htwkalender-pwa/service/fetch/fetchService.go

package fetch

import (
	"fmt"
	"github.com/labstack/echo/v5"
	"github.com/pocketbase/pocketbase"
	"github.com/pocketbase/pocketbase/apis"
	"github.com/pocketbase/pocketbase/models"
	"golang.org/x/net/html"
	"htwk-planner/model"
	"htwk-planner/service/date"
	"htwk-planner/service/db"
	"io"
	"net/http"
	"regexp"
	"strconv"
	"strings"
	"time"
)

func FetchHTWK(c echo.Context, app *pocketbase.PocketBase) error {

	var seminarGroups []model.SeminarGroup

	//seminarGroupsLabel := []string{
	//	"22EIM-MET",
	//}

	seminarGroupsLabel := db.GetAllCourses(app)
	for _, seminarGroupLabel := range seminarGroupsLabel {

		result, getError := getPlanHTML("ws", seminarGroupLabel)

		if getError == nil {
			seminarGroup := parseSeminarGroup(result)
			seminarGroups = append(seminarGroups, seminarGroup)
		}
	}

	//Save SeminarGroups to DB
	print("Saving SeminarGroups to DB...\n")

	collection, dbError := findCollection(app, "events")
	if dbError != nil {
		return apis.NewNotFoundError("Collection not found", dbError)
	}

	dbError = db.SaveEvents(seminarGroups, collection, app)
	if dbError != nil {
		return apis.NewApiError(400, "Could not save Event into database", dbError)
	}

	return c.JSON(http.StatusOK, seminarGroups)
}

func findCollection(app *pocketbase.PocketBase, collectionName string) (*models.Collection, error) {
	collection, dbError := app.Dao().FindCollectionByNameOrId(collectionName)
	return collection, dbError
}

func parseSeminarGroup(result string) model.SeminarGroup {
	doc, err := html.Parse(strings.NewReader(result))
	if err != nil {
		fmt.Printf("Error occurred while parsing the HTML document: %s\n", err.Error())
		return model.SeminarGroup{}
	}

	table := findFirstTable(doc)
	eventTables := getEventTables(doc)
	allDayLabels := getAllDayLabels(doc)

	if eventTables == nil || allDayLabels == nil {
		return model.SeminarGroup{}
	}

	eventsWithCombinedWeeks := toEvents(eventTables, allDayLabels)
	splitEventsByWeekVal := splitEventsByWeek(eventsWithCombinedWeeks)
	events := splitEventsBySingleWeek(splitEventsByWeekVal)
	semesterString := findFirstSpanWithClass(table, "header-0-2-0").FirstChild.Data
	semester, year := extractSemesterAndYear(semesterString)
	events = convertWeeksToDates(events, semester, year)
	var seminarGroup = model.SeminarGroup{
		University: findFirstSpanWithClass(table, "header-1-0-0").FirstChild.Data,
		Course:     findFirstSpanWithClass(table, "header-2-0-1").FirstChild.Data,
		Events:     events,
	}
	return seminarGroup
}

func convertWeeksToDates(events []model.Event, semester string, year string) []model.Event {
	var newEvents []model.Event
	eventYear, _ := strconv.Atoi(year)

	// for each event we need to calculate the start and end date based on the week and the year
	for _, event := range events {
		eventWeek, _ := strconv.Atoi(event.Week)
		eventDay, _ := date.GetDateFromWeekNumber(eventYear, eventWeek, event.Day)
		start := addTimeToDate(eventDay, event.Start)
		end := addTimeToDate(eventDay, event.End)
		newEvent := event
		newEvent.Start = start.String()
		newEvent.End = end.String()
		newEvent.Semester = semester
		newEvents = append(newEvents, newEvent)
	}
	return newEvents
}

func addTimeToDate(date time.Time, timeString string) time.Time {
	//convert time string to time
	timeParts := strings.Split(timeString, ":")
	hour, _ := strconv.Atoi(timeParts[0])
	minute, _ := strconv.Atoi(timeParts[1])

	return time.Date(date.Year(), date.Month(), date.Day(), hour, minute, 0, 0, time.UTC)
}

func extractSemesterAndYear(semesterString string) (string, string) {
	winterPattern := "Wintersemester"
	summerPattern := "Sommersemester"

