feat: implement TMDb enrichment for movie profiles

- Add SqliteMovieProfileRepository for managing movie profiles in SQLite.
- Create TmdbEnrichmentClient to fetch movie details from TMDb API.
- Implement enrichment event handling with EnrichmentHandler.
- Introduce periodic jobs for cleaning up expired import sessions and checking for stale movie profiles.
- Update application context to include movie profile repository.
- Add API endpoint to retrieve movie profiles.
- Extend domain models with new structures for movie enrichment (Genre, Keyword, CastMember, CrewMember, MovieProfile).
- Modify event system to include MovieEnrichmentRequested event.
- Enhance tests to cover new functionality and ensure stability.
This commit is contained in:
2026-05-12 13:23:41 +02:00
parent c696a3b780
commit 38d13fbff1
30 changed files with 1193 additions and 30 deletions

15
Cargo.lock generated
View File

@@ -5152,6 +5152,20 @@ version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index" source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20" checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20"
[[package]]
name = "tmdb-enrichment"
version = "0.1.0"
dependencies = [
"anyhow",
"async-trait",
"chrono",
"domain",
"reqwest 0.13.3",
"serde",
"serde_json",
"tracing",
]
[[package]] [[package]]
name = "tokio" name = "tokio"
version = "1.52.1" version = "1.52.1"
@@ -6331,6 +6345,7 @@ dependencies = [
"sqlite-event-queue", "sqlite-event-queue",
"sqlite-federation", "sqlite-federation",
"sqlx", "sqlx",
"tmdb-enrichment",
"tokio", "tokio",
"tracing", "tracing",
"tracing-subscriber", "tracing-subscriber",

View File

@@ -21,6 +21,7 @@ members = [
"crates/adapters/nats", "crates/adapters/nats",
"crates/api-types", "crates/api-types",
"crates/application", "crates/application",
"crates/adapters/tmdb-enrichment",
"crates/domain", "crates/domain",
"crates/presentation", "crates/presentation",
"crates/tui", "crates/tui",
@@ -55,6 +56,7 @@ csv = "1"
api-types = { path = "crates/api-types" } api-types = { path = "crates/api-types" }
domain = { path = "crates/domain" } domain = { path = "crates/domain" }
tmdb-enrichment = { path = "crates/adapters/tmdb-enrichment" }
application = { path = "crates/application" } application = { path = "crates/application" }
presentation = { path = "crates/presentation" } presentation = { path = "crates/presentation" }
auth = { path = "crates/adapters/auth" } auth = { path = "crates/adapters/auth" }

View File

@@ -29,6 +29,7 @@ COPY crates/adapters/postgres-event-queue/Cargo.toml crates/adapters/postgres-ev
COPY crates/adapters/template-askama/Cargo.toml crates/adapters/template-askama/Cargo.toml COPY crates/adapters/template-askama/Cargo.toml crates/adapters/template-askama/Cargo.toml
COPY crates/api-types/Cargo.toml crates/api-types/Cargo.toml COPY crates/api-types/Cargo.toml crates/api-types/Cargo.toml
COPY crates/application/Cargo.toml crates/application/Cargo.toml COPY crates/application/Cargo.toml crates/application/Cargo.toml
COPY crates/adapters/tmdb-enrichment/Cargo.toml crates/adapters/tmdb-enrichment/Cargo.toml
COPY crates/domain/Cargo.toml crates/domain/Cargo.toml COPY crates/domain/Cargo.toml crates/domain/Cargo.toml
COPY crates/presentation/Cargo.toml crates/presentation/Cargo.toml COPY crates/presentation/Cargo.toml crates/presentation/Cargo.toml
COPY crates/tui/Cargo.toml crates/tui/Cargo.toml COPY crates/tui/Cargo.toml crates/tui/Cargo.toml

View File

@@ -7,6 +7,7 @@ A self-hosted, server-side rendered movie logging system with a full REST API. B
- Log movies with a TMDB/OMDb ID or manual title/year/director, with a 05 rating - Log movies with a TMDB/OMDb ID or manual title/year/director, with a 05 rating
- Immutable append-only viewing ledger (tracks re-watches) - Immutable append-only viewing ledger (tracks re-watches)
- Background poster fetching and storage (local filesystem or S3-compatible) - Background poster fetching and storage (local filesystem or S3-compatible)
- Movie enrichment via TMDb — full cast, crew, genres, keywords, runtime, budget/revenue, ratings; fetched automatically on movie discovery and refreshed every 30 days; exposed via `GET /api/v1/movies/{id}/profile`
- RSS/Atom feed for public subscription (global and per-user) - RSS/Atom feed for public subscription (global and per-user)
- JWT authentication via cookie (HTML) or Bearer token (REST API) - JWT authentication via cookie (HTML) or Bearer token (REST API)
- ActivityPub federation — follow/unfollow remote users, accept/reject/remove followers, federated reviews broadcast as `Note` objects with `#MoviesDiary` + `#MovieTitle` hashtags, paginated outbox, boost/Announce tracking, NodeInfo discovery endpoint, shared inbox delivery, actor profile sync (bio, avatar, discoverable) - ActivityPub federation — follow/unfollow remote users, accept/reject/remove followers, federated reviews broadcast as `Note` objects with `#MoviesDiary` + `#MovieTitle` hashtags, paginated outbox, boost/Announce tracking, NodeInfo discovery endpoint, shared inbox delivery, actor profile sync (bio, avatar, discoverable)
@@ -37,6 +38,7 @@ adapters/
poster-fetcher — downloads poster images poster-fetcher — downloads poster images
image-storage — stores images (posters + user avatars) on local filesystem or S3-compatible storage image-storage — stores images (posters + user avatars) on local filesystem or S3-compatible storage
poster-sync — event handler: triggers poster fetch+store on MovieDiscovered poster-sync — event handler: triggers poster fetch+store on MovieDiscovered
tmdb-enrichment — event handler: fetches full movie profile (cast, crew, genres, keywords, box office) from TMDb on MovieEnrichmentRequested; resolves IMDb IDs automatically
template-askama — Askama HTML rendering template-askama — Askama HTML rendering
rss — RSS/Atom feed generation rss — RSS/Atom feed generation
export — CSV and JSON diary serialization export — CSV and JSON diary serialization
@@ -74,6 +76,9 @@ JWT_SECRET=change-me
# OMDb metadata # OMDb metadata
OMDB_API_KEY=your-key OMDB_API_KEY=your-key
# TMDb metadata + enrichment (optional — enables full cast/crew/genre data)
# TMDB_API_KEY=your-key
# Public base URL (used for ActivityPub actor URLs and canonical links) # Public base URL (used for ActivityPub actor URLs and canonical links)
BASE_URL=https://yourdomain.example.com BASE_URL=https://yourdomain.example.com

View File

@@ -39,6 +39,10 @@ pub enum EventPayload {
review_id: String, review_id: String,
user_id: String, user_id: String,
}, },
MovieEnrichmentRequested {
movie_id: String,
external_metadata_id: String,
},
} }
impl EventPayload { impl EventPayload {
@@ -50,6 +54,7 @@ impl EventPayload {
EventPayload::MovieDeleted { .. } => "MovieDeleted", EventPayload::MovieDeleted { .. } => "MovieDeleted",
EventPayload::UserUpdated { .. } => "UserUpdated", EventPayload::UserUpdated { .. } => "UserUpdated",
EventPayload::ReviewDeleted { .. } => "ReviewDeleted", EventPayload::ReviewDeleted { .. } => "ReviewDeleted",
EventPayload::MovieEnrichmentRequested { .. } => "MovieEnrichmentRequested",
} }
} }
} }
@@ -103,6 +108,12 @@ impl From<&DomainEvent> for EventPayload {
review_id: review_id.value().to_string(), review_id: review_id.value().to_string(),
user_id: user_id.value().to_string(), user_id: user_id.value().to_string(),
}, },
DomainEvent::MovieEnrichmentRequested { movie_id, external_metadata_id } => {
EventPayload::MovieEnrichmentRequested {
movie_id: movie_id.value().to_string(),
external_metadata_id: external_metadata_id.clone(),
}
}
} }
} }
} }
@@ -154,6 +165,12 @@ impl TryFrom<EventPayload> for DomainEvent {
user_id: UserId::from_uuid(parse_uuid(&user_id, "user_id")?), user_id: UserId::from_uuid(parse_uuid(&user_id, "user_id")?),
}) })
} }
EventPayload::MovieEnrichmentRequested { movie_id, external_metadata_id } => {
Ok(DomainEvent::MovieEnrichmentRequested {
movie_id: MovieId::from_uuid(parse_uuid(&movie_id, "movie_id")?),
external_metadata_id,
})
}
} }
} }
} }

