px7sihb6k
/
coeditor
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

增加图片识别后端接口

Browse Source
master
linlnf 3 months ago
parent 8758733bc9
commit 6507272923
15 changed files with 226 additions and 37 deletions
Show Stats Download Patch File Download Diff File

20
GinSkeleton/api_doc.md
Unescape View File

@ -191,7 +191,7 @@ Authorization|Headers|string|必填|Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.
参数字段|参数属性|类型|选项|默认值
---|---|---|---|---
pic|form-data|string|必填|""
pic|form-data|string|必填|"" (示例内容在 storage/app/test/文字识别.txt 中)
> 返回示例：
```json
@ -202,4 +202,22 @@ pic|form-data|string|必填|""
},
"msg": "Success"
}
```
#### 请求语音识别
> <font color=#FF4500>*post*，/admin/ai_recognition/voc_recognition</font>
参数字段|参数属性|类型|选项|默认值
---|---|---|---|---
voc|form-data|string|必填|"" (示例内容在 storage/app/test/A13_221.txt 中)
> 返回示例：
```json
{
"code": 200,
"data": {
"words": "韩国的基本目标是射箭三块金牌只到三块金牌，羽毛球，两块金牌，以及举重等12块金牌。\n"
},
"msg": "Success"
}
```

2
GinSkeleton/app/global/consts/consts.go
Unescape View File

@ -81,4 +81,6 @@ const (
// 模型功能按相关
PicRecognitionFailMsg string = "图片文字识别失败"
PicRecognitionFailCode int = -400450
VocRecognitionFailMsg string = "语音识别失败"
VocRecognitionFailCode int = -400451
)

1
GinSkeleton/app/global/my_errors/my_errors.go
Unescape View File

@ -72,4 +72,5 @@ const (
ErrorBaiduCEGetTokenFail string = "访问百度智能云时，获得access_token 出现问题："
ErrorBaiduCEUseOCRFail string = "使用百度智能云OCR接口失败："
ErrorBaiduCEPostFail string = "使用百度智能云接口失败(POST通用)："
ErrorBaiduCEUseVOPFail string = "使用百度智能云VOP接口失败："
)

9
GinSkeleton/app/http/controller/web/ai_recognition_controller.go
Unescape View File

@ -21,3 +21,12 @@ func (u *AiRecognition) PicRecognition(context *gin.Context) {
response.Fail(context, consts.PicRecognitionFailCode, consts.PicRecognitionFailMsg, "")
}
}
// 语音文字识别
func (u *AiRecognition) VocRecognition(context *gin.Context) {
if r, recogWords := ai_model_cli.RequestVOP(context); r {
response.Success(context, consts.CurdStatusOkMsg, recogWords)
} else {
response.Fail(context, consts.VocRecognitionFailCode, consts.VocRecognitionFailMsg, "")
}
}

2
GinSkeleton/app/http/validator/common/register_validator/web_register_validator.go
Unescape View File

@ -51,4 +51,6 @@ func WebRegisterValidator() {
// ai 识别功能相关
key = consts.ValidatorPrefix + "PicRecognition"
containers.Set(key, ai_recognition.PicRecognition{})
key = consts.ValidatorPrefix + "VocRecognition"
containers.Set(key, ai_recognition.VocRecognition{})
}

57
GinSkeleton/app/http/validator/web/ai_recognition/pic_recognition.go
Unescape View File

@ -1,10 +1,14 @@
package ai_recognition
import (
"bytes"
"encoding/base64"
"goskeleton/app/global/consts"
"goskeleton/app/http/controller/web"
"goskeleton/app/http/validator/core/data_transfer"
"goskeleton/app/utils/response"
"image"
"math"
"github.com/gin-gonic/gin"
)
@ -25,37 +29,36 @@ func (p PicRecognition) CheckParams(context *gin.Context) {
response.ErrorSystem(context, "PicRecognition表单参数验证器json化失败", "")
return
}
// TODO:linlnf
// pic := p.Pic
// // 先进行 URL 解码
pic := p.Pic
// 先进行 URL 解码
// decodedPic, err := url.QueryUnescape(pic)
// if err != nil {
// response.ErrorSystem(context, "PicRecognition表单参数验证器URL解码失败", "")
// return
// }
// // 再进行 Base64 解码
// decodedData, err := base64.StdEncoding.DecodeString(decodedPic)
// if err != nil {
// response.ErrorSystem(context, "PicRecognition表单参数验证器Base64解码失败", "")
// return
// }
// // 检查大小不超过 10M（10 * 1024 * 1024 字节）
// if len(decodedData) > 10*1024*1024 {
// response.ErrorSystem(context, "PicRecognition表单参数验证器图片大小超过 10M", "")
// return
// }
// // 将字节数据转换为图像以检查格式和尺寸
// img, _, err := image.Decode(bytes.NewReader(decodedData))
// if err != nil {
// return
// }
// bounds := img.Bounds()
// minSide := math.Min(float64(bounds.Dx()), float64(bounds.Dy()))
// maxSide := math.Max(float64(bounds.Dx()), float64(bounds.Dy()))
// // 最短边至少 15px，最长边最大 8192px
// if minSide < 15 || maxSide > 8192 {
// response.ErrorSystem(context, "PicRecognition表单参数验证器图片尺寸过大或过小", "")
// return
// }
// 再进行 Base64 解码
decodedData, err := base64.StdEncoding.DecodeString(pic)
if err != nil {
response.ErrorSystem(context, "PicRecognition表单参数验证器Base64解码失败", "")
return
}
// 检查大小不超过 10M（10 * 1024 * 1024 字节）
if len(decodedData) > 10*1024*1024 {
response.ErrorSystem(context, "PicRecognition表单参数验证器图片大小超过 10M", "")
return
}
// 将字节数据转换为图像以检查格式和尺寸
img, _, err := image.Decode(bytes.NewReader(decodedData))
if err != nil {
return
}
bounds := img.Bounds()
minSide := math.Min(float64(bounds.Dx()), float64(bounds.Dy()))
maxSide := math.Max(float64(bounds.Dx()), float64(bounds.Dy()))
// 最短边至少 15px，最长边最大 8192px
if minSide < 15 || maxSide > 8192 {
response.ErrorSystem(context, "PicRecognition表单参数验证器图片尺寸过大或过小", "")
return
}
(&web.AiRecognition{}).PicRecognition(extraAddBindDataContext)
}

