eta_api/services/wechat_platform.go

package services

import (
	"bytes"
	"encoding/json"
	"eta/eta_api/cache"
	"eta/eta_api/models"
	"eta/eta_api/models/rag"
	"eta/eta_api/services/elastic"
	"eta/eta_api/services/llm"
	"eta/eta_api/services/llm/facade"
	"eta/eta_api/utils"
	"eta/eta_api/utils/llm/eta_llm/eta_llm_http"
	"fmt"
	html2 "golang.org/x/net/html"
	"html"
	"os"
	"path"
	"regexp"
	"strconv"
	"strings"
	"time"
)

// AddWechatPlatform
// @Description: 添加新的公众号
// @param item
func AddWechatPlatform(item *rag.WechatPlatform) {
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("公众号入库后查找最新记录失败，err:%v", err)
		}
	}()
	if item.FakeId != `` {
		return
	}

	if item.ArticleLink == `` {
		return
	}

	articleLink := item.ArticleLink

	articleDetail, err := llm.SearchByWechatArticle(item.ArticleLink)
	if err != nil {
		return
	}

	if articleDetail.Appuin == `` {
		err = fmt.Errorf("文章内未匹配到公众号唯一标识")
		return
	}

	wechatPlatform := new(rag.WechatPlatform)
	// 查找是否存在这个公众号id的
	wechatPlatformInfo, tmpErr := wechatPlatform.GetByFakeID(articleDetail.Appuin)
	if tmpErr != nil && !utils.IsErrNoRow(tmpErr) {
		err = tmpErr
		return
	}
	if tmpErr == nil {
		// 如果找到了，那么需要将当前的给移除掉
		err = item.Del()
		if err != nil {
			return
		}

		// 并将查出来的微信公众号摘出来的数据重新赋值
		item = wechatPlatformInfo

	} else if utils.IsErrNoRow(tmpErr) {
		// 如果没找到，那么就变更当前的信息
		item.FakeId = articleDetail.Appuin
		item.Nickname = articleDetail.Nickname
		//item.Alias = req.Alias
		item.RoundHeadImg = articleDetail.RoundHeadImg
		//item.ServiceType = req.ServiceType
		item.Signature = articleDetail.ProfileSignature
		//item.Verified = verified
		item.ModifyTime = time.Now()

		err = item.Update([]string{rag.WechatPlatformColumns.FakeID, rag.WechatPlatformColumns.Nickname, rag.WechatPlatformColumns.RoundHeadImg, rag.WechatPlatformColumns.Signature, rag.WechatPlatformColumns.ModifyTime})
		if err != nil {
			return
		}

		// 修改公众号头像
		go replaceWechatPlatformPic(item)
	}

	// 把刚搜索的文章加入到文章库中
	AddWechatArticle(item, articleLink, articleDetail, nil)

	BeachAddWechatArticle(item, 10)
	fmt.Println("公众号入库完成")

	return
}

// AddWechatArticle
// @Description: 添加公众号文章入库
// @author: Roc
// @datetime 2025-03-05 13:24:14
// @param item *rag.WechatPlatform
// @param link string
// @param articleDetail WechatArticleDataResp
func AddWechatArticle(item *rag.WechatPlatform, articleLink string, articleDetail llm.WechatArticleDataResp, articleMenu *llm.ArticleMenu) {
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("公众号文章入库失败，文章链接：%s ，err:%v", articleLink, err)
		}
	}()
	obj := new(rag.WechatArticle)

	_, err = obj.GetByLink(articleLink)
	if err == nil {
		// 文章已经入库了，不需要重复入库
		return
	}

	// 如果不是 ErrNoRow 的时候，那么就是查询数据库出问题了，需要直接返回
	if !utils.IsErrNoRow(err) {
		return
	}

	// 这个时候，说明数据库中没有这个文章，那么需要文章入库
	err = nil

	var publishAt time.Time
	if articleDetail.CreateAt != `` {
		createAtInt, tmpErr := strconv.Atoi(articleDetail.CreateAt)
		if tmpErr == nil {
			publishAt = time.Unix(int64(createAtInt), 1000)
		}
	} else if articleMenu != nil {
		publishAt = time.Unix(int64(articleMenu.UpdateTime), 1000)
	}

	content := articleDetail.HtmlContent
	// 图片下载下来到本地，如果成功了，那么就用新的
	tmpContent, err := ReplaceHtmlImg(content)
	if tmpContent != `` {
		content = tmpContent
	}

	obj = &rag.WechatArticle{
		WechatArticleId:  0,
		WechatPlatformId: item.WechatPlatformId,
		FakeId:           item.FakeId,
		Title:            articleDetail.Title,
		Link:             articleLink,
		CoverUrl:         articleDetail.CoverUrl,
		Description:      articleDetail.Desc,
		Content:          html.EscapeString(content),
		TextContent:      articleDetail.TextContent,
		Country:          articleDetail.CountryName,
		Province:         articleDetail.ProvinceName,
		City:             articleDetail.CityName,
		//Abstract:          "",
		//ArticleCreateTime: createAt,
		ModifyTime: time.Now(),
		CreateTime: time.Now(),
	}
	if !publishAt.IsZero() {
		obj.ArticleCreateTime = publishAt
	}

	if articleMenu != nil {
		obj.Title = articleMenu.Title
		//obj.Link = articleMenu.Link
		obj.CoverUrl = articleMenu.Cover
		obj.Description = articleMenu.Digest
	}
	err = obj.Create()

