fix:#82 added event parser

This commit is contained in:
masterElmar
2023-12-12 17:12:46 +01:00
parent 31c27635d3
commit a714e92cd9
2 changed files with 163 additions and 9 deletions

View File

@@ -2,7 +2,11 @@ package main
import (
"errors"
"github.com/google/uuid"
"github.com/pocketbase/pocketbase/tools/types"
"htwkalender/model"
"net/http"
"regexp"
"strconv"
"strings"
"sync"
@@ -18,10 +22,160 @@ import (
func main() {
var sportCourseLinks = fetchAllAvailableSportCourses()
events := fetchHTWKSportCourses(sportCourseLinks)
sportEntries := fetchHTWKSportCourses(sportCourseLinks)
for _, event := range sportEntries {
println(event.Title)
}
events := formatEntriesToEvents(sportEntries)
for _, event := range events {
println(event.Title)
println(event.Name)
}
}
func formatEntriesToEvents(entries []SportEntry) []model.Event {
var events []model.Event
for i, entry := range entries {
eventStarts, eventEnds := calculateEventStarts(entry.Details.DateRange.Start, entry.Details.DateRange.End, entry.Details.Cycle)
for j := range eventStarts {
start, _ := types.ParseDateTime(eventStarts[j].In(time.UTC))
end, _ := types.ParseDateTime(eventEnds[j].In(time.UTC))
var event = model.Event{
UUID: uuid.NewSHA1(uuid.NameSpaceDNS, []byte(entry.Title+strconv.FormatInt(int64(i), 10))).String(),
Day: entry.Details.DateRange.Start.Weekday().String(),
Week: strconv.Itoa(23),
Start: start,
End: end,
Name: entry.Title,
EventType: entry.Details.Type,
Prof: entry.Details.CourseLead.Name,
Rooms: entry.Details.Location.Name,
Notes: entry.AdditionalNote,
BookedAt: "",
Course: "Sport",
Semester: checkSemester(entry.Details.DateRange.Start),
}
events = append(events, event)
}
}
return events
}
func calculateEventStarts(start time.Time, end time.Time, cycle string) ([]time.Time, []time.Time) {
// start is the begin of the cycle e.g. 01.04.2020
// end is the end of the cycle e.g. 30.09.2020
// cycle is the day and timespan (e.g. "Mo 18:00-20:00")
// check if start is before end
if start.After(end) {
return nil, nil
}
// check if cycle is valid
if !checkCycle(cycle) {
return nil, nil
}
var weekDay = cycle[0:2]
// match weekday to time.Weekday (e.g. "Mo" -> time.Monday)
var weekDayInt int
switch weekDay {
case "Mo":
weekDayInt = 1
case "Di":
weekDayInt = 2
case "Mi":
weekDayInt = 3
case "Do":
weekDayInt = 4
case "Fr":
weekDayInt = 5
case "Sa":
weekDayInt = 6
case "So":
weekDayInt = 0
}
// get every date matching the weekday in the cycle between start and end
var eventDates []time.Time
for d := start; d.Before(end); d = d.AddDate(0, 0, 1) {
if d.Weekday() == time.Weekday(weekDayInt) {
eventDates = append(eventDates, d)
}
}
// add hours and minutes to the dates in eventDates
// array of tuple of start and end times
var eventStartsWithTime []time.Time
var eventEndWithTime []time.Time
for _, eventStart := range eventDates {
timeRegExp, _ := regexp.Compile("[0-9]{2}:[0-9]{2}")
times := timeRegExp.FindAllString(cycle, 2)
startHour, _ := strconv.Atoi(times[0][0:2])
startMinute, _ := strconv.Atoi(times[0][3:5])
endHour, _ := strconv.Atoi(times[1][0:2])
endMinute, _ := strconv.Atoi(times[1][3:5])
eventStartsWithTime = append(eventStartsWithTime, time.Date(eventStart.Year(), eventStart.Month(), eventStart.Day(), startHour, startMinute, 0, 0, eventStart.Location()))
eventEndWithTime = append(eventEndWithTime, time.Date(eventStart.Year(), eventStart.Month(), eventStart.Day(), endHour, endMinute, 0, 0, eventStart.Location()))
}
return eventStartsWithTime, eventEndWithTime
}
func checkCycle(cycle string) bool {
// check if cycle is valid
if len(cycle) < 12 {
return false
}
// check if cycle has a weekday
weekDay := cycle[0:2]
if weekDay != "Mo" && weekDay != "Di" && weekDay != "Mi" && weekDay != "Do" && weekDay != "Fr" && weekDay != "Sa" && weekDay != "So" {
return false
}
// check if cycle has a timespan
timeSpan := cycle[3:12]
if len(timeSpan) != 9 {
return false
}
// check if timespan has a start and end time
startTime := timeSpan[0:5]
endTime := timeSpan[6:9]
if len(startTime) != 5 || len(endTime) != 3 {
return false
}
// check if start time is before end time
if startTime > endTime {
return false
}
return true
}
// check if ws or ss
func checkSemester(date time.Time) string {
if date.Month() >= 4 && date.Month() <= 9 {
return "ss"
} else {
return "ws"
}
}
@@ -52,7 +206,7 @@ func fetchAllAvailableSportCourses() []string {
// fetchAllHTWKSportCourses fetches all sport courses from the given links.
// to speed up the process, it uses multithreading.
func fetchHTWKSportCourses(links []string) []Event {
func fetchHTWKSportCourses(links []string) []SportEntry {
//multithreaded webpage requests to speed up the process
@@ -75,7 +229,7 @@ func fetchHTWKSportCourses(links []string) []Event {
}
wg.Wait()
var events []Event
var events []SportEntry
for _, doc := range htmlPageArray {
if doc != nil {
@@ -111,15 +265,15 @@ func htmlRequest(url string) (*goquery.Document, error) {
// If the sport course exists, it will return the sport course.
// goquery is used to parse the html. The html structure is not very consistent, so it is hard to parse.
// May be improved in the future.
func fetchHtwkSportCourse(doc *goquery.Document) ([]Event, error) {
var events []Event
func fetchHtwkSportCourse(doc *goquery.Document) ([]SportEntry, error) {
var events []SportEntry
if doc.Find("h1").Text() == "Aktuelle Sportangebote" {
return nil, errors.New("not a sport course page")
}
doc.Find(".eventHead").Each(func(i int, s *goquery.Selection) {
var event Event
var event SportEntry
var details EventDetails
fullTitle := strings.TrimSpace(s.Find("h3").Text())

View File

@@ -4,8 +4,8 @@ import "time"
// MODELS
// Event represents the overall event details.
type Event struct {
// SportEntry represents the overall event details.
type SportEntry struct {
Title string
Details EventDetails
AdditionalNote string