43
GinSkeleton/app/http/validator/web/ai_recognition/voc_recognition.go
Unescape View File

@ -0,0 +1,43 @@
package ai_recognition
import (
"encoding/base64"
"goskeleton/app/global/consts"
"goskeleton/app/http/controller/web"
"goskeleton/app/http/validator/core/data_transfer"
"goskeleton/app/utils/response"
"github.com/gin-gonic/gin"
)
type VocRecognition struct {
Voc string `form:"voc" json:"voc" binding:"required"` // 必填、对于文本,表示它的长度>=1
}
func (v VocRecognition) CheckParams(context *gin.Context) {
if err := context.ShouldBind(&v); err != nil {
// 将表单参数验证器出现的错误直接交给错误翻译器统一处理即可
response.ValidatorError(context, err)
return
}
// 该函数主要是将本结构体的字段（成员）按照 consts.ValidatorPrefix+ json标签对应的 键 => 值 形式绑定在上下文，便于下一步（控制器）可以直接通过 context.Get(键) 获取相关值
extraAddBindDataContext := data_transfer.DataAddContext(v, consts.ValidatorPrefix, context)
if extraAddBindDataContext == nil {
response.ErrorSystem(context, "VocRecognition表单参数验证器json化失败", "")
return
}
voc := v.Voc
// 先进行 URL 解码
// decodedVoc, err := url.QueryUnescape(voc)
// if err != nil {
// response.ErrorSystem(context, "VocRecognition表单参数验证器URL解码失败", "")
// return
// }
// 再进行 Base64 解码
_, err := base64.StdEncoding.DecodeString(voc)
if err != nil {
response.ErrorSystem(context, "VocRecognition表单参数验证器Base64解码失败", "")
return
}
(&web.AiRecognition{}).VocRecognition(extraAddBindDataContext)
}

10
GinSkeleton/app/service/ai_model_cli/ocr_cli.go
Unescape View File

@ -4,7 +4,6 @@ import (
"goskeleton/app/global/my_errors"
"goskeleton/app/global/variable"
"goskeleton/app/utils/baidubce"
"goskeleton/app/utils/files"
"net/http"
"net/url"
"strings"
@ -19,8 +18,9 @@ import (
func RequestOCR(context *gin.Context) (r bool, c interface{}) {
path := "https://aip.baidubce.com/rest/2.0/ocr/v1/accurate_basic?access_token=" + baidubce.GetAccessToken()
// 获得图片数据，需要 base64 和urlencode处理
// linlnf: test
content, _ := files.GetFileBase64("storage/app/test/文字识别.png")
// TODO:linlnf
content := context.PostForm("pic")
// 组装 body 数据，参数在该网址：https://cloud.baidu.com/doc/OCR/s/1k3h7y3db
data := make(url.Values)
data.Set("detect_direction", "false")
@ -52,7 +52,7 @@ func RequestOCR(context *gin.Context) (r bool, c interface{}) {
return false, nil
}
return true, gin.H{
"words": decodeBody2Str(mbody),
"words": decodeOCRBody2Str(mbody),
}
}
@ -60,7 +60,7 @@ func RequestOCR(context *gin.Context) (r bool, c interface{}) {
* 将返回的内容中的所有识别出的文字信息进行组合
* @return 识别出的信息的内容字符串
*/
func decodeBody2Str(mbody map[string]interface{}) (str string) {
func decodeOCRBody2Str(mbody map[string]interface{}) (str string) {
words, ok := mbody["words_result"]
if ok {
sliceWords, ok := words.([]interface{})

