Merge pull request #8 from west2-online/courses

重构选课列表解析

course.go

@@ -1,6 +1,9 @@
 package jwch
 
 import (
+	"regexp"
+	"sort"
+	"strconv"
 	"strings"
 
 	"github.com/west2-online/jwch/constants"
@@ -73,21 +76,239 @@ func (s *Student) GetSemesterCourses(term, viewState, eventValidation string) ([
 			return nil, errno.HTMLParseError.WithMessage("get course info failed")
 		}
 
+		// 解析调课信息
+		/*
+			06周 星期3:5-6节  调至  09周 星期1:7-8节  旗山西1-206
+		*/
+		courseInfo11 := strings.Split(utils.InnerTextWithBr(info[11]), "\n")
+		adjustRegex := regexp.MustCompile(`(\d{2})周 星期(\d):(\d{1,2})-(\d{1,2})节[\s ]*调至[\s ]*(\d{2})周 星期(\d):(\d{1,2})-(\d{1,2})节[\s ]*(\S*)`)
+		adjustRules := []CourseAdjustRule{}
+
+		for i := 0; i < len(courseInfo11); i++ {
+			courseInfo11[i] = strings.TrimSpace(courseInfo11[i])
+
+			if courseInfo11[i] == "" { // 空行
+				continue
+			}
+
+			adjustMatchArr := adjustRegex.FindStringSubmatch(courseInfo11[i])
+
+			if len(adjustMatchArr) < 10 {
+				return nil, errno.HTMLParseError.WithMessage("get course adjust failed")
+			}
+
+			adjustRules = append(adjustRules, CourseAdjustRule{
+				OldWeek:       utils.SafeAtoi(adjustMatchArr[1]),
+				OldWeekday:    utils.SafeAtoi(adjustMatchArr[2]),
+				OldStartClass: utils.SafeAtoi(adjustMatchArr[3]),
+				OldEndClass:   utils.SafeAtoi(adjustMatchArr[4]),
+
+				NewWeek:       utils.SafeAtoi(adjustMatchArr[5]),
+				NewWeekday:    utils.SafeAtoi(adjustMatchArr[6]),
+				NewStartClass: utils.SafeAtoi(adjustMatchArr[7]),
+				NewEndClass:   utils.SafeAtoi(adjustMatchArr[8]),
+				NewLocation:   adjustMatchArr[9],
+			})
+		}
+
+		// 解析上课时间、地点，融合调课信息
+		/*
+			05-18 星期1:3-4节 铜盘A110
+			05-17 星期3:1-2节 铜盘A110
+			05-17 星期5:3-4节 铜盘A110
+		*/
+		courseInfo8 := strings.Split(utils.InnerTextWithBr(info[8]), "\n")
+		scheduleRules := []CourseScheduleRule{}
+
+		for i := 0; i < len(courseInfo8); i++ {
+			courseInfo8[i] = strings.TrimSpace(courseInfo8[i])
+
+			if courseInfo8[i] == "" { // 空行
+				continue
+			}
+
+			lineData := strings.Fields(courseInfo8[i])
+
+			if len(lineData) < 3 {
+				return nil, errno.HTMLParseError.WithMessage("get course info failed")
+			}
+
+			if strings.Contains(lineData[0], "周") { // 处理整周的课程，比如军训
+				/*
+					03周  星期1  -  04周  星期7
+					[0] 03周
+					[1] 星期1
+					[2] -
+					[3] 04周
+					[4] 星期7
+				*/
+				startWeek, _ := strconv.Atoi(strings.TrimSuffix(lineData[0], "周"))
+				endWeek, _ := strconv.Atoi(strings.TrimSuffix(lineData[3], "周"))
+				startWeekday, _ := strconv.Atoi(strings.TrimPrefix(lineData[1], "星期"))
+				endWeekday, _ := strconv.Atoi(strings.TrimPrefix(lineData[4], "星期"))
+
+				for weekday := 1; weekday <= 7; weekday++ {
+					curStartWeek := startWeek
+					curEndWeek := endWeek
+
+					if weekday < startWeekday {
+						curStartWeek++
