//! Keyword-less catalog discovery: TMDB's popular / trending / top-rated lists //! → [`Card`]s. //! //! The card tier's *browse* surface, as opposed to its search surface. A client //! with a home page (meta-watch's wall) cannot ask "what should I show?" through //! a text query — there is no text. This branch answers a query whose intent is //! carried entirely by structured filters, by calling TMDB's catalog endpoints. //! //! ## Query dialect //! //! Deliberately identical to the indexer feeder's older `discovery.rs`, which //! answers the same dialect with metadata-only *seeds*. Same words, so a client //! switches tiers by swapping `fileType:video` for `fileType:card` and nothing //! else: //! //! - **mode** (one of) — `popular:true` and `top_rated:true` are served by //! `/discover/{movie,tv}` sorted by popularity / rating (rather than the //! `/{movie,tv}/popular` list endpoints, which cannot carry the adult gate — //! see `DISCOVERY_EXCLUDED_KEYWORDS`); `trending:true` → the real //! `/trending/{movie,tv}/week`, the one mode with no `/discover` equivalent //! and therefore the one row that stays ungated. //! - **kind** (required) — `contentKind:movie` → films, //! `contentKind:series` (or `episode`/`tvshow`/`tv`) → shows. Also what routes //! the query here at all, alongside `fileType:card`. //! - **anime** (optional) — `anime:true` narrows `/discover` to TMDB's *anime* //! keyword, sorted to match the mode. Anime trending has no keyword-filtered //! endpoint either, so it approximates with popularity — and stays gated. //! //! ## One call, complete cards //! //! A TMDB catalog list already carries `title`, `overview`, `poster_path` and //! `year` per entry — everything [`Card::to_record`] needs to emit a renderable //! card. So a row costs **one** TMDB call, not one plus twenty by-id lookups. //! //! The price is a thinner id bag: `tvdbid`/`imdbid` come from `external_ids`, //! which a list response has no room for, so a discovery card carries only its //! `tmdbid`. That is the id consumers actually key on (it is what the detail //! page's phase-2 `tmdbid:` query is built from), and the full bag is resolved //! anyway the moment someone clicks the card — `card_by_id` is a different code //! path and still fetches it. Paying twenty extra calls per row up front to //! pre-fill ids nobody reads until a click would defeat the branch's purpose. //! //! ## Why nothing is cached here //! //! A catalog list is **mutable by nature** — "popular this week" is the one //! thing that must not be frozen. The SDK's redb tables are permanent-hit caches //! sized for immutable data (`external_ids` is explicitly "cached forever"), so //! persisting a popularity list there would pin the wall to whatever it looked //! like on first boot. The layer above already de-duplicates the repeat cost: //! the gateway persists each emitted card to meta-core and its search-coverage //! gate then skips a repeated `(upstream, query)` for an hour, serving meta-core //! instead. Every call here is budget-gated, so a burst cannot run away. use std::collections::HashMap; use std::sync::Arc; use meta_feeder_sdk::plugin::GatewayQuery; use tracing::debug; use crate::card::{Card, CardSource}; use crate::consts::{ AIRING_EXCLUDED_GENRES, DISCOVERY_EXCLUDED_KEYWORDS, DISCOVERY_MAX_PAGES, DISCOVERY_MIN_VOTES, }; use crate::resolve::{dedup_titles, Resolver}; use crate::tmdb_client::{title_names, TmdbHit, TmdbKind}; /// TMDB keyword id for "anime". `/discover` is the only catalog endpoint that /// takes a keyword filter, which is why `anime:true` switches endpoints rather /// than adding a parameter. const TMDB_ANIME_KEYWORD: &str = "210024"; /// Results TMDB returns per catalog page — used to translate a desired card /// count into a page count. const TMDB_PAGE_SIZE: usize = 20; /// The catalog-list flavour a discovery query asks for. #[derive(Clone, Copy, PartialEq, Eq, Debug)] pub(crate) enum DiscoveryMode { Popular, Trending, TopRated, /// What came out **today** — the day-by-day row. TV: shows with an episode /// airing today; film: releases from the last fortnight. /// /// Date-bounded rather than availability-bounded on purpose. "Is a file out /// yet" is an indexer question, and asking it per show per day is exactly /// the load that wedges a Prowlarr; an air date is a fact TMDB already /// holds, costs the same single call as every other row, and makes the row /// roll over by itself at midnight with nothing scheduled. Airing, } impl DiscoveryMode { /// The query-filter key that selects this mode. /// /// It is also echoed onto every emitted record — not here, but by /// [`Card::to_record`]'s trailing filter echo, which stamps back every /// filter the query carried that the card did not already set. That echo is /// load-bearing: both the gateway dispatcher and meta-search's consumer-side /// `record_matches` re-apply the query's filters to each record, and a /// *missing* field fails the match, so an un-echoed `popular` would see /// every card dropped at one of the two tiers. fn marker(self) -> &'static str { match self { DiscoveryMode::Popular => "popular", DiscoveryMode::Trending => "trending", DiscoveryMode::TopRated => "top_rated", DiscoveryMode::Airing => "airing", } } /// `sort_by` value for the `/discover` (anime) path. `/discover` has no /// "trending", so trending approximates to popularity there — the non-anime /// path still uses the real `/trending` endpoint. fn discover_sort(self) -> &'static str { match self { DiscoveryMode::TopRated => "vote_average.desc", // Airing sorts by popularity too: the date bound is what selects the // row, so the sort only decides who leads it. DiscoveryMode::Popular | DiscoveryMode::Trending | DiscoveryMode::Airing => { "popularity.desc" } } } } /// True when this is a keyword-less catalog query this branch should answer. /// /// Three conditions, and each excludes a query that belongs to a *different* /// branch of `handle_query`: no free text (that is the principal search), no /// `tmdbid:` filter (that is the direct by-id lookup a deep link uses), and a /// truthy mode marker. Note `contentKind` alone deliberately does **not** select /// this branch — a text query carries it too. pub(crate) fn is_discovery_query(query: &GatewayQuery) -> bool { query.free_text.trim().is_empty() && !query.filters.contains_key("tmdbid") && discovery_mode(query).is_some() } /// First truthy mode marker on the query, in precedence order. fn discovery_mode(query: &GatewayQuery) -> Option { for mode in [ // Airing leads: it is the most specific claim a row can make, and a // caller that sends `airing:true popular:true` means "today's, by // popularity", not "popular, ever". DiscoveryMode::Airing, DiscoveryMode::Trending, DiscoveryMode::Popular, DiscoveryMode::TopRated, ] { if filter_is_true(query, mode.marker()) { return Some(mode); } } None } /// True iff `query.filters[key]` carries a truthy value. fn filter_is_true(query: &GatewayQuery, key: &str) -> bool { query .filters .get(key) .is_some_and(|v| v.iter().any(|s| s.eq_ignore_ascii_case("true"))) } /// Map the query's `contentKind` filter to a TMDB media kind. /// /// `series` is the canonical spelling for a card (it describes a whole work), /// but `episode`/`tvshow`/`tv` are accepted so the dialect stays a superset of /// the indexer feeder's — a client that has not yet switched its row queries /// still resolves the right endpoint here, even though its records will then be /// dropped downstream for the mismatch. fn discovery_kind(query: &GatewayQuery) -> Option { let values = query.filters.get("contentKind")?; for v in values { match v.trim().to_ascii_lowercase().as_str() { "movie" => return Some(TmdbKind::Movie), "series" | "episode" | "tvshow" | "tv" => return Some(TmdbKind::Tv), _ => {} } } None } /// The genre slugs a `genres:` filter requested, folded for comparison. /// /// Multi-valued = OR, matching the filter semantics on both re-validating tiers. fn requested_genres(query: &GatewayQuery) -> Vec { query .filters .get("genres") .map(|vs| { vs.iter() .map(|v| genre_fold(v)) .filter(|v| !v.is_empty()) .collect() }) .unwrap_or_default() } /// Fold a genre label for comparison: lowercase, alphanumerics only. /// /// Must stay identical to `genres_filter_matches`'s fold in the SDK and in /// meta-search — those two re-apply the filter to every record this emits, so a /// divergence here means the row's own results get dropped downstream. fn genre_fold(s: &str) -> String { s.chars() .filter(|c| c.is_alphanumeric()) .flat_map(|c| c.to_lowercase()) .collect() } /// Resolve the requested genre slugs to TMDB genre ids via the id → name table. /// Unknown slugs