89
GinSkeleton/app/service/ai_model_cli/vop_cli.go
Unescape View File

@ -0,0 +1,89 @@
package ai_model_cli
import (
"bytes"
"encoding/base64"
"encoding/json"
"goskeleton/app/global/my_errors"
"goskeleton/app/global/variable"
"goskeleton/app/utils/baidubce"
"net/http"
"github.com/gin-gonic/gin"
)
type BaiduBCEVOP struct {
Format string `json:"format"`
Rate int `json:"rate"`
Channel int `json:"channel"`
Cuid string `json:"cuid"`
DevPid int `json:"dev_pid"`
Speech string `json:"speech"`
Len int `json:"len"`
Token string `json:"token"`
}
/**
* 向百度智能云ocr(基础精准版)发送消息, 获得识别消息
* @return 识别出的信息的内容
*/
func RequestVOP(context *gin.Context) (r bool, c interface{}) {
path := "https://vop.baidu.com/pro_api"
// 获得语音数据，需要 base64 和urlencode处理
// TODO:linlnf
content := context.PostForm("voc")
// 再进行 Base64 解码
decodedVoc, _ := base64.StdEncoding.DecodeString(content)
// 组装 body 数据，参数在该网址：https://cloud.baidu.com/doc/SPEECH/s/4lbxdz34z
jsonData, _ := json.Marshal(BaiduBCEVOP{
Format: "pcm",
Rate: 16000,
Channel: 1,
Cuid: "kRxqyFGTgJOtBU4b5LnhWEvX6g8EU4Z7",
DevPid: 80001,
Speech: content,
Len: len(decodedVoc),
Token: baidubce.GetAccessToken(),
})
client := &http.Client{}
// 请求数据
req, err := http.NewRequest("POST", path, bytes.NewBuffer(jsonData))
if err != nil {
variable.ZapLog.Error(my_errors.ErrorBaiduCEUseVOPFail + err.Error())
return false, nil
}
req.Header.Add("Content-Type", "application/json")
req.Header.Add("Accept", "application/json")
res, err := client.Do(req)
if err != nil {
variable.ZapLog.Error(my_errors.ErrorBaiduCEUseVOPFail + err.Error())
return false, nil
}
defer res.Body.Close()
//从res中提取识别出的信息
mbody, err := baidubce.DecodeResBody(res, "vop")
if err != nil {
variable.ZapLog.Error(my_errors.ErrorBaiduCEUseVOPFail + err.Error())
return false, nil
}
return true, gin.H{
"words": decodeBody2Str(mbody),
}
}
/*
* 将返回的内容中的所有识别出的文字信息进行组合
* @return 识别出的信息的内容字符串
*/
func decodeBody2Str(mbody map[string]interface{}) (str string) {
words, ok := mbody["result"].([]interface{})
if ok {
for _, word := range words {
str += word.(string) + "\n"
}
return str
}
return ""
}

20
GinSkeleton/app/utils/baidubce/baidubce.go
Unescape View File

@ -51,7 +51,25 @@ func DecodeResBody(res *http.Response, flag string) (mbody map[string]interface{
return nil, err
}
mbody = jbody.(map[string]interface{})
err_str, _ := mbody["error_msg"].(string)
var err_str string
switch flag {
case "vop":
err_str1, _ := mbody["err_msg"].(string)
if err_str1 == "success." {
err_str = ""
} else {
err_str = err_str1
}
default:
err_str2, _ := mbody["error_msg"].(string)
if err_str2 == "" {
err_str = ""
} else {
err_str = err_str2
}
}
if err_str != "" {
variable.ZapLog.Error(my_errors.ErrorBaiduCEPostFail + flag + ", " + err_str)
err = errors.New(err_str)

6
GinSkeleton/app/utils/files/baseInfo.go
Unescape View File

@ -50,10 +50,10 @@ func GetFilesMimeByFp(fp multipart.File) string {
}
// 读取本地文件，进行Base64编码
func GetFileBase64(filePath string) (string, error) {
func GetFileBase64(filePath string) (int, string, error) {
data, err := os.ReadFile(filePath)
if err != nil {
return "", err
return 0, "", err
}
return base64.StdEncoding.EncodeToString(data), nil
return len(data), base64.StdEncoding.EncodeToString(data), nil
}

2
GinSkeleton/routers/web.go
Unescape View File

@ -67,6 +67,8 @@ func InitWebRouter_Co() *gin.Engine {
{
// 请求图片文字识别
aiRecognition.POST("pic_recognition", validatorFactory.Create(consts.ValidatorPrefix+"PicRecognition"))
// 请求语音识别
aiRecognition.POST("voc_recognition", validatorFactory.Create(consts.ValidatorPrefix+"VocRecognition"))
}
// 【需要token】中间件验证的路由

1
GinSkeleton/storage/app/test/A13_221.txt
View File

File diff suppressed because one or more lines are too long

BIN
GinSkeleton/storage/app/test/A13_221.wav
View File

Binary file not shown.

1
GinSkeleton/storage/app/test/文字识别.txt
View File

File diff suppressed because one or more lines are too long
Write Preview
Loading…
Cancel
Save