+					}
+
+					if weekday > endWeekday {
+						curEndWeek--
+					}
+
+					if curStartWeek > curEndWeek {
+						continue
+					}
+
+					scheduleRules = append(scheduleRules, CourseScheduleRule{
+						Location:   "",
+						StartClass: 1,
+						EndClass:   8,
+						StartWeek:  curStartWeek,
+						EndWeek:    curEndWeek,
+						Weekday:    weekday,
+						Single:     true,
+						Double:     true,
+						Adjust:     false,
+					})
+				}
+
+				continue
+			}
+
+			/*
+				08-16 星期5:7-8节 铜盘A508
+				[0] 08-16
+				[1] 星期5:7-8节
+				[2] 铜盘A508
+			*/
+			/*
+				02-14 星期1:1-2节(双) 旗山西1-206
+				[0] 02-14
+				[1] 星期1:1-2节(双)
+				[2] 旗山西1-206
+			*/
+			/*
+				01-13 星期1:3-4节(单) 旗山西1-206
+				[0] 01-13
+				[1] 星期1:3-4节(单)
+				[2] 旗山西1-206
+			*/
+
+			// 是不是用正则表达式更好一点？
+			weekInfo := strings.SplitN(lineData[0], "-", 2)    // [8, 16]
+			dayInfo := strings.SplitN(lineData[1], ":", 2)     // ["星期5", "7-8节"] or ["星期1", "1-2节(双)"]
+			classBasicInfo := strings.Split(dayInfo[1], "节")   // ["7-8", ""] or ["1-2", "(双)"]
+			classInfo := strings.Split(classBasicInfo[0], "-") // ["7", "8"]
+			location := lineData[2]
+			startClass := utils.SafeAtoi(classInfo[0])
+			endClass := utils.SafeAtoi(classInfo[1])
+			startWeek := utils.SafeAtoi(weekInfo[0])
+			endWeek := utils.SafeAtoi(weekInfo[1])
+			weekDay := utils.SafeAtoi(strings.TrimPrefix(dayInfo[0], "星期"))
+			single := !strings.Contains(classBasicInfo[1], "双")
+			double := !strings.Contains(classBasicInfo[1], "单")
+
+			if len(adjustRules) == 0 {
+				scheduleRules = append(scheduleRules, CourseScheduleRule{
+					Location:   location,
+					StartClass: startClass,
+					EndClass:   endClass,
+					StartWeek:  startWeek,
+					EndWeek:    endWeek,
+					Weekday:    weekDay,
+					Single:     single,
+					Double:     double,
+				})
+			} else {
+				startWeek := utils.SafeAtoi(weekInfo[0])
+				endWeek := utils.SafeAtoi(weekInfo[1])
+				startClass := utils.SafeAtoi(classInfo[0])
+				endClass := utils.SafeAtoi(classInfo[1])
+				removedWeeks := []int{}
+
+				for _, adjustRule := range adjustRules {
+					// 匹配是否是对应的调课信息
+					if adjustRule.OldWeek < startWeek ||
+						adjustRule.OldWeek > endWeek ||
+						adjustRule.OldStartClass != startClass ||
+						adjustRule.OldEndClass != endClass ||
+						adjustRule.OldWeekday != weekDay {
+						continue
+					}
+
+					// 记录被去掉的周次
+					removedWeeks = append(removedWeeks, adjustRule.OldWeek)
+
+					// 添加新的课程信息
+					scheduleRules = append(scheduleRules, CourseScheduleRule{
+						Location:   adjustRule.NewLocation,
+						StartClass: adjustRule.NewStartClass,
+						EndClass:   adjustRule.NewEndClass,
+						StartWeek:  adjustRule.NewWeek,
+						EndWeek:    adjustRule.NewWeek,
+						Weekday:    adjustRule.NewWeekday,
+						Single:     true,
+						Double:     true,
+						Adjust:     true, // 调课
+					})
+				}
+
+				sort.Ints(removedWeeks)
+				// 去掉被调课的周次
+				curStartWeek := startWeek
+
+				for _, removedWeek := range removedWeeks {
+					if removedWeek == curStartWeek {
+						curStartWeek++
+
+						continue
+					}
+
+					scheduleRules = append(scheduleRules, CourseScheduleRule{
+						Location:   location,
+						StartClass: startClass,
+						EndClass:   endClass,
+						StartWeek:  curStartWeek,
+						EndWeek:    removedWeek - 1,
+						Weekday:    weekDay,
+						Single:     single,
+						Double:     double,
+						Adjust:     false,
+					})
+
+					curStartWeek = removedWeek + 1
+				}
+
+				if curStartWeek <= endWeek {
+					scheduleRules = append(scheduleRules, CourseScheduleRule{
+						Location:   location,
+						StartClass: startClass,
+						EndClass:   endClass,
+						StartWeek:  curStartWeek,
+						EndWeek:    endWeek,
+						Weekday:    weekDay,
+						Single:     single,
+						Double:     double,
+						Adjust:     false,
+					})
+				}
+			}
+		}
+
 		// TODO: performance optimization
 		res = append(res, &Course{
-			Type:          htmlquery.OutputHTML(info[0], false),
-			Name:          htmlquery.OutputHTML(info[1], false),
-			Syllabus:      "https://jwcjwxt2.fzu.edu.cn:81" + safeExtractRegex(`javascript:pop1\('(.*?)&`, safeExtractionValue(info[2], "a", "href", 0)),
-			LessonPlan:    "https://jwcjwxt2.fzu.edu.cn:81" + safeExtractRegex(`javascript:pop1\('(.*?)&`, safeExtractionValue(info[2], "a", "href", 1)),
-			PaymentStatus: safeExtractionFirst(info[3], "font"),
-			Credits:       safeExtractionFirst(info[4], "span"),
-			ElectiveType:  utils.GetChineseCharacter(htmlquery.OutputHTML(info[5], false)),
-			ExamType:      utils.GetChineseCharacter(htmlquery.OutputHTML(info[6], false)),
-			Teacher:       htmlquery.OutputHTML(info[7], false),
-			Classroom:     strings.TrimSpace(htmlquery.InnerText(info[8])),
-			ExamTime:      strings.TrimSpace(htmlquery.InnerText(info[9])),
-			Remark:        htmlquery.OutputHTML(info[10], false),
-			Adjust:        htmlquery.OutputHTML(info[11], false),
+			Name:             htmlquery.OutputHTML(info[1], false),
+			Syllabus:         "https://jwcjwxt2.fzu.edu.cn:81" + safeExtractRegex(`javascript:pop1\('(.*?)&`, safeExtractionValue(info[2], "a", "href", 0)),
+			LessonPlan:       "https://jwcjwxt2.fzu.edu.cn:81" + safeExtractRegex(`javascript:pop1\('(.*?)&`, safeExtractionValue(info[2], "a", "href", 1)),
+			Teacher:          htmlquery.OutputHTML(info[7], false),
+			ScheduleRules:    scheduleRules,
+			RawScheduleRules: strings.Join(courseInfo8, "\n"),
+			RawAdjust:        strings.Join(courseInfo11, "\n"),
+			Remark:           htmlquery.OutputHTML(info[10], false),
 		})
 	}
 

model.go

@@ -47,19 +47,39 @@ type Term struct {
 
 // 课程信息
 type Course struct {
-	Type          string `json:"type"`          // 修读类别
-	Name          string `json:"name"`          // 课程名称
-	PaymentStatus string `json:"paymentstatus"` // 缴费状态
-	Syllabus      string `json:"syllabus"`      // 课程大纲
-	LessonPlan    string `json:"lessonplan"`    // 课程计划
-	Credits       string `json:"credit"`        // 学分
-	ElectiveType  string `json:"electivetype"`  // 选课类型
-	ExamType      string `json:"examtype"`      // 考试类别
-	Teacher       string `json:"teacher"`       // 任课教师
-	Classroom     string `json:"classroom"`     // 上课时间地点
-	ExamTime      string `json:"examtime"`      // 考试时间地点
-	Remark        string `json:"remark"`        // 备注
-	Adjust        string `json:"adjust"`        // 调课信息
+	Name             string               `json:"name"`             // 课程名称
+	Syllabus         string               `json:"syllabus"`         // 课程大纲
+	LessonPlan       string               `json:"lessonplan"`       // 课程计划
+	Teacher          string               `json:"teacher"`          // 任课教师
+	ScheduleRules    []CourseScheduleRule `json:"scheduleRules"`    // 上课时间地点规则
+	RawScheduleRules string               `json:"rawScheduleRules"` // 上课时间地点（原始文本）
+	RawAdjust        string               `json:"rawAdjust"`        // 调课信息（原始文本）
+	Remark           string               `json:"remark"`           // 备注
+}
+
+type CourseScheduleRule struct {
+	Location   string `json:"location"`   // 上课地点
+	StartClass int    `json:"startClass"` // 开始节数
+	EndClass   int    `json:"endClass"`   // 结束节数
+	StartWeek  int    `json:"startWeek"`  // 开始周
+	EndWeek    int    `json:"endWeek"`    // 结束周
+	Weekday    int    `json:"weekday"`    // 星期几
+	Single     bool   `json:"single"`     // 单周 (PS: 为啥不用 odd)
+	Double     bool   `json:"double"`     // 双周 (PS: 为啥不用 even)
+	Adjust     bool   `json:"adjust"`     // 调课
+}
+
+type CourseAdjustRule struct {
+	OldWeek       int `json:"oldWeek"`       // 原-周次
+	OldWeekday    int `json:"oldWeekday"`    // 原-星期几
+	OldStartClass int `json:"oldStartClass"` // 原-开始节数
+	OldEndClass   int `json:"oldEndClass"`   // 原-结束节数
+
+	NewWeek       int    `json:"newWeek"`       // 新-周次
+	NewWeekday    int    `json:"newWeekday"`    // 新-星期几
+	NewStartClass int    `json:"newStartClass"` // 新-开始节数
+	NewEndClass   int    `json:"newEndClass"`   // 新-结束节数
+	NewLocation   string `json:"newLocation"`   // 新-上课地点
 }
 
 type Mark struct {

utils/utils.go

@@ -7,9 +7,11 @@ import (
 	"encoding/hex"
 	"encoding/json"
 	"fmt"
+	"golang.org/x/net/html"
 	"net/http"
 	"os"
 	"reflect"
+	"strconv"
 )
 
 func SaveData(filePath string, data []byte) error {
@@ -108,3 +110,31 @@ func JSONUnmarshalFromFile(filePath string, v any) error {
 
 	return json.Unmarshal(data, v)
 }
+
+func InnerTextWithBr(n *html.Node) string {
+	if n.Type == html.TextNode {
+		return n.Data
+	}
+
+	if n.Type == html.ElementNode && n.Data == "br" {
+		return "\n"
+	}
+
+	var buf bytes.Buffer
+
+	for c := n.FirstChild; c != nil; c = c.NextSibling {
+		buf.WriteString(InnerTextWithBr(c))
+	}
+
+	return buf.String()
+}
+
+func SafeAtoi(s string) int {
+	n, err := strconv.Atoi(s)
+
+	if err != nil {
+		return 0
+	}
+
+	return n
+}