View File

@@ -6,8 +6,9 @@ pub fn event_to_subject(prefix: &str, event: &DomainEvent) -> String {
DomainEvent::ReviewUpdated { .. } => "review.updated", DomainEvent::ReviewUpdated { .. } => "review.updated",
DomainEvent::ReviewDeleted { .. } => "review.deleted", DomainEvent::ReviewDeleted { .. } => "review.deleted",
DomainEvent::MovieDiscovered { .. } => "movie.discovered", DomainEvent::MovieDiscovered { .. } => "movie.discovered",
DomainEvent::MovieDeleted { .. } => "movie.deleted", DomainEvent::MovieDeleted { .. } => "movie.deleted",
DomainEvent::UserUpdated { .. } => "user.updated", DomainEvent::UserUpdated { .. } => "user.updated",
DomainEvent::MovieEnrichmentRequested { .. } => "movie.enrichment.requested",
}; };
format!("{prefix}.{suffix}") format!("{prefix}.{suffix}")
} }

View File

@@ -0,0 +1,54 @@
CREATE TABLE IF NOT EXISTS movie_profiles (
movie_id TEXT PRIMARY KEY NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_id BIGINT NOT NULL,
imdb_id TEXT,
overview TEXT,
tagline TEXT,
runtime_minutes INTEGER,
budget_usd BIGINT,
revenue_usd BIGINT,
vote_average DOUBLE PRECISION,
vote_count INTEGER,
original_language TEXT,
collection_name TEXT,
enriched_at TIMESTAMPTZ NOT NULL
);
CREATE TABLE IF NOT EXISTS movie_genres (
movie_id TEXT NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_id INTEGER NOT NULL,
name TEXT NOT NULL,
PRIMARY KEY (movie_id, tmdb_id)
);
CREATE TABLE IF NOT EXISTS movie_keywords (
movie_id TEXT NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_id INTEGER NOT NULL,
name TEXT NOT NULL,
PRIMARY KEY (movie_id, tmdb_id)
);
CREATE TABLE IF NOT EXISTS movie_cast (
movie_id TEXT NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_person_id BIGINT NOT NULL,
name TEXT NOT NULL,
character TEXT NOT NULL,
billing_order INTEGER NOT NULL,
profile_path TEXT,
PRIMARY KEY (movie_id, tmdb_person_id)
);
CREATE TABLE IF NOT EXISTS movie_crew (
movie_id TEXT NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_person_id BIGINT NOT NULL,
name TEXT NOT NULL,
job TEXT NOT NULL,
department TEXT NOT NULL,
profile_path TEXT,
PRIMARY KEY (movie_id, tmdb_person_id, job)
);
CREATE INDEX IF NOT EXISTS idx_movie_cast_person ON movie_cast (tmdb_person_id);
CREATE INDEX IF NOT EXISTS idx_movie_crew_person ON movie_crew (tmdb_person_id);
CREATE INDEX IF NOT EXISTS idx_movie_genres_name ON movie_genres (name);
CREATE INDEX IF NOT EXISTS idx_movie_keywords_name ON movie_keywords (name);

View File

@@ -15,6 +15,7 @@ use sqlx::PgPool;
mod import_profile; mod import_profile;
mod import_session; mod import_session;
mod models; mod models;
mod profile;
mod users; mod users;
use models::{ use models::{
@@ -24,6 +25,7 @@ use models::{
pub use import_profile::PostgresImportProfileRepository; pub use import_profile::PostgresImportProfileRepository;
pub use import_session::PostgresImportSessionRepository; pub use import_session::PostgresImportSessionRepository;
pub use profile::PostgresMovieProfileRepository;
pub use users::PostgresUserRepository; pub use users::PostgresUserRepository;
fn format_year_month(ym: &str) -> String { fn format_year_month(ym: &str) -> String {
@@ -865,6 +867,7 @@ pub async fn wire(database_url: &str) -> anyhow::Result<(
std::sync::Arc<dyn domain::ports::UserRepository>, std::sync::Arc<dyn domain::ports::UserRepository>,
std::sync::Arc<dyn domain::ports::ImportSessionRepository>, std::sync::Arc<dyn domain::ports::ImportSessionRepository>,
std::sync::Arc<dyn domain::ports::ImportProfileRepository>, std::sync::Arc<dyn domain::ports::ImportProfileRepository>,
std::sync::Arc<dyn domain::ports::MovieProfileRepository>,
)> { )> {
use anyhow::Context; use anyhow::Context;
@@ -880,6 +883,7 @@ pub async fn wire(database_url: &str) -> anyhow::Result<(
let import_session_repo = std::sync::Arc::new(PostgresImportSessionRepository::new(pool.clone())); let import_session_repo = std::sync::Arc::new(PostgresImportSessionRepository::new(pool.clone()));
let import_profile_repo = std::sync::Arc::new(PostgresImportProfileRepository::new(pool.clone())); let import_profile_repo = std::sync::Arc::new(PostgresImportProfileRepository::new(pool.clone()));
let movie_profile_repo = std::sync::Arc::new(PostgresMovieProfileRepository::new(pool.clone()));
Ok(( Ok((
pool.clone(), pool.clone(),
@@ -890,5 +894,6 @@ pub async fn wire(database_url: &str) -> anyhow::Result<(
std::sync::Arc::new(PostgresUserRepository::new(pool)) as _, std::sync::Arc::new(PostgresUserRepository::new(pool)) as _,
import_session_repo as _, import_session_repo as _,
import_profile_repo as _, import_profile_repo as _,
movie_profile_repo as _,
)) ))
} }

