feat:#52 added new fetcher from new endpoint

backend/service/fetch/htmlDownloader.go

@@ -0,0 +1,44 @@
+package fetch
+
+import (
+	"fmt"
+	"io"
+	"net/http"
+)
+
+// getPlanHTML Get the HTML document from the specified URL
+
+func GetHTML(url string) (string, error) {
+
+	// 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
+
+}
+
+func toUtf8(iso88591Buf []byte) string {
+	buf := make([]rune, len(iso88591Buf))
+	for i, b := range iso88591Buf {
+		buf[i] = rune(b)
+	}
+	return string(buf)
+}

backend/service/fetch/v1/fetchSeminarEventService.go

@@ -1,11 +1,11 @@
-package fetch
+package v1
 
 import (
 	"fmt"
 	"htwkalender/model"
 	"htwkalender/service/date"
 	"htwkalender/service/db"
-	"io"
+	"htwkalender/service/fetch"
 	"net/http"
 	"regexp"
 	"strconv"
@@ -30,7 +30,7 @@ func GetSeminarEvents(c echo.Context, app *pocketbase.PocketBase) error {
 
 	seminarGroups = ReplaceEmptyEventNames(seminarGroups)
 
-	savedRecords, dbError := db.SaveEvents(seminarGroups, app)
+	savedRecords, dbError := db.SaveSeminarGroupEvents(seminarGroups, app)
 
 	if dbError != nil {
 		return apis.NewNotFoundError("Events could not be saved", dbError.Error())
@@ -64,13 +64,15 @@ func GetSeminarGroupsEventsFromHTML(seminarGroupsLabel []string) []model.Seminar
 	var seminarGroups []model.SeminarGroup
 	for _, seminarGroupLabel := range seminarGroupsLabel {
 
-		result, getError := getPlanHTML("ss", seminarGroupLabel)
+		ssUrl := "https://stundenplan.htwk-leipzig.de/" + string("ss") + "/Berichte/Text-Listen;Studenten-Sets;name;" + seminarGroupLabel + "?template=sws_semgrp&weeks=1-65"
+		result, getError := fetch.GetHTML(ssUrl)
 		if getError == nil {
 			seminarGroup := parseSeminarGroup(result)
 			seminarGroups = append(seminarGroups, seminarGroup)
 		}
 
-		result, getError = getPlanHTML("ws", seminarGroupLabel)
+		wsUrl := "https://stundenplan.htwk-leipzig.de/" + string("ws") + "/Berichte/Text-Listen;Studenten-Sets;name;" + seminarGroupLabel + "?template=sws_semgrp&weeks=1-65"
+		result, getError = fetch.GetHTML(wsUrl)
 		if getError == nil {
 			seminarGroup := parseSeminarGroup(result)
 			seminarGroups = append(seminarGroups, seminarGroup)
@@ -80,7 +82,6 @@ func GetSeminarGroupsEventsFromHTML(seminarGroupsLabel []string) []model.Seminar
 }
 
 func splitEventType(events []model.Event) []model.Event {
-
 	for i, event := range events {
 		matched, _ := regexp.Match("^([VPS])([wp])$", []byte(event.EventType))
 		if matched {
@@ -90,7 +91,6 @@ func splitEventType(events []model.Event) []model.Event {
 			events[i] = event
 		}
 	}
-
 	return events
 }
 
@@ -133,7 +133,6 @@ func generateUUIDs(events []model.Event, course string) []model.Event {
 		events[i].UUID = hash.String()
 	}
 	return events
-
 }
 
 // convertWeeksToDates converts the week and year to a date
@@ -230,7 +229,6 @@ func toEvents(tables [][]*html.Node, days []string) []model.Event {
 		}
 
 	}
-
 	return events
 }
 
@@ -296,40 +294,3 @@ func splitWeekRange(weekRange string) []string {
 
 	return weeks
 }
-
-func toUtf8(iso88591Buf []byte) string {
-	buf := make([]rune, len(iso88591Buf))
-	for i, b := range iso88591Buf {
-		buf[i] = rune(b)
-	}
-	return string(buf)
-}
-
-// getPlanHTML Get the HTML document from the specified URL
-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
-
-}

backend/service/fetch/v1/fetchSeminarEventService_test.go

@@ -1,4 +1,4 @@
-package fetch
+package v1
 
 import (
 	"fmt"

backend/service/fetch/v1/fetchSeminarGroupService.go

@@ -1,4 +1,4 @@
-package fetch
+package v1
 
 import (
 	"encoding/xml"

backend/service/fetch/v1/htmlParsingFunctions.go

@@ -1,4 +1,4 @@
-package fetch
+package v1
 
 import (
 	"golang.org/x/net/html"
@@ -182,7 +182,6 @@ func findTableData(node *html.Node) []*html.Node {
 			child = child.NextSibling
 		}
 	}
-
 	return tableData
 }
 

backend/service/fetch/v2/eventParser.go

@@ -0,0 +1,60 @@
+package v2
+
+import (
+	"github.com/pocketbase/pocketbase/tools/types"
+	"golang.org/x/net/html"
+	"htwkalender/model"
+	"htwkalender/service/date"
+	"regexp"
+	"strings"
+)
+
+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 {
+				start, _ := types.ParseDateTime(date.CreateTimeFromHourAndMinuteString(getTextContent(tableData[1])))
+				end, _ := types.ParseDateTime(date.CreateTimeFromHourAndMinuteString(getTextContent(tableData[2])))
+
+				courses := getTextContent(tableData[7])
+
+				if len(courses) > 0 {
+					for _, course := range strings.Split(courses, " ") {
+						events = append(events, model.Event{
+							Day:       days[table],
+							Week:      getTextContent(tableData[0]),
+							Start:     start,
+							End:       end,
+							Name:      getTextContent(tableData[3]),
+							EventType: getTextContent(tableData[4]),
+							Notes:     getTextContent(tableData[5]),
+							Prof:      getTextContent(tableData[6]),
+							Rooms:     getTextContent(tableData[8]),
+							BookedAt:  getTextContent(tableData[10]),
+							Course:    course,
+						})
+					}
+				}
+			}
+		}
+
+	}
+	return events
+}
+
+func splitEventType(events []model.Event) []model.Event {
+	for i, event := range events {
+		matched, _ := regexp.Match("^([VPS])([wp])$", []byte(event.EventType))
+		if matched {
+			eventType := event.EventType
+			event.EventType = eventType[0:1]
+			event.Compulsory = eventType[1:2]
+			events[i] = event
+		}
+	}
+	return events
+}