are dropped; if none resolve the caller omits `with_genres` /// rather than sending an empty filter that TMDB would reject. fn with_genres_ids(wanted: &[String], names: &HashMap) -> Vec { let mut ids: Vec = names .iter() .filter(|(_, name)| wanted.contains(&genre_fold(name))) .map(|(id, _)| *id) .collect(); ids.sort_unstable(); ids } /// How far back the film side of [`DiscoveryMode::Airing`] reaches, in days. /// /// A series airs an episode on a given day, so "today" is a real set. Films do /// not: a single day's releases are usually empty, and a viewer opening the app /// on a Tuesday still means "what's new". A fortnight is the smallest window /// that keeps the row populated without turning it into "this year". const AIRING_MOVIE_WINDOW_DAYS: i64 = 13; /// Today, UTC, as `YYYY-MM-DD` — the bound [`DiscoveryMode::Airing`] is built /// from. UTC rather than a local zone: the feeder serves peers in every zone, /// and TMDB's air dates are themselves zone-less calendar dates. fn today_utc_days() -> i64 { std::time::SystemTime::now() .duration_since(std::time::UNIX_EPOCH) .map(|d| (d.as_secs() / 86_400) as i64) .unwrap_or(0) } /// `YYYY-MM-DD` for a count of days since 1970-01-01. /// /// Hand-rolled (Howard Hinnant's civil-from-days) rather than pulling in a date /// crate for one format call — and split from [`today_utc_days`] so the whole /// thing is testable against known dates instead of "whatever today is". fn fmt_days(days: i64) -> String { let z = days + 719_468; let era = if z >= 0 { z } else { z - 146_096 } / 146_097; let doe = (z - era * 146_097) as i64; // [0, 146096] let yoe = (doe - doe / 1460 + doe / 36_524 - doe / 146_096) / 365; // [0, 399] let y = yoe + era * 400; let doy = doe - (365 * yoe + yoe / 4 - yoe / 100); // [0, 365] let mp = (5 * doy + 2) / 153; // [0, 11] let d = doy - (153 * mp + 2) / 5 + 1; // [1, 31] let m = if mp < 10 { mp + 3 } else { mp - 9 }; // [1, 12] let y = if m <= 2 { y + 1 } else { y }; format!("{y:04}-{m:02}-{d:02}") } /// The `/discover` date bounds for [`DiscoveryMode::Airing`], given today. /// /// TV bounds `air_date` to the single day, which is what makes the row /// "today's episodes" and what makes it roll over on its own. Film bounds /// `primary_release_date` to [`AIRING_MOVIE_WINDOW_DAYS`] back — see that /// constant for why the two kinds cannot share a window. fn airing_bounds(kind: TmdbKind, today: i64) -> String { match kind { TmdbKind::Tv => { let d = fmt_days(today); // ⚠ `without_genres` rides with the date bound, and only here. A row // of "aired today" is otherwise the daily-broadcast schedule — talk, // news, soap, reality — which is exactly what the viewer did not // mean. See [`AIRING_EXCLUDED_GENRES`]. The other modes keep every // genre: a *popular* row has no reason to hide a talk show. format!( "air_date.gte={d}&air_date.lte={d}&without_genres={AIRING_EXCLUDED_GENRES}&" ) } TmdbKind::Movie => format!( "primary_release_date.gte={}&primary_release_date.lte={}&", fmt_days(today - AIRING_MOVIE_WINDOW_DAYS), fmt_days(today), ), } } /// Build the `path_and_query` for `TmdbClient::discovery_list` from the mode, /// kind, anime flag, requested genre ids, and 1-based `page`. fn build_path_and_query( mode: DiscoveryMode, kind: TmdbKind, anime: bool, genre_ids: &[u32], page: u32, ) -> String { let seg = match kind { TmdbKind::Movie => "movie", TmdbKind::Tv => "tv", }; // Trending is the one mode with no `/discover` equivalent: `/trending` is a // distinct, TMDB-computed "this week" signal, not a sort order. So it keeps // the list endpoint — and, being a list endpoint, **cannot carry the adult // gate below**. The anime variant still routes through `/discover` (there is // no keyword-filtered trending at all), approximating it with popularity. // …and a genre-scoped row can't use it either: `/trending` takes no // `with_genres`, and returning an unfiltered trending page for an "Action" // row would fill it with whatever is trending. `/discover` sorted by // popularity is the closest filterable approximation. if matches!(mode, DiscoveryMode::Trending) && !anime && genre_ids.is_empty() { return format!