schedule_converter/parser/empty_classroom.go

package parser

import (
	"fmt"
	"strings"

	"schedule_converter/models"

	"github.com/PuerkitoBio/goquery"
)

// ParseEmptyClassroomHTML 解析空教室查询结果 HTML，返回空教室条目列表。
func ParseEmptyClassroomHTML(html, semester, weekStart, weekEnd string) ([]models.EmptyClassroomEntry, error) {
	if html == "" {
		return nil, nil
	}

	// Clean nbsp entities
	cleaned := strings.ReplaceAll(html, "&nbsp", "")
	cleaned = strings.ReplaceAll(cleaned, "&", "&")

	doc, err := goquery.NewDocumentFromReader(strings.NewReader(cleaned))
	if err != nil {
		return nil, err
	}

	dayNames := [7]string{"星期一", "星期二", "星期三", "星期四", "星期五", "星期六", "星期日"}
	periodLabels := [6]string{"1,2", "3,4", "5,6", "7,8", "9,10", "11,12"}

	var classrooms []models.EmptyClassroomEntry

	// Find the main data table with class="dataTable"
	doc.Find("table.dataTable").Each(func(_ int, table *goquery.Selection) {
		rows := table.Find("tr")
		if rows.Length() < 3 {
			return
		}

		// Verify this is the classroom table by checking for weekday headers
		hasWeekdayHeader := false
		rows.First().Find("th").Each(func(_ int, th *goquery.Selection) {
			text := strings.TrimSpace(th.Text())
			for _, day := range dayNames {
				if strings.Contains(text, day) {
					hasWeekdayHeader = true
					break
				}
			}
		})
		if !hasWeekdayHeader {
			return
		}

		// Build column mapping: column index -> (dayIndex, periodIndex)
		colMapping := make(map[int][2]int)

		// Start from column 1 (skip first column which is classroom name)
		colIdx := 1
		for dayIdx := 0; dayIdx < 7; dayIdx++ {
			for periodIdx := 0; periodIdx < 6; periodIdx++ {
				colMapping[colIdx] = [2]int{dayIdx, periodIdx}
				colIdx++
			}
		}

		// Process data rows (skip first 2 header rows)
		rows.Each(func(rowIdx int, row *goquery.Selection) {
			if rowIdx < 2 {
				return
			}

			cells := row.Find("td")
			if cells.Length() < 2 {
				return
			}

			// Get classroom name from first cell
			classroomName := strings.TrimSpace(cells.First().Text())
			// Filter out invalid classroom names
			if classroomName == "" || len(classroomName) < 2 {
				return
			}

			// Skip non-classroom entries like "线上网络授课"
			if strings.Contains(classroomName, "线上") || strings.Contains(classroomName, "网络") {
				return
			}

			// Collect empty slots for this classroom
			var slots []string
			weekdaySet := make(map[int]bool)
			var weekdays []string

			// Check each time slot column
			for colIdx := 1; colIdx < cells.Length(); colIdx++ {
				cell := cells.Eq(colIdx)

				// Get mapping for this column
				mapping, exists := colMapping[colIdx]
				if !exists {
					continue
				}
				dayIdx := mapping[0]
				periodIdx := mapping[1]

				// Check if cell is empty (no kjs_icon div)
				isEmpty := true
				cell.Find("div.kjs_icon").Each(func(_ int, div *goquery.Selection) {
					isEmpty = false
				})

				// Also check if cell has any content
				cellText := strings.TrimSpace(cell.Text())
				if cellText != "" {
					isEmpty = false
				}

				if isEmpty {
					dayName := dayNames[dayIdx]
					periodText := periodLabels[periodIdx]
					slot := fmt.Sprintf("%s 第%s节", dayName, periodText)
					slots = append(slots, slot)

					if !weekdaySet[dayIdx] {
						weekdaySet[dayIdx] = true
						weekdays = append(weekdays, dayName)
					}
				}
			}

			// Create entry if there are empty slots
			if len(slots) > 0 {
				weekStr := ""
				if weekStart != "" && weekEnd != "" {
					weekStr = fmt.Sprintf("%s-%s周", weekStart, weekEnd)
				}
				entry := models.EmptyClassroomEntry{
					Classroom: classroomName,
					Weekday:   strings.Join(weekdays, ", "),
					Periods:   strings.Join(slots, "\n"),
					Term:      semester,
					Weeks:     weekStr,
				}
				classrooms = append(classrooms, entry)
			}
		})
	})

	return classrooms, nil
}

// TestParseEmptyClassroomHTML is a test helper function that exports the parser for testing
func TestParseEmptyClassroomHTML(html, semester, weekStart, weekEnd string) ([]models.EmptyClassroomEntry, error) {
	return ParseEmptyClassroomHTML(html, semester, weekStart, weekEnd)
}