schedule_converter/parser/score.go

package parser

import (
	"strings"

	"schedule_converter/models"

	"github.com/PuerkitoBio/goquery"
)

// ParseGradesHTML 解析成绩页面 HTML，返回成绩条目列表。
func ParseGradesHTML(html string) ([]models.GradeEntry, error) {
	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
	if err != nil {
		return nil, err
	}

	var grades []models.GradeEntry

	doc.Find("table.bot_line").Each(func(_ int, table *goquery.Selection) {
		rows := table.Find("tr")
		if rows.Length() < 2 {
			return
		}

		rows.Each(func(idx int, row *goquery.Selection) {
			if idx == 0 {
				return
			}
			cells := row.Find("td")
			if cells.Length() < 8 {
				return
			}

			texts := getCellTexts(cells)
			if len(texts) < 9 {
				return
			}

			isMidterm := len(texts) == 10

			entry := parseGradeRowTexts(texts, detectRetake(row))

			if isMidterm && entry.Score == "" {
				entry.Score = safeGet(texts, 8)
				if len(texts) > 9 {
					entry.GradePoint = safeGet(texts, 9)
				}
			}

			if entry.Name != "" && len(entry.Name) >= 2 {
				if entry.Score == "" {
					entry.Score = "-"
				}
				grades = append(grades, entry)
			}
		})
	})

	return grades, nil
}

// ParseGPAHTML 解析学分绩页面 HTML，返回 GPA 概要与成绩条目列表。
func ParseGPAHTML(html string) (*models.GPASummary, []models.GradeEntry, error) {
	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
	if err != nil {
		return nil, nil, err
	}

	summary := &models.GPASummary{}

	doc.Find("span").Each(func(_ int, span *goquery.Selection) {
		id, _ := span.Attr("id")
		text := strings.TrimSpace(span.Text())
		switch id {
		case "pjxfj":
			summary.AverageGPA = text
		case "zrs":
			summary.Rank = text
		case "kskxfj":
			summary.ExamTotalGPA = text
		case "kskxfs":
			summary.ExamTotalCredits = text
		case "kckxfj":
			summary.FailCredits = text
		case "xqs":
			summary.Semesters = text
		}
	})

	var grades []models.GradeEntry

	doc.Find("table.bot_line").Each(func(_ int, table *goquery.Selection) {
		rows := table.Find("tr")
		if rows.Length() < 2 {
			return
		}

		rows.Each(func(idx int, row *goquery.Selection) {
			if idx == 0 {
				return
			}
			cells := row.Find("td")
			if cells.Length() < 10 {
				return
			}

			texts := getCellTexts(cells)

			entry := parseGradeRowTexts(texts, detectRetake(row))

			if entry.Name != "" && len(entry.Name) >= 2 {
				grades = append(grades, entry)
			}
		})
	})

	return summary, grades, nil
}

func parseGradeRowTexts(texts []string, isRetake bool) models.GradeEntry {
	entry := models.GradeEntry{IsRetake: isRetake}

	// 详细成绩/GPA 页面格式：
	// 序号、学期、学院、课程代码、课程名称、课程性质、课程类别、学分、是否考试课、是否学位课、成绩标记、成绩、最终成绩、绩点1、绩点、录入时间、备注。
	if len(texts) >= 15 && looksLikeDetailedGradeRow(texts) {
		entry.Term = safeGet(texts, 1)
		entry.Code = safeGet(texts, 3)
		entry.Name = safeGet(texts, 4)
		entry.Nature = safeGet(texts, 5)
		entry.Category = safeGet(texts, 6)
		entry.Credit = safeGet(texts, 7)
		entry.IsExam = safeGet(texts, 8)
		entry.Score = firstNonEmpty(safeGet(texts, 12), safeGet(texts, 11))
		entry.GradePoint = firstNonEmpty(safeGet(texts, 14), safeGet(texts, 13))
		return entry
	}

	// 常规成绩页面格式：学年、学期、课程代码、课程名称、课程性质、课程类别、学分、成绩、绩点、...
	if len(texts) >= 10 {
		entry.Term = safeGet(texts, 1)
		entry.Code = safeGet(texts, 2)
		entry.Name = safeGet(texts, 3)
		entry.Nature = safeGet(texts, 4)
		entry.Category = safeGet(texts, 5)
		entry.Credit = safeGet(texts, 6)
		entry.Score = safeGet(texts, 7)
		entry.GradePoint = safeGet(texts, 8)
		return entry
	}

	entry.Term = safeGet(texts, 1)
	entry.Code = safeGet(texts, 3)
	entry.Name = safeGet(texts, 4)
	entry.Credit = safeGet(texts, 7)
	entry.Score = safeGet(texts, 8)
	if len(texts) > 9 {
		entry.GradePoint = safeGet(texts, 9)
	}
	return entry
}

func looksLikeDetailedGradeRow(texts []string) bool {
	return safeGet(texts, 3) != "" && safeGet(texts, 4) != "" && safeGet(texts, 7) != "" && (safeGet(texts, 8) == "是" || safeGet(texts, 8) == "否")
}

func firstNonEmpty(values ...string) string {
	for _, value := range values {
		if strings.TrimSpace(value) != "" {
			return value
		}
	}
	return ""
}

// detectRetake 检测成绩行是否包含补考标记。
// 教务系统在成绩单元格内以 <font color="#0066ff">补考</font> 标记补考，
// 也会在整行文本中出现"补考"字样。
func detectRetake(row *goquery.Selection) bool {
	// 方式1：查找包含"补考"文字的 <font> 标签
	if row.Find("font:contains('补考')").Length() > 0 {
		return true
	}
	// 方式2：整行纯文本包含"补考"
	rowText := strings.TrimSpace(row.Text())
	return strings.Contains(rowText, "补考")
}

func getCellTexts(cells *goquery.Selection) []string {
	var texts []string
	cells.Each(func(_ int, cell *goquery.Selection) {
		text := strings.TrimSpace(cell.Text())
		texts = append(texts, text)
	})
	return texts
}

func safeGet(arr []string, idx int) string {
	if idx < len(arr) {
		return arr[idx]
	}
	return ""
}