	// 修改文章封面图
	go replaceWechatArticleCoverPic(obj)

	// 文章入库成功后，需要将相关信息入摘要库
	go cache.AddWechatArticleLlmOpToCache(obj.WechatArticleId, ``)

}

// BeachAddWechatArticle
// @Description: 批量添加公众号文章
// @param item
// @param num
// @return err
func BeachAddWechatArticle(item *rag.WechatPlatform, num int) {
	var err error
	defer func() {
		//fmt.Println("公众号文章批量入库完成")
		if err != nil {
			utils.FileLog.Error("公众号文章批量入库失败，err:%v", err)
			fmt.Println("公众号文章批量入库失败，err:", err)
		}
	}()
	if item.FakeId == `` {
		return
	}

	wechatArticleObj := new(rag.WechatArticle)

	// 获取公众号的文章列表
	articleListResp, err := llm.SearchByWechatArticleList(item.FakeId, num)
	if err != nil {
		return
	}
	for _, articleMenu := range articleListResp.List {
		// 判断文章是否已经入库，如果已经入库了，那么就过滤，不去重复查询微信了
		_, err = wechatArticleObj.GetByLink(articleMenu.Link)
		if err == nil {
			// 文章已经入库了，不需要重复入库
			continue
		}
		if !utils.IsErrNoRow(err) {
			return
		}
		err = nil

		articleDetail, tmpErr := llm.SearchByWechatArticle(articleMenu.Link)
		if tmpErr != nil {
			err = tmpErr
			return
		}

		// 把刚搜索的文章加入到指标库
		AddWechatArticle(item, articleMenu.Link, articleDetail, &articleMenu)

		//time.Sleep(10 * time.Second)

		// 随机休眠，至少大于10s
		sleepTimeInt := utils.GetRandInt(10, 20)
		if sleepTimeInt < 10 {
			sleepTimeInt = 10
		}
		time.Sleep(time.Duration(sleepTimeInt) * time.Second)
	}
	return
}

//
//// GenerateArticleAbstract
//// @Description: 文章摘要生成
//// @author: Roc
//// @datetime 2025-03-10 16:17:53
//// @param item *rag.WechatArticle
//func GenerateArticleAbstract(item *rag.WechatArticle, forceGenerate bool) {
//	var err error
//	defer func() {
//		if err != nil {
//			utils.FileLog.Error("文章转临时文件失败，err:%v", err)
//			fmt.Println("文章转临时文件失败，err:", err)
//		}
//	}()
//
//	// 内容为空，那就不需要生成摘要
//	if item.TextContent == `` {
//		return
//	}
//
//	abstractObj := rag.WechatArticleAbstract{}
//	tmpAbstractItem, err := abstractObj.GetByWechatArticleId(item.WechatArticleId)
//	// 如果找到了，同时不是强制生成，那么就直接处理到知识库中
//	if err == nil && !forceGenerate {
//		// 摘要已经生成，不需要重复生成，只需要重新加入到向量库中
//		WechatArticleAbstractToKnowledge(item, tmpAbstractItem, false)
//
//		return
//	}
//	if !utils.IsErrNoRow(err) {
//		return
//	}
//
//	//开始对话
//	abstract, addArticleChatRecordList, tmpErr := getAnswerByContent(item.WechatArticleId, utils.AI_ARTICLE_SOURCE_ETA_REPORT)
//	if tmpErr != nil {
//		err = fmt.Errorf("LLM对话失败，Err:" + tmpErr.Error())
//		return
//	}
//
//	// 添加问答记录
//	if len(addArticleChatRecordList) > 0 {
//		recordObj := rag.WechatArticleChatRecord{}
//		err = recordObj.CreateInBatches(addArticleChatRecordList)
//		if err != nil {
//			return
//		}
//	}
//
//	if abstract != `` {
//		if abstract == `sorry` || strings.Index(abstract, `根据已知信息无法回答该问题`) == 0 {
//			item.AbstractStatus = 2
//			item.ModifyTime = time.Now()
//			err = item.Update([]string{"AbstractStatus", "ModifyTime"})
//			return
//		}
//		item.AbstractStatus = 1
//		item.ModifyTime = time.Now()
//		err = item.Update([]string{"AbstractStatus", "ModifyTime"})
//
//		abstractItem := &rag.WechatArticleAbstract{
//			WechatArticleAbstractId: 0,
//			WechatArticleId:         item.WechatArticleId,
//			Content:                 abstract,
//			Version:                 0,
//			VectorKey:               "",
//			ModifyTime:              time.Now(),
//			CreateTime:              time.Now(),
//		}
//		err = abstractItem.Create()
//		if err != nil {
//			return
//		}
//
//		// 数据入ES库
//		go AddOrEditEsWechatArticleAbstract(abstractItem.WechatArticleAbstractId)
//
//		WechatArticleAbstractToKnowledge(item, abstractItem, false)
//	}
//}

// GenerateArticleAbstract
// @Description: 文章摘要生成（默认提示词批量生成）
// @author: Roc
// @datetime 2025-03-10 16:17:53
// @param item *rag.WechatArticle
func GenerateWechatArticleAbstract(item *rag.WechatArticle, forceGenerate bool) {
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("文章转临时文件失败，err:%v", err)
			fmt.Println("文章转临时文件失败，err:", err)
		}
	}()

