2022-08-12 19:55:28 +02:00
|
|
|
use scraper::{Html, Selector};
|
|
|
|
|
2022-08-12 21:20:39 +02:00
|
|
|
mod models;
|
|
|
|
|
2022-08-12 19:55:28 +02:00
|
|
|
#[tokio::main]
|
|
|
|
async fn main() -> Result<(), Box<dyn std::error::Error>> {
|
|
|
|
//let url = "https://informatique.up8.edu/licence-iv/edt/l3.html";
|
|
|
|
|
|
|
|
// Get raw html
|
|
|
|
//let html = reqwest::get(url).await?.text().await?;
|
|
|
|
let html = include_str!("../target/debug.html");
|
|
|
|
|
|
|
|
// Parse document
|
2022-08-13 15:00:34 +02:00
|
|
|
let document = Html::parse_document(html);
|
2022-08-12 19:55:28 +02:00
|
|
|
|
2022-08-13 15:00:34 +02:00
|
|
|
// Selectors
|
|
|
|
let sel_table = Selector::parse("table").unwrap();
|
2022-08-14 11:20:32 +02:00
|
|
|
let sel_tr = Selector::parse("tr").unwrap();
|
2022-08-13 15:00:34 +02:00
|
|
|
let sel_tbody = Selector::parse("tbody").unwrap();
|
2022-08-14 11:20:32 +02:00
|
|
|
let sel_th = Selector::parse("th").unwrap();
|
|
|
|
let sel_td = Selector::parse("td").unwrap();
|
|
|
|
let sel_em = Selector::parse("em").unwrap();
|
|
|
|
let sel_small = Selector::parse("small").unwrap();
|
|
|
|
let sel_strong = Selector::parse("strong").unwrap();
|
2022-08-12 21:20:39 +02:00
|
|
|
|
2022-08-13 15:00:34 +02:00
|
|
|
// Find the timetable
|
|
|
|
let raw_timetable = document.select(&sel_table).next().unwrap();
|
2022-08-12 21:20:39 +02:00
|
|
|
|
|
|
|
// Find the slots available for the timetable
|
2022-08-13 15:00:34 +02:00
|
|
|
let raw_schedules = raw_timetable.select(&sel_tr).next().unwrap();
|
|
|
|
|
|
|
|
// Find availables schedules
|
|
|
|
let mut schedules = Vec::new();
|
|
|
|
for time in raw_schedules.select(&sel_th) {
|
|
|
|
schedules.push(time.inner_html());
|
|
|
|
}
|
|
|
|
|
|
|
|
// Find the timetable values
|
|
|
|
let raw_timetable_values = raw_timetable.select(&sel_tbody).next().unwrap();
|
|
|
|
|
|
|
|
// For each days
|
|
|
|
let mut timetable = Vec::new();
|
|
|
|
for day in raw_timetable_values.select(&sel_tr) {
|
|
|
|
let mut courses_vec = Vec::new();
|
2022-08-14 11:20:32 +02:00
|
|
|
let mut location_tracker = 0;
|
2022-08-13 15:00:34 +02:00
|
|
|
for course in day.select(&sel_td) {
|
|
|
|
if course.inner_html() == "—" {
|
|
|
|
courses_vec.push(None);
|
2022-08-14 11:20:32 +02:00
|
|
|
location_tracker += 1;
|
2022-08-13 15:00:34 +02:00
|
|
|
} else {
|
|
|
|
courses_vec.push(Some(models::Course {
|
2022-08-14 11:20:32 +02:00
|
|
|
name: course.select(&sel_em).next().unwrap().inner_html(),
|
|
|
|
professor: match course
|
|
|
|
.select(&sel_small)
|
|
|
|
.next()
|
|
|
|
.unwrap()
|
|
|
|
.inner_html()
|
|
|
|
.split("<br>")
|
|
|
|
.next()
|
|
|
|
{
|
|
|
|
Some(data) => {
|
|
|
|
if data.contains("</strong>") {
|
|
|
|
// This is the room, so there is no professor assigned
|
|
|
|
// to this courses yet
|
|
|
|
None
|
|
|
|
} else {
|
|
|
|
Some(data.to_string())
|
|
|
|
}
|
|
|
|
}
|
|
|
|
None => None,
|
|
|
|
},
|
|
|
|
room: course.select(&sel_strong).next().unwrap().inner_html(),
|
|
|
|
start: location_tracker,
|
|
|
|
size: match course.value().attr("colspan") {
|
|
|
|
Some(i) => i.parse().unwrap(),
|
|
|
|
None => 1,
|
|
|
|
},
|
2022-08-13 15:00:34 +02:00
|
|
|
}));
|
2022-08-14 11:20:32 +02:00
|
|
|
|
|
|
|
match &courses_vec[courses_vec.len() - 1] {
|
|
|
|
Some(course) => location_tracker += course.size,
|
|
|
|
None => location_tracker += 1,
|
|
|
|
}
|
2022-08-13 15:00:34 +02:00
|
|
|
}
|
|
|
|
}
|
2022-08-14 11:20:32 +02:00
|
|
|
println!("\n");
|
2022-08-13 15:00:34 +02:00
|
|
|
|
|
|
|
timetable.push(models::Day {
|
|
|
|
name: day.select(&sel_th).next().unwrap().inner_html(),
|
|
|
|
courses: courses_vec,
|
|
|
|
})
|
|
|
|
}
|
2022-08-12 19:55:28 +02:00
|
|
|
|
2022-08-13 15:00:34 +02:00
|
|
|
// TODO: Make fn who chacke if timetable is bell built (time consistency)
|
2022-08-13 17:27:35 +02:00
|
|
|
if !check_timetable_consistency(&schedules, &timetable) {
|
|
|
|
panic!("Error when building the timetable.");
|
|
|
|
}
|
2022-08-12 19:55:28 +02:00
|
|
|
|
|
|
|
Ok(())
|
2022-08-12 19:01:29 +02:00
|
|
|
}
|
2022-08-13 17:27:35 +02:00
|
|
|
|
|
|
|
/// Check if the timetable is well built
|
|
|
|
fn check_timetable_consistency(schedules: &Vec<String>, timetable: &Vec<models::Day>) -> bool {
|
2022-08-14 11:20:32 +02:00
|
|
|
let mut checker = true;
|
|
|
|
for day in timetable {
|
|
|
|
let mut i = 0;
|
|
|
|
for course in &day.courses {
|
|
|
|
match course {
|
|
|
|
Some(course_it) => {
|
|
|
|
// Checks the consistency of course start times
|
|
|
|
if i != course_it.start {
|
|
|
|
checker = false;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
// Keep the track of how many courses are in the day
|
|
|
|
i += course_it.size
|
|
|
|
}
|
|
|
|
None => i += 1,
|
|
|
|
}
|
|
|
|
}
|
|
|
|
// The counter should be the same as the amount of possible hours of the day
|
|
|
|
if i != schedules.len() {
|
|
|
|
checker = false;
|
|
|
|
break;
|
|
|
|
}
|
2022-08-13 17:27:35 +02:00
|
|
|
}
|
|
|
|
|
2022-08-14 11:20:32 +02:00
|
|
|
checker
|
2022-08-13 17:27:35 +02:00
|
|
|
}
|