View File

@@ -0,0 +1,236 @@
use async_trait::async_trait;
use chrono::{DateTime, Utc};
use domain::{
errors::DomainError,
models::{CastMember, CrewMember, Genre, Keyword, MovieProfile},
ports::MovieProfileRepository,
value_objects::MovieId,
};
use sqlx::{PgPool, Row};
pub struct PostgresMovieProfileRepository {
pool: PgPool,
}
impl PostgresMovieProfileRepository {
pub fn new(pool: PgPool) -> Self {
Self { pool }
}
fn map_err(e: sqlx::Error) -> DomainError {
tracing::error!("Database error: {:?}", e);
DomainError::InfrastructureError("Database operation failed".into())
}
}
#[async_trait]
impl MovieProfileRepository for PostgresMovieProfileRepository {
async fn upsert(&self, p: &MovieProfile) -> Result<(), DomainError> {
let movie_id = p.movie_id.value().to_string();
let mut tx = self.pool.begin().await.map_err(Self::map_err)?;
sqlx::query(
r#"INSERT INTO movie_profiles
(movie_id, tmdb_id, imdb_id, overview, tagline, runtime_minutes,
budget_usd, revenue_usd, vote_average, vote_count,
original_language, collection_name, enriched_at)
VALUES ($1,$2,$3,$4,$5,$6,$7,$8,$9,$10,$11,$12,$13)
ON CONFLICT(movie_id) DO UPDATE SET
tmdb_id=EXCLUDED.tmdb_id, imdb_id=EXCLUDED.imdb_id,
overview=EXCLUDED.overview, tagline=EXCLUDED.tagline,
runtime_minutes=EXCLUDED.runtime_minutes,
budget_usd=EXCLUDED.budget_usd, revenue_usd=EXCLUDED.revenue_usd,
vote_average=EXCLUDED.vote_average, vote_count=EXCLUDED.vote_count,
original_language=EXCLUDED.original_language,
collection_name=EXCLUDED.collection_name,
enriched_at=EXCLUDED.enriched_at"#,
)
.bind(&movie_id)
.bind(p.tmdb_id as i64)
.bind(&p.imdb_id)
.bind(&p.overview)
.bind(&p.tagline)
.bind(p.runtime_minutes.map(|v| v as i32))
.bind(p.budget_usd)
.bind(p.revenue_usd)
.bind(p.vote_average)
.bind(p.vote_count.map(|v| v as i32))
.bind(&p.original_language)
.bind(&p.collection_name)
.bind(p.enriched_at)
.execute(&mut *tx)
.await
.map_err(Self::map_err)?;
sqlx::query("DELETE FROM movie_genres WHERE movie_id = $1")
.bind(&movie_id)
.execute(&mut *tx).await.map_err(Self::map_err)?;
for g in &p.genres {
sqlx::query("INSERT INTO movie_genres (movie_id, tmdb_id, name) VALUES ($1,$2,$3) ON CONFLICT DO NOTHING")
.bind(&movie_id).bind(g.tmdb_id as i32).bind(&g.name)
.execute(&mut *tx).await.map_err(Self::map_err)?;
}
sqlx::query("DELETE FROM movie_keywords WHERE movie_id = $1")
.bind(&movie_id)
.execute(&mut *tx).await.map_err(Self::map_err)?;
for k in &p.keywords {
sqlx::query("INSERT INTO movie_keywords (movie_id, tmdb_id, name) VALUES ($1,$2,$3) ON CONFLICT DO NOTHING")
.bind(&movie_id).bind(k.tmdb_id as i32).bind(&k.name)
.execute(&mut *tx).await.map_err(Self::map_err)?;
}
sqlx::query("DELETE FROM movie_cast WHERE movie_id = $1")
.bind(&movie_id)
.execute(&mut *tx).await.map_err(Self::map_err)?;
for c in &p.cast {
sqlx::query(
"INSERT INTO movie_cast \
(movie_id, tmdb_person_id, name, character, billing_order, profile_path) \
VALUES ($1,$2,$3,$4,$5,$6) ON CONFLICT DO NOTHING",
)
.bind(&movie_id).bind(c.tmdb_person_id as i64).bind(&c.name)
.bind(&c.character).bind(c.billing_order as i32).bind(&c.profile_path)
.execute(&mut *tx).await.map_err(Self::map_err)?;
}
sqlx::query("DELETE FROM movie_crew WHERE movie_id = $1")
.bind(&movie_id)
.execute(&mut *tx).await.map_err(Self::map_err)?;
for cr in &p.crew {
sqlx::query(
"INSERT INTO movie_crew \
(movie_id, tmdb_person_id, name, job, department, profile_path) \
VALUES ($1,$2,$3,$4,$5,$6) ON CONFLICT DO NOTHING",
)
.bind(&movie_id).bind(cr.tmdb_person_id as i64).bind(&cr.name)
.bind(&cr.job).bind(&cr.department).bind(&cr.profile_path)
.execute(&mut *tx).await.map_err(Self::map_err)?;
}
tx.commit().await.map_err(Self::map_err)
}
async fn get_by_movie_id(&self, id: &MovieId) -> Result<Option<MovieProfile>, DomainError> {
let movie_id = id.value().to_string();
let row = sqlx::query(
"SELECT tmdb_id, imdb_id, overview, tagline, runtime_minutes, budget_usd,
revenue_usd, vote_average, vote_count, original_language,
collection_name, enriched_at
FROM movie_profiles WHERE movie_id = $1",
)
.bind(&movie_id)
.fetch_optional(&self.pool)
.await
.map_err(Self::map_err)?;
let row = match row {
Some(r) => r,
None => return Ok(None),
};
let enriched_at: DateTime<Utc> = row.try_get("enriched_at")
.map_err(|_| DomainError::InfrastructureError("invalid enriched_at".into()))?;
let genres = sqlx::query("SELECT tmdb_id, name FROM movie_genres WHERE movie_id = $1")
.bind(&movie_id)
.fetch_all(&self.pool).await.map_err(Self::map_err)?
.into_iter()
.map(|r| Genre {
tmdb_id: r.try_get::<i32, _>("tmdb_id").unwrap_or(0) as u32,
name: r.try_get("name").unwrap_or_default(),
})
.collect();
let keywords = sqlx::query("SELECT tmdb_id, name FROM movie_keywords WHERE movie_id = $1")
.bind(&movie_id)
.fetch_all(&self.pool).await.map_err(Self::map_err)?
.into_iter()
.map(|r| Keyword {
tmdb_id: r.try_get::<i32, _>("tmdb_id").unwrap_or(0) as u32,
name: r.try_get("name").unwrap_or_default(),
})
.collect();
let cast = sqlx::query(
"SELECT tmdb_person_id, name, character, billing_order, profile_path \
FROM movie_cast WHERE movie_id = $1 ORDER BY billing_order",
)
.bind(&movie_id)
.fetch_all(&self.pool).await.map_err(Self::map_err)?
.into_iter()
.map(|r| CastMember {
tmdb_person_id: r.try_get::<i64, _>("tmdb_person_id").unwrap_or(0) as u64,
name: r.try_get("name").unwrap_or_default(),
character: r.try_get("character").unwrap_or_default(),
billing_order: r.try_get::<i32, _>("billing_order").unwrap_or(0) as u32,
profile_path: r.try_get("profile_path").ok(),
})
.collect();
let crew = sqlx::query(
"SELECT tmdb_person_id, name, job, department, profile_path \
FROM movie_crew WHERE movie_id = $1",
)
.bind(&movie_id)
.fetch_all(&self.pool).await.map_err(Self::map_err)?
.into_iter()
.map(|r| CrewMember {
tmdb_person_id: r.try_get::<i64, _>("tmdb_person_id").unwrap_or(0) as u64,
name: r.try_get("name").unwrap_or_default(),
job: r.try_get("job").unwrap_or_default(),
department: r.try_get("department").unwrap_or_default(),
profile_path: r.try_get("profile_path").ok(),
})
.collect();
Ok(Some(MovieProfile {
movie_id: id.clone(),
tmdb_id: row.try_get::<i64, _>("tmdb_id").unwrap_or(0) as u64,
imdb_id: row.try_get("imdb_id").ok(),
overview: row.try_get("overview").ok(),
tagline: row.try_get("tagline").ok(),
runtime_minutes: row.try_get::<Option<i32>, _>("runtime_minutes").ok().flatten().map(|v| v as u32),
budget_usd: row.try_get("budget_usd").ok(),
revenue_usd: row.try_get("revenue_usd").ok(),
vote_average: row.try_get("vote_average").ok(),
vote_count: row.try_get::<Option<i32>, _>("vote_count").ok().flatten().map(|v| v as u32),
original_language: row.try_get("original_language").ok(),
collection_name: row.try_get("collection_name").ok(),
genres,
keywords,
cast,
crew,
enriched_at,
}))
}
async fn list_stale(&self) -> Result<Vec<(MovieId, String)>, DomainError> {
let threshold = Utc::now() - chrono::Duration::days(30);
let rows = sqlx::query(
r#"SELECT m.id, m.external_metadata_id
FROM movies m
LEFT JOIN movie_profiles p ON p.movie_id = m.id
WHERE m.external_metadata_id IS NOT NULL
AND (p.movie_id IS NULL OR p.enriched_at < $1)
ORDER BY p.enriched_at ASC NULLS FIRST"#,
)
.bind(threshold)
.fetch_all(&self.pool)
.await
.map_err(Self::map_err)?;
Ok(rows
.into_iter()
.filter_map(|r| {
let ext_id: Option<String> = r.try_get("external_metadata_id").ok()?;
let ext_id = ext_id?;
let id_str: String = r.try_get("id").ok()?;
let movie_id = id_str.parse::<uuid::Uuid>().ok().map(MovieId::from_uuid)?;
Some((movie_id, ext_id))
})
.collect())
}
}

