finish 试卷库

backend/global/file.go

@@ -371,8 +371,10 @@ func (f *FileDatabase) AddFile(lstid int, reg *Regex, istemp bool, progress func
 					}
 					v := make(map[string]uint8, len(words)*2)
 					for _, word := range words {
-						if word != "" && word != "\n" && word != " " {
-							v[word]++
+						if word != "" && !strings.Contains("\n 。，、的是使,.（）()1234567890ABCDEFGHIJKLMNOPQRSTUVWXYZ", word) {
+							if v[word] == 0 { // 二值化
+								v[word] = 1
+							}
 						}
 					}
 					data, err := json.Marshal(v)
@@ -448,9 +450,7 @@ func (f *FileDatabase) AddFile(lstid int, reg *Regex, istemp bool, progress func
 				}
 			}
 			majorq.Sub = append(majorq.Sub, QuestionJSON{
-				Name:   queidstr,
-				Points: 0, //TODO: fill sub points
-				Rate:   r,
+				Name: queidstr, //TODO: fill sub points
 			})
 		}
 		filequestions = append(filequestions, majorq)
@@ -552,18 +552,20 @@ func (f *FileDatabase) DelFile(lstid, uid int, istemp bool) error {
 }
 
 // GetFile get analyzed file's structure from List(ID)
-func (f *FileDatabase) GetFile(lstid, uid int) (*File, int64, error) {
+func (f *FileDatabase) GetFile(lstid, uid int) (file *File, sz int64, istemp bool, err error) {
 	f.mu.RLock()
 	defer f.mu.RUnlock()
 	lst, err := sql.Find[List](&f.db, FileTableList, "WHERE ID="+strconv.Itoa(lstid))
 	if err != nil {
-		return nil, 0, err
+		return
 	}
 	if lst.HasntAnalyzed {
-		return nil, 0, ErrHasntAnalyzed
+		err = ErrHasntAnalyzed
+		return
 	}
 	if lst.IsTemp && lst.Uploader != uid {
-		return nil, 0, ErrNoGetFileStatusPermission
+		err = ErrNoGetFileStatusPermission
+		return
 	}
 	ftable := ""
 	if lst.IsTemp {
@@ -571,9 +573,9 @@ func (f *FileDatabase) GetFile(lstid, uid int) (*File, int64, error) {
 	} else {
 		ftable = FileTableFile
 	}
-	file, err := sql.Find[File](&f.db, ftable, "WHERE ListID="+strconv.Itoa(lstid))
+	filestruct, err := sql.Find[File](&f.db, ftable, "WHERE ListID="+strconv.Itoa(lstid))
 	if err != nil {
-		return nil, 0, err
+		return
 	}
-	return &file, lst.Size, nil
+	return &filestruct, lst.Size, lst.IsTemp, nil
 }

backend/global/question.go

@@ -17,7 +17,6 @@ import (
 type QuestionJSON struct {
 	Name   string         `json:"name"`             // Name is name or Question ID
 	Points int            `json:"points,omitempty"` // Points is sum of subs' points or self
-	Rate   float64        `json:"rate,omitempty"`   // Rate is the avg(non-leaf) or max(leaf) similarity
 	Sub    []QuestionJSON `json:"sub,omitempty"`
 }
 
@@ -88,6 +87,43 @@ type Question struct {
 	Dup    []byte // Dup is json of {queid: rate, ...}
 }
 
+// GetQuestion by id
+func (f *FileDatabase) GetQuestion(id int64, istemp bool) (Question, error) {
+	f.mu.RLock()
+	defer f.mu.RUnlock()
+	qtable := ""
+	if istemp {
+		qtable = FileTableTempQuestion
+	} else {
+		qtable = FileTableQuestion
+	}
+	return sql.Find[Question](&f.db, qtable, "WHERE ID="+strconv.FormatInt(id, 10))
+}
+
+// GetQuestionHex by hexid
+func (f *FileDatabase) GetQuestionHex(hexid string, istemp bool) (q Question, err error) {
+	idb, err := hex.DecodeString(hexid)
+	if err != nil {
+		return
+	}
+	return f.GetQuestion(int64(binary.LittleEndian.Uint64(idb)), istemp)
+}
+
+// MaxDuplicateRate parse q.Dup and get the max rate
+func (q *Question) MaxDuplicateRate() float64 {
+	dupmap := make(map[string]float64, 64)
+	if err := json.Unmarshal(q.Dup, &dupmap); err != nil {
+		return 0
+	}
+	r := 0.0
+	for _, v := range dupmap {
+		if v > r {
+			r = v
+		}
+	}
+	return r
+}
+
 // GetDuplicateRate calc q & que's dup rate
 func (q *Question) GetDuplicateRate(que *Question) (float64, error) {
 	v1, v2 := make(map[string]uint8, 64), make(map[string]uint8, 64)

