robin/ugc-scaper
1
0
Fork 0
mirror of https://codeberg.org/icewind/ugc-scaper.git synced 2026-06-03 18:24:10 +02:00
Code Issues Projects Releases Packages Wiki Activity

improve date matching

Browse source
This commit is contained in:
Robin Appelman 2025-04-20 00:21:48 +02:00
commit 8dfd870f82
2 changed files with 25 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

29
archiver/src/archive.rs
View file

@ -7,13 +7,22 @@ use thiserror::Error;
use time::format_description::FormatItem; use time::format_description::FormatItem;
use time::macros::format_description; use time::macros::format_description;
use time::parsing::Parsed; use time::parsing::Parsed;
use time::Date; use time::{Date, Duration};
use tokio_stream::Stream; use tokio_stream::Stream;
use tracing::{debug};
use ugc_scraper_types::{ use ugc_scraper_types::{
Class, GameMode, MapHistory, MatchInfo, Membership, MembershipRole, NameChange, Player, Record, Class, GameMode, MapHistory, MatchInfo, Membership, MembershipRole, NameChange, Player, Record,
Region, RosterHistory, SteamID, Team, Region, RosterHistory, SteamID, Team,
}; };
const MATCH_DATE_FORMAT: &[FormatItem<'static>] = format_description!(
"[weekday case_sensitive:false repr:short], [month repr:short] [day padding:none] [year]"
);
const MATCH_DATE_FORMAT2: &[FormatItem<'static>] = format_description!(
"[weekday case_sensitive:false repr:short] [month repr:short] [day padding:none] [year]"
);
const MATCH_DATE_FORMATS: &[&[FormatItem<'static>]] = &[MATCH_DATE_FORMAT, MATCH_DATE_FORMAT2];
#[derive(Debug, Error)] #[derive(Debug, Error)]
pub enum ArchiveError { pub enum ArchiveError {
#[error("Invalid database {url}: {error:#}")] #[error("Invalid database {url}: {error:#}")]
@ -505,8 +514,6 @@ impl Archive {
&self, &self,
match_info: &MatchInfo, match_info: &MatchInfo,
) -> Result<Option<Date>, ArchiveError> { ) -> Result<Option<Date>, ArchiveError> {
const MATCH_DATE_FORMAT: &[FormatItem<'static>] = format_description!("[weekday case_sensitive:false repr:short], [month repr:short] [day padding:none] [year]");
const MATCH_DATE_FORMAT2: &[FormatItem<'static>] = format_description!("[weekday case_sensitive:false repr:short] [month repr:short] [day padding:none] [year]");
if let Ok(match_date) = parse_old_match_date(&match_info.default_date) { if let Ok(match_date) = parse_old_match_date(&match_info.default_date) {
return Ok(Some(match_date)); return Ok(Some(match_date));
} else if !match_info.format.is_tf2() { } else if !match_info.format.is_tf2() {
@ -531,12 +538,12 @@ impl Archive {
let Some(match_date) = try_date_formats( let Some(match_date) = try_date_formats(
match_info.default_date.as_str(), match_info.default_date.as_str(),
possible_date.year(), possible_date.year(),
&[MATCH_DATE_FORMAT, MATCH_DATE_FORMAT2], MATCH_DATE_FORMATS,
) else { ) else {
break; break;
}; };
if match_date == possible_date { if (match_date - possible_date).abs() < Duration::days(7) {
return Ok(Some(possible_date)); return Ok(Some(possible_date));
} }
} }
@ -605,9 +612,19 @@ fn test_parse_old_match_date() {
fn try_date_formats(date: &str, year: i32, formats: &[&[FormatItem<'static>]]) -> Option<Date> { fn try_date_formats(date: &str, year: i32, formats: &[&[FormatItem<'static>]]) -> Option<Date> {
for format in formats { for format in formats {
if let Ok(match_date) = Date::parse(&format!("{} {}", date, year), format) { match Date::parse(&format!("{} {}", date, year), format) {
Ok(match_date) => {
return Some(match_date); return Some(match_date);
}
Err(e) => {
debug!(error = ?e, year, date, "date format not matching");
}
}; };
} }
None None
} }
#[test]
fn test_parse_date() {
assert!(try_date_formats("Sun Oct 06", 2019, MATCH_DATE_FORMATS).is_some());
}

1
archiver/src/main.rs
View file

@ -227,6 +227,7 @@ async fn fixup_matches(client: &UgcClient, archive: &Archive) -> MainResult {
|| match_info.format == GameMode::Fours || match_info.format == GameMode::Fours
|| match_info.format == GameMode::Ultiduo) || match_info.format == GameMode::Ultiduo)
{ {
dbg!(match_info.default_date);
error!("failed to parse match date"); error!("failed to parse match date");
panic!(); panic!();
} }