View File

@@ -0,0 +1,54 @@
CREATE TABLE IF NOT EXISTS movie_profiles (
movie_id TEXT PRIMARY KEY NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_id INTEGER NOT NULL,
imdb_id TEXT,
overview TEXT,
tagline TEXT,
runtime_minutes INTEGER,
budget_usd INTEGER,
revenue_usd INTEGER,
vote_average REAL,
vote_count INTEGER,
original_language TEXT,
collection_name TEXT,
enriched_at TEXT NOT NULL
);
CREATE TABLE IF NOT EXISTS movie_genres (
movie_id TEXT NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_id INTEGER NOT NULL,
name TEXT NOT NULL,
PRIMARY KEY (movie_id, tmdb_id)
);
CREATE TABLE IF NOT EXISTS movie_keywords (
movie_id TEXT NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_id INTEGER NOT NULL,
name TEXT NOT NULL,
PRIMARY KEY (movie_id, tmdb_id)
);
CREATE TABLE IF NOT EXISTS movie_cast (
movie_id TEXT NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_person_id INTEGER NOT NULL,
name TEXT NOT NULL,
character TEXT NOT NULL,
billing_order INTEGER NOT NULL,
profile_path TEXT,
PRIMARY KEY (movie_id, tmdb_person_id)
);
CREATE TABLE IF NOT EXISTS movie_crew (
movie_id TEXT NOT NULL REFERENCES movies(id) ON DELETE CASCADE,
tmdb_person_id INTEGER NOT NULL,
name TEXT NOT NULL,
job TEXT NOT NULL,
department TEXT NOT NULL,
profile_path TEXT,
PRIMARY KEY (movie_id, tmdb_person_id, job)
);
CREATE INDEX IF NOT EXISTS idx_movie_cast_person ON movie_cast (tmdb_person_id);
CREATE INDEX IF NOT EXISTS idx_movie_crew_person ON movie_crew (tmdb_person_id);
CREATE INDEX IF NOT EXISTS idx_movie_genres_name ON movie_genres (name);
CREATE INDEX IF NOT EXISTS idx_movie_keywords_name ON movie_keywords (name);

View File

@@ -16,6 +16,7 @@ mod import_profile;
mod import_session; mod import_session;
mod migrations; mod migrations;
mod models; mod models;
mod profile;
mod users; mod users;
use models::{ use models::{
@@ -25,6 +26,7 @@ use models::{
pub use import_profile::SqliteImportProfileRepository; pub use import_profile::SqliteImportProfileRepository;
pub use import_session::SqliteImportSessionRepository; pub use import_session::SqliteImportSessionRepository;
pub use profile::SqliteMovieProfileRepository;
pub use users::SqliteUserRepository; pub use users::SqliteUserRepository;
fn format_year_month(ym: &str) -> String { fn format_year_month(ym: &str) -> String {
@@ -854,6 +856,7 @@ pub async fn wire(database_url: &str) -> anyhow::Result<(
std::sync::Arc<dyn domain::ports::UserRepository>, std::sync::Arc<dyn domain::ports::UserRepository>,
std::sync::Arc<dyn domain::ports::ImportSessionRepository>, std::sync::Arc<dyn domain::ports::ImportSessionRepository>,
std::sync::Arc<dyn domain::ports::ImportProfileRepository>, std::sync::Arc<dyn domain::ports::ImportProfileRepository>,
std::sync::Arc<dyn domain::ports::MovieProfileRepository>,
)> { )> {
use std::str::FromStr; use std::str::FromStr;
use anyhow::Context; use anyhow::Context;
@@ -876,6 +879,7 @@ pub async fn wire(database_url: &str) -> anyhow::Result<(
let import_session_repo = std::sync::Arc::new(SqliteImportSessionRepository::new(pool.clone())); let import_session_repo = std::sync::Arc::new(SqliteImportSessionRepository::new(pool.clone()));
let import_profile_repo = std::sync::Arc::new(SqliteImportProfileRepository::new(pool.clone())); let import_profile_repo = std::sync::Arc::new(SqliteImportProfileRepository::new(pool.clone()));
let movie_profile_repo = std::sync::Arc::new(SqliteMovieProfileRepository::new(pool.clone()));
Ok(( Ok((
pool.clone(), pool.clone(),
@@ -886,6 +890,7 @@ pub async fn wire(database_url: &str) -> anyhow::Result<(
std::sync::Arc::new(SqliteUserRepository::new(pool)) as _, std::sync::Arc::new(SqliteUserRepository::new(pool)) as _,
import_session_repo as _, import_session_repo as _,
import_profile_repo as _, import_profile_repo as _,
movie_profile_repo as _,
)) ))
} }

