This commit is contained in:
48
src/fetching/caching.rs
Normal file
48
src/fetching/caching.rs
Normal file
@@ -0,0 +1,48 @@
|
||||
use std::{collections::HashMap, path::Path};
|
||||
|
||||
use reqwest::{Client, ClientBuilder};
|
||||
|
||||
use crate::{dance_info::DanceInfo, worldsdc::DanceInfoError};
|
||||
|
||||
use super::DanceInfoParser;
|
||||
|
||||
struct CachingFetcher {
|
||||
hitcache: Vec<(u32, String)>,
|
||||
errorcache: Vec<(u32, String)>,
|
||||
client: Client,
|
||||
}
|
||||
|
||||
#[derive(thiserror::Error, Debug)]
|
||||
enum CachingFetcherCreationError {
|
||||
#[error("Could not create client: {0}")]
|
||||
ClientError(#[from] reqwest::Error),
|
||||
}
|
||||
|
||||
impl CachingFetcher {
|
||||
pub fn new(cachepath: &Path) -> Result<Self, CachingFetcherCreationError> {
|
||||
let client = ClientBuilder::new().build()?;
|
||||
|
||||
Ok(Self {
|
||||
hitcache: vec![],
|
||||
errorcache: vec![],
|
||||
client,
|
||||
})
|
||||
}
|
||||
pub async fn fetch(&mut self, id: u32) -> Result<DanceInfo, DanceInfoError> {
|
||||
let mut params = HashMap::new();
|
||||
params.insert("q", id.to_string());
|
||||
let response = self
|
||||
.client
|
||||
.request(
|
||||
reqwest::Method::POST,
|
||||
"https://points.worldsdc.com/lookup2020/find",
|
||||
)
|
||||
.form(¶ms)
|
||||
.send()
|
||||
.await
|
||||
.map_err(DanceInfoError::Request)?;
|
||||
|
||||
let x: DanceInfoParser = response.json().await.map_err(DanceInfoError::JsonParse)?;
|
||||
Ok(x.into())
|
||||
}
|
||||
}
|
||||
39
src/fetching/mod.rs
Normal file
39
src/fetching/mod.rs
Normal file
@@ -0,0 +1,39 @@
|
||||
use crate::dance_info::DanceInfo;
|
||||
use crate::fetching::types::DanceInfoError;
|
||||
|
||||
mod scoringdance;
|
||||
mod worldsdc;
|
||||
pub mod types;
|
||||
|
||||
use async_trait::async_trait;
|
||||
use std::sync::Arc;
|
||||
|
||||
#[async_trait]
|
||||
pub trait WsdcFetcher: Send + Sync {
|
||||
async fn fetch(&self, id: u32) -> Result<DanceInfo, DanceInfoError>;
|
||||
}
|
||||
|
||||
pub struct WorldsdcFetcher;
|
||||
pub struct ScoringDanceFetcher;
|
||||
|
||||
#[async_trait]
|
||||
impl WsdcFetcher for WorldsdcFetcher {
|
||||
async fn fetch(&self, id: u32) -> Result<DanceInfo, DanceInfoError> {
|
||||
worldsdc::fetch_wsdc_info_wsdc(id).await
|
||||
}
|
||||
}
|
||||
|
||||
#[async_trait]
|
||||
impl WsdcFetcher for ScoringDanceFetcher {
|
||||
async fn fetch(&self, id: u32) -> Result<DanceInfo, DanceInfoError> {
|
||||
scoringdance::fetch_wsdc_info_scoring_dance(id).await
|
||||
}
|
||||
}
|
||||
|
||||
/// Convenience alias for a shared, dynamic fetcher
|
||||
pub type DynWsdcFetcher = Arc<dyn WsdcFetcher>;
|
||||
|
||||
/// Back-compat helper that uses the `WorldsdcFetcher`.
|
||||
pub async fn fetch_wsdc_info(id: u32) -> Result<DanceInfo, DanceInfoError> {
|
||||
WorldsdcFetcher.fetch(id).await
|
||||
}
|
||||
203
src/fetching/scoringdance.rs
Normal file
203
src/fetching/scoringdance.rs
Normal file
@@ -0,0 +1,203 @@
|
||||
use std::str::FromStr;
|
||||
|
||||
use reqwest::ClientBuilder;
|
||||
use scraper::{ElementRef, Html, Selector};
|
||||
|
||||
use crate::{
|
||||
app_signature,
|
||||
dance_info::{CompState, DanceInfo, DanceRank, DanceRole},
|
||||
fetching::DanceInfoError,
|
||||
};
|
||||
#[derive(thiserror::Error, Debug)]
|
||||
pub enum ScoringParseError {
|
||||
#[error("Could not parse: {0}")]
|
||||
ParseMismatch(String),
|
||||
}
|
||||
|
||||
fn parse_card(t: ElementRef) -> Result<(String, Vec<Vec<String>>), ScoringParseError> {
|
||||
#[allow(clippy::unwrap_used)]
|
||||
let title_selector = Selector::parse("div.card-header").unwrap();
|
||||
#[allow(clippy::unwrap_used)]
|
||||
let table_selector = Selector::parse("div.card-body > table").unwrap();
|
||||
#[allow(clippy::unwrap_used)]
|
||||
let row_selector = Selector::parse("tr").unwrap();
|
||||
#[allow(clippy::unwrap_used)]
|
||||
let cell_selector = Selector::parse("th,td").unwrap();
|
||||
let table = t
|
||||
.select(&table_selector)
|
||||
.next()
|
||||
.ok_or_else(|| ScoringParseError::ParseMismatch("table".to_owned()))?;
|
||||
let title = t
|
||||
.select(&title_selector)
|
||||
.next()
|
||||
.ok_or_else(|| ScoringParseError::ParseMismatch("title".to_owned()))?
|
||||
.text()
|
||||
.collect::<Vec<_>>()
|
||||
.join("")
|
||||
.trim()
|
||||
.to_owned();
|
||||
|
||||
let parsed_table = table
|
||||
.select(&row_selector)
|
||||
.map(|row| {
|
||||
row.select(&cell_selector)
|
||||
.map(|v| v.text().collect::<String>().trim().to_string())
|
||||
.collect::<Vec<_>>()
|
||||
})
|
||||
.collect::<Vec<_>>();
|
||||
Ok((title, parsed_table))
|
||||
}
|
||||
|
||||
fn parse_details(d: &[Vec<String>]) -> Result<(String, String), ScoringParseError> {
|
||||
let first_name_row = d.iter().find(|v| {
|
||||
v.first()
|
||||
.is_some_and(|v| v.to_lowercase().contains("first name"))
|
||||
});
|
||||
let last_name_row = d.iter().find(|v| {
|
||||
v.first()
|
||||
.is_some_and(|v| v.to_lowercase().contains("last name"))
|
||||
});
|
||||
let first_name = first_name_row
|
||||
.ok_or_else(|| ScoringParseError::ParseMismatch("first_name".to_owned()))?
|
||||
.last()
|
||||
.ok_or_else(|| ScoringParseError::ParseMismatch("first_name value".to_owned()))?;
|
||||
let last_name = last_name_row
|
||||
.ok_or_else(|| ScoringParseError::ParseMismatch("last_name".to_owned()))?
