Files
arr/crates/arr-parse/tests/parse.rs
T
Miguel Palhas 8d79c7c378 fix(parse): pack titles and degenerate ranges
COMPLETE before the season tag was weak junk, so it stayed in
the title and exact-title matching failed silently (#135). It
now closes the boundary wherever it lands.

S01-S01 produced no claim because the dash-range branch required
last > season; a degenerate range is one season's pack, matching
the rule match_episode lifted.
2026-08-23 18:02:39 +01:00

474 lines
16 KiB
Rust

//! Table-driven checks over representative release names. The full fixture
//! corpus pulled from real indexers is issue #9; these rows pin the parser's
//! behaviour per shape of name.
use arr_parse::{
parse, Codec, Edition, EpisodeClaim, HdrMarker, LanguageMarker, NameClaims, Resolution, Source,
};
use serde as _;
struct Case {
name: &'static str,
want: NameClaims,
}
#[allow(clippy::unnecessary_wraps)]
fn s(v: &str) -> Option<String> {
Some(v.to_string())
}
#[allow(clippy::too_many_lines)]
fn cases() -> Vec<Case> {
vec![
Case {
name: "Movie.Title.2024.2160p.UHD.BluRay.DV.HDR10.x265.TrueHD-GROUP",
want: NameClaims {
title: s("Movie Title"),
year: Some(2024),
resolution: Some(Resolution::P2160),
source: Some(Source::BluRay),
codec: Some(Codec::X265),
hdr: vec![HdrMarker::DolbyVision, HdrMarker::Hdr10],
group: s("GROUP"),
..NameClaims::default()
},
},
Case {
// "Web" before the year is title, not a source claim.
name: "Charlottes.Web.2006.1080p.BluRay.x264-GRP",
want: NameClaims {
title: s("Charlottes Web"),
year: Some(2006),
resolution: Some(Resolution::P1080),
source: Some(Source::BluRay),
codec: Some(Codec::X264),
group: s("GRP"),
..NameClaims::default()
},
},
Case {
name: "Dune Part Two (2024) [2160p] [WEB-DL] [HDR10+] [YTS.MX]",
want: NameClaims {
title: s("Dune Part Two"),
year: Some(2024),
resolution: Some(Resolution::P2160),
source: Some(Source::WebDl),
hdr: vec![HdrMarker::Hdr10Plus],
group: s("YTS.MX"),
..NameClaims::default()
},
},
Case {
name: "O.Auto.da.Compadecida.2000.NACIONAL.1080p.WEB-DL.x264-BRA",
want: NameClaims {
title: s("O Auto da Compadecida"),
year: Some(2000),
resolution: Some(Resolution::P1080),
source: Some(Source::WebDl),
codec: Some(Codec::X264),
languages: vec![LanguageMarker::PtBr],
group: s("BRA"),
..NameClaims::default()
},
},
Case {
// Repeated pt-BR markers dedupe to one claim.
name: "Filme.2023.Dublado.PT-BR.1080p.WEBRip.x264",
want: NameClaims {
title: s("Filme"),
year: Some(2023),
resolution: Some(Resolution::P1080),
source: Some(Source::WebRip),
codec: Some(Codec::X264),
languages: vec![LanguageMarker::PtBr],
..NameClaims::default()
},
},
Case {
name: "Filme.Infantil.2023.1080p.WEB-DL.Dual.Áudio.x264",
want: NameClaims {
title: s("Filme Infantil"),
year: Some(2023),
resolution: Some(Resolution::P1080),
source: Some(Source::WebDl),
codec: Some(Codec::X264),
languages: vec![LanguageMarker::Dual],
..NameClaims::default()
},
},
Case {
name: "Movie.2024.MULTi.1080p.WEB.x265-GRP",
want: NameClaims {
title: s("Movie"),
year: Some(2024),
resolution: Some(Resolution::P1080),
source: Some(Source::WebDl),
codec: Some(Codec::X265),
languages: vec![LanguageMarker::Multi],
group: s("GRP"),
..NameClaims::default()
},
},
Case {
// Season/episode tags bound the title; no year needed.
name: "Serie.S01E02.720p.HDTV.x264-LOL",
want: NameClaims {
title: s("Serie"),
resolution: Some(Resolution::P720),
source: Some(Source::Hdtv),
codec: Some(Codec::X264),
group: s("LOL"),
episode: Some(EpisodeClaim::Episodes {
season: 1,
episodes: vec![2],
}),
..NameClaims::default()
},
},
Case {
// Remux beats the BluRay marker riding along.
name: "Movie.2024.2160p.BluRay.REMUX.HEVC.TrueHD.7.1.Atmos-FGT",
want: NameClaims {
title: s("Movie"),
year: Some(2024),
resolution: Some(Resolution::P2160),
source: Some(Source::Remux),
codec: Some(Codec::X265),
group: s("FGT"),
..NameClaims::default()
},
},
Case {
name: "Movie.2024.Extended.1080p.BluRay.x264",
want: NameClaims {
title: s("Movie"),
year: Some(2024),
resolution: Some(Resolution::P1080),
source: Some(Source::BluRay),
codec: Some(Codec::X264),
editions: vec![Edition::Extended],
..NameClaims::default()
},
},
Case {
// Edition between title and year is stripped off the title tail.