View File

@@ -0,0 +1,240 @@
use async_trait::async_trait;
use chrono::{DateTime, Utc};
use domain::{
errors::DomainError,
models::{CastMember, CrewMember, Genre, Keyword, MovieProfile},
ports::MovieProfileRepository,
value_objects::MovieId,
};
use sqlx::{Row, SqlitePool};
pub struct SqliteMovieProfileRepository {
pool: SqlitePool,
}
impl SqliteMovieProfileRepository {
pub fn new(pool: SqlitePool) -> Self {
Self { pool }
}
fn map_err(e: sqlx::Error) -> DomainError {
tracing::error!("Database error: {:?}", e);
DomainError::InfrastructureError("Database operation failed".into())
}
}
#[async_trait]
impl MovieProfileRepository for SqliteMovieProfileRepository {
async fn upsert(&self, p: &MovieProfile) -> Result<(), DomainError> {
let movie_id = p.movie_id.value().to_string();
let enriched_at = p.enriched_at.to_rfc3339();
let mut tx = self.pool.begin().await.map_err(Self::map_err)?;
sqlx::query(
r#"INSERT INTO movie_profiles
(movie_id, tmdb_id, imdb_id, overview, tagline, runtime_minutes,
budget_usd, revenue_usd, vote_average, vote_count,
original_language, collection_name, enriched_at)
VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)
ON CONFLICT(movie_id) DO UPDATE SET
tmdb_id=excluded.tmdb_id, imdb_id=excluded.imdb_id,
overview=excluded.overview, tagline=excluded.tagline,
runtime_minutes=excluded.runtime_minutes,
budget_usd=excluded.budget_usd, revenue_usd=excluded.revenue_usd,
vote_average=excluded.vote_average, vote_count=excluded.vote_count,
original_language=excluded.original_language,
collection_name=excluded.collection_name,
enriched_at=excluded.enriched_at"#,
)
.bind(&movie_id)
.bind(p.tmdb_id as i64)
.bind(&p.imdb_id)
.bind(&p.overview)
.bind(&p.tagline)
.bind(p.runtime_minutes.map(|v| v as i64))
.bind(p.budget_usd)
.bind(p.revenue_usd)
.bind(p.vote_average)
.bind(p.vote_count.map(|v| v as i64))
.bind(&p.original_language)
.bind(&p.collection_name)
.bind(&enriched_at)
.execute(&mut *tx)
.await
.map_err(Self::map_err)?;
sqlx::query("DELETE FROM movie_genres WHERE movie_id = ?")
.bind(&movie_id)
.execute(&mut *tx).await.map_err(Self::map_err)?;
for g in &p.genres {
sqlx::query("INSERT OR IGNORE INTO movie_genres (movie_id, tmdb_id, name) VALUES (?,?,?)")
.bind(&movie_id).bind(g.tmdb_id as i64).bind(&g.name)
.execute(&mut *tx).await.map_err(Self::map_err)?;
}
sqlx::query("DELETE FROM movie_keywords WHERE movie_id = ?")
.bind(&movie_id)
.execute(&mut *tx).await.map_err(Self::map_err)?;
for k in &p.keywords {
sqlx::query("INSERT OR IGNORE INTO movie_keywords (movie_id, tmdb_id, name) VALUES (?,?,?)")
.bind(&movie_id).bind(k.tmdb_id as i64).bind(&k.name)
.execute(&mut *tx).await.map_err(Self::map_err)?;
}
sqlx::query("DELETE FROM movie_cast WHERE movie_id = ?")
.bind(&movie_id)
.execute(&mut *tx).await.map_err(Self::map_err)?;
for c in &p.cast {
sqlx::query(
"INSERT OR IGNORE INTO movie_cast \
(movie_id, tmdb_person_id, name, character, billing_order, profile_path) \
VALUES (?,?,?,?,?,?)",
)
.bind(&movie_id).bind(c.tmdb_person_id as i64).bind(&c.name)
.bind(&c.character).bind(c.billing_order as i64).bind(&c.profile_path)
.execute(&mut *tx).await.map_err(Self::map_err)?;
}
sqlx::query("DELETE FROM movie_crew WHERE movie_id = ?")
.bind(&movie_id)
.execute(&mut *tx).await.map_err(Self::map_err)?;
for cr in &p.crew {
sqlx::query(
"INSERT OR IGNORE INTO movie_crew \
(movie_id, tmdb_person_id, name, job, department, profile_path) \
VALUES (?,?,?,?,?,?)",
)
.bind(&movie_id).bind(cr.tmdb_person_id as i64).bind(&cr.name)
.bind(&cr.job).bind(&cr.department).bind(&cr.profile_path)
.execute(&mut *tx).await.map_err(Self::map_err)?;
}
tx.commit().await.map_err(Self::map_err)
}
async fn get_by_movie_id(&self, id: &MovieId) -> Result<Option<MovieProfile>, DomainError> {
let movie_id = id.value().to_string();
let row = sqlx::query(
"SELECT tmdb_id, imdb_id, overview, tagline, runtime_minutes, budget_usd,
revenue_usd, vote_average, vote_count, original_language,
collection_name, enriched_at
FROM movie_profiles WHERE movie_id = ?",
)
.bind(&movie_id)
.fetch_optional(&self.pool)
.await
.map_err(Self::map_err)?;
let row = match row {
Some(r) => r,
None => return Ok(None),
};
let enriched_at_str: String = row.try_get("enriched_at")
.map_err(|_| DomainError::InfrastructureError("invalid enriched_at".into()))?;
let enriched_at: DateTime<Utc> = enriched_at_str
.parse()
.map_err(|_| DomainError::InfrastructureError("invalid enriched_at".into()))?;
let genres = sqlx::query("SELECT tmdb_id, name FROM movie_genres WHERE movie_id = ?")
.bind(&movie_id)
.fetch_all(&self.pool).await.map_err(Self::map_err)?
.into_iter()
.map(|r| Genre {
tmdb_id: r.try_get::<i64, _>("tmdb_id").unwrap_or(0) as u32,
name: r.try_get("name").unwrap_or_default(),
})
.collect();
let keywords = sqlx::query("SELECT tmdb_id, name FROM movie_keywords WHERE movie_id = ?")
.bind(&movie_id)
.fetch_all(&self.pool).await.map_err(Self::map_err)?
.into_iter()
.map(|r| Keyword {
tmdb_id: r.try_get::<i64, _>("tmdb_id").unwrap_or(0) as u32,
name: r.try_get("name").unwrap_or_default(),
})
.collect();
let cast = sqlx::query(
"SELECT tmdb_person_id, name, character, billing_order, profile_path \
FROM movie_cast WHERE movie_id = ? ORDER BY billing_order",
)
.bind(&movie_id)
.fetch_all(&self.pool).await.map_err(Self::map_err)?
.into_iter()
.map(|r| CastMember {
tmdb_person_id: r.try_get::<i64, _>("tmdb_person_id").unwrap_or(0) as u64,
name: r.try_get("name").unwrap_or_default(),
character: r.try_get("character").unwrap_or_default(),
billing_order: r.try_get::<i64, _>("billing_order").unwrap_or(0) as u32,
profile_path: r.try_get("profile_path").ok(),
})
.collect();
let crew = sqlx::query(
"SELECT tmdb_person_id, name, job, department, profile_path \
FROM movie_crew WHERE movie_id = ?",
)
.bind(&movie_id)
.fetch_all(&self.pool).await.map_err(Self::map_err)?
.into_iter()
.map(|r| CrewMember {
tmdb_person_id: r.try_get::<i64, _>("tmdb_person_id").unwrap_or(0) as u64,
name: r.try_get("name").unwrap_or_default(),
job: r.try_get("job").unwrap_or_default(),
department: r.try_get("department").unwrap_or_default(),
profile_path: r.try_get("profile_path").ok(),
})
.collect();
Ok(Some(MovieProfile {
movie_id: id.clone(),
tmdb_id: row.try_get::<i64, _>("tmdb_id").unwrap_or(0) as u64,
imdb_id: row.try_get("imdb_id").ok(),
overview: row.try_get("overview").ok(),
tagline: row.try_get("tagline").ok(),
runtime_minutes: row.try_get::<Option<i64>, _>("runtime_minutes").ok().flatten().map(|v| v as u32),
budget_usd: row.try_get("budget_usd").ok(),
revenue_usd: row.try_get("revenue_usd").ok(),
vote_average: row.try_get("vote_average").ok(),
vote_count: row.try_get::<Option<i64>, _>("vote_count").ok().flatten().map(|v| v as u32),
original_language: row.try_get("original_language").ok(),
collection_name: row.try_get("collection_name").ok(),
genres,
keywords,
cast,
crew,
enriched_at,
}))
}
async fn list_stale(&self) -> Result<Vec<(MovieId, String)>, DomainError> {
let threshold = (Utc::now() - chrono::Duration::days(30)).to_rfc3339();
let rows = sqlx::query(
r#"SELECT m.id, m.external_metadata_id
FROM movies m
LEFT JOIN movie_profiles p ON p.movie_id = m.id
WHERE m.external_metadata_id IS NOT NULL
AND (p.movie_id IS NULL OR p.enriched_at < ?)
ORDER BY p.enriched_at ASC"#,
)
.bind(&threshold)
.fetch_all(&self.pool)
.await
.map_err(Self::map_err)?;
Ok(rows
.into_iter()
.filter_map(|r| {
let ext_id: Option<String> = r.try_get("external_metadata_id").ok()?;
let ext_id = ext_id?;
let id_str: String = r.try_get("id").ok()?;
let movie_id = id_str.parse::<uuid::Uuid>().ok().map(MovieId::from_uuid)?;
Some((movie_id, ext_id))
})
.collect())
}
}

View File

@@ -0,0 +1,14 @@
[package]
name = "tmdb-enrichment"
version = "0.1.0"
edition = "2024"
[dependencies]
domain = { workspace = true }
reqwest = { workspace = true }
serde = { workspace = true }
serde_json = { workspace = true }
anyhow = { workspace = true }
async-trait = { workspace = true }
tracing = { workspace = true }
chrono = { workspace = true }

