eta_api/services/semantic_analysis.go

package services

import (
	saModel "eta_gn/eta_api/models/semantic_analysis"
	"eta_gn/eta_api/services/alarm_msg"
	"eta_gn/eta_api/utils"
	"fmt"
	"github.com/PuerkitoBio/goquery"
	"html"
	"sort"
	"strings"
)

// LoadSaDocContent2Section 读取文档内容为段落, 以<p>标签划分
func LoadSaDocContent2Section(content string) (sections []string, err error) {
	if content == `` {
		return
	}
	defer func() {
		if err != nil {
			fmt.Println(err.Error())
		}
	}()

	doc, e := goquery.NewDocumentFromReader(strings.NewReader(content))
	if e != nil {
		err = fmt.Errorf("读取html内容失败, Err: %s", e.Error())
		return
	}
	doc.Find("p").Each(func(i int, s *goquery.Selection) {
		h, e := s.Html()
		if e != nil {
			err = fmt.Errorf("读取html标签失败, Err: %s", e.Error())
			return
		}
		t := strings.TrimSpace(s.Text())
		if t != "" {
			sections = append(sections, fmt.Sprintf(`<p>%s</p>`, h))
		}
	})
	return
}

// GetSaCompareTableData 获取文档比对表格数据
func GetSaCompareTableData(labelIds, docIds, secIds []int, contentMap map[string]int) (resp *saModel.SaCompareSaveResp, err error) {
	resp = new(saModel.SaCompareSaveResp)

	// 文档列表
	docMap := make(map[int]*saModel.SaDoc)
	if len(docIds) > 0 {
		docOB := new(saModel.SaDoc)
		docCond := fmt.Sprintf(` AND %s IN (%s)`, saModel.SaDocColumns.SaDocId, utils.GetOrmInReplace(len(docIds)))
		docPars := make([]interface{}, 0)
		docPars = append(docPars, docIds)
		docFields := []string{saModel.SaDocColumns.SaDocId, saModel.SaDocColumns.Title, saModel.SaDocColumns.Theme}
		docItems, e := docOB.GetItemsByCondition(docCond, docPars, docFields, "")
		if e != nil {
			err = fmt.Errorf("获取比对文档列表失败, Err: %s", e.Error())
			return
		}
		for _, d := range docItems {
			docMap[d.SaDocId] = d
		}
	}

	// 段落map
	secMap := make(map[int]string)
	if len(secIds) > 0 {
		secOB := new(saModel.SaDocSection)
		secCond := fmt.Sprintf(` AND %s IN (%s)`, saModel.SaDocSectionColumns.SaDocSectionId, utils.GetOrmInReplace(len(secIds)))
		secPars := make([]interface{}, 0)
		secPars = append(secPars, secIds)
		secFields := []string{saModel.SaDocSectionColumns.SaDocSectionId, saModel.SaDocSectionColumns.DocId, saModel.SaDocSectionColumns.Content}
		secItems, e := secOB.GetItemsByCondition(secCond, secPars, secFields, "")
		if e != nil {
			err = fmt.Errorf("获取比对文档列表失败, Err: %s", e.Error())
			return
		}
		for _, s := range secItems {
			secMap[s.SaDocSectionId] = html.UnescapeString(s.Content)
		}
	}

	// 标签map
	labelMap := make(map[int]string)
	labelOB := new(saModel.SaLabel)
	labelCond := ``
	labelPars := make([]interface{}, 0)
	labelFields := []string{saModel.SaLabelColumns.SaLabelId, saModel.SaLabelColumns.LabelName}
	labelItems, e := labelOB.GetItemsByCondition(labelCond, labelPars, labelFields, "")
	if e != nil {
		err = fmt.Errorf("获取标签列表失败, Err: %s", e.Error())
		return
	}
	for _, l := range labelItems {
		labelMap[l.SaLabelId] = l.LabelName
	}

	// 表头信息, 按照docIds的顺序不然会乱
	for _, d := range docIds {
		dv := docMap[d]
		if dv != nil {
			resp.TitleList = append(resp.TitleList, dv.Title)
			//resp.ThemeList = append(resp.ThemeList, dv.Theme)
		}
	}