|
||||
.last()
|
||||
.ok_or_else(|| ScoringParseError::ParseMismatch("first_name value".to_owned()))?;
|
||||
Ok((first_name.clone(), last_name.clone()))
|
||||
}
|
||||
|
||||
fn parse_stats(
|
||||
d: &[Vec<String>],
|
||||
) -> Result<(DanceRole, CompState, Option<CompState>), ScoringParseError> {
|
||||
let chapters = d.chunk_by(|_, b| b.len() != 1).map(|v| {
|
||||
let (a, b) = v.split_first().unwrap();
|
||||
let a = a.first().unwrap();
|
||||
let b = b
|
||||
.iter()
|
||||
.map(|v| [v.first().unwrap(), v.last().unwrap()])
|
||||
.collect::<Vec<_>>();
|
||||
(a, b)
|
||||
});
|
||||
|
||||
let (all_time, rest) = chapters.partition::<Vec<_>, _>(|(a, _)| a.to_lowercase() == "all time");
|
||||
let all_time = &all_time.split_first().unwrap().0.1;
|
||||
let mut sorted_chapters = rest
|
||||
.into_iter()
|
||||
.map(|(chapter, items)| {
|
||||
let rank: DanceRank = serde_plain::from_str(chapter).map_err(|_| chapter.to_owned())?;
|
||||
Ok::<(DanceRank, Vec<[&String; 2]>), String>((rank, items))
|
||||
})
|
||||
.filter_map(|v| match v {
|
||||
Ok(v) => Some(v),
|
||||
Err(e) => {
|
||||
tracing::warn!("Unknown chapter in parsed html: {e}");
|
||||
None
|
||||
}
|
||||
})
|
||||
.collect::<Vec<_>>();
|
||||
sorted_chapters.sort_by_key(|(a, _)| *a);
|
||||
sorted_chapters.reverse();
|
||||
let leader_rank = sorted_chapters.iter().find_map(|(rank, items)| {
|
||||
items
|
||||
.iter()
|
||||
.find(|[a, _]| a.to_lowercase().contains("points leader"))
|
||||
.map(|[_, points]| (rank, points))
|
||||
});
|
||||
let follower_rank = sorted_chapters.iter().find_map(|(rank, items)| {
|
||||
items
|
||||
.iter()
|
||||
.find(|[a, _]| a.to_lowercase().contains("points follower"))
|
||||
.map(|[_, points]| (rank, points))
|
||||
});
|
||||
let primary_role: DanceRole = all_time
|
||||
.iter()
|
||||
.find_map(|[key, value]| {
|
||||
if !key.to_lowercase().contains("primary role") {
|
||||
return None;
|
||||
}
|
||||
Some(value)
|
||||
})
|
||||
.map(|arg0: &&std::string::String| DanceRole::from_str(arg0.as_str()).unwrap())
|
||||
.unwrap();
|
||||
|
||||
let ((rank, points), non_d) = match primary_role {
|
||||
DanceRole::Leader => (leader_rank.unwrap(), follower_rank),
|
||||
DanceRole::Follower => (follower_rank.unwrap(), leader_rank),
|
||||
};
|
||||
let dominant_comp = CompState {
|
||||
points: points.parse().unwrap(),
|
||||
rank: *rank,
|
||||
};
|
||||
let non_dominant_comp = non_d.map(|(rank, points)| CompState {
|
||||
points: points.parse().unwrap(),
|
||||
rank: *rank,
|
||||
});
|
||||
Ok((primary_role, dominant_comp, non_dominant_comp))
|
||||
|
||||
// dbg!(chapters.collect::<Vec<_>>());
|
||||
}
|
||||
|
||||
fn extract_tables(html: &str) -> Result<Vec<(String, Vec<Vec<String>>)>, ScoringParseError> {
|
||||
dbg!(&html);
|
||||
let document = Html::parse_document(html);
|
||||
let card_selector = Selector::parse("div:has( > div.card-header)").unwrap();
|
||||
document
|
||||
.select(&card_selector)
|
||||
.inspect(|v| {
|
||||
dbg!(&v);
|
||||
})
|
||||
.map(parse_card)
|
||||
.collect()
|
||||
}
|
||||
|
||||
fn parse_info(html: &str) -> Result<DanceInfo, ScoringParseError> {
|
||||
let tables = extract_tables(html)?;
|
||||
dbg!(&tables);
|
||||
let details = &tables
|
||||
.iter()
|
||||
.find(|(v, _)| v.to_lowercase().contains("detail"))
|
||||
.ok_or_else(|| ScoringParseError::ParseMismatch("detail card".to_owned()))?