View File

@@ -0,0 +1,211 @@
use std::sync::Arc;
use async_trait::async_trait;
use chrono::Utc;
use domain::{
errors::DomainError,
events::DomainEvent,
models::{CastMember, CrewMember, Genre, Keyword, MovieProfile},
ports::{EventHandler, MovieEnrichmentClient, MovieProfileRepository},
value_objects::MovieId,
};
use serde::Deserialize;
// ── TMDb enrichment client ───────────────────────────────────────────────────
pub struct TmdbEnrichmentClient {
api_key: String,
http: reqwest::Client,
}
impl TmdbEnrichmentClient {
pub fn from_env() -> Result<Self, DomainError> {
let api_key = std::env::var("TMDB_API_KEY").map_err(|_| {
DomainError::InfrastructureError("TMDB_API_KEY is not set".into())
})?;
Ok(Self { api_key, http: reqwest::Client::new() })
}
fn base(&self, path: &str) -> String {
format!("https://api.themoviedb.org/3{}", path)
}
async fn get<T: for<'de> Deserialize<'de>>(&self, url: &str, extra: &[(&str, &str)]) -> Result<T, DomainError> {
let mut req = self.http.get(url).query(&[("api_key", self.api_key.as_str())]);
for (k, v) in extra {
req = req.query(&[(k, v)]);
}
req.send().await
.map_err(|e| DomainError::InfrastructureError(e.to_string()))?
.error_for_status()
.map_err(|e| DomainError::InfrastructureError(e.to_string()))?
.json::<T>().await
.map_err(|e| DomainError::InfrastructureError(e.to_string()))
}
async fn resolve_tmdb_id(&self, external_id: &str) -> Result<u64, DomainError> {
if let Some(numeric) = external_id.strip_prefix("tmdb:") {
return numeric.parse::<u64>()
.map_err(|_| DomainError::InfrastructureError(format!("Invalid tmdb id: {numeric}")));
}
// Assume IMDb ID (tt…) — use /find
#[derive(Deserialize)]
struct FindResult { id: u64 }
#[derive(Deserialize)]
struct FindResponse { movie_results: Vec<FindResult> }
let url = self.base(&format!("/find/{}", external_id));
let resp: FindResponse = self.get(&url, &[("external_source", "imdb_id")]).await?;
resp.movie_results
.into_iter()
.next()
.map(|r| r.id)
.ok_or_else(|| DomainError::NotFound(format!("TMDb: no movie for {external_id}")))
}
}
#[async_trait]
impl MovieEnrichmentClient for TmdbEnrichmentClient {
async fn fetch_profile(&self, movie_id: MovieId, external_metadata_id: &str) -> Result<MovieProfile, DomainError> {
let tmdb_id = self.resolve_tmdb_id(external_metadata_id).await?;
#[derive(Deserialize)]
struct GenreDto { id: u32, name: String }
#[derive(Deserialize)]
struct CollectionDto { name: String }
#[derive(Deserialize)]
struct CastDto {
id: u64,
name: String,
character: String,
order: u32,
profile_path: Option<String>,
}
#[derive(Deserialize)]
struct CrewDto {
id: u64,
name: String,
job: String,
department: String,
profile_path: Option<String>,
}
#[derive(Deserialize)]
struct Credits { cast: Vec<CastDto>, crew: Vec<CrewDto> }
#[derive(Deserialize)]
struct KeywordDto { id: u32, name: String }
#[derive(Deserialize)]
struct Keywords { keywords: Vec<KeywordDto> }
#[derive(Deserialize)]
struct Details {
imdb_id: Option<String>,
overview: Option<String>,
tagline: Option<String>,
runtime: Option<u32>,
budget: Option<i64>,
revenue: Option<i64>,
vote_average: Option<f64>,
vote_count: Option<u32>,
original_language: Option<String>,
genres: Vec<GenreDto>,
belongs_to_collection: Option<CollectionDto>,
credits: Credits,
keywords: Keywords,
}
let url = self.base(&format!("/movie/{}", tmdb_id));
let d: Details = self.get(&url, &[("append_to_response", "credits,keywords")]).await?;
Ok(MovieProfile {
movie_id,
tmdb_id,
imdb_id: d.imdb_id.filter(|s| !s.is_empty()),
overview: d.overview.filter(|s| !s.is_empty()),
tagline: d.tagline.filter(|s| !s.is_empty()),
runtime_minutes: d.runtime,
budget_usd: d.budget.filter(|&v| v > 0),
revenue_usd: d.revenue.filter(|&v| v > 0),
vote_average: d.vote_average,
vote_count: d.vote_count,
original_language: d.original_language,
collection_name: d.belongs_to_collection.map(|c| c.name),
genres: d.genres.into_iter().map(|g| Genre { tmdb_id: g.id, name: g.name }).collect(),
keywords: d.keywords.keywords.into_iter()
.map(|k| Keyword { tmdb_id: k.id, name: k.name })
.collect(),
cast: d.credits.cast.into_iter().map(|c| CastMember {
tmdb_person_id: c.id,
name: c.name,
character: c.character,
billing_order: c.order,
profile_path: c.profile_path,
}).collect(),
crew: d.credits.crew.into_iter().map(|c| CrewMember {
tmdb_person_id: c.id,
name: c.name,
job: c.job,
department: c.department,
profile_path: c.profile_path,
}).collect(),
enriched_at: Utc::now(),
})
}
}
// ── Enrichment event handler ─────────────────────────────────────────────────
pub struct EnrichmentHandler {
pub enrichment_client: Arc<dyn MovieEnrichmentClient>,
pub profile_repo: Arc<dyn MovieProfileRepository>,
}
#[async_trait]
impl EventHandler for EnrichmentHandler {
async fn handle(&self, event: &DomainEvent) -> Result<(), DomainError> {
let (movie_id, external_metadata_id) = match event {
DomainEvent::MovieEnrichmentRequested { movie_id, external_metadata_id } => {
(movie_id.clone(), external_metadata_id.clone())
}
_ => return Ok(()),
};
// Skip if profile is fresh (checked by the repo's list_stale, but guard here too)
if let Ok(Some(existing)) = self.profile_repo.get_by_movie_id(&movie_id).await {
let age = Utc::now() - existing.enriched_at;
if age.num_days() < 30 {
tracing::debug!(
movie_id = %movie_id.value(),
"skipping enrichment — profile is {} days old",
age.num_days()
);
return Ok(());
}
}
tracing::info!(movie_id = %movie_id.value(), external_id = %external_metadata_id, "enriching movie");
match self.enrichment_client.fetch_profile(movie_id.clone(), &external_metadata_id).await {
Ok(profile) => {
self.profile_repo.upsert(&profile).await?;
tracing::info!(
movie_id = %movie_id.value(),
genres = profile.genres.len(),
cast = profile.cast.len(),
crew = profile.crew.len(),
"enrichment stored"
);
}
Err(DomainError::NotFound(msg)) => {
tracing::warn!(movie_id = %movie_id.value(), "TMDb lookup found nothing: {msg}");
}
Err(e) => return Err(e),
}
Ok(())
}
}

View File

@@ -1,6 +1,58 @@
use serde::{Deserialize, Serialize}; use serde::{Deserialize, Serialize};
use uuid::Uuid; use uuid::Uuid;
// ── Movie profile (enrichment) ────────────────────────────────────────────────
#[derive(Debug, Clone, Serialize, Deserialize, utoipa::ToSchema)]
pub struct GenreDto {
pub tmdb_id: u32,
pub name: String,
}
#[derive(Debug, Clone, Serialize, Deserialize, utoipa::ToSchema)]
pub struct KeywordDto {
pub tmdb_id: u32,
pub name: String,
}
#[derive(Debug, Clone, Serialize, Deserialize, utoipa::ToSchema)]
pub struct CastMemberDto {
pub tmdb_person_id: u64,
pub name: String,
pub character: String,
pub billing_order: u32,
pub profile_path: Option<String>,
}
#[derive(Debug, Clone, Serialize, Deserialize, utoipa::ToSchema)]
pub struct CrewMemberDto {
pub tmdb_person_id: u64,
pub name: String,
pub job: String,
pub department: String,
pub profile_path: Option<String>,
}
#[derive(Debug, Clone, Serialize, Deserialize, utoipa::ToSchema)]
pub struct MovieProfileResponse {
pub tmdb_id: u64,
pub imdb_id: Option<String>,
pub overview: Option<String>,
pub tagline: Option<String>,
pub runtime_minutes: Option<u32>,
pub budget_usd: Option<i64>,
pub revenue_usd: Option<i64>,
pub vote_average: Option<f64>,
pub vote_count: Option<u32>,
pub original_language: Option<String>,
pub collection_name: Option<String>,
pub genres: Vec<GenreDto>,
pub keywords: Vec<KeywordDto>,
pub cast: Vec<CastMemberDto>,
pub crew: Vec<CrewMemberDto>,
pub enriched_at: String,
}
#[derive(Debug, Clone, Serialize, Deserialize, utoipa::ToSchema)] #[derive(Debug, Clone, Serialize, Deserialize, utoipa::ToSchema)]
pub struct MovieDto { pub struct MovieDto {
pub id: Uuid, pub id: Uuid,

View File

@@ -4,7 +4,7 @@ use domain::ports::{
AuthService, DiaryExporter, DiaryRepository, DocumentParser, EventPublisher, AuthService, DiaryExporter, DiaryRepository, DocumentParser, EventPublisher,
ImageStorage, ImageStorage,
ImportProfileRepository, ImportSessionRepository, ImportProfileRepository, ImportSessionRepository,
MetadataClient, MovieRepository, PasswordHasher, PosterFetcherClient, MetadataClient, MovieProfileRepository, MovieRepository, PasswordHasher, PosterFetcherClient,
ReviewRepository, StatsRepository, UserRepository, ReviewRepository, StatsRepository, UserRepository,
}; };
@@ -27,5 +27,6 @@ pub struct AppContext {
pub user_repository: Arc<dyn UserRepository>, pub user_repository: Arc<dyn UserRepository>,
pub import_session_repository: Arc<dyn ImportSessionRepository>, pub import_session_repository: Arc<dyn ImportSessionRepository>,
pub import_profile_repository: Arc<dyn ImportProfileRepository>, pub import_profile_repository: Arc<dyn ImportProfileRepository>,
pub movie_profile_repository: Arc<dyn MovieProfileRepository>,
pub config: AppConfig, pub config: AppConfig,
} }