	winterMatch := strings.Contains(semesterString, winterPattern)
	summerMatch := strings.Contains(semesterString, summerPattern)

	semester := ""
	semesterShortcut := ""

	if winterMatch {
		semester = "Wintersemester"
		semesterShortcut = "ws"
	} else if summerMatch {
		semester = "Sommersemester"
		semesterShortcut = "ss"
	} else {
		return "", ""
	}

	yearPattern := `\d{4}`
	combinedPattern := semester + `\s` + yearPattern
	re := regexp.MustCompile(combinedPattern)
	match := re.FindString(semesterString)
	year := ""

	if match != "" {
		reYear := regexp.MustCompile(yearPattern)
		year = reYear.FindString(match)
	}
	return semesterShortcut, year
}

func toEvents(tables [][]*html.Node, days []string) []model.Event {
	var events []model.Event

	for table := range tables {
		for row := range tables[table] {

			tableData := findTableData(tables[table][row])
			if len(tableData) > 0 {
				events = append(events, model.Event{
					Day:       days[table],
					Week:      getTextContent(tableData[0]),
					Start:     getTextContent(tableData[1]),
					End:       getTextContent(tableData[2]),
					Name:      getTextContent(tableData[3]),
					EventType: getTextContent(tableData[4]),
					Prof:      getTextContent(tableData[5]),
					Rooms:     getTextContent(tableData[6]),
					Notes:     getTextContent(tableData[7]),
					BookedAt:  getTextContent(tableData[8]),
				})
			}
		}

	}

	return events
}

func splitEventsByWeek(events []model.Event) []model.Event {
	var newEvents []model.Event

	for _, event := range events {
		weeks := strings.Split(event.Week, ",")
		for _, week := range weeks {
			newEvent := event
			newEvent.Week = strings.TrimSpace(week)
			newEvents = append(newEvents, newEvent)
		}
	}
	return newEvents
}

func splitEventsBySingleWeek(events []model.Event) []model.Event {
	var newEvents []model.Event

	for _, event := range events {
		if strings.Contains(event.Week, "-") {
			weeks := splitWeekRange(event.Week)
			for _, week := range weeks {
				newEvent := event
				newEvent.Week = week
				newEvents = append(newEvents, newEvent)
			}
		} else {
			newEvents = append(newEvents, event)
		}
	}
	return newEvents
}

func splitWeekRange(weekRange string) []string {
	parts := strings.Split(weekRange, "-")
	if len(parts) != 2 {
		return nil // Invalid format
	}

	start, errStart := strconv.Atoi(strings.TrimSpace(parts[0]))
	end, errEnd := strconv.Atoi(strings.TrimSpace(parts[1]))

	if errStart != nil || errEnd != nil {
		return nil // Error converting to integers
	}

	var weeks []string
	for i := start; i <= end; i++ {
		weeks = append(weeks, strconv.Itoa(i))
	}

	return weeks
}

func toUtf8(iso88591Buf []byte) string {
	buf := make([]rune, len(iso88591Buf))
	for i, b := range iso88591Buf {
		buf[i] = rune(b)
	}
	return string(buf)
}

func getPlanHTML(semester string, matrikel string) (string, error) {
	url := "https://stundenplan.htwk-leipzig.de/" + string(semester) + "/Berichte/Text-Listen;Studenten-Sets;name;" + matrikel + "?template=sws_semgrp&weeks=1-65"

	// Send GET request
	response, err := http.Get(url)
	if err != nil {
		fmt.Printf("Error occurred while making the request: %s\n", err.Error())
		return "", err
	}
	defer func(Body io.ReadCloser) {
		err := Body.Close()
		if err != nil {
			return
		}
	}(response.Body)

	// Read the response body
	body, err := io.ReadAll(response.Body)

	if err != nil {
		fmt.Printf("Error occurred while reading the response: %s\n", err.Error())
		return "", err
	}

	return toUtf8(body), err

}

// Find the first <table> element in the HTML document
func findFirstTable(node *html.Node) *html.Node {
	if node.Type == html.ElementNode && node.Data == "table" {
		return node
	}
	// Traverse child nodes recursively
	for child := node.FirstChild; child != nil; child = child.NextSibling {
		found := findFirstTable(child)
		if found != nil {
			return found
		}
	}
	return nil
}