|
||||
.1;
|
||||
let stats = &tables
|
||||
.iter()
|
||||
.find(|(v, _)| v.to_lowercase().contains("stats"))
|
||||
.ok_or_else(|| ScoringParseError::ParseMismatch("stats card".to_owned()))?
|
||||
.1;
|
||||
|
||||
let (dominant_role, dominant_role_comp, non_dominant_role_comp) = parse_stats(stats)?;
|
||||
let (firstname, lastname) = parse_details(details)?;
|
||||
|
||||
Ok(DanceInfo {
|
||||
firstname,
|
||||
lastname,
|
||||
dominant_role,
|
||||
dominant_role_comp,
|
||||
non_dominant_role_comp,
|
||||
})
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn test_parse_table() {
|
||||
dbg!(parse_info(include_str!("../../test_data/2026-01-07_robert.html")));
|
||||
}
|
||||
|
||||
pub async fn fetch_wsdc_info_scoring_dance(id: u32) -> Result<DanceInfo, DanceInfoError> {
|
||||
let client = ClientBuilder::new()
|
||||
.user_agent(app_signature())
|
||||
.build()
|
||||
.map_err(DanceInfoError::ClientBuild)?;
|
||||
|
||||
let url = format!("https://scoring.dance/enUS/wsdc/registry/{id}.html");
|
||||
let request = client
|
||||
.request(reqwest::Method::GET, url)
|
||||
.build()
|
||||
.map_err(DanceInfoError::RequestBuild)?;
|
||||
let response = client
|
||||
.execute(request)
|
||||
.await
|
||||
.map_err(DanceInfoError::Request)?;
|
||||
|
||||
parse_info(response.text().await.unwrap().as_str()).map_err(DanceInfoError::HtmlParse)
|
||||
}
|
||||
67
src/fetching/types.rs
Normal file
67
src/fetching/types.rs
Normal file
@@ -0,0 +1,67 @@
|
||||
use crate::dance_info::{CompState, DanceInfo, DanceRank, DanceRole};
|
||||
|
||||
#[derive(thiserror::Error, Debug)]
|
||||
pub enum DanceInfoError {
|
||||
#[error("Failed to build client: {0}")]
|
||||
ClientBuild(reqwest::Error),
|
||||
#[error("Failed to build request: {0}")]
|
||||
RequestBuild(reqwest::Error),
|
||||
#[error("Request error: {0:#?}")]
|
||||
Request(reqwest::Error),
|
||||
#[error("Failed to parse response: {0}")]
|
||||
JsonParse(reqwest::Error),
|
||||
#[error("Failed to parse html: {0}")]
|
||||
HtmlParse(#[from] super::scoringdance::ScoringParseError),
|
||||
}
|
||||
|
||||
#[derive(serde::Deserialize, Debug)]
|
||||
enum OptionalDanceRank {
|
||||
#[serde(rename = "N/A")]
|
||||
NotAvailable,
|
||||
#[serde(untagged)]
|
||||
Rank(DanceRank),
|
||||
}
|
||||
|
||||
#[derive(serde::Deserialize, Debug)]
|
||||
enum OptionalDancePoints {
|
||||
#[serde(rename = "N/A")]
|
||||
NotAvailable,
|
||||
#[serde(untagged)]
|
||||
Points(u16),
|
||||
}
|
||||
|
||||
#[derive(serde::Deserialize, Debug)]
|
||||
pub struct DanceInfoParser {
|
||||
pub dancer_first: String,
|
||||
pub dancer_last: String,
|
||||
pub short_dominate_role: DanceRole,
|
||||
#[allow(dead_code)]
|
||||
pub short_non_dominate_role: DanceRole,
|
||||
pub dominate_role_highest_level_points: u16,
|
||||
pub dominate_role_highest_level: DanceRank,
|
||||