("trending/{seg}/week?page={page}"); } // Everything else goes through `/discover`, including the non-anime rows // that used to use `/{seg}/popular` and `/{seg}/top_rated`. // // Those fixed list endpoints accept neither `without_keywords` nor // `vote_count.gte`, and measurement showed they genuinely need them: three // hentai titles reached positions 4, 5 and 17 of the *non-anime* `tv/popular` // row for the same traffic-metric reason described on // [`DISCOVERY_EXCLUDED_KEYWORDS`]. `/discover` with `sort_by=popularity.desc` // is the filterable equivalent, so one code path now carries the gate for // every mode that can hold it. // // Top-rated floors higher than the shared gate: sorting by rating puts a // single 10.0 vote on top, which a 50-vote gate doesn't stop. let votes = match mode { DiscoveryMode::TopRated => 200, // ⚠ No vote floor on the airing row, and that is the point. The shared // gate exists to keep unrated residue off a popularity ranking; here the // date bound already decides membership, and a floor would filter out // exactly what the row is for — a series whose first season is airing // now has not had time to collect 50 votes. DiscoveryMode::Airing => 0, _ => DISCOVERY_MIN_VOTES, }; let dates = match mode { DiscoveryMode::Airing => airing_bounds(kind, today_utc_days()), _ => String::new(), }; let keyword = if anime { format!("with_keywords={TMDB_ANIME_KEYWORD}&") } else { String::new() }; // `with_genres` is comma-joined = AND on TMDB's side, pipe-joined = OR. A // multi-valued filter means OR everywhere else in the DSL, so join with `|` // to keep the wire semantics and the upstream semantics saying the same // thing. URL-encoded, since a bare `|` is not valid in a query string. let genres = if genre_ids.is_empty() { String::new() } else { let joined = genre_ids .iter() .map(|id| id.to_string()) .collect::>() .join("%7C"); format!("with_genres={joined}&") }; format!( "discover/{seg}?{keyword}{genres}{dates}sort_by={}&include_adult=false\ &without_keywords={DISCOVERY_EXCLUDED_KEYWORDS}&vote_count.gte={votes}&page={page}", mode.discover_sort(), ) } /// Project one catalog hit into a [`Card`], with no further network calls. /// /// `source_id` uses the same `"tv:{id}"` / `"movie:{id}"` shape /// `Resolver::card_by_id` produces, so a discovery card's `record_id` — and /// therefore its `0x1007` locator CID — is byte-identical to the one the by-id /// path derives for the same work. That is what lets a card the wall surfaced /// and a card the search box surfaced converge on one record instead of two. fn card_from_hit(hit: &TmdbHit, kind: TmdbKind, genres: &HashMap) -> Card { let seg = match kind { TmdbKind::Movie => "movie", TmdbKind::Tv => "tv", }; let mut titles = vec![hit.title.clone()]; if let Some(o) = &hit.original_title { titles.push(o.clone()); } titles.extend(hit.alt_titles.iter().cloned()); Card { source: CardSource::Tmdb, kind, source_id: format!("{seg}:{}", hit.tmdbid), tmdbid: Some(hit.tmdbid), // Not available on a list response — see the module doc. Resolved on // the click-time `card_by_id` path instead. tvdb_id: None, imdb_id: None, title: hit.title.clone(), guard_titles: Arc::new(dedup_titles(titles)), // A list hit carries no AKAs, so this is title + original only; the // click-time `card_by_id` card adds the AKAs under the same CID. names: title_names( &hit.title, hit.original_title.as_deref(), hit.original_language.as_deref(), &hit.akas, ), overview: hit.overview.clone().filter(|s| !s.trim().is_empty()), poster_path: hit.poster_path.clone().filter(|s| !s.trim().is_empty()), // A list hit carries bare genre ids; the table names them. genres: hit.genre_ids.iter().filter_map(|id| genres.get(id).cloned()).collect(), year: hit.year, // A list response carries no season structure. `seasonCount` is simply // omitted from the record (`Card::to_record` skips it at `seasons == 0`). seasons: 0, season_summaries: Arc::new(Vec::new()), // A list response carries no `images` append, so a discovery card files // no `posters/*` set; the click-time `card_by_id` card adds it under the // same CID. posters: Arc::new(Vec::new()), } } /// Resolve a keyword-less catalog query to up to `max_results` cards, walking /// TMDB pages until the cap is reached or a page comes back empty. /// /// Best-effort throughout: no mode or no `contentKind` yields `[]`, and a /// rate-limit or miss breaks the walk and returns what was already gathered. /// The branch never errors — an empty row is the failure mode, matching /// `Resolver::cards_for_text`. pub(crate) async