mirror of
https://codeberg.org/icewind/ugc-scaper.git
synced 2026-06-04 02:34:11 +02:00
historical seasons
This commit is contained in:
parent
43bfba6307
commit
e20a9cdae9
8 changed files with 3064 additions and 3 deletions
|
|
@ -6,12 +6,14 @@ use time::macros::format_description;
|
|||
|
||||
mod player;
|
||||
mod player_details;
|
||||
mod seasons;
|
||||
mod team;
|
||||
mod team_matches;
|
||||
mod team_roster_history;
|
||||
|
||||
pub use player::*;
|
||||
pub use player_details::*;
|
||||
pub use seasons::*;
|
||||
pub use team::*;
|
||||
pub use team_matches::*;
|
||||
pub use team_roster_history::*;
|
||||
|
|
|
|||
84
src/parser/seasons.rs
Normal file
84
src/parser/seasons.rs
Normal file
|
|
@ -0,0 +1,84 @@
|
|||
use super::Parser;
|
||||
use crate::data::{Season, Seasons};
|
||||
use crate::parser::{select_text, ElementExt};
|
||||
use crate::{ParseError, Result};
|
||||
use scraper::{Html, Selector};
|
||||
|
||||
const SELECTOR_MENU: &str = ".sub-menu";
|
||||
const SELECTOR_NAME: &str = ".mega-menu-sub-title";
|
||||
const SELECTOR_SEASON_LINK: &str = "ul[id$=\"seasons\"] a[href^=\"rankings_\"]";
|
||||
|
||||
pub struct SeasonsParser {
|
||||
selector_menu: Selector,
|
||||
selector_name: Selector,
|
||||
selector_link: Selector,
|
||||
}
|
||||
|
||||
impl Default for SeasonsParser {
|
||||
fn default() -> Self {
|
||||
Self::new()
|
||||
}
|
||||
}
|
||||
|
||||
impl SeasonsParser {
|
||||
pub fn new() -> Self {
|
||||
SeasonsParser {
|
||||
selector_menu: Selector::parse(SELECTOR_MENU).unwrap(),
|
||||
selector_name: Selector::parse(SELECTOR_NAME).unwrap(),
|
||||
selector_link: Selector::parse(SELECTOR_SEASON_LINK).unwrap(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl Parser for SeasonsParser {
|
||||
type Output = Vec<Seasons>;
|
||||
|
||||
fn parse(&self, document: &str) -> Result<Self::Output> {
|
||||
let document = Html::parse_document(document);
|
||||
|
||||
document
|
||||
.select(&self.selector_menu)
|
||||
.filter(|item| item.select(&self.selector_link).next().is_some())
|
||||
.filter(|item| item.select(&self.selector_name).next().is_some())
|
||||
.map(|item| {
|
||||
let name = select_text(item, &self.selector_name)
|
||||
.ok_or(ParseError::EmptyText {
|
||||
role: "game mode name",
|
||||
selector: SELECTOR_NAME,
|
||||
})?
|
||||
.trim_end_matches(" Menu");
|
||||
|
||||
let seasons = item
|
||||
.select(&self.selector_link)
|
||||
.map(|link| {
|
||||
let text = link
|
||||
.first_text()
|
||||
.ok_or(ParseError::EmptyText {
|
||||
role: "season name",
|
||||
selector: SELECTOR_SEASON_LINK,
|
||||
})?
|
||||
.trim_end_matches(" Final Standings")
|
||||
.trim_end_matches(" Final Rank")
|
||||
.trim_end_matches(" Final Ranks");
|
||||
let link = link.attr("href").ok_or(ParseError::EmptyText {
|
||||
role: "season link",
|
||||
selector: SELECTOR_SEASON_LINK,
|
||||
})?;
|
||||
let id = link
|
||||
.trim_start_matches("rankings_")
|
||||
.trim_end_matches(".cfm");
|
||||
Ok(Season {
|
||||
name: text.to_string(),
|
||||
id: id.to_string(),
|
||||
})
|
||||
})
|
||||
.collect::<Result<_>>()?;
|
||||
|
||||
Ok(Seasons {
|
||||
mode: name.to_string(),
|
||||
seasons,
|
||||
})
|
||||
})
|
||||
.collect::<Result<Vec<_>>>()
|
||||
}
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue