yolo/arr
3
0
Fork 0
Code Issues 12 Pull Requests Actions Packages Projects Releases Wiki Activity

Merge #123: episode and season-pack matching in arr-core

Browse Source 
Closes #123
This commit is contained in:
Miguel Palhas
2026-08-23 17:10:05 +01:00
parent 8b8e1dcec9 e6d3ea4964
commit 237696417b
2 changed files with 432 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
crates/arr-core/src/lib.rs
+4 -1
View File
@@ -15,7 +15,10 @@ pub mod tracking;
pub use arr_parse::NameClaims as ParsedRelease;
pub use grabbing::{season_grab_mode, SeasonGrabFacts, SeasonGrabMode};
pub use matching::{match_movie, MatchKind, MovieMatch, ReleaseIds, WantedMovie};
pub use matching::{
match_episode, match_movie, EpisodeMatch, MatchKind, MatchShape, MovieMatch, ReleaseIds,
WantedEpisode, WantedMovie,
};
pub use score::{Score, ScoreWeights};
pub use status::{derive_series_status, SeriesStatus};
crates/arr-core/src/matching.rs
+428 -3
View File
@@ -14,9 +14,13 @@
//! - a title match needs the years to agree, and a wanted title with a known
//! year is never matched by a release that does not state one;
//! - anything that matches two wanted titles matches neither.
//!
//! The TV side ([`match_episode`]) answers a narrower question about one
//! release and one wanted episode: is this its file, or a pack for exactly
//! the season holding it?
use crate::MovieId;
use arr_parse::NameClaims;
use crate::{EpisodeId, MovieId};
use arr_parse::{EpisodeClaim, NameClaims};
/// A wanted title, in the shape matching needs.
#[derive(Clone, Debug, Eq, PartialEq)]
@@ -150,6 +154,113 @@ fn unique(mut matches: impl Iterator<Item = MovieMatch>) -> Option<MovieMatch> {
matches.next().is_none().then_some(first)
}
/// One wanted episode, in the shape matching needs. The ids are the
/// series', not the episode's: trackers attach them to the show.
#[derive(Clone, Debug, Eq, PartialEq)]
pub struct WantedEpisode {
pub id: EpisodeId,
/// TMDB id of the series, compared against the one the tracker attached
/// to the item.
pub tmdb_id: Option<u32>,
/// `IMDb` id of the series, as TMDB spells it (`tt11301866`). Absent
/// until the metadata refresh has run.
pub imdb_id: Option<String>,
pub title: String,
pub season: u32,
pub episode: u32,
}
/// Which file shape satisfied a wanted episode.
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
pub enum MatchShape {
/// A release holding exactly this one episode.
SingleEpisode,
/// A pack for exactly the season this episode belongs to.
SeasonPack,
}
/// What tied a release to a wanted episode, and what it is.
#[derive(Clone, Copy, Debug, Eq, PartialEq)]
pub struct EpisodeMatch {
pub episode: EpisodeId,
pub kind: MatchKind,
pub shape: MatchShape,
}
/// Whether a release satisfies one wanted episode, and as what.
///
/// Identity is decided as [`match_movie`] decides it — a supplied id that
/// matches nothing wanted ends the comparison, a junk id falls through to
/// the title — except the title lane compares the series title alone. TV
/// release names rarely state a year at all, and when they do it is the
/// season's air year rather than the series', so the year rule films live
/// by would starve this lane; exact title comparison is what keeps two
/// series sharing a word apart.
///
/// Only two shapes satisfy anything: a file holding exactly this episode
/// and a pack for exactly its season. Multi-episode files and multi-season
/// packs are left to the operator, who reaches them through the episode and
/// season decks (§6.2).
#[must_use]
pub fn match_episode(
wanted: &WantedEpisode,
ids: &ReleaseIds,
claims: &NameClaims,
) -> Option<EpisodeMatch> {
let imdb_id = ids.imdb_id.as_deref().and_then(imdb_key);
let kind = if ids.tmdb_id.is_some() || imdb_id.is_some() {
if ids.tmdb_id.is_some() && wanted.tmdb_id == ids.tmdb_id {
MatchKind::TmdbId
} else if imdb_id.is_some() && wanted.imdb_id.as_deref().and_then(imdb_key) == imdb_id {
MatchKind::ImdbId
} else {
return None;
}
} else {
let title = match_key(claims.title.as_deref()?);
if title.is_empty() || title != match_key(&wanted.title) {
return None;
}
MatchKind::TitleAndYear
};
// Whatever the name or id says, nothing without a TV tag holds an
// episode, and no daily-show air date is a numbered one.
let claim = claims.episode.as_ref()?;
let shape = if is_this_seasons_pack(claim, wanted.season) {
MatchShape::SeasonPack
} else if is_single_episode(claim, wanted.season, wanted.episode) {
MatchShape::SingleEpisode
} else {
return None;
};
Some(EpisodeMatch {
episode: wanted.id,
kind,
shape,
})
}
/// A pack for exactly this season. Multi-season packs are never selected
/// automatically: they drag other seasons in with them.
fn is_this_seasons_pack(claim: &EpisodeClaim, season: u32) -> bool {
match claim {
EpisodeClaim::Season { season: claimed } => *claimed == season,
EpisodeClaim::Seasons { first, last } => first == last && *first == season,
EpisodeClaim::Episodes { .. } | EpisodeClaim::Daily { .. } => false,
}
}
/// A release holding exactly this one episode. Multi-episode files are left
/// to manual grabs: two episodes automatically picking the same torrent
/// would fight over one `grabs` row.
fn is_single_episode(claim: &EpisodeClaim, season: u32, episode: u32) -> bool {
matches!(
claim,
EpisodeClaim::Episodes { episodes, .. } if episodes.len() == 1
) && claim.covers(season, episode)
}
/// The form two titles are compared in: lower case, accents folded to ASCII,
/// `&` spelled out, and every run of anything else one space.
///
@@ -202,7 +313,10 @@ fn fold(character: char) -> Option<&'static str> {
#[cfg(test)]
mod tests {
use super::{match_key, match_movie, MatchKind, MovieMatch, ReleaseIds, WantedMovie};
use super::{
match_episode, match_key, match_movie, EpisodeId, EpisodeMatch, MatchKind, MatchShape,
MovieMatch, ReleaseIds, WantedEpisode, WantedMovie,
};
use crate::MovieId;
fn wanted() -> Vec<WantedMovie> {
@@ -440,4 +554,315 @@ mod tests {
})
);
}
fn wanted_episode() -> WantedEpisode {
WantedEpisode {
id: EpisodeId(7),
tmdb_id: Some(934_644),
imdb_id: Some("tt11301866".to_owned()),
title: "Fallout".to_owned(),
season: 1,
episode: 3,
}
}
fn episode_ids(tmdb_id: Option<u32>, imdb_id: Option<&str>) -> ReleaseIds {
ids(tmdb_id, imdb_id)
}
/// `(release name, expected shape)` for releases that satisfy the
/// wanted `Fallout` S01E03, matched on title alone.
#[test]
fn real_single_episode_releases_match_their_episode() {
let cases = [
(
"Fallout.2024.S01E03.1080p.WEB-DL.DDP5.1.H.264-GROUP",
MatchShape::SingleEpisode,
),
(
"Fallout.S01E03.2160p.WEB-DL.DV.HDR10Plus.HEVC-GROUP",
MatchShape::SingleEpisode,
),
(
"Fallout.S01E03.720p.AMZN.WEBRip.DDP5.1.x264-GROUP",
MatchShape::SingleEpisode,
),
];
for (name, shape) in cases {
assert_eq!(
match_episode(
&wanted_episode(),
&ReleaseIds::default(),
&arr_parse::parse(name),
),
Some(EpisodeMatch {
episode: EpisodeId(7),
kind: MatchKind::TitleAndYear,
shape,
}),
"{name}"
);
}
}
/// The same series, offered as a pack for exactly its season.
#[test]
fn a_pack_for_this_season_matches_every_episode_in_it() {
let cases = [
(
"Fallout.S01.2160p.WEB-DL.DDP5.1.Atmos.H.265-GROUP",
MatchShape::SeasonPack,
),
(
"Fallout.Season.1.1080p.BluRay.x265-GROUP",
MatchShape::SeasonPack,
),
(
"Fallout.S01.1080p.BluRay.x264-GROUP",
MatchShape::SeasonPack,
),
];
for (name, shape) in cases {
assert_eq!(
match_episode(
&wanted_episode(),
&ReleaseIds::default(),
&arr_parse::parse(name),
),
Some(EpisodeMatch {
episode: EpisodeId(7),
kind: MatchKind::TitleAndYear,
shape,
}),
"{name}"
);
}
}
/// Punctuation, case and accents in the series title do not block the
/// match.
#[test]
fn mangled_series_titles_still_match() {
let wanted = WantedEpisode {
id: EpisodeId(8),
tmdb_id: None,
imdb_id: None,
title: "Ms. Marvel".to_owned(),
season: 1,
episode: 3,
};
assert_eq!(
match_episode(
&wanted,
&ReleaseIds::default(),
&arr_parse::parse("Ms.Marvel.S01E03.1080p.DSNY.WEB-DL.DDP5.1.Atmos.H.264-GROUP"),
)
.map(|matched| matched.shape),
Some(MatchShape::SingleEpisode)
);
}
/// Two series that share a word are different series. Both pairs are
/// real shows.
#[test]
fn another_series_sharing_a_word_is_not_a_match() {
let lone_star = WantedEpisode {
id: EpisodeId(9),
tmdb_id: None,
imdb_id: None,
title: "9-1-1: Lone Star".to_owned(),
season: 4,
episode: 1,
};
assert_eq!(
match_episode(
&lone_star,
&ReleaseIds::default(),
&arr_parse::parse("9-1-1.S07E01.1080p.AMZN.WEB-DL.DDP5.1.H.264-GROUP"),
),
None
);
let fallout = wanted_episode();
assert_eq!(
match_episode(
&fallout,
&ReleaseIds::default(),
&arr_parse::parse("The.Fallout.2021.S01E03.1080p.WEB-DL"),
),
None
);
}
/// A tracker-supplied id identifies the series whatever the name says.
#[test]
fn a_supplied_id_matches_whatever_the_name_says_tv() {
for (tmdb_id, imdb_id) in [(Some(934_644), None), (None, Some("tt11301866"))] {
assert_eq!(
match_episode(
&wanted_episode(),
&episode_ids(tmdb_id, imdb_id),
&arr_parse::parse("Totally.Unrelated.Name.S01.1080p.WEB-DL-GROUP"),
),
Some(EpisodeMatch {
episode: EpisodeId(7),
kind: if tmdb_id.is_some() {
MatchKind::TmdbId
} else {
MatchKind::ImdbId
},
shape: MatchShape::SeasonPack,
}),
"{tmdb_id:?} {imdb_id:?}"
);
}
}
/// Torznab definitions spell the same id several ways; all of them are
/// the one series.
#[test]
fn an_imdb_id_matches_however_the_tracker_spells_it_tv() {
for spelling in ["tt11301866", "11301866", "tt011301866"] {
assert_eq!(
match_episode(
&wanted_episode(),
&episode_ids(None, Some(spelling)),
&arr_parse::parse("Unrelated.Name.S01E03.1080p.WEB-DL"),
)
.map(|matched| matched.kind),
Some(MatchKind::ImdbId),
"{spelling}"
);
}
}
/// Same rule as films: the tracker said which series this is and it is
/// not ours, so the title does not get a vote.
#[test]
fn a_supplied_id_that_is_not_wanted_ends_the_comparison_tv() {
assert_eq!(
match_episode(
&wanted_episode(),
&episode_ids(Some(1), None),
&arr_parse::parse("Fallout.S01E03.1080p.WEB-DL"),
),
None
);
assert_eq!(
match_episode(
&wanted_episode(),
&episode_ids(None, Some("tt0000001")),
&arr_parse::parse("Fallout.S01E03.1080p.WEB-DL"),
),
None
);
}
/// A placeholder or a non-id is no id at all, so the title lane still
/// runs.
#[test]
fn a_junk_imdb_id_falls_through_to_the_title_tv() {
for junk in ["", "tt", "tt0000000", "n/a"] {
assert_eq!(
match_episode(
&wanted_episode(),
&episode_ids(None, Some(junk)),
&arr_parse::parse("Fallout.S01E03.1080p.WEB-DL"),
)
.map(|matched| matched.shape),
Some(MatchShape::SingleEpisode),
"{junk}"
);
}
}
/// Everything that is not this exact episode or this exact season's
/// pack satisfies nothing, whoever issued it.
#[test]
fn every_other_shape_satisfies_nothing() {
let refusals = [
// Wrong episode, wrong season.
"Fallout.S01E04.1080p.WEB-DL-GROUP",
"Fallout.S02E03.1080p.WEB-DL-GROUP",
// Multi-episode files fight over one grabs row; manual only.
"Fallout.S01E02E03.1080p.WEB-DL-GROUP",
"Fallout.S01E02-E04.1080p.WEB-DL-GROUP",
// Another season's pack, and a multi-season pack.
"Fallout.S02.2160p.WEB-DL-GROUP",
"Fallout.S01-S03.1080p.BluRay-GROUP",
// A daily-show air date is not a numbered episode.
"Last.Week.Tonight.With.John.Oliver.2024.05.13.1080p.WEB-DL",
// No TV tag at all: a film, not an episode of anything.
"A.Complete.Unknown.2024.2160p.WEB-DL-GROUP",
];
for name in refusals {
assert_eq!(
match_episode(
&wanted_episode(),
&ReleaseIds::default(),
&arr_parse::parse(name),
),
None,
"{name}"
);
}
}
/// `S01-S01` does not parse as a claim (`arr-parse` takes only a real
/// range), but the lifted rule is what a degenerate range would mean:
/// one season's pack, not a multi-season one.
#[test]
fn a_degenerate_season_range_is_this_seasons_pack() {
use super::{is_single_episode, is_this_seasons_pack};
use arr_parse::EpisodeClaim::{Daily, Episodes, Season, Seasons};
assert!(is_this_seasons_pack(&Seasons { first: 1, last: 1 }, 1));
assert!(!is_this_seasons_pack(&Seasons { first: 1, last: 3 }, 1));
assert!(!is_this_seasons_pack(
&Episodes {
season: 1,
episodes: vec![3]
},
1
));
assert!(!is_this_seasons_pack(&Season { season: 2 }, 1));
assert!(!is_this_seasons_pack(
&Daily {
year: 2024,
month: 5,
day: 13
},
1
));
// A single episode of another season is still not this one.
assert!(!is_single_episode(
&Episodes {
season: 2,
episodes: vec![3]
},
1,
3
));
}
/// A daily show's own episode, dated but unnumbered, satisfies nothing:
/// there is no numbered wanted episode it can be filed against.
#[test]
fn a_daily_claim_never_satisfies_a_numbered_wanted_episode() {
let wanted = WantedEpisode {
id: EpisodeId(10),
tmdb_id: Some(46_298),
imdb_id: None,
title: "Last Week Tonight with John Oliver".to_owned(),
season: 11,
episode: 17,
};
assert_eq!(
match_episode(
&wanted,
&ReleaseIds::default(),
&arr_parse::parse("Last.Week.Tonight.With.John.Oliver.2024.05.13.1080p.WEB-DL"),
),
None
);
}
}