fn cards_for_discovery( resolver: &Resolver, query: &GatewayQuery, max_results: usize, ) -> Vec { let (Some(mode), Some(kind)) = (discovery_mode(query), discovery_kind(query)) else { return Vec::new(); }; let anime = filter_is_true(query, "anime"); // Fetched once per process, not per row (see `Resolver::genre_names`). let gmap = resolver.genre_names(kind).await; // A `genres:` filter narrows the row upstream via `with_genres`, so the row // is a full page OF that genre rather than a page filtered down to it. let wanted = requested_genres(query); let genre_ids = with_genres_ids(&wanted, &gmap); if !wanted.is_empty() && genre_ids.is_empty() { // Asked for a genre this kind doesn't have (e.g. `genres:kids` on the // movie side, where TMDB files it as "Family"). Returning an unfiltered // page would silently fill the row with the wrong thing. debug!( target: "meta-share::gateway", upstream = "tmdb", ?wanted, "discovery: no TMDB genre id matched; row left empty" ); return Vec::new(); } // One page of headroom over the arithmetic minimum: `out` counts *displayable* // cards, and a catalog page reliably contains a few entries with no overview // or no poster (newly-added or non-English works), which are declined at // projection. Without the spare page a 20-card row comes back at 17. let pages = (max_results.div_ceil(TMDB_PAGE_SIZE).max(1) as u32 + 1).min(DISCOVERY_MAX_PAGES); let mut out: Vec = Vec::new(); for page in 1..=pages { if out.len() >= max_results { break; } let path_and_query = build_path_and_query(mode, kind, anime, &genre_ids, page); let Some(hits) = resolver.discovery_page(kind, &path_and_query).await else { break; // budget exhausted, 429, or miss — keep what we have }; if hits.is_empty() { break; // walked past the last catalog page } out.extend( hits.iter() .map(|h| card_from_hit(h, kind, &gmap)) .filter(Card::is_displayable), ); } out.truncate(max_results); debug!( target: "meta-share::gateway", upstream = "tmdb", mode = mode.marker(), anime, count = out.len(), "resolved discovery cards" ); out } #[cfg(test)] mod tests { use super::*; use std::collections::BTreeMap; fn query(filters: &[(&str, &str)]) -> GatewayQuery { let mut map: BTreeMap> = BTreeMap::new(); for (k, v) in filters { map.entry(k.to_string()).or_default().push(v.to_string()); } GatewayQuery { raw_text: String::new(), free_text: String::new(), filters: map, ranges: Vec::new(), negations: Vec::new(), } } #[test] fn selects_only_keyword_less_mode_queries() { assert!(is_discovery_query(&query(&[ ("popular", "true"), ("contentKind", "series"), ("fileType", "card"), ]))); // A mode marker but with free text → the principal search owns it. let mut with_text = query(&[("popular", "true"), ("contentKind", "series")]); with_text.free_text = "frieren".to_string(); assert!(!is_discovery_query(&with_text)); // An explicit id → the by-id lookup owns it. assert!(!is_discovery_query(&query(&[ ("popular", "true"), ("tmdbid", "95479"), ]))); // contentKind alone must NOT select this branch. assert!(!is_discovery_query(&query(&[ ("contentKind", "series"), ("fileType", "card"), ]))); // A falsey marker is not a marker. assert!(!is_discovery_query(&query(&[("popular", "false")]))); } /// `contentKind:series` is the canonical card spelling; the indexer feeder's /// older `episode` spelling still resolves the same endpoint. #[test] fn content_kind_maps_series_and_the_legacy_spellings() { assert_eq!( discovery_kind(&query(&[("contentKind", "series")])), Some(TmdbKind::Tv) ); assert_eq!( discovery_kind(&query(&[("contentKind", "episode")])), Some(TmdbKind::Tv) ); assert_eq!( discovery_kind(&query(&[("contentKind", "movie")])), Some(TmdbKind::Movie) ); assert_eq!(discovery_kind(&query(&[("contentKind", "audiobook")])), None); assert_eq!(discovery_kind(&query(&[])), None); } #[test] fn builds_the_catalog_endpoints() { // Popular / top-rated route through /discover (so the gate applies); // only the sort and the vote floor differ. let pop = build_path_and_query(DiscoveryMode::Popular, TmdbKind::Movie, false, &[], 1); assert!(pop.starts_with("discover/movie?"), "{pop}"); assert!(pop.contains("sort_by=popularity.desc"), "{pop}"); assert!(pop.contains("page=1"), "{pop}"); let paged = build_path_and_query(DiscoveryMode::Popular, TmdbKind::Tv, false, &[], 2); assert!(paged.starts_with("discover/tv?"), "{paged}"); assert!