View File

@@ -0,0 +1,60 @@
use std::time::Duration;
use async_trait::async_trait;
use domain::{errors::DomainError, events::DomainEvent, ports::PeriodicJob};
use crate::context::AppContext;
pub struct ImportSessionCleanupJob {
ctx: AppContext,
}
impl ImportSessionCleanupJob {
pub fn new(ctx: AppContext) -> Self {
Self { ctx }
}
}
#[async_trait]
impl PeriodicJob for ImportSessionCleanupJob {
fn interval(&self) -> Duration {
Duration::from_secs(3600)
}
async fn run(&self) -> Result<(), DomainError> {
let n = crate::use_cases::cleanup_expired_import_sessions::execute(&self.ctx).await?;
tracing::info!("import session cleanup: removed {} expired sessions", n);
Ok(())
}
}
pub struct EnrichmentStalenessJob {
ctx: AppContext,
}
impl EnrichmentStalenessJob {
pub fn new(ctx: AppContext) -> Self {
Self { ctx }
}
}
#[async_trait]
impl PeriodicJob for EnrichmentStalenessJob {
fn interval(&self) -> Duration {
Duration::from_secs(3600)
}
async fn run(&self) -> Result<(), DomainError> {
let stale = self.ctx.movie_profile_repository.list_stale().await?;
if stale.is_empty() {
return Ok(());
}
tracing::info!("enrichment scan: {} stale movies", stale.len());
for (movie_id, external_metadata_id) in stale {
let event = DomainEvent::MovieEnrichmentRequested { movie_id, external_metadata_id };
self.ctx.event_publisher.publish(&event).await?;
}
Ok(())
}
}

View File

@@ -1,4 +1,5 @@
pub mod commands; pub mod commands;
pub mod jobs;
pub mod worker; pub mod worker;
pub mod config; pub mod config;
pub mod context; pub mod context;

View File

@@ -50,6 +50,14 @@ async fn publish_events(
} }
} }
if let Some(ext_id) = movie.external_metadata_id() {
let enrichment_event = DomainEvent::MovieEnrichmentRequested {
movie_id: movie.id().clone(),
external_metadata_id: ext_id.value().to_string(),
};
ctx.event_publisher.publish(&enrichment_event).await?;
}
ctx.event_publisher.publish(&review_event).await?; ctx.event_publisher.publish(&review_event).await?;
Ok(()) Ok(())
} }

View File

@@ -96,6 +96,7 @@ mod tests {
DomainEvent::ReviewDeleted { .. } => "review_deleted", DomainEvent::ReviewDeleted { .. } => "review_deleted",
DomainEvent::MovieDeleted { .. } => "movie_deleted", DomainEvent::MovieDeleted { .. } => "movie_deleted",
DomainEvent::UserUpdated { .. } => "user_updated", DomainEvent::UserUpdated { .. } => "user_updated",
DomainEvent::MovieEnrichmentRequested { .. } => "movie_enrichment_requested",
}; };
self.calls.lock().unwrap().push(label); self.calls.lock().unwrap().push(label);
Ok(()) Ok(())

View File

@@ -37,6 +37,10 @@ pub enum DomainEvent {
review_id: ReviewId, review_id: ReviewId,
user_id: UserId, user_id: UserId,
}, },
MovieEnrichmentRequested {
movie_id: MovieId,
external_metadata_id: String,
},
} }
#[async_trait] #[async_trait]

View File

@@ -1,4 +1,4 @@
use chrono::{NaiveDateTime, Utc}; use chrono::{DateTime, NaiveDateTime, Utc};
use crate::{ use crate::{
errors::DomainError, errors::DomainError,
@@ -490,3 +490,56 @@ mod tests {
assert_eq!(user.avatar_path(), None); assert_eq!(user.avatar_path(), None);
} }
} }
// ── Movie enrichment ───────────────────────────────────────────────────────────
#[derive(Clone, Debug)]
pub struct Genre {
pub tmdb_id: u32,
pub name: String,
}
#[derive(Clone, Debug)]
pub struct Keyword {
pub tmdb_id: u32,
pub name: String,
}
#[derive(Clone, Debug)]
pub struct CastMember {
pub tmdb_person_id: u64,
pub name: String,
pub character: String,
pub billing_order: u32,
pub profile_path: Option<String>,
}
#[derive(Clone, Debug)]
pub struct CrewMember {
pub tmdb_person_id: u64,
pub name: String,
pub job: String,
pub department: String,
pub profile_path: Option<String>,
}
#[derive(Clone, Debug)]
pub struct MovieProfile {
pub movie_id: MovieId,
pub tmdb_id: u64,
pub imdb_id: Option<String>,
pub overview: Option<String>,
pub tagline: Option<String>,
pub runtime_minutes: Option<u32>,
pub budget_usd: Option<i64>,
pub revenue_usd: Option<i64>,
pub vote_average: Option<f64>,
pub vote_count: Option<u32>,
pub original_language: Option<String>,
pub collection_name: Option<String>,
pub genres: Vec<Genre>,
pub keywords: Vec<Keyword>,
pub cast: Vec<CastMember>,
pub crew: Vec<CrewMember>,
pub enriched_at: DateTime<Utc>,
}

View File

@@ -6,8 +6,8 @@ use crate::{
events::{DomainEvent, EventEnvelope}, events::{DomainEvent, EventEnvelope},
models::{ models::{
AnnotatedRow, DiaryEntry, DiaryFilter, ExportFormat, FeedEntry, FieldMapping, AnnotatedRow, DiaryEntry, DiaryFilter, ExportFormat, FeedEntry, FieldMapping,
FileFormat, ImportError, ImportProfile, ImportSession, Movie, MovieStats, ParsedFile, FileFormat, ImportError, ImportProfile, ImportSession, Movie, MovieProfile, MovieStats,
Review, ReviewHistory, User, UserStats, UserSummary, UserTrends, ParsedFile, Review, ReviewHistory, User, UserStats, UserSummary, UserTrends,
collections::{PageParams, Paginated}, collections::{PageParams, Paginated},
}, },
value_objects::{ value_objects::{
@@ -217,6 +217,31 @@ pub trait EventHandler: Send + Sync {
async fn handle(&self, event: &DomainEvent) -> Result<(), DomainError>; async fn handle(&self, event: &DomainEvent) -> Result<(), DomainError>;
} }
#[async_trait]
pub trait PeriodicJob: Send + Sync {
fn interval(&self) -> std::time::Duration;
async fn run(&self) -> Result<(), DomainError>;
}
#[async_trait]
pub trait MovieProfileRepository: Send + Sync {
async fn upsert(&self, profile: &MovieProfile) -> Result<(), DomainError>;
async fn get_by_movie_id(&self, id: &MovieId) -> Result<Option<MovieProfile>, DomainError>;
/// Returns (movie_id, external_metadata_id) for movies with no profile or a stale one
/// (enriched_at older than 30 days).
async fn list_stale(&self) -> Result<Vec<(MovieId, String)>, DomainError>;
}
#[async_trait]
pub trait MovieEnrichmentClient: Send + Sync {
/// Resolves an external ID (TMDb or IMDb) and fetches the full movie profile.
async fn fetch_profile(
&self,
movie_id: MovieId,
external_metadata_id: &str,
) -> Result<MovieProfile, DomainError>;
}
#[async_trait] #[async_trait]
pub trait ImportSessionRepository: Send + Sync { pub trait ImportSessionRepository: Send + Sync {
async fn create(&self, session: &ImportSession) -> Result<(), DomainError>; async fn create(&self, session: &ImportSession) -> Result<(), DomainError>;