	// 内容为空，那就不需要生成摘要
	if item.TextContent == `` {
		return
	}

	questionObj := rag.Question{}
	questionList, err := questionObj.GetListByCondition(``, ` AND is_default = 1 `, []interface{}{}, 0, 100)
	if err != nil {
		err = fmt.Errorf("获取问题列表失败，Err:" + err.Error())
		return
	}

	// 没问题就不生成了
	if len(questionList) <= 0 {
		return
	}

	for _, question := range questionList {
		GenerateWechatArticleAbstractByQuestion(item, question, forceGenerate)
	}

	return
}

// GenerateArticleAbstractByQuestion
// @Description: 文章摘要生成（根据提示词生成）
// @author: Roc
// @datetime 2025-03-10 16:17:53
// @param item *rag.WechatArticle
func GenerateWechatArticleAbstractByQuestion(item *rag.WechatArticle, question *rag.Question, forceGenerate bool) {
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("摘要生成失败，err:%v", err)
			fmt.Println("摘要生成失败，err:", err)
		}
	}()

	// 内容为空，那就不需要生成摘要
	if item.TextContent == `` {
		return
	}

	abstractObj := rag.WechatArticleAbstract{}
	abstractItem, err := abstractObj.GetByWechatArticleIdAndQuestionId(item.WechatArticleId, question.QuestionId)
	// 如果找到了，同时不是强制生成，那么就直接处理到知识库中
	if err == nil && !forceGenerate {
		// 摘要已经生成，不需要重复生成，只需要重新加入到向量库中
		WechatArticleAbstractToKnowledge(item, abstractItem, false)

		return
	}

	//你现在是一名资深的期货行业分析师，请基于以下的问题进行汇总总结，如果不能正常总结出来，那么就只需要回复我：sorry
	questionStr := fmt.Sprintf(`%s\n%s`, `你现在是一名资深的期货行业分析师，请基于以下的问题进行汇总总结，如果不能正常总结出来，那么就只需要回复我：sorry。以下是问题：`, question.QuestionContent)
	//开始对话
	abstract, industryTags, _, tmpErr := getAnswerByContent(item.WechatArticleId, utils.AI_ARTICLE_SOURCE_WECHAT, questionStr)
	if tmpErr != nil {
		err = fmt.Errorf("LLM对话失败，Err:" + tmpErr.Error())
		return
	}

	if abstract == `` {
		return
	}

	var tagIdJsonStr string
	// 标签ID
	{
		tagIdList := make([]int, 0)
		tagIdMap := make(map[int]bool)

		if abstractItem != nil && abstractItem.Tags != `` {
			tmpErr = json.Unmarshal([]byte(abstractItem.Tags), &tagIdList)
			if tmpErr != nil {
				utils.FileLog.Info(fmt.Sprintf("json.Unmarshal 失败，标签数据：%s，Err:%s", abstractItem.Tags, tmpErr.Error()))
			} else {
				for _, tagId := range tagIdList {
					tagIdMap[tagId] = true
				}
			}
		}
		for _, tagName := range industryTags {
			tagId, tmpErr := GetTagIdByName(tagName)
			if tmpErr != nil {
				utils.FileLog.Info(fmt.Sprintf("获取标签ID失败，标签名称：%s，Err:%s", tagName, tmpErr.Error()))
			}
			if _, ok := tagIdMap[tagId]; !ok {
				tagIdList = append(tagIdList, tagId)
				tagIdMap[tagId] = true
			}
		}
		//for _, tagName := range varietyTags {
		//	tagId, tmpErr := GetTagIdByName(tagName)
		//	if tmpErr != nil {
		//		utils.FileLog.Info(fmt.Sprintf("获取标签ID失败，标签名称：%s，Err:%s", tagName, tmpErr.Error()))
		//	}
		//	if _, ok := tagIdMap[tagId]; !ok {
		//		tagIdList = append(tagIdList, tagId)
		//		tagIdMap[tagId] = true
		//	}
		//}

		tagIdJsonByte, err := json.Marshal(tagIdList)
		if err != nil {
			utils.FileLog.Info(fmt.Sprintf("标签ID序列化失败，Err:%s", tmpErr.Error()))
		} else {
			tagIdJsonStr = string(tagIdJsonByte)
		}
	}

	if abstract == `sorry` || strings.Index(abstract, `根据已知信息无法回答该问题`) == 0 {
		item.AbstractStatus = 2
		item.ModifyTime = time.Now()
		err = item.Update([]string{"AbstractStatus", "ModifyTime"})
		return
	}
	item.AbstractStatus = 1
	item.ModifyTime = time.Now()
	err = item.Update([]string{"AbstractStatus", "ModifyTime"})

	if abstractItem == nil || abstractItem.WechatArticleAbstractId <= 0 {
		abstractItem = &rag.WechatArticleAbstract{
			WechatArticleAbstractId: 0,
			WechatArticleId:         item.WechatArticleId,
			Content:                 abstract,
			Version:                 1,
			VectorKey:               "",
			ModifyTime:              time.Now(),
			CreateTime:              time.Now(),
			QuestionId:              question.QuestionId,
			Tags:                    tagIdJsonStr,
			QuestionContent:         question.QuestionContent,
		}
		err = abstractItem.Create()
	} else {
		// 添加历史记录
		rag.AddArticleAbstractHistoryByWechatArticleAbstract(abstractItem)

		abstractItem.Content = abstract
		abstractItem.Version++
		abstractItem.ModifyTime = time.Now()
		abstractItem.Tags = ""
		abstractItem.QuestionContent = question.QuestionContent
		err = abstractItem.Update([]string{"content", "version", "modify_time", "tags", "question_content"})
	}

	if err != nil {
		return
	}

	// 数据入ES库
	go AddOrEditEsWechatArticleAbstract(abstractItem.WechatArticleAbstractId)