pub non_dominate_role_highest_level_points: OptionalDancePoints,
|
||||
pub non_dominate_role_highest_level: OptionalDanceRank,
|
||||
}
|
||||
|
||||
impl From<DanceInfoParser> for DanceInfo {
|
||||
fn from(value: DanceInfoParser) -> Self {
|
||||
let non_dominant_role_comp = if let OptionalDanceRank::Rank(r) =
|
||||
value.non_dominate_role_highest_level
|
||||
&& let OptionalDancePoints::Points(l) = value.non_dominate_role_highest_level_points
|
||||
{
|
||||
Some(CompState { rank: r, points: l })
|
||||
} else {
|
||||
None
|
||||
};
|
||||
Self {
|
||||
firstname: value.dancer_first,
|
||||
lastname: value.dancer_last,
|
||||
dominant_role: value.short_dominate_role,
|
||||
dominant_role_comp: CompState {
|
||||
rank: value.dominate_role_highest_level,
|
||||
points: value.dominate_role_highest_level_points,
|
||||
},
|
||||
non_dominant_role_comp,
|
||||
}
|
||||
}
|
||||
}
|
||||
61
src/fetching/worldsdc.rs
Normal file
61
src/fetching/worldsdc.rs
Normal file
@@ -0,0 +1,61 @@
|
||||
use std::collections::HashMap;
|
||||
|
||||
use crate::{
|
||||
app_signature,
|
||||
dance_info::{CompState, DanceInfo, DanceRank, DanceRole},
|
||||
fetching::{DanceInfoError, types::DanceInfoParser, scoringdance::{self, fetch_wsdc_info_scoring_dance}},
|
||||
};
|
||||
use reqwest::ClientBuilder;
|
||||
|
||||
pub async fn fetch_wsdc_info_wsdc(id: u32) -> Result<DanceInfo, DanceInfoError> {
|
||||
let client = ClientBuilder::new()
|
||||
.user_agent(app_signature())
|
||||
.build()
|
||||
.map_err(DanceInfoError::ClientBuild)?;
|
||||
|
||||
let mut params = HashMap::new();
|
||||
|
||||
let url = if cfg!(test) {
|
||||
// "https://o5grQU3Y.free.beeceptor.com/lookup2020/find"
|
||||
"http://localhost:8000"
|
||||
} else {
|
||||
"https://points.worldsdc.com/lookup2020/find"
|
||||
};
|
||||
params.insert("num", id.to_string());
|
||||
let request = client
|
||||
.request(reqwest::Method::POST, url)
|
||||
.form(¶ms)
|
||||
.build()
|
||||
.map_err(DanceInfoError::RequestBuild)?;
|
||||
let response = client
|
||||
.execute(request)
|
||||
.await
|
||||
.map_err(DanceInfoError::Request)?;
|
||||
|
||||
let x: DanceInfoParser = response.json().await.map_err(DanceInfoError::JsonParse)?;
|
||||
Ok(x.into())
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
#![allow(clippy::unwrap_used, reason = "Allow unwrap in tests")]
|
||||
use crate::fetching::fetch_wsdc_info;
|
||||
|
||||
#[test]
|
||||
#[ignore = "Only run when the mock api is setup"]
|
||||
fn test_fetch_wsdc() {
|
||||
let rt = match tokio::runtime::Builder::new_current_thread()
|
||||
.enable_all()
|
||||
.build()
|
||||
{
|
||||
Ok(o) => o,
|
||||
Err(e) => {
|
||||
tracing::error!("Could not start runtime: {e}");
|
||||
return;
|
||||
}
|
||||
};
|
||||
let x = rt.block_on(fetch_wsdc_info(7));
|
||||
dbg!(&x);
|
||||
x.unwrap();
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user