(paged.contains("page=2"), "{paged}"); let top = build_path_and_query(DiscoveryMode::TopRated, TmdbKind::Movie, false, &[], 1); assert!(top.contains("sort_by=vote_average.desc"), "{top}"); assert!(top.contains("vote_count.gte=200"), "{top}"); // Trending keeps the real endpoint — see the dedicated test. assert_eq!( build_path_and_query(DiscoveryMode::Trending, TmdbKind::Tv, false, &[], 1), "trending/tv/week?page=1" ); } #[test] fn anime_switches_to_the_keyword_filtered_discover_endpoint() { let p = build_path_and_query(DiscoveryMode::Popular, TmdbKind::Tv, true, &[], 1); assert!(p.starts_with("discover/tv?"), "{p}"); assert!(p.contains("with_keywords=210024"), "{p}"); assert!(p.contains("sort_by=popularity.desc"), "{p}"); let top = build_path_and_query(DiscoveryMode::TopRated, TmdbKind::Movie, true, &[], 1); assert!(top.contains("sort_by=vote_average.desc"), "{top}"); // Sorting by rating puts a 1-vote 10.0 on top, so top-rated floors higher. assert!(top.contains("vote_count.gte=200"), "{top}"); } /// The gate is not anime-specific. Three hentai titles were measured at /// positions 4, 5 and 17 of the **non-anime** `tv/popular` row, which is why /// the plain popular/top-rated modes were moved off the fixed list endpoints /// (which cannot carry `without_keywords`) and onto `/discover`. #[test] fn non_anime_popular_and_top_rated_also_gate() { for mode in [DiscoveryMode::Popular, DiscoveryMode::TopRated] { let p = build_path_and_query(mode, TmdbKind::Tv, false, &[], 1); assert!(p.starts_with("discover/tv?"), "{p}"); assert!(!p.contains("with_keywords="), "no anime narrowing here: {p}"); assert!(p.contains("without_keywords="), "{p}"); assert!(p.contains("vote_count.gte="), "{p}"); } } /// Trending is the documented exception: `/trending` is a TMDB-computed /// signal with no `/discover` sort equivalent, so the non-anime trending row /// keeps the list endpoint and cannot carry the gate. #[test] fn non_anime_trending_keeps_the_real_trending_endpoint() { let p = build_path_and_query(DiscoveryMode::Trending, TmdbKind::Tv, false, &[], 1); assert_eq!(p, "trending/tv/week?page=1"); // …but the anime variant has no keyword-filtered trending at all, so it // still routes through /discover and stays gated. let a = build_path_and_query(DiscoveryMode::Trending, TmdbKind::Tv, true, &[], 1); assert!(a.starts_with("discover/tv?"), "{a}"); assert!(a.contains("without_keywords="), "{a}"); } /// A genre-scoped row narrows **upstream** via `with_genres`, so the row is a /// full page of that genre rather than a page filtered down to a handful. #[test] fn a_genre_row_narrows_upstream() { let p = build_path_and_query(DiscoveryMode::Popular, TmdbKind::Tv, false, &[10759], 1); assert!(p.contains("with_genres=10759"), "{p}"); // Multi-valued = OR, so pipe-joined (comma would mean AND to TMDB). let multi = build_path_and_query(DiscoveryMode::Popular, TmdbKind::Tv, false, &[16, 35], 1); assert!(multi.contains("with_genres=16%7C35"), "{multi}"); // No genre filter ⇒ the parameter is absent, not empty. let none = build_path_and_query(DiscoveryMode::Popular, TmdbKind::Tv, false, &[], 1); assert!(!none.contains("with_genres"), "{none}"); } /// A genre row can't use `/trending` (it takes no `with_genres`), so it falls /// back to the filterable `/discover` rather than returning an unfiltered /// trending page under a genre heading. #[test] fn a_genre_scoped_trending_row_falls_back_to_discover() { let p = build_path_and_query(DiscoveryMode::Trending, TmdbKind::Tv, false, &[10759], 1); assert!(p.starts_with("discover/tv?"), "{p}"); assert!(p.contains("with_genres=10759"), "{p}"); } /// Slugs resolve to TMDB ids through the id → name table, folded on both /// sides so the DSL never has to spell a space or an ampersand. #[test] fn genre_slugs_resolve_to_tmdb_ids() { let names = HashMap::from([ (16u32, "Animation".to_string()), (10759, "Action & Adventure".to_string()), (10765, "Sci-Fi & Fantasy".to_string()), ]); let want = |s: &str| with_genres_ids(&[genre_fold(s)], &names); assert_eq!(want("action-adventure"), vec![10759]); assert_eq!(want("Action & Adventure"), vec![10759]); assert_eq!(want("scififantasy"), vec![10765]); assert_eq!(want("animation"), vec![16]); assert!