48 lines
1.3 KiB
Rust
48 lines
1.3 KiB
Rust
use mwbot::{Bot, parsoid::WikinodeIterator as _};
|
|
|
|
fn extract_number_from_url(url: &str) -> Option<u32> {
|
|
// Split the URL into parts using '/'
|
|
let parts: Vec<&str> = url.split('/').collect();
|
|
|
|
// Iterate over the parts to find the one that ends with ".html"
|
|
for part in parts {
|
|
if part.ends_with(".html") {
|
|
// Remove the ".html" suffix
|
|
let number_str = part.trim_end_matches(".html");
|
|
// Parse the remaining string into an integer
|
|
return number_str.parse().ok();
|
|
}
|
|
}
|
|
|
|
None
|
|
}
|
|
|
|
pub async fn get_wsdc_id(bot: &Bot, page: &mwbot::Page) -> Option<u32> {
|
|
let x = page.html().await.unwrap().into_mutable();
|
|
for w in &x.filter_external_links() {
|
|
if let Some(id) = extract_number_from_url(&w.target()) {
|
|
return Some(id);
|
|
}
|
|
}
|
|
None
|
|
}
|
|
|
|
pub async fn get_description(bot: &Bot, page: &mwbot::Page) -> Option<String> {
|
|
let x = page.html().await.unwrap().into_mutable();
|
|
for w in &x.iter_sections() {
|
|
dbg!(w);
|
|
let Some(h) = w.heading() else {
|
|
println!("No heading");
|
|
continue;
|
|
};
|
|
dbg!(&h);
|
|
let Some(t) = h.as_text() else {
|
|
println!("No text");
|
|
continue;
|
|
};
|
|
|
|
dbg!(t.borrow());
|
|
}
|
|
None
|
|
}
|