	// 标签列表
	respLabel := make([]*saModel.SaCompareSaveRespLabel, 0)
	for _, l := range labelIds {
		v := new(saModel.SaCompareSaveRespLabel)
		v.SaLabelId = l
		v.LabelName = labelMap[l]
		// 文档列表
		docList := make([]*saModel.SaCompareSaveRespDoc, 0)
		for _, d := range docIds {
			vd := new(saModel.SaCompareSaveRespDoc)
			vd.SaDocId = d
			dv := docMap[d]
			if dv != nil {
				vd.Title = dv.Title
			}
			secList := make([]*saModel.SaCompareSaveRespSection, 0)
			// 段落列表
			for _, s := range secIds {
				k := fmt.Sprintf("%d-%d-%d", l, d, s)
				if contentMap[k] == 0 {
					continue
				}
				vs := new(saModel.SaCompareSaveRespSection)
				vs.SaDocSectionId = s
				vs.Content = secMap[s]
				secList = append(secList, vs)
			}
			vd.SectionList = secList
			docList = append(docList, vd)
		}
		v.DocList = docList
		respLabel = append(respLabel, v)
	}
	resp.LabelList = respLabel
	return
}

// GetSaCompareDetailByDocIds 根据文档IDs获取比对详情信息
// compareId大于0: docIds为比对关联的文档IDs, 段落标签展示当前标签、历史标签、Ta的标签
// compareId等于0: docIds为选择的文档IDs, 段落标签展示历史标签、Ta的标签
func GetSaCompareDetailByDocIds(docIds []int, compareId, sysAdminId int) (detail *saModel.SaCompareDetail, err error) {
	if len(docIds) == 0 {
		return
	}
	// 获取文档信息
	docOB := new(saModel.SaDoc)
	docCond := fmt.Sprintf(` AND %s IN (%s)`, saModel.SaDocColumns.SaDocId, utils.GetOrmInReplace(len(docIds)))
	docPars := make([]interface{}, 0)
	docPars = append(docPars, docIds)
	docs, e := docOB.GetItemsByCondition(docCond, docPars, []string{}, "")
	if e != nil {
		err = fmt.Errorf("获取文档信息失败, Err: %s", e.Error())
		return
	}

	// 获取文档段落信息
	secOB := new(saModel.SaDocSection)
	secCond := fmt.Sprintf(` AND %s IN (%s)`, saModel.SaDocSectionColumns.DocId, utils.GetOrmInReplace(len(docIds)))
	secPars := make([]interface{}, 0)
	secPars = append(secPars, docIds)
	secsItems, e := secOB.GetItemsByCondition(secCond, secPars, []string{}, "doc_id ASC, sort ASC")
	secsMap := make(map[int][]*saModel.SaDocSection)
	for _, s := range secsItems {
		if secsMap[s.DocId] == nil {
			secsMap[s.DocId] = make([]*saModel.SaDocSection, 0)
		}
		secsMap[s.DocId] = append(secsMap[s.DocId], s)
	}

	// 获取文档打的标签
	compLabs := make([]*saModel.SaCompareLabel, 0)
	compLabOB := new(saModel.SaCompareLabel)
	compLabCond := fmt.Sprintf(` AND %s IN (%s)`, saModel.SaCompareLabelColumns.DocId, utils.GetOrmInReplace(len(docIds)))
	compLabPars := make([]interface{}, 0)
	compLabPars = append(compLabPars, docIds)
	compLabQuery, e := compLabOB.GetItemsByCondition(compLabCond, compLabPars, []string{}, "")
	if e != nil {
		err = fmt.Errorf("获取比对标签失败, Err: %s", e.Error())
		return
	}
	compLabs = compLabQuery

	// 历史搜索关键词
	keywordsOB := new(saModel.SaCompareSearchKeyword)
	keywordsCond := fmt.Sprintf(` AND %s = ?`, saModel.SaCompareSearchKeywordColumns.CompareId)
	keywordsPars := make([]interface{}, 0)
	keywordsPars = append(keywordsPars, compareId)
	keywordsItems, e := keywordsOB.GetItemsByCondition(keywordsCond, keywordsPars, []string{}, "")
	if e != nil {
		err = fmt.Errorf("获取历史搜索关键词失败, Err: %s", e.Error())
		return
	}
	keywords := make([]string, 0)
	for _, k := range keywordsItems {
		keywords = append(keywords, k.Keyword)
	}