// Find the first <span> element with the specified class attribute value
func findFirstSpanWithClass(node *html.Node, classValue string) *html.Node {
	if node.Type == html.ElementNode && node.Data == "span" {
		if hasClassAttribute(node, classValue) {
			return node
		}
	}

	// Traverse child nodes recursively
	for child := node.FirstChild; child != nil; child = child.NextSibling {
		found := findFirstSpanWithClass(child, classValue)
		if found != nil {
			return found
		}
	}

	return nil
}

// Check if the specified element has the specified class attribute value
func hasClassAttribute(node *html.Node, classValue string) bool {
	for _, attr := range node.Attr {
		if attr.Key == "class" && strings.Contains(attr.Val, classValue) {
			return true
		}
	}
	return false
}

// Get Tables with days
func getEventTables(node *html.Node) [][]*html.Node {
	var eventTables [][]*html.Node

	tables := findTables(node)
	// get all tables with events
	for events := range tables {
		rows := findTableRows(tables[events])
		// check that a first row exists
		if len(rows) > 0 {
			rows = rows[1:]
			eventTables = append(eventTables, rows)
		}
	}

	return eventTables

}

// Get Tables with days
func getAllDayLabels(node *html.Node) []string {
	paragraphs := findParagraphs(node)
	var dayArray []string

	for _, p := range paragraphs {
		label := getDayLabel(p)
		if label != "" {
			dayArray = append(dayArray, label)
		}
	}
	return dayArray
}

// Find all <p> elements in the HTML document
func findParagraphs(node *html.Node) []*html.Node {
	var paragraphs []*html.Node

	if node.Type == html.ElementNode && node.Data == "p" {
		paragraphs = append(paragraphs, node)
	}

	for child := node.FirstChild; child != nil; child = child.NextSibling {
		paragraphs = append(paragraphs, findParagraphs(child)...)
	}

	return paragraphs
}

// Find all <tr> elements in <tbody>, excluding the first one
func findTableRows(node *html.Node) []*html.Node {
	var tableRows []*html.Node

	if node.Type == html.ElementNode && node.Data == "tbody" {
		child := node.FirstChild
		for child != nil {
			if child.Type == html.ElementNode && child.Data == "tr" {
				tableRows = append(tableRows, child)
			}
			child = child.NextSibling
		}
	}

	// Traverse child nodes recursively
	for child := node.FirstChild; child != nil; child = child.NextSibling {
		var tableRowElement = findTableRows(child)
		if tableRowElement != nil {
			tableRows = append(tableRows, tableRowElement...)
		}
	}

	if tableRows == nil {
		return []*html.Node{}
	} else {
		return tableRows
	}
}

// Find all <p> elements in the HTML document
func findTables(node *html.Node) []*html.Node {
	var tables []*html.Node

	if node.Type == html.ElementNode && node.Data == "table" {
		tables = append(tables, node)
	}

	for child := node.FirstChild; child != nil; child = child.NextSibling {
		tables = append(tables, findDayTables(child)...)
	}

	return tables
}

// Find all <p> elements in the HTML document
func findDayTables(node *html.Node) []*html.Node {
	var tables []*html.Node

	for child := node.FirstChild; child != nil; child = child.NextSibling {
		tables = append(tables, findDayTables(child)...)
	}

	if node.Type == html.ElementNode && node.Data == "table" && hasClassAttribute(node, "spreadsheet") {
		tables = append(tables, node)
	}

	return tables
}

// Get the text content of the specified node and its descendants
func getDayLabel(node *html.Node) string {

	child := node.FirstChild
	if child != nil {
		if child.Type == html.ElementNode && child.Data == "span" {
			if child.FirstChild != nil {
				return child.FirstChild.Data
			}
		}
	}
	return ""
}

// Find all <td> elements in the current <tr>
func findTableData(node *html.Node) []*html.Node {
	var tableData []*html.Node

	if node.Type == html.ElementNode && node.Data == "tr" {
		child := node.FirstChild
		for child != nil {
			if child.Type == html.ElementNode && child.Data == "td" {
				tableData = append(tableData, child)
			}
			child = child.NextSibling
		}
	}

	return tableData
}

func getTextContent(node *html.Node) string {
	var textContent string

	if node.Type == html.TextNode {
		textContent = node.Data
	}

	for child := node.FirstChild; child != nil; child = child.NextSibling {
		textContent += getTextContent(child)
	}

	return textContent
}