
inputPath := filepath.Join(tmpDir, "input.png")
|
||
|
||
if imageData != "" {
|
||
data := strings.TrimSpace(imageData)
|
||
if idx := strings.Index(data, "base64,"); idx >= 0 {
|
||
data = data[idx+7:]
|
||
}
|
||
decoded, err := base64.StdEncoding.DecodeString(data)
|
||
if err != nil {
|
||
return map[string]interface{}{"error": fmt.Sprintf("base64解码失败: %v", err)}, nil
|
||
}
|
||
if err := os.WriteFile(inputPath, decoded, 0644); err != nil {
|
||
return map[string]interface{}{"error": fmt.Sprintf("写入临时文件失败: %v", err)}, nil
|
||
}
|
||
} else {
|
||
client := &http.Client{Timeout: 30 * time.Second}
|
||
resp, err := client.Get(imageURL)
|
||
if err != nil {
|
||
return map[string]interface{}{"error": fmt.Sprintf("下载图片失败: %v", err)}, nil
|
||
}
|
||
defer resp.Body.Close()
|
||
if resp.StatusCode != http.StatusOK {
|
||
return map[string]interface{}{"error": fmt.Sprintf("下载图片返回状态码 %d", resp.StatusCode)}, nil
|
||
}
|
||
data, err := io.ReadAll(resp.Body)
|
||
if err != nil {
|
||
return map[string]interface{}{"error": fmt.Sprintf("读取图片数据失败: %v", err)}, nil
|
||
}
|
||
if err := os.WriteFile(inputPath, data, 0644); err != nil {
|
||
return map[string]interface{}{"error": fmt.Sprintf("写入临时文件失败: %v", err)}, nil
|
||
}
|
||
}
|
||
|
||
if language == "" {
|
||
language = "chi_sim+eng"
|
||
}
|
||
|
||
outputPath := filepath.Join(tmpDir, "output")
|
||
|
||
argsList := []string{inputPath, outputPath, "-l", language, "--psm", "3"}
|
||
cmd := exec.Command("tesseract", argsList...)
|
||
var stderr strings.Builder
|
||
cmd.Stderr = &stderr
|
||
if err := cmd.Run(); err != nil {
|
||
return map[string]interface{}{"error": fmt.Sprintf("OCR识别失败: %v (stderr: %s)", err, stderr.String())}, nil
|
||
}
|
||
|
||
resultFile := outputPath + ".txt"
|
||
text, err := os.ReadFile(resultFile)
|
||
if err != nil {
|
||
return map[string]interface{}{"error": fmt.Sprintf("读取OCR结果失败: %v", err)}, nil
|
||
}
|
||
|
||
recognized := strings.TrimSpace(string(text))
|
||
if recognized == "" {
|
||
return map[string]interface{}{"text": "", "message": "未识别出文字内容"}, nil
|
||
}
|
||
|
||
return map[string]interface{}{
|
||
"text": recognized,
|
||
"length": len(recognized),
|
||
"language": language,
|
||
}, nil
|
||
}
|
||
|
||
func NewPluginFactory(name string, config map[string]interface{}) (sdk.Plugin, error) {
|
||
return &Plugin{name: name}, nil
|
||
} |