	WechatArticleAbstractToKnowledge(item, abstractItem, false)
}

// DelDoc
// @Description: 删除摘要向量库
// @author: Roc
// @datetime 2025-03-12 16:55:05
// @param wechatArticleAbstractList []*rag.WechatArticleAbstract
// @return err error
func DelDoc(wechatArticleAbstractList []*rag.WechatArticleAbstract) (err error) {
	defer func() {
		if err != nil {
			utils.FileLog.Error("删除摘要向量库文件失败，err:%v", err)
			fmt.Println("删除摘要向量库文件失败，err:", err)
		}
	}()

	vectorKeyList := make([]string, 0)
	wechatArticleAbstractIdList := make([]int, 0)

	for _, v := range wechatArticleAbstractList {
		if v.VectorKey == `` {
			continue
		}
		vectorKeyList = append(vectorKeyList, v.VectorKey)
		wechatArticleAbstractIdList = append(wechatArticleAbstractIdList, v.WechatArticleAbstractId)
	}

	// 没有就不删除
	if len(vectorKeyList) <= 0 {
		return
	}

	_, err = llm.DelDocsToKnowledge(models.BusinessConfMap[models.KnowledgeBaseName], vectorKeyList)
	if err != nil {
		err = fmt.Errorf("删除LLM摘要向量库文件失败，Err:" + err.Error())
		return
	}
	//fmt.Println(resp)
	obj := rag.WechatArticleAbstract{}
	err = obj.DelVectorKey(wechatArticleAbstractIdList)

	return
}

// DelLlmDoc
// @Description: 删除摘要向量库
// @author: Roc
// @datetime 2025-03-12 16:55:05
// @param wechatArticleAbstractList []*rag.WechatArticleAbstract
// @return err error
func DelLlmDoc(vectorKeyList []string, wechatArticleAbstractIdList []int) (err error) {
	defer func() {
		if err != nil {
			utils.FileLog.Error("删除摘要向量库文件失败，err:%v", err)
			fmt.Println("删除摘要向量库文件失败，err:", err)
		}
	}()

	// 没有就不删除
	if len(vectorKeyList) <= 0 {
		return
	}

	_, err = llm.DelDocsToKnowledge(models.BusinessConfMap[models.KnowledgeBaseName], vectorKeyList)
	if err != nil {
		err = fmt.Errorf("删除LLM摘要向量库文件失败，Err:" + err.Error())
		return
	}
	//fmt.Println(resp)
	obj := rag.WechatArticleAbstract{}
	err = obj.DelVectorKey(wechatArticleAbstractIdList)

	return
}

func getAnswerByContent(articleId int, source int, questionStr string) (answer string, industryTags, varietyTags []string, err error) {
	//addArticleChatRecordList = make([]*rag.WechatArticleChatRecord, 0)

	result, err := facade.AIGCBaseOnPromote(facade.AIGC{
		Promote:   questionStr,
		Source:    source,
		ArticleId: articleId,
		LLMModel:  `deepseek-r1:32b`,
	})
	if err != nil {
		return
	}

	// JSON字符串转字节
	//answerByte, err := json.Marshal(result)
	//if err != nil {
	//	return
	//}
	//originalAnswer := string(answerByte)

	// 提取 </think> 后面的内容
	thinkEndIndex := strings.Index(result.Answer, "</think>")
	if thinkEndIndex != -1 {
		answer = strings.TrimSpace(result.Answer[thinkEndIndex+len("</think>"):])
	} else {
		answer = result.Answer
	}

	answer = strings.TrimSpace(answer)

	// 提取标签
	industryTags, varietyTags = extractLabels(answer)

	//// 待入库的数据
	//addArticleChatRecordList = append(addArticleChatRecordList, &rag.WechatArticleChatRecord{
	//	WechatArticleChatRecordId: 0,
	//	WechatArticleId:           articleId,
	//	ChatUserType:              "user",
	//	Content:                   questionStr,
	//	SendTime:                  time.Now(),
	//	CreatedTime:               time.Now(),
	//	UpdateTime:                time.Now(),
	//}, &rag.WechatArticleChatRecord{
	//	WechatArticleChatRecordId: 0,
	//	WechatArticleId:           articleId,
	//	ChatUserType:              "assistant",
	//	Content:                   originalAnswer,
	//	SendTime:                  time.Now(),
	//	CreatedTime:               time.Now(),
	//	UpdateTime:                time.Now(),
	//})

	return
}

func getAnswerByContentBak(wechatArticleId int, docId string) (answer string, addArticleChatRecordList []*rag.WechatArticleChatRecord, err error) {
	historyList := make([]eta_llm_http.HistoryContent, 0)
	addArticleChatRecordList = make([]*rag.WechatArticleChatRecord, 0)

	questionObj := rag.Question{}
	questionList, err := questionObj.GetListByCondition(``, ``, []interface{}{}, 0, 100)
	if err != nil {
		err = fmt.Errorf("获取问题列表失败，Err:" + err.Error())
		return
	}