(want("horror").is_empty(), "unknown slug resolves to nothing"); } /// The fold must stay identical to the one both re-validating tiers use, or /// a row's own results get dropped downstream. #[test] fn genre_fold_matches_the_consumer_side_rule() { assert_eq!(genre_fold("Action & Adventure"), "actionadventure"); assert_eq!(genre_fold("action-adventure"), "actionadventure"); assert_eq!(genre_fold("Sci-Fi & Fantasy"), "scififantasy"); assert_eq!(genre_fold("Science Fiction"), "sciencefiction"); assert_eq!(genre_fold(" "), ""); } /// The adult-content gate on anime rows, and *why* it is two filters. /// /// `include_adult=false` is not enough on its own — measured live, every /// offending title returns `adult: false` (TMDB reserves that flag for /// pornography), so the keyword exclusion carries the real load and the vote /// floor sweeps up the erotic-tagged residue behind it. Dropping either one /// puts hentai back on the home page. #[test] fn anime_rows_gate_adult_content_three_ways() { for mode in [ DiscoveryMode::Popular, DiscoveryMode::Trending, DiscoveryMode::TopRated, ] { for kind in [TmdbKind::Tv, TmdbKind::Movie] { let p = build_path_and_query(mode, kind, true, &[], 1); assert!(p.contains("include_adult=false"), "{p}"); assert!( p.contains(&format!("without_keywords={DISCOVERY_EXCLUDED_KEYWORDS}")), "{p}" ); assert!(p.contains("vote_count.gte="), "{p}"); } } } fn genre_table() -> HashMap { HashMap::from([(16, "Animation".to_string()), (18, "Drama".to_string())]) } fn hit() -> TmdbHit { TmdbHit { akas: Vec::new(), tmdbid: 95479, title: "Frieren: Beyond Journey's End".to_string(), original_title: Some("葬送のフリーレン".to_string()), original_language: Some("ja".to_string()), overview: Some("A mage reflects on her long life.".to_string()), year: Some(2023), poster_path: Some("/poster.jpg".to_string()), genre_ids: vec![16], alt_titles: Vec::new(), posters: Vec::new(), } } /// A list hit must carry everything a renderable card needs, so the branch /// costs one TMDB call per page and no by-id follow-ups. #[test] fn a_list_hit_becomes_a_complete_card() { let c = card_from_hit(&hit(), TmdbKind::Tv, &genre_table()); assert_eq!(c.tmdbid, Some(95479)); assert!(c.overview.is_some(), "overview is what to_record requires"); assert!(c.poster_path.is_some(), "poster is what to_record requires"); assert_eq!(c.year, Some(2023)); assert_eq!(c.content_kind(), "series"); // Genres come from the id table — the category axis a card used to lack. assert_eq!(c.genres, vec!["Animation".to_string()]); // The id bag beyond tmdbid needs external_ids — deliberately absent. assert_eq!(c.tvdb_id, None); assert_eq!(c.imdb_id, None); } /// The locator CID is derived from `record_id`, so a discovery card and a /// by-id card for the same work must address identically or the two tiers /// would publish two records for one work. #[test] fn record_id_matches_the_by_id_path() { assert_eq!(card_from_hit(&hit(), TmdbKind::Tv, &genre_table()).record_id(), "tmdb:tv:95479"); assert_eq!( card_from_hit(&hit(), TmdbKind::Movie, &genre_table()).record_id(), "tmdb:movie:95479" ); } /// The whole point of the branch: `contentKind` on the emitted record is the /// *card's* kind, never an echo of the query filter. A row asking /// `contentKind:episode` must still produce `series` here — which is exactly /// why a client's row query has to ask for `series` (the mismatch is then /// caught by `record_matches` downstream rather than silently mislabelling). /// **The invariant this whole branch rests on.** The mode/anime markers are /// stamped onto the record by `Card::to_record`'s trailing filter echo, not /// by anything here — and both the gateway dispatcher and meta-search /// re-apply the query's filters to every record, where a *missing* field /// fails the match. So if that echo is ever "tidied up", every discovery /// card is silently dropped in transit and the wall goes blank with no /// error. This test fails that day. #[test] fn a_discovery_card_survives_the_consumer_re_filter() { let q = query(&[ ("popular", "true"), ("contentKind", "series"), ("fileType", "card"), ("anime", "true"), ]); let card = card_from_hit(&hit(), TmdbKind::Tv, &genre_table()); let tmdb = crate::tmdb_client::TmdbClient::new("token".to_string()); let rec = card.to_record(&tmdb, &q.filters).expect("displayable card emits"); assert_eq!(rec.fields.get("popular").map(String::as_str), Some("true")); assert_eq!