name: "Movie.Directors.Cut.2024.1080p.BluRay",
want: NameClaims {
title: s("Movie"),
year: Some(2024),
resolution: Some(Resolution::P1080),
source: Some(Source::BluRay),
editions: vec![Edition::DirectorsCut],
..NameClaims::default()
},
},
Case {
// First year-like token is the title, second is the year.
name: "2012.2009.1080p.BluRay.x264",
want: NameClaims {
title: s("2012"),
year: Some(2009),
resolution: Some(Resolution::P1080),
source: Some(Source::BluRay),
codec: Some(Codec::X264),
..NameClaims::default()
},
},
Case {
name: "Blade.Runner.2049.2017.2160p.WEB-DL.DDP5.1.HDR.HEVC-XEBEC",
want: NameClaims {
title: s("Blade Runner 2049"),
year: Some(2017),
resolution: Some(Resolution::P2160),
source: Some(Source::WebDl),
codec: Some(Codec::X265),
hdr: vec![HdrMarker::Hdr],
group: s("XEBEC"),
..NameClaims::default()
},
},
Case {
name: "Movie.2024.HDTS.x264-ETRG",
want: NameClaims {
title: s("Movie"),
year: Some(2024),
source: Some(Source::Telesync),
codec: Some(Codec::X264),
group: s("ETRG"),
..NameClaims::default()
},
},
Case {
name: "Movie.2024.1080p.HMAX.WEB-DL.DD5.1.H.264-GROUP",
want: NameClaims {
title: s("Movie"),
year: Some(2024),
resolution: Some(Resolution::P1080),
source: Some(Source::WebDl),
codec: Some(Codec::X264),
group: s("GROUP"),
..NameClaims::default()
},
},
Case {
// Spaced dash before the group.
name: "Movie Title 2024 1080p BluRay x265 - VXT",
want: NameClaims {
title: s("Movie Title"),
year: Some(2024),
resolution: Some(Resolution::P1080),
source: Some(Source::BluRay),
codec: Some(Codec::X265),
group: s("VXT"),
..NameClaims::default()
},
},
Case {
// Filename with a container extension.
name: "Movie.Title.2024.720p.WEBRip.x264.mkv",
want: NameClaims {
title: s("Movie Title"),
year: Some(2024),
resolution: Some(Resolution::P720),
source: Some(Source::WebRip),
codec: Some(Codec::X264),
..NameClaims::default()
},
},
Case {
// No markers at all: the whole thing is a title claim.
name: "Some Random Words",
want: NameClaims {
title: s("Some Random Words"),
..NameClaims::default()
},
},
Case {
// Marker-only garbage parses partially, title absent.
name: "1080p",
want: NameClaims {
resolution: Some(Resolution::P1080),
..NameClaims::default()
},
},
Case {
name: "",
want: NameClaims::default(),
},
Case {
// Dolby Vision spelled out, dotted.
name: "Movie.2024.2160p.WEB-DL.Dolby.Vision.HDR10.HEVC-GRP",
want: NameClaims {
title: s("Movie"),
year: Some(2024),
resolution: Some(Resolution::P2160),
source: Some(Source::WebDl),
codec: Some(Codec::X265),
hdr: vec![HdrMarker::DolbyVision, HdrMarker::Hdr10],
group: s("GRP"),
..NameClaims::default()
},
},
Case {
// Hyphenated title survives when nothing marks a group.
name: "Spider-Man.2002.1080p.BluRay.x264-SECTOR7",
want: NameClaims {
title: s("Spider-Man"),
year: Some(2002),
resolution: Some(Resolution::P1080),
source: Some(Source::BluRay),
codec: Some(Codec::X264),
group: s("SECTOR7"),
..NameClaims::default()
},
},
Case {
name: "www.Torrenting.com - Movie.2024.1080p.WEB-DL.x264",
want: NameClaims {
title: s("Movie"),
year: Some(2024),
resolution: Some(Resolution::P1080),
source: Some(Source::WebDl),
codec: Some(Codec::X264),
..NameClaims::default()
},
},
Case {
// Season pack, no episode.
name: "Show.Name.S02.2160p.WEB-DL.DDP5.1.DV.HDR10.H.265-NTb",
want: NameClaims {
title: s("Show Name"),
resolution: Some(Resolution::P2160),
source: Some(Source::WebDl),
codec: Some(Codec::X265),
hdr: vec![HdrMarker::DolbyVision, HdrMarker::Hdr10],
group: s("NTb"),
episode: Some(EpisodeClaim::Season { season: 2 }),
..NameClaims::default()
},
},
Case {
// `COMPLETE` before the tag closes the title like any strong
// marker, wherever the group puts it.