backend/paper.go

@@ -41,6 +41,7 @@ type filelist struct {
 type filestatus struct {
 	Name         string        `json:"name"`
 	Size         float64       `json:"size"`
+	Rate         float64       `json:"rate"`
 	Questions    []question    `json:"questions"`
 	Duplications []duplication `json:"duplications"` // Duplications length == 10
 }
@@ -308,12 +309,12 @@ func init() {
 			writeresult(w, codeError, nil, err.Error(), typeError)
 			return
 		}
-		file, sz, err := global.FileDB.GetFile(id, *user.ID)
+		file, sz, istemp, err := global.FileDB.GetFile(id, *user.ID)
 		if err != nil {
 			writeresult(w, codeError, nil, err.Error(), typeError)
 			return
 		}
-		qs, ds, err := parseFileQuestions(file.Questions)
+		qs, ds, filerate, err := parseFileQuestions(file.Questions, istemp)
 		if err != nil {
 			writeresult(w, codeError, nil, err.Error(), typeError)
 			return
@@ -321,6 +322,7 @@ func init() {
 		writeresult(w, codeSuccess, &filestatus{
 			Name:         file.Class + ".docx",
 			Size:         float64(sz) / 1024 / 1024, // MB
+			Rate:         filerate * 100,
 			Questions:    qs,
 			Duplications: ds,
 		}, messageOk, typeSuccess)

backend/question.go

@@ -1,7 +1,10 @@
 package backend
 
 import (
+	"container/heap"
 	"encoding/json"
+	"math"
+	"strconv"
 
 	"github.com/fumiama/paper-manager/backend/global"
 )
@@ -17,21 +20,78 @@ type duplication struct {
 	Name    string `json:"name"`
 }
 
-func parseFileQuestions(qb []byte) ([]question, []duplication, error) {
+type duplications []duplication
+
+func (d *duplications) Len() int {
+	return len(*d)
+}
+
+// Less is actually more for a big-top heap
+func (d *duplications) Less(i, j int) bool {
+	return (*d)[i].Percent > (*d)[j].Percent
+}
+
+func (d *duplications) Swap(i, j int) {
+	(*d)[i], (*d)[j] = (*d)[j], (*d)[i]
+}
+
+func (d *duplications) Push(x any) {
+	*d = append(*d, x.(duplication))
+}
+
+func (d *duplications) Pop() any {
+	if d.Len() == 0 {
+		return nil
+	}
+	i := d.Len() - 1
+	x := (*d)[i]
+	*d = (*d)[:i]
+	return x
+}
+
+func parseFileQuestions(qb []byte, istemp bool) ([]question, []duplication, float64, error) {
 	ques := make([]global.QuestionJSON, 0, 16)
 	qs := make([]question, 0, 16)
-	ds := make([]duplication, 0, 16)
-	err := json.Unmarshal(qb, &qs)
+	err := json.Unmarshal(qb, &ques)
 	if err != nil {
-		return nil, nil, err
+		return nil, nil, 0, err
 	}
+	dh := make(duplications, 0, 16)
+	heap.Init(&dh)
+	sum := 0.0
+	cnt := 0
 	for _, q := range ques {
 		qs = append(qs, question{
 			Count: len(q.Sub),
 			Point: q.Points,
 			Name:  q.Name,
 		})
-		// TODO: use heap to get top 10 ds
+		for i, subq := range q.Sub {
+			qstruct, err := global.FileDB.GetQuestionHex(subq.Name, istemp)
+			if err != nil {
+				continue
+			}
+			p := qstruct.MaxDuplicateRate()
+			heap.Push(&dh, duplication{
+				Percent: int(math.Round(p * 100)),
+				Name:    q.Name + "." + strconv.Itoa(i+1),
+			})
+			sum += p
+			cnt++
+		}
 	}
-	return nil, nil, nil
+	i := dh.Len()
+	ds := make([]duplication, 10)
+	if i > 10 {
+		i = 10
+	} else {
+		for j := i; j < 10; j++ {
+			ds[j] = duplication{Name: "N/A"}
+		}
+	}
+	for i--; i >= 0; i-- {
+		ds[i] = heap.Pop(&dh).(duplication)
+	}
+
+	return qs, ds, sum / float64(cnt), nil
 }

frontend/vben/mock/page/file.ts

@@ -1,5 +1,5 @@
 import { MockMethod } from 'vite-plugin-mock'
-import { resultError, resultSuccess, getRequestToken, requestParams } from '../_util'
+// import { resultError, resultSuccess, getRequestToken, requestParams } from '../_util'
 
 export default [
   /*{
@@ -16,7 +16,7 @@ export default [
       })
     },
   },*/
-  {
+  /*{
     url: '/api/getFileStatus',
     timeout: 500,
     method: 'get',
@@ -49,5 +49,5 @@ export default [
         ],
       })
     },
-  },
+  },*/
 ] as MockMethod[]

frontend/vben/src/api/page/model/fileModel.ts

@@ -16,6 +16,7 @@ export interface Duplication {
 export interface FileStatus {
   name: string
   size: number
+  rate: number
   questions: Question[]
   duplications: Duplication[]
 }

frontend/vben/src/views/page/file/index.vue

@@ -1,7 +1,9 @@
 <template>
   <PageWrapper :title="t('routes.filelist.file') + ': ' + docxNameRef">
     <template #headerContent>
-      <a-button type="primary" @click="downloadDocx"> 下载试卷 ({{ docxSizeRef }}MB) </a-button>
+      <a-button type="primary" @click="downloadDocx">
+        下载试卷 ({{ docxSizeRef.toFixed(2) }}MB)
+      </a-button>
     </template>
     <div ref="chartRef" :style="{ height, width }"></div>
     <div class="docxWrap" :style="{ width }">
@@ -131,7 +133,7 @@
                   },
                 },
                 {
-                  text: '重复率前十',
+                  text: '重复率: ' + ret.rate.toFixed(2) + '%, 前十如下',
                   left: '40%',
                   top: '1%',
                   textStyle: {