backend/service/fetch/v2/fetcher.go

@@ -0,0 +1,95 @@
+package v2
+
+import (
+	"github.com/google/uuid"
+	"github.com/labstack/echo/v5"
+	"github.com/pocketbase/pocketbase"
+	"golang.org/x/net/html"
+	"htwkalender/model"
+	"htwkalender/service/db"
+	"htwkalender/service/fetch"
+	"strconv"
+	"strings"
+)
+
+func ParseEventsFromRemote(c echo.Context, app *pocketbase.PocketBase) error {
+
+	url := "https://stundenplan.htwk-leipzig.de/ws/Berichte/Text-Listen;Veranstaltungsarten;name;Vp%0AVw%0AV%0ASp%0ASw%0AS%0APp%0APw%0AP%0AZV%0ATut%0ASperr%0Apf%0Awpf%0Afak%0A%0A?&template=sws_modul&weeks=1-65&combined=yes"
+
+	// Fetch Webpage from URL
+	webpage, err := fetch.GetHTML(url)
+	if err != nil {
+		return err
+	}
+
+	// Parse HTML to Node Tree
+	doc, err2 := parseHTML(err, webpage)
+	if err2 != nil {
+		return err2
+	}
+
+	// Get all event tables and all day labels
+	eventTables := getEventTables(doc)
+	allDayLabels := getAllDayLabels(doc)
+
+	if eventTables == nil || allDayLabels == nil {
+		return err
+	}
+
+	eventsWithCombinedWeeks := toEvents(eventTables, allDayLabels)
+
+	if eventsWithCombinedWeeks == nil {
+		return err
+	}
+
+	splitEventsByWeekVal := splitEventsByWeek(eventsWithCombinedWeeks)
+	events := splitEventsBySingleWeek(splitEventsByWeekVal)
+
+	if events == nil {
+		return err
+	}
+
+	// Search Semester String in Page Head
+	table := findFirstTable(doc)
+	semesterString := findFirstSpanWithClass(table, "header-0-2-0").FirstChild.Data
+
+	semester, year := extractSemesterAndYear(semesterString)
+	events = convertWeeksToDates(events, semester, year)
+	events = generateUUIDs(events)
+	events = splitEventType(events)
+
+	var seminarGroup = model.SeminarGroup{
+		University: findFirstSpanWithClass(table, "header-1-0-0").FirstChild.Data,
+		Events:     events,
+	}
+
+	if seminarGroup.Events == nil && seminarGroup.University == "" {
+		return err
+	}
+
+	savedRecords, dbError := db.SaveEvents(events, app)
+
+	if dbError != nil {
+		return dbError
+	} else {
+		savedRecordsLength := strconv.FormatInt(int64(len(savedRecords)), 10)
+		return c.JSON(200, "Successfully saved "+savedRecordsLength+" events")
+	}
+}
+
+func parseHTML(err error, webpage string) (*html.Node, error) {
+	doc, err := html.Parse(strings.NewReader(webpage))
+	if err != nil {
+		return nil, err
+	}
+	return doc, nil
+}
+
+func generateUUIDs(events []model.Event) []model.Event {
+	for i, event := range events {
+		// generate a hash value from the event name, course and semester
+		hash := uuid.NewSHA1(uuid.NameSpaceOID, []byte(event.Name+event.Course))
+		events[i].UUID = hash.String()
+	}
+	return events
+}

backend/service/fetch/v2/htmlParsingFunctions.go

@@ -0,0 +1,324 @@
+package v2
+
+import (
+	"github.com/pocketbase/pocketbase/tools/types"
+	"golang.org/x/net/html"
+	"htwkalender/model"
+	"htwkalender/service/date"
+	"regexp"
+	"strconv"
+	"strings"
+	"time"
+)
+
+// 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 {
+
+	// Check if the current node is a <span> element with the specified class attribute value
+	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...)
+		}
+	}
+
+	// check if tableRows is nil
+	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
+}
+
+// Get the text content of the specified node and its descendants
+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
+}
+
+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 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 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 := replaceTimeForDate(eventDay, event.Start.Time())
+		end := replaceTimeForDate(eventDay, event.End.Time())
+
+		//Check if end is before start
+		if end.Before(start) {
+			end = end.AddDate(0, 0, 1)
+		}
+
+		newEvent := event
+		newEvent.Start, _ = types.ParseDateTime(start.In(time.UTC))
+		newEvent.End, _ = types.ParseDateTime(end.In(time.UTC))
+		newEvent.Semester = semester
+		newEvents = append(newEvents, newEvent)
+	}
+	return newEvents
+}
+
+// replaceTimeForDate replaces hour, minute, second, nsec for the selected date
+func replaceTimeForDate(date time.Time, replacementTime time.Time) time.Time {
+	return time.Date(date.Year(), date.Month(), date.Day(), replacementTime.Hour(), replacementTime.Minute(), replacementTime.Second(), replacementTime.Nanosecond(), date.Location())
+}