(rec.fields.get("anime").map(String::as_str), Some("true")); assert_eq!(rec.fields.get("fileType").map(String::as_str), Some("card")); assert_eq!( rec.fields.get("contentKind").map(String::as_str), Some("series"), "the card's own kind must win over the query filter" ); assert!(rec.fields.contains_key("description/eng")); assert!(rec.fields.contains_key("poster_url")); // Genres ride as a key-set, never the legacy comma-joined value // (METADATA_KEYS §14.12 — new writers must not add csv-set fields). assert_eq!(rec.fields.get("genres/Animation").map(String::as_str), Some("true")); assert!(!rec.fields.contains_key("genres"), "must not write the legacy csv-set"); // The thin id bag: resolved lazily on the click-time by-id path. assert!(!rec.fields.contains_key("tvdbid")); assert!(!rec.fields.contains_key("imdbid")); assert!(!rec.fields.contains_key("seasonCount")); assert!( meta_feeder_sdk::query_eval::record_matches(&rec.fields, &q), "a discovery card must survive the same filter re-application the \ dispatcher and meta-search perform" ); } /// A hit TMDB has not written up yet is declined at projection — which is /// why the page walk counts displayable cards, not raw hits. #[test] fn an_incomplete_hit_is_not_displayable() { let mut no_overview = hit(); no_overview.overview = None; assert!(!card_from_hit(&no_overview, TmdbKind::Tv, &genre_table()).is_displayable()); let mut no_poster = hit(); no_poster.poster_path = None; assert!(!card_from_hit(&no_poster, TmdbKind::Tv, &genre_table()).is_displayable()); assert!(card_from_hit(&hit(), TmdbKind::Tv, &genre_table()).is_displayable()); } #[test] fn content_kind_comes_from_the_card_not_the_filter() { let tv = card_from_hit(&hit(), TmdbKind::Tv, &genre_table()); assert_eq!(tv.content_kind(), "series"); let movie = card_from_hit(&hit(), TmdbKind::Movie, &genre_table()); assert_eq!(movie.content_kind(), "movie"); } /// The day-by-day row: what selects it, and what it asks TMDB for. /// /// Bounding by air date rather than by availability is the whole design — /// see [`DiscoveryMode::Airing`]. These pin the three parts that make the /// row roll over on its own: the date maths, the bounds per kind, and the /// fact that the vote floor is lifted (a season airing now has no votes /// yet, and the shared floor would empty the row it is meant to fill). #[test] fn airing_row_is_bounded_by_date_not_by_votes() { // Date maths, against known days-since-epoch. assert_eq!(fmt_days(0), "1970-01-01"); assert_eq!(fmt_days(19_723), "2024-01-01"); // a leap year's first day assert_eq!(fmt_days(19_782), "2024-02-29"); // the leap day itself assert_eq!(fmt_days(20_453), "2025-12-31"); // TV: exactly one day, so "today's episodes" is literal. let tv = airing_bounds(TmdbKind::Tv, 19_723); assert!(tv.starts_with("air_date.gte=2024-01-01&air_date.lte=2024-01-01&"), "{tv}"); // The daily-broadcast genres are excluded from this row, and only this // row — a talk show airs every day and would otherwise fill it. assert!(tv.contains(&format!("without_genres={AIRING_EXCLUDED_GENRES}")), "{tv}"); assert!( !build_path_and_query(DiscoveryMode::Popular, TmdbKind::Tv, false, &[], 1) .contains("without_genres"), "a popular row must keep every genre" ); // Film: a fortnight back, because a single day of releases is empty. let movie = airing_bounds(TmdbKind::Movie, 19_723); assert!(movie.contains("primary_release_date.gte=2023-12-19"), "{movie}"); assert!(movie.contains("primary_release_date.lte=2024-01-01"), "{movie}"); // The marker selects the mode, and outranks a co-sent `popular`. assert_eq!( discovery_mode(&query(&[("airing", "true"), ("popular", "true")])), Some(DiscoveryMode::Airing) ); // The built query carries the date bound and lifts the vote floor. let url = build_path_and_query(DiscoveryMode::Airing, TmdbKind::Tv, false, &[], 1); assert!(url.starts_with("discover/tv?"), "{url}"); assert!(url.contains("air_date.gte="), "{url}"); assert!(url.contains("vote_count.gte=0"), "{url}"); // Still gated for adult residue like every other /discover row. assert!(url.contains("without_keywords="), "{url}"); // A popular row is unchanged: no date bound, floor intact. let pop = build_path_and_query(DiscoveryMode::Popular, TmdbKind::Tv, false, &[], 1); assert!(!pop.contains("air_date."), "{pop}"); assert!(pop.contains(&format!("vote_count.gte={DISCOVERY_MIN_VOTES}")), "{pop}"); } }