	// 段落标签Map
	secLabelMap, isMineMap := formatCompareLabelStatusGroupSection(compLabs, compareId, sysAdminId)

	partSecMap := make(map[int][]*saModel.SaCompareLabel)
	partSecExistMap := make(map[string]bool)
	// 头部标签列表-包含多个文档中引用的所有标签并进行去重
	tabLabelIds := make([]int, 0)
	tabLabels := make([]*saModel.SaCompareDetailHeadLabel, 0)
	for _, l := range compLabs {
		if !utils.InArrayByInt(tabLabelIds, l.LabelId) {
			t := new(saModel.SaCompareDetailHeadLabel)
			t.LabelId = l.LabelId
			t.LabelName = l.LabelName
			t.IsMine = isMineMap[l.LabelId]
			tabLabelIds = append(tabLabelIds, l.LabelId)
			tabLabels = append(tabLabels, t)
		}
		// 文档片段Map
		if l.IsPart != 1 {
			continue
		}
		if partSecMap[l.DocId] == nil {
			partSecMap[l.DocId] = make([]*saModel.SaCompareLabel, 0)
		}
		ek := fmt.Sprintf("%d-%s", l.SectionId, utils.MD5(l.Content))
		if partSecExistMap[ek] {
			continue
		}
		partSecMap[l.DocId] = append(partSecMap[l.DocId], l)
	}

	// 详情
	detail = new(saModel.SaCompareDetail)
	docList := make([]*saModel.SaCompareDetailDoc, 0)
	for _, d := range docs {
		dv := new(saModel.SaCompareDetailDoc)
		dv.DocId = d.SaDocId
		dv.Title = d.Title
		dv.Theme = d.Theme
		dv.ClassifyName = d.ClassifyName
		// 整段
		secList := make([]*saModel.SaCompareDetailSection, 0)
		secs := secsMap[d.SaDocId]
		if secs != nil {
			for _, s := range secs {
				sv := new(saModel.SaCompareDetailSection)
				sv.SectionId = s.SaDocSectionId
				sv.Content = html.UnescapeString(s.Content)
				sv.Sort = s.Sort
				sv.LabelList = secLabelMap[fmt.Sprintf("%d-%s", s.SaDocSectionId, utils.MD5(``))]
				secList = append(secList, sv)
			}
		}
		// 片段
		parts := partSecMap[d.SaDocId]
		if parts != nil {
			for _, p := range parts {
				pv := new(saModel.SaCompareDetailSection)
				pv.SectionId = p.SectionId
				pv.Content = html.UnescapeString(p.Content)
				pv.IsPart = 1
				pv.StartIndex = p.StartIndex
				pv.EndIndex = p.EndIndex
				pv.LabelList = secLabelMap[fmt.Sprintf("%d-%s", p.SectionId, utils.MD5(p.Content))]
				fmt.Println("kkk", fmt.Sprintf("%d-%s", p.SectionId, utils.MD5(p.Content)))
				secList = append(secList, pv)
			}
		}

		dv.SectionList = secList
		docList = append(docList, dv)
	}
	detail.HeadLabel = tabLabels
	detail.DocList = docList
	detail.KeywordsList = keywords
	return
}

// formatCompareLabelStatusGroupSection 根据段落格式化段落标签的状态
func formatCompareLabelStatusGroupSection(compLabels []*saModel.SaCompareLabel, compareId, sysAdminId int) (labelMap map[string][]*saModel.SaCompareDetailFormatLabel, isMineMap map[int]int) {
	labelMap = make(map[string][]*saModel.SaCompareDetailFormatLabel)
	repeatMap := make(map[string][]int)
	thisMap := make(map[string]int)
	historyMap := make(map[string]int)
	otherMap := make(map[string]int)
	isMineMap = make(map[int]int) // 用于判断标签是否自己曾经使用或者当前使用过

	for _, l := range compLabels {
		// 判断段落标签的三种状态
		m := utils.MD5(l.Content)
		k := fmt.Sprintf("%d-%s-%d", l.SectionId, m, l.LabelId)
		if l.CompareId == compareId {
			thisMap[k] = 1
			if l.SysAdminId == sysAdminId {
				isMineMap[l.LabelId] = 1
			}
		}
		if l.CompareId != compareId && l.SysAdminId == sysAdminId {
			historyMap[k] = 1
			isMineMap[l.LabelId] = 1
		}
		if l.SysAdminId != sysAdminId {
			otherMap[k] = 1
		}