	// 没问题就不生成了
	if len(questionList) <= 0 {
		return
	}

	//你现在是一名资深的期货行业分析师，请基于以下的问题进行汇总总结，如果不能正常总结出来，那么就只需要回复我：sorry
	questionStrList := []string{`你现在是一名资深的期货行业分析师，请基于以下的问题进行汇总总结，如果不能正常总结出来，那么就只需要回复我：sorry。以下是问题：`}
	for _, v := range questionList {
		questionStrList = append(questionStrList, v.QuestionContent)
	}
	questionStr := strings.Join(questionStrList, "\n")

	originalAnswer, result, err := llm.ChatByFile(docId, questionStr, historyList)
	fmt.Println(result)
	if err != nil {
		err = fmt.Errorf("LLM对话失败，Err:" + err.Error())
		return
	}

	// 提取 </think> 后面的内容
	thinkEndIndex := strings.Index(result.Answer, "</think>")
	if thinkEndIndex != -1 {
		answer = strings.TrimSpace(result.Answer[thinkEndIndex+len("</think>"):])
	} else {
		answer = result.Answer
	}

	answer = strings.TrimSpace(answer)

	// 待入库的数据
	addArticleChatRecordList = append(addArticleChatRecordList, &rag.WechatArticleChatRecord{
		WechatArticleChatRecordId: 0,
		WechatArticleId:           wechatArticleId,
		ChatUserType:              "user",
		Content:                   questionStr,
		SendTime:                  time.Now(),
		CreatedTime:               time.Now(),
		UpdateTime:                time.Now(),
	}, &rag.WechatArticleChatRecord{
		WechatArticleChatRecordId: 0,
		WechatArticleId:           wechatArticleId,
		ChatUserType:              "assistant",
		Content:                   originalAnswer,
		SendTime:                  time.Now(),
		CreatedTime:               time.Now(),
		UpdateTime:                time.Now(),
	})

	return
}

// ArticleToKnowledge
// @Description: 原文入向量库
// @author: Roc
// @datetime 2025-03-10 16:13:16
// @param item *rag.WechatArticle
func ArticleToKnowledge(item *rag.WechatArticle) {
	if item.TextContent == `` {
		return
	}
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("上传文章原文到知识库失败，err:%v", err)
			fmt.Println("上传文章原文到知识库失败，err:", err)
		}
	}()

	// 生成临时文件
	//dateDir := time.Now().Format("20060102")
	//uploadDir :=   "./static/ai/article/" + dateDir
	uploadDir := "./static/ai/article"
	err = os.MkdirAll(uploadDir, utils.DIR_MOD)
	if err != nil {
		err = fmt.Errorf("存储目录创建失败,Err:" + err.Error())
		return
	}
	//fileName := utils.RemoveSpecialChars(item.Title) + `.md`
	fileName := utils.MD5(item.Title) + `.md`
	tmpFilePath := uploadDir + "/" + fileName
	err = utils.SaveToFile(item.TextContent, tmpFilePath)
	if err != nil {
		err = fmt.Errorf("生成临时文件失败，Err:" + err.Error())
		return
	}
	defer func() {
		os.Remove(tmpFilePath)
	}()

	knowledgeArticleName := models.BusinessConfMap[models.KnowledgeArticleName]
	// 上传临时文件到LLM
	uploadFileResp, err := llm.UploadDocsToKnowledge(tmpFilePath, knowledgeArticleName)
	if err != nil {
		err = fmt.Errorf("上传文章原文到知识库失败，Err:" + err.Error())
		return
	}

	if len(uploadFileResp.FailedFiles) > 0 {
		for _, v := range uploadFileResp.FailedFiles {
			err = fmt.Errorf("上传文章原文到知识库失败，Err:" + v)
		}
	}

	item.VectorKey = tmpFilePath
	item.ModifyTime = time.Now()
	err = item.Update([]string{"vector_key", "modify_time"})

}

// WechatArticleAbstractToKnowledge
// @Description: 摘要入向量库
// @author: Roc
// @datetime 2025-03-10 16:14:59
// @param wechatArticleItem *rag.WechatArticle
// @param abstractItem *rag.WechatArticleAbstract
func WechatArticleAbstractToKnowledge(wechatArticleItem *rag.WechatArticle, abstractItem *rag.WechatArticleAbstract, isReUpload bool) {
	if abstractItem.Content == `` {
		return
	}
	// 已经生成了，那就不处理了
	if abstractItem.VectorKey != `` && !isReUpload {
		return
	}
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("摘要入向量库失败，err:%v", err)
			fmt.Println("摘要入向量库失败，err:", err)
		}

		// 数据入ES库
		go AddOrEditEsWechatArticleAbstract(abstractItem.WechatArticleAbstractId)
	}()

