anufriev.g.a
/
l9_stud_bot
Archived
1
0
Fork 0
Code Issues Pull Requests Packages Projects 1 Releases 5 Activity

Добавлено: полный парсинг расписания с подгруппами

This commit is contained in:
far-galaxy 2023-02-24 12:47:47 +04:00
parent 804b767dbb
commit fef468f695
3 changed files with 131 additions and 89 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

88
modules/ssau_parser/parser.go
View File

@ -9,7 +9,6 @@ import (
"net/http"
"net/url"
"strings"
"time"
"github.com/PuerkitoBio/goquery"
)
@ -79,17 +78,12 @@ func FindInRasp(query string) (RaspItems, error) {
}
} else {
return nil, fmt.Errorf("Responce: %s", resp.Status)
return nil, fmt.Errorf("responce: %s", resp.Status)
}
return list, nil
}
type Times struct {
Begin time.Time
End time.Time
}
func Connect(uri string, week int) (*goquery.Document, error) {
client := http.Client{}
@ -111,83 +105,3 @@ func Connect(uri string, week int) (*goquery.Document, error) {
return doc, nil
}
type Lesson struct {
Begin time.Time
End time.Time
Name string
}
type Shedule struct {
SpecName string
Week int
Lessons [][]Lesson
}
func Parse(doc *goquery.Document) (*Shedule, error) {
spec := doc.Find(".info-block__description div").First().Text()[1:]
log.Println(spec)
var raw_dates []string
doc.Find(".schedule__head-date").Each(func(i int, s *goquery.Selection) {
sh_date := s.Text()
raw_dates = append(raw_dates, sh_date)
})
var raw_times []string
doc.Find(".schedule__time-item").Each(func(i int, s *goquery.Selection) {
sh_time := s.Text() + "+04"
raw_times = append(raw_times, sh_time)
})
var lessons []string
doc.Find(".schedule__item:not(.schedule__head)").Each(func(i int, s *goquery.Selection) {
lesson := s.Text()
lessons = append(lessons, lesson)
})
var shedule [][]Lesson
for t := 0; t < len(raw_times); t += 2 {
var time_line []Lesson
for d, date := range raw_dates {
begin_raw := date + raw_times[t]
begin, err := time.Parse(" 02.01.2006 15:04 -07", begin_raw)
if err != nil {
return nil, err
}
end_raw := date + raw_times[t+1]
end, err := time.Parse(" 02.01.2006 15:04 -07", end_raw)
if err != nil {
return nil, err
}
idx := (len(raw_times)-2)*t/2 + d
lesson := Lesson{
Begin: begin,
End: end,
Name: lessons[idx],
}
time_line = append(time_line, lesson)
}
shedule = append(shedule, time_line)
}
return &Shedule{SpecName: spec, Lessons: shedule}, nil
}
/*
type Lesson struct {
Type string
Name string
Place string
TeacherID int64
Comment string
}
func parseLesson(l *goquery.Selection) {
var lesson Lesson
d, _ := l.Find("div.schedule__discipline").Attr("class")
t := strings.Split(d, " ")
lesson.Type = t[len(t)-1]
lesson.Name = l.Find("div.schedule__discipline").First().Text()
}
*/

125
modules/ssau_parser/shedule.go Normal file
View File

@ -0,0 +1,125 @@
package ssau_parser
import (
"log"
"strconv"
"strings"
"time"
"github.com/PuerkitoBio/goquery"
)
type Lesson struct {
Begin time.Time
End time.Time
SubLessons []SubLesson
}
type SubLesson struct {
Type string
Name string
Place string
TeacherID int64
Comment string
SubGroup string
}
type Shedule struct {
SpecName string
Week int
Lessons [][]Lesson
}
func Parse(doc *goquery.Document) (*Shedule, error) {
spec := doc.Find(".info-block__description div").First().Text()[1:]
log.Println(spec)
var raw_dates []string
doc.Find(".schedule__head-date").Each(func(i int, s *goquery.Selection) {
sh_date := s.Text()
raw_dates = append(raw_dates, sh_date)
})
var raw_times []string
doc.Find(".schedule__time-item").Each(func(i int, s *goquery.Selection) {
sh_time := s.Text() + "+04"
raw_times = append(raw_times, sh_time)
})
var lessons [][]SubLesson
doc.Find(".schedule__item:not(.schedule__head)").Each(func(i int, s *goquery.Selection) {
sl := ParseSubLesson(s)
lessons = append(lessons, sl)
})
var shedule [][]Lesson
for t := 0; t < len(raw_times); t += 2 {
var time_line []Lesson
for d, date := range raw_dates {
begin_raw := date + raw_times[t]
begin, err := time.Parse(" 02.01.2006 15:04 -07", begin_raw)
if err != nil {
return nil, err
}
end_raw := date + raw_times[t+1]
end, err := time.Parse(" 02.01.2006 15:04 -07", end_raw)
if err != nil {
return nil, err
}
idx := (len(raw_times)-2)*t/2 + d
lesson := Lesson{
Begin: begin,
End: end,
SubLessons: lessons[idx],
}
time_line = append(time_line, lesson)
}
shedule = append(shedule, time_line)
}
return &Shedule{SpecName: spec, Lessons: shedule}, nil
}
var types = [4]string{"lect", "lab", "pract", "other"}
func ParseSubLesson(s *goquery.Selection) []SubLesson {
var subs []SubLesson
s.Find(".schedule__lesson").Each(func(j int, l *goquery.Selection) {
var sublesson SubLesson
name := l.Find("div.schedule__discipline").First()
sublesson.Name = name.Text()[1:]
l_type := name.AttrOr("class", "lesson-color-type-4")
t := strings.Split(l_type, " ")
l_type = t[len(t)-1]
type_idx, err := strconv.ParseInt(l_type[len(l_type)-1:], 0, 8)
if err != nil {
type_idx = 4
}
sublesson.Type = types[type_idx-1]
teacher := l.Find(".schedule__teacher a").AttrOr("href", "/rasp?staffId=")
teacherId, err := strconv.ParseInt(teacher[14:], 0, 64)
if err != nil {
teacherId = 0
}
sublesson.TeacherID = teacherId
subgroup := l.Find(".schedule__groups span").First().Text()
if subgroup == " " {
subgroup = ""
}
sublesson.SubGroup = subgroup
place := l.Find("div.schedule__place").First().Text()
if len(place) > 2 {
place = place[1:]
}
sublesson.Place = place
sublesson.Comment = l.Find("div.schedule__comment").First().Text()
subs = append(subs, sublesson)
})
return subs
}

7
modules/ssau_parser/ssau_parser_test.go
View File

@ -26,7 +26,7 @@ func TestConnect(t *testing.T) {
}
func TestParse(t *testing.T) {
list, err := FindInRasp("2405")
list, err := FindInRasp("2207")
if err != nil {
t.Error(err)
}
@ -35,5 +35,8 @@ func TestParse(t *testing.T) {
if err != nil {
t.Error(err)
}
Parse(doc)
_, err = Parse(doc)
if err != nil {
t.Error(err)
}
}