		k2 := fmt.Sprintf("%d-%s", l.SectionId, m)
		// 判断每段落内的标签是否重复添加
		if repeatMap[k2] == nil {
			repeatMap[k2] = make([]int, 0)
		}
		if utils.InArrayByInt(repeatMap[k2], l.LabelId) {
			continue
		}
		repeatMap[k2] = append(repeatMap[k2], l.LabelId)

		// 初始化段落标签
		if labelMap[k2] == nil {
			labelMap[k2] = make([]*saModel.SaCompareDetailFormatLabel, 0)
		}
		labelMap[k2] = append(labelMap[k2], &saModel.SaCompareDetailFormatLabel{
			LabelId:   l.LabelId,
			LabelName: l.LabelName,
		})
	}
	for s, l := range labelMap {
		for _, v := range l {
			k := fmt.Sprintf("%s-%d", s, v.LabelId)
			v.IsThis = thisMap[k]
			v.IsHistory = historyMap[k]
			v.IsOther = otherMap[k]
		}
	}
	return
}

// HandleElasticSaDocAndSection Elastic-新增/编辑文档和段落
func HandleElasticSaDocAndSection(saDoc *saModel.SaDoc, sections []*saModel.SaDocSection, delIds []int) (err error) {
	defer func() {
		if err != nil {
			alarm_msg.SendAlarmMsg(fmt.Sprintf("Elastic-语义分析文档, Err: %s", err.Error()), 2)
		}
	}()

	indexName := utils.EsSemanticAnalysisDocIndexName
	content := ``

	// 段落
	items := make([]*saModel.ElasticSaDoc, 0)
	for _, s := range sections {
		h := html.UnescapeString(s.Content)
		content += h
		items = append(items, &saModel.ElasticSaDoc{
			SaDocId:        s.DocId,
			SaDocSectionId: s.SaDocSectionId,
			ClassifyId:     saDoc.ClassifyId,
			ClassifyName:   saDoc.ClassifyName,
			Title:          saDoc.Title,
			Theme:          saDoc.Theme,
			BodyContent:    h,
			Author:         saDoc.SysAdminName,
			CoverImg:       saDoc.CoverImg,
			CreateTime:     saDoc.CreateTime.Format(utils.FormatDateTime),
		})
	}

	// 文档
	docId := fmt.Sprintf("%d-0", saDoc.SaDocId)
	item := &saModel.ElasticSaDoc{
		SaDocId:        saDoc.SaDocId,
		SaDocSectionId: 0,
		ClassifyId:     saDoc.ClassifyId,
		ClassifyName:   saDoc.ClassifyName,
		Title:          saDoc.Title,
		Theme:          saDoc.Theme,
		BodyContent:    content,
		Author:         saDoc.SysAdminName,
		CoverImg:       saDoc.CoverImg,
		CreateTime:     saDoc.CreateTime.Format(utils.FormatDateTime),
	}

	// 新增/更新
	if e := EsAddOrEditSaDoc(indexName, docId, item); e != nil {
		err = fmt.Errorf("新增/更新ES语义分析文档失败, Err: %s", e.Error())
		return
	}
	for _, v := range items {
		docId = fmt.Sprintf("%d-%d", v.SaDocId, v.SaDocSectionId)
		if e := EsAddOrEditSaDoc(indexName, docId, v); e != nil {
			err = fmt.Errorf("新增/更新ES语义分析文档段落失败, Err: %s", e.Error())
			return
		}
	}

	// 删除段落
	if len(delIds) > 0 {
		for _, d := range delIds {
			docId = fmt.Sprintf("%d-%d", saDoc.SaDocId, d)
			if e := EsDeleteData(indexName, docId); e != nil && !strings.Contains(e.Error(), "404") {
				err = fmt.Errorf("删除ES语义分析文档段落失败, Err: %s", e.Error())
				return
			}
		}
	}
	return
}