	// 生成临时文件
	//dateDir := time.Now().Format("20060102")
	//uploadDir :=  + "./static/ai/article/" + dateDir
	uploadDir := "./static/ai/abstract"
	err = os.MkdirAll(uploadDir, utils.DIR_MOD)
	if err != nil {
		err = fmt.Errorf("存储目录创建失败,Err:" + err.Error())
		return
	}
	fileName := utils.MD5(fmt.Sprintf("%d_%d", utils.AI_ARTICLE_SOURCE_WECHAT, wechatArticleItem.WechatArticleId)) + `.md`
	tmpFilePath := uploadDir + "/" + fileName
	err = utils.SaveToFile(abstractItem.Content, tmpFilePath)
	if err != nil {
		err = fmt.Errorf("生成临时文件失败，Err:" + err.Error())
		return
	}
	defer func() {
		os.Remove(tmpFilePath)
	}()

	knowledgeArticleName := models.BusinessConfMap[models.KnowledgeBaseName]
	// 上传临时文件到LLM
	uploadFileResp, err := llm.UploadDocsToKnowledge(tmpFilePath, knowledgeArticleName)
	if err != nil {
		err = fmt.Errorf("上传文章原文到知识库失败，Err:" + err.Error())
		return
	}

	if len(uploadFileResp.FailedFiles) > 0 {
		for _, v := range uploadFileResp.FailedFiles {
			err = fmt.Errorf("上传文章原文到知识库失败，Err:" + v)
		}
	}

	abstractItem.VectorKey = tmpFilePath
	abstractItem.ModifyTime = time.Now()
	err = abstractItem.Update([]string{"vector_key", "modify_time"})

}

// replaceWechatPlatformPic
// @Description: 替换公众号头像
// @author: Roc
// @datetime 2025-03-11 09:38:24
// @param item *rag.WechatPlatform
func replaceWechatPlatformPic(item *rag.WechatPlatform) {
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("替换公众号头像失败，err:%v", err)
			fmt.Println("替换公众号头像失败，err:", err)
		}
	}()
	if item.RoundHeadImg == `` {
		return
	}
	resourceUrl, err := downloadWxPicAndUploadToOss(item.RoundHeadImg, `head_img`)
	if err != nil {
		return
	}
	item.RoundHeadImg = resourceUrl
	err = item.Update([]string{"round_head_img"})

}

// replaceWechatArticleCoverPic
// @Description: 替换文章封面图
// @author: Roc
// @datetime 2025-03-11 09:38:35
// @param item *rag.WechatArticle
func replaceWechatArticleCoverPic(item *rag.WechatArticle) {
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("替换公众号头像失败，err:%v", err)
			fmt.Println("替换公众号头像失败，err:", err)
		}

		// 数据入ES库
		AddOrEditEsWechatArticle(item.WechatArticleId)
	}()
	if item.CoverUrl == `` {
		return
	}
	resourceUrl, err := downloadWxPicAndUploadToOss(item.CoverUrl, `cover_url`)
	if err != nil {
		return
	}
	item.CoverUrl = resourceUrl
	err = item.Update([]string{"cover_url"})

}

// replaceWechatArticlePic
// @Description: 替换文章内容图
// @author: Roc
// @datetime 2025-03-11 09:38:35
// @param item *rag.WechatArticle
func ReplaceWechatArticlePic(item *rag.WechatArticle) {
	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("替换公众号头像失败，err:%v", err)
			fmt.Println("替换公众号头像失败，err:", err)
		}
	}()
	if item.Content == `` {
		return
	}

	content, err := ReplaceHtmlImg(html.UnescapeString(item.Content))
	if err != nil {
		return
	}
	item.Content = html.EscapeString(content)
	err = item.Update([]string{"content"})

	return
}

// downloadWxPicAndUploadToOss
// @Description: 下载微信图片并上传到OSS
// @author: Roc
// @datetime 2025-03-11 09:28:49
// @param wxPicUrl string
// @return resourceUrl string
// @return err error
func downloadWxPicAndUploadToOss(wxPicUrl, source string) (resourceUrl string, err error) {
	localFilePath, err := utils.DownloadWxImage(wxPicUrl)
	if err != nil {
		return
	}
	defer func() {
		os.Remove(localFilePath)
	}()
	ossClient := NewOssClient()
	if ossClient == nil {
		err = fmt.Errorf(`初始化OSS服务失败`)
		return
	}
	ext := path.Ext(localFilePath)
	fileName := fmt.Sprintf(`%s%s%s`, time.Now().Format(utils.FormatShortDateTimeUnSpace), utils.GetRandStringNoSpecialChar(16), ext)
	//savePath := utils.UploadDir + `wx/wx_article/` + time.Now().Format("200601/20060102/") + fileName
	savePath := fmt.Sprintf(`%swx/%s/%s%s`, utils.UploadDir, source, time.Now().Format("200601/20060102/"), fileName)
	resourceUrl, err = ossClient.UploadFile(fileName, localFilePath, savePath)
	if err != nil {
		err = fmt.Errorf("文件上传失败，Err:" + err.Error())
		return
	}

	return

}

// ReplaceHtmlImg
// @Description: 将html中的图片替换成自己的
// @author: Roc
// @datetime 2025-03-11 14:32:00
// @param htmlStr string
// @return newHtml string
// @return err error
func ReplaceHtmlImg(htmlStr string) (newHtml string, err error) {
	doc, err := html2.Parse(strings.NewReader(htmlStr))
	if err != nil {
		return
	}
	if err != nil {
		return
	}
	handleNode(doc)