View File

@@ -358,6 +358,12 @@ mod tests {
async fn delete(&self, _: &domain::value_objects::ImportProfileId) -> Result<(), DomainError> { panic!() } async fn delete(&self, _: &domain::value_objects::ImportProfileId) -> Result<(), DomainError> { panic!() }
} }
#[async_trait::async_trait] #[async_trait::async_trait]
impl domain::ports::MovieProfileRepository for Panic {
async fn upsert(&self, _: &domain::models::MovieProfile) -> Result<(), DomainError> { panic!() }
async fn get_by_movie_id(&self, _: &domain::value_objects::MovieId) -> Result<Option<domain::models::MovieProfile>, DomainError> { Ok(None) }
async fn list_stale(&self) -> Result<Vec<(domain::value_objects::MovieId, String)>, DomainError> { Ok(vec![]) }
}
#[async_trait::async_trait]
impl domain::ports::DiaryExporter for Panic { impl domain::ports::DiaryExporter for Panic {
async fn serialize_entries( async fn serialize_entries(
&self, &self,
@@ -483,6 +489,7 @@ mod tests {
user_repository: Arc::clone(&repo) as _, user_repository: Arc::clone(&repo) as _,
import_session_repository: Arc::clone(&repo) as _, import_session_repository: Arc::clone(&repo) as _,
import_profile_repository: Arc::clone(&repo) as _, import_profile_repository: Arc::clone(&repo) as _,
movie_profile_repository: Arc::clone(&repo) as _,
auth_service, auth_service,
config: AppConfig { config: AppConfig {
allow_registration: false, allow_registration: false,

View File

@@ -36,9 +36,10 @@ use api_types::{
BlockedDomainResponse, FollowRequest, RemoteActorDto, BlockedDomainResponse, FollowRequest, RemoteActorDto,
}; };
use api_types::{ use api_types::{
ActivityFeedQueryParams, ActivityFeedResponse, DiaryEntryDto, DiaryQueryParams, DiaryResponse, ActivityFeedQueryParams, ActivityFeedResponse, CastMemberDto, CrewMemberDto, DiaryEntryDto,
DirectorStatDto, ExportQueryParams, FeedEntryDto, LogReviewRequest, LoginRequest, LoginResponse, DiaryQueryParams, DiaryResponse, DirectorStatDto, ExportQueryParams, FeedEntryDto,
MonthActivityDto, MonthlyRatingDto, MovieDetailResponse, MovieDto, MovieStatsDto, GenreDto, KeywordDto, LogReviewRequest, LoginRequest, LoginResponse, MonthActivityDto,
MonthlyRatingDto, MovieDetailResponse, MovieDto, MovieProfileResponse, MovieStatsDto,
PaginationQueryParams, ProfileResponse, RegisterRequest, ReviewDto, ReviewHistoryResponse, PaginationQueryParams, ProfileResponse, RegisterRequest, ReviewDto, ReviewHistoryResponse,
SocialFeedResponse, SocialReviewDto, UserProfileQueryParams, UserProfileResponse, UserStatsDto, SocialFeedResponse, SocialReviewDto, UserProfileQueryParams, UserProfileResponse, UserStatsDto,
UserSummaryDto, UserTrendsDto, UsersResponse, UserSummaryDto, UserTrendsDto, UsersResponse,
@@ -293,6 +294,52 @@ pub async fn get_movie_detail(
})) }))
} }
#[utoipa::path(
get, path = "/api/v1/movies/{id}/profile",
params(("id" = Uuid, Path, description = "Movie ID")),
responses(
(status = 200, body = MovieProfileResponse),
(status = 404, description = "No profile found for this movie"),
)
)]
pub async fn get_movie_profile(
State(state): State<AppState>,
Path(movie_id): Path<Uuid>,
) -> impl IntoResponse {
let id = domain::value_objects::MovieId::from_uuid(movie_id);
match state.app_ctx.movie_profile_repository.get_by_movie_id(&id).await {
Ok(Some(p)) => Json(MovieProfileResponse {
tmdb_id: p.tmdb_id,
imdb_id: p.imdb_id,
overview: p.overview,
tagline: p.tagline,
runtime_minutes: p.runtime_minutes,
budget_usd: p.budget_usd,
revenue_usd: p.revenue_usd,
vote_average: p.vote_average,
vote_count: p.vote_count,
original_language: p.original_language,
collection_name: p.collection_name,
genres: p.genres.into_iter().map(|g| GenreDto { tmdb_id: g.tmdb_id, name: g.name }).collect(),
keywords: p.keywords.into_iter().map(|k| KeywordDto { tmdb_id: k.tmdb_id, name: k.name }).collect(),
cast: p.cast.into_iter().map(|c| CastMemberDto {
tmdb_person_id: c.tmdb_person_id, name: c.name, character: c.character,
billing_order: c.billing_order, profile_path: c.profile_path,
}).collect(),
crew: p.crew.into_iter().map(|c| CrewMemberDto {
tmdb_person_id: c.tmdb_person_id, name: c.name, job: c.job,
department: c.department, profile_path: c.profile_path,
}).collect(),
enriched_at: p.enriched_at.to_rfc3339(),
}).into_response(),
Ok(None) => StatusCode::NOT_FOUND.into_response(),
Err(e) => {
tracing::error!("get_movie_profile: {:?}", e);
StatusCode::INTERNAL_SERVER_ERROR.into_response()
}
}
}
#[utoipa::path( #[utoipa::path(
get, path = "/api/v1/profile", get, path = "/api/v1/profile",
responses( responses(

View File

@@ -49,17 +49,17 @@ async fn wire_dependencies() -> anyhow::Result<(AppState, axum::Router)> {
let poster_fetcher = poster_fetcher::create()?; let poster_fetcher = poster_fetcher::create()?;
let image_storage = image_storage::create()?; let image_storage = image_storage::create()?;
let (movie_repository, review_repository, diary_repository, stats_repository, user_repository, import_session_repository, import_profile_repository, db_pool) = let (movie_repository, review_repository, diary_repository, stats_repository, user_repository, import_session_repository, import_profile_repository, movie_profile_repository, db_pool) =
match backend.as_str() { match backend.as_str() {
#[cfg(feature = "postgres")] #[cfg(feature = "postgres")]
"postgres" => { "postgres" => {
let (pool, m, r, d, s, u, is, ip) = postgres::wire(&database_url).await?; let (pool, m, r, d, s, u, is, ip, mp) = postgres::wire(&database_url).await?;
(m, r, d, s, u, is, ip, DbPool::Postgres(pool)) (m, r, d, s, u, is, ip, mp, DbPool::Postgres(pool))
} }
#[cfg(feature = "sqlite")] #[cfg(feature = "sqlite")]
_ => { _ => {
let (pool, m, r, d, s, u, is, ip) = sqlite::wire(&database_url).await?; let (pool, m, r, d, s, u, is, ip, mp) = sqlite::wire(&database_url).await?;
(m, r, d, s, u, is, ip, DbPool::Sqlite(pool)) (m, r, d, s, u, is, ip, mp, DbPool::Sqlite(pool))
} }
#[cfg(not(feature = "sqlite"))] #[cfg(not(feature = "sqlite"))]
_ => anyhow::bail!("DATABASE_BACKEND={backend} is not supported by this build (sqlite feature is not enabled)"), _ => anyhow::bail!("DATABASE_BACKEND={backend} is not supported by this build (sqlite feature is not enabled)"),
@@ -161,6 +161,7 @@ async fn wire_dependencies() -> anyhow::Result<(AppState, axum::Router)> {
user_repository, user_repository,
import_session_repository: import_session_repository as Arc<dyn ImportSessionRepository>, import_session_repository: import_session_repository as Arc<dyn ImportSessionRepository>,
import_profile_repository: import_profile_repository as Arc<dyn ImportProfileRepository>, import_profile_repository: import_profile_repository as Arc<dyn ImportProfileRepository>,
movie_profile_repository,
config: app_config, config: app_config,
}; };
@@ -185,6 +186,7 @@ enum DbPool {
Postgres(sqlx::PgPool), Postgres(sqlx::PgPool),
} }
#[