name: "Bluey.COMPLETE.S01.1080p.WEB-DL",
want: NameClaims {
title: s("Bluey"),
resolution: Some(Resolution::P1080),
source: Some(Source::WebDl),
episode: Some(EpisodeClaim::Season { season: 1 }),
..NameClaims::default()
},
},
Case {
// The trailing placement already worked; it keeps working.
name: "Bluey.S01.COMPLETE.1080p.WEB-DL",
want: NameClaims {
title: s("Bluey"),
resolution: Some(Resolution::P1080),
source: Some(Source::WebDl),
episode: Some(EpisodeClaim::Season { season: 1 }),
..NameClaims::default()
},
},
Case {
// A degenerate range is one season's pack.
name: "Fallout.S01-S01.1080p.BluRay.x264-GROUP",
want: NameClaims {
title: s("Fallout"),
resolution: Some(Resolution::P1080),
source: Some(Source::BluRay),
codec: Some(Codec::X264),
group: s("GROUP"),
episode: Some(EpisodeClaim::Seasons { first: 1, last: 1 }),
..NameClaims::default()
},
},
Case {
name: "Movie.2019.4K.HDR.DV.2160p.BDRemux.Ita.Eng.x265-NAHOM",
want: NameClaims {
title: s("Movie"),
year: Some(2019),
resolution: Some(Resolution::P2160),
source: Some(Source::Remux),
codec: Some(Codec::X265),
hdr: vec![HdrMarker::Hdr, HdrMarker::DolbyVision],
languages: vec![LanguageMarker::Italian, LanguageMarker::English],
group: s("NAHOM"),
..NameClaims::default()
},
},
]
}
#[test]
fn corpus_rows_parse_as_expected() {
let mut failures = Vec::new();
for Case { name, want } in cases() {
let got = parse(name);
if got != want {
failures.push(format!("{name:?}\n got: {got:?}\n want: {want:?}"));
}
}
assert!(
failures.is_empty(),
"{} mismatches:\n{}",
failures.len(),
failures.join("\n")
);
}
#[test]
fn garbage_never_panics() {
let inputs = [
"",
" ",
"....",
"----",
"[[[]]]",
"()()()",
"ção áéíóú 日本語 🎬🎬🎬",
"-.-.-.-[]()s01e999999999999999999999",
"Movie.2024.1080p-",
"a",
&"x.".repeat(5000),
];
for input in inputs {
let _ = parse(input);
}
}
#[test]
fn episode_tags_separate_packs_from_single_episodes() {
let single = parse("Show.S01E02.1080p.WEB-DL.x264-GRP")
.episode
.expect("S01E02 is an episode tag");
assert_eq!(
single,
EpisodeClaim::Episodes {
season: 1,
episodes: vec![2]
}
);
assert!(!single.is_season_pack());
assert!(!single.is_multi_episode());
assert!(single.covers(1, 2));
assert!(!single.covers(1, 3));
assert!(single.covers_season(1));
let multi = parse("Show.S01E02-E04.1080p.WEB-DL.x264-GRP")
.episode
.expect("S01E02-E04 is an episode tag");
assert!(multi.is_multi_episode());
assert!(!multi.is_season_pack());
assert!(multi.covers(1, 3));
assert!(!multi.covers(1, 5));
let pack = parse("Show.S01.1080p.WEB-DL.x264-GRP")
.episode
.expect("S01 is a season tag");
assert!(pack.is_season_pack());
assert!(
pack.covers(1, 7),
"a season pack covers every episode in it"
);
assert!(!pack.covers(2, 7));
let seasons = parse("Show.S01-S03.1080p.WEB-DL.x264-GRP")
.episode
.expect("S01-S03 is a season range");
assert!(seasons.is_season_pack());
assert!(seasons.covers_season(2));
assert!(!seasons.covers_season(4));
let daily = parse("Show.2024.05.13.1080p.WEB.h264-GRP")
.episode
.expect("a dotted air date is a daily tag");
assert!(daily.covers_date(2024, 5, 13));
assert!(!daily.covers_date(2024, 5, 14));
assert!(!daily.covers(1, 1), "a daily episode has no season number");
}
#[test]
fn non_tv_names_claim_no_episode() {
for name in [
"Dune.Part.Two.2024.2160p.WEB-DL.H.265-FLUX",
"Movie.1999.1080p.BluRay.x264-GRP",
"Show.2024.13.45.1080p.WEB-DL.x264-GRP",
"Blade.Runner.2049.2017.2160p.UHD.BluRay.x265-TERMiNAL",
] {
assert_eq!(parse(name).episode, None, "{name}");
}
}
#[test]
fn normalisation_is_separator_and_case_blind() {
let key = arr_parse::normalise("Dune.Part.Two.2024.2160p.WEB-DL.DDP5.1.Atmos");
assert_eq!(key, "dune.part.two.2024.2160p.web.dl.ddp5.1.atmos");
assert_eq!(
arr_parse::normalise("dune part two [2024] 2160p WEB DL DDP5 1 atmos "),
key,
"the same release under different separators is the same blacklist key"
);
assert_eq!(arr_parse::normalise("!!!"), "");
}