	// 将处理后的HTML节点重新渲染为HTML字符串
	var buf bytes.Buffer
	if err = html2.Render(&buf, doc); err != nil {
		fmt.Println(err)
		return
	}
	newHtml = buf.String()

	return
}

// handleNode
// @Description: html节点处理
// @author: Roc
// @datetime 2025-03-11 14:32:45
// @param n *html2.Node
func handleNode(n *html2.Node) {
	if n.Type == html2.ElementNode {
		if n.Data == "img" {
			for k, attr := range n.Attr {
				// 新增代码：如果标签是img且存在data-src属性，则将data-src的值赋给src
				if n.Data == "img" && attr.Key == "src" {
					resourceUrl, tmpErr := downloadWxPicAndUploadToOss(attr.Val, `article`)
					if tmpErr != nil {
						continue
					}
					attr.Val = resourceUrl
				}
				n.Attr[k] = attr
			}
		}

	}
	for c := n.FirstChild; c != nil; c = c.NextSibling {
		handleNode(c)
	}
}

// AddOrEditEsWechatPlatformId
// @Description: 批量处理某个公众号下的文章到ES
// @author: Roc
// @datetime 2025-03-13 11:01:28
// @param articleId int
func AddOrEditEsWechatPlatformId(wechatPlatformId int) {
	if utils.EsWechatArticleName == `` {
		return
	}
	obj := rag.WechatArticle{}
	list, _ := obj.GetListByCondition(` wechat_article_id `, ` AND wechat_platform_id = ? `, []interface{}{wechatPlatformId}, 0, 1000000)
	for _, item := range list {
		AddOrEditEsWechatArticle(item.WechatArticleId)
	}
}

// AddOrEditEsWechatArticle
// @Description: 新增/编辑微信文章入ES
// @author: Roc
// @datetime 2025-03-13 11:01:28
// @param articleId int
func AddOrEditEsWechatArticle(articleId int) {
	if utils.EsWechatArticleName == `` {
		return
	}

	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("添加公众号微信信息到ES失败，err:%v", err)
			fmt.Println("添加公众号微信信息到ES失败，err:", err)
		}
	}()
	obj := rag.WechatArticle{}
	articleInfo, err := obj.GetById(articleId)
	if err != nil {
		err = fmt.Errorf("获取公众号文章信息失败，Err:" + err.Error())
		return
	}
	platformObj := rag.WechatPlatform{}
	platformInfo, err := platformObj.GetById(articleInfo.WechatPlatformId)
	if err != nil {
		err = fmt.Errorf("获取公众号平台信息失败，Err:" + err.Error())
		return
	}

	esItem := elastic.WechatArticleAndPlatform{
		WechatArticleId:  articleInfo.WechatArticleId,
		WechatPlatformId: articleInfo.WechatPlatformId,
		FakeId:           articleInfo.FakeId,
		Title:            articleInfo.Title,
		Link:             articleInfo.Link,
		CoverUrl:         articleInfo.CoverUrl,
		Description:      articleInfo.Description,
		//Content:          articleInfo.Content,
		//TextContent: articleInfo.TextContent,
		//AbstractStatus:          articleInfo.AbstractStatus,
		Country:           articleInfo.Country,
		Province:          articleInfo.Province,
		City:              articleInfo.City,
		ArticleCreateTime: articleInfo.ArticleCreateTime,
		IsDeleted:         articleInfo.IsDeleted,
		ModifyTime:        articleInfo.ModifyTime,
		CreateTime:        articleInfo.CreateTime,
		Nickname:          platformInfo.Nickname,
		Alias:             platformInfo.Alias,
		RoundHeadImg:      platformInfo.RoundHeadImg,
	}

	err = elastic.WechatArticleEsAddOrEdit(strconv.Itoa(articleInfo.WechatArticleId), esItem)
}

// AddOrEditEsWechatArticleAbstract
// @Description: 新增/编辑微信文章摘要入ES
// @author: Roc
// @datetime 2025-03-13 14:13:47
// @param articleAbstractId int
func AddOrEditEsWechatArticleAbstract(articleAbstractId int) {
	if utils.EsWechatArticleAbstractName == `` {
		return
	}

	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("添加公众号微信信息到ES失败，err:%v", err)
			fmt.Println("添加公众号微信信息到ES失败，err:", err)
		}
	}()
	obj := rag.WechatArticleAbstract{}
	abstractInfo, err := obj.GetById(articleAbstractId)
	if err != nil {
		err = fmt.Errorf("获取公众号文章信息失败，Err:" + err.Error())
		return
	}
	articleObj := rag.WechatArticle{}
	articleInfo, err := articleObj.GetById(abstractInfo.WechatArticleId)
	if err != nil {
		err = fmt.Errorf("获取公众号文章信息失败，Err:" + err.Error())
		return
	}

	// 标签ID
	tagIdList := make([]int, 0)
	if abstractInfo.Tags != `` {
		err = json.Unmarshal([]byte(abstractInfo.Tags), &tagIdList)
		if err != nil {
			err = fmt.Errorf("报告标签ID转int失败，Err:" + err.Error())
			utils.FileLog.Info(fmt.Sprintf("json.Unmarshal 报告标签ID转int失败，标签数据：%s，Err:%s", abstractInfo.Tags, err.Error()))
		}
	}

	esItem := elastic.WechatArticleAbstractItem{
		WechatArticleAbstractId: abstractInfo.WechatArticleAbstractId,
		WechatArticleId:         abstractInfo.WechatArticleId,
		WechatPlatformId:        articleInfo.WechatPlatformId,
		Abstract:                abstractInfo.Content,
		QuestionId:              abstractInfo.QuestionId,
		Version:                 abstractInfo.Version,
		VectorKey:               abstractInfo.VectorKey,
		ModifyTime:              articleInfo.ModifyTime,
		CreateTime:              articleInfo.CreateTime,
		Title:                   articleInfo.Title,
		Link:                    articleInfo.Link,
		TagIdList:               tagIdList,
	}

	err = elastic.WechatArticleAbstractEsAddOrEdit(strconv.Itoa(articleAbstractId), esItem)
}

// DelEsWechatArticleAbstract
// @Description: 删除ES中的微信文章摘要
// @author: Roc
// @datetime 2025-03-13 14:13:47
// @param articleAbstractId int
func DelEsWechatArticleAbstract(articleAbstractId int) {
	if utils.EsWechatArticleAbstractName == `` {
		return
	}

	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("删除公众号微信信息到ES失败，err:%v", err)
			fmt.Println("删除公众号微信信息到ES失败，err:", err)
		}
	}()

	err = elastic.WechatArticleAbstractEsDel(strconv.Itoa(articleAbstractId))
}

// AddOrEditEsRagQuestion
// @Description: 新增/编辑知识库问题入ES
// @author: Roc
// @datetime 2025-03-28 11:25:50
// @param questionId int
func AddOrEditEsRagQuestion(questionId int) {
	if utils.EsWechatArticleName == `` {
		return
	}

	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("添加公众号微信信息到ES失败，err:%v", err)
			fmt.Println("添加公众号微信信息到ES失败，err:", err)
		}
	}()
	obj := rag.Question{}
	questionInfo, err := obj.GetByID(questionId)
	if err != nil {
		err = fmt.Errorf("获取公众号文章信息失败，Err:" + err.Error())
		return
	}

	esItem := elastic.RagQuestionItem{
		QuestionId:      questionInfo.QuestionId,
		QuestionTitle:   questionInfo.QuestionTitle,
		QuestionContent: questionInfo.QuestionContent,
		Sort:            questionInfo.Sort,
		SysUserId:       questionInfo.SysUserId,
		SysUserRealName: questionInfo.SysUserRealName,
		ModifyTime:      questionInfo.ModifyTime,
		CreateTime:      questionInfo.CreateTime,
	}

	err = elastic.RagQuestionEsAddOrEdit(strconv.Itoa(questionInfo.QuestionId), esItem)
}

// DelEsRagQuestion
// @Description: 删除ES中的知识库问题
// @author: Roc
// @datetime 2025-03-28 11:26:40
// @param questionId int
func DelEsRagQuestion(questionId int) {
	if utils.EsWechatArticleAbstractName == `` {
		return
	}

	var err error
	defer func() {
		if err != nil {
			utils.FileLog.Error("添加公众号微信信息到ES失败，err:%v", err)
			fmt.Println("添加公众号微信信息到ES失败，err:", err)
		}
	}()

	err = elastic.RagQuestionEsDel(strconv.Itoa(questionId))
}

// extractLabels
// @Description: 提取摘要中的标签
// @author: Roc
// @datetime 2025-04-18 17:16:05
// @param text string
// @return industryTags []string
// @return varietyTags []string
func extractLabels(text string) (industryTags []string, varietyTags []string) {
	reIndustry := regexp.MustCompile(`行业标签((?:【[^】]*】)+)`)
	industryMatch := reIndustry.FindStringSubmatch(text)
	if len(industryMatch) > 1 {
		industryContent := industryMatch[1]
		reSplit := regexp.MustCompile(`【([^】]*)】`)
		industryTags = make([]string, 0)
		for _, m := range reSplit.FindAllStringSubmatch(industryContent, -1) {
			if len(m) > 1 {
				industryTags = append(industryTags, m[1])
			}
		}
	}

	reVariety := regexp.MustCompile(`品种标签((?:【[^】]*】)+)`)
	varietyMatch := reVariety.FindStringSubmatch(text)
	if len(varietyMatch) > 1 {
		varietyContent := varietyMatch[1]
		reSplit := regexp.MustCompile(`【([^】]*)】`)
		varietyTags = make([]string, 0)
		for _, m := range reSplit.FindAllStringSubmatch(varietyContent, -1) {
			if len(m) > 1 {
				varietyTags = append(varietyTags, m[1])
			}
		}
	}
	return
}

var aiAbstractTagMap = map[string]int{}

// GetTagIdByName
// @Description: 获取标签ID
// @author: Roc
// @datetime 2025-04-18 17:25:46
// @param tagName string
// @return tagId int
// @return err error
func GetTagIdByName(tagName string) (tagId int, err error) {
	tagName = strings.TrimSpace(tagName)
	tagId, ok := aiAbstractTagMap[tagName]
	if ok {
		return
	}

	obj := rag.Tag{}
	item, err := obj.GetByCondition(fmt.Sprintf(` AND  %s = ? `, rag.TagColumns.TagName), []interface{}{tagName})
	if err != nil {
		if !utils.IsErrNoRow(err) {
			err = fmt.Errorf("获取标签失败，Err:" + err.Error())
			return
		}

		item = &rag.Tag{
			TagId:      0,
			TagName:    tagName,
			Sort:       0,
			ModifyTime: time.Now(),
			CreateTime: time.Now(),
		}
		err = item.Create()
		if err != nil {
			err = fmt.Errorf("添加标签失败，Err:" + err.Error())
			return
		}
	}

	tagId = item.TagId
	aiAbstractTagMap[tagName] = tagId

	return
}