// DeleteElasticSaDocAndSection Elastic-删除文档和段落
func DeleteElasticSaDocAndSection(saDocId int, secIds []int) (err error) {
	defer func() {
		if err != nil {
			alarm_msg.SendAlarmMsg(fmt.Sprintf("Elastic-语义分析文档, Err: %s", err.Error()), 2)
		}
	}()
	indexName := utils.EsSemanticAnalysisDocIndexName

	docId := fmt.Sprintf("%d-0", saDocId)
	if e := EsDeleteData(indexName, docId); e != nil && !strings.Contains(e.Error(), "404") {
		err = fmt.Errorf("删除ES语义分析文档失败, Err: %s", e.Error())
		return
	}

	if len(secIds) > 0 {
		for _, d := range secIds {
			docId = fmt.Sprintf("%d-%d", saDocId, d)
			if e := EsDeleteData(indexName, docId); e != nil && !strings.Contains(e.Error(), "404") {
				err = fmt.Errorf("删除ES语义分析文档段落失败, Err: %s", e.Error())
				return
			}
		}
	}
	return
}

// FormatCompareLabels2TableData 格式化比对标签为表格数据
func FormatCompareLabels2TableData(compareLabels []*saModel.SaCompareLabelItem) (resp *saModel.SaCompareSaveResp, err error) {
	resp = new(saModel.SaCompareSaveResp)
	resp.LabelList = make([]*saModel.SaCompareSaveRespLabel, 0)

	// 取出文档作为X轴, 标签作为Y轴
	labelMap := make(map[int]*saModel.SaCompareLabelItem)
	docMap := make(map[int]*saModel.SaCompareLabelItem)
	secMap := make(map[string][]*saModel.SaCompareLabelItem)
	for _, v := range compareLabels {
		// 标签-Y轴
		if labelMap[v.LabelId] == nil {
			labelMap[v.LabelId] = v
			resp.LabelList = append(resp.LabelList, &saModel.SaCompareSaveRespLabel{
				SaLabelId: v.LabelId,
				LabelName: v.LabelName,
				DocList:   make([]*saModel.SaCompareSaveRespDoc, 0),
			})
		}
		// 文档-X轴
		if docMap[v.DocId] == nil {
			docMap[v.DocId] = v
			resp.TitleList = append(resp.TitleList, v.Title)
		}
		// 标签ID-文档ID作为key写入map, 后续段落匹配
		k := fmt.Sprintf("%d-%d", v.LabelId, v.DocId)
		if secMap[k] == nil {
			secMap[k] = make([]*saModel.SaCompareLabelItem, 0)
		}
		secMap[k] = append(secMap[k], v)
	}

	// 填充标签数据
	secExistMap := make(map[string]bool)
	for _, l := range resp.LabelList {
		docs := make([]*saModel.SaCompareSaveRespDoc, 0)
		for _, d := range docMap {
			dv := new(saModel.SaCompareSaveRespDoc)
			dv.SaDocId = d.DocId
			dv.Title = d.Title

			// 文档段落
			k := fmt.Sprintf("%d-%d", l.SaLabelId, d.DocId)
			secs := make([]*saModel.SaCompareSaveRespSection, 0)
			secList := secMap[k]
			if secList != nil && len(secList) > 0 {
				for _, s := range secList {
					sv := new(saModel.SaCompareSaveRespSection)
					sv.SaDocSectionId = s.SectionId
					content := html.UnescapeString(s.SectionContent)
					if s.CompareContent != "" {
						sv.IsPart = 1
						content = html.UnescapeString(s.CompareContent)
					}
					sv.Content = content
					// 同标签同文档同段落中的整段/片段去重
					ek := fmt.Sprintf("%d-%d-%d-%s", l.SaLabelId, d.DocId, s.SectionId, utils.MD5(content))
					fmt.Println(ek)
					if secExistMap[ek] {
						fmt.Println("跳过", ek)
						continue
					}
					secExistMap[ek] = true
					secs = append(secs, sv)
				}
			}

			dv.SectionList = secs
			docs = append(docs, dv)
		}
		// 标签对应的文档列表排序, 与resp.TitleList排序保持一致
		sort.Slice(docs, func(i, j int) bool {
			return docs[j].SaDocId > docs[i].SaDocId
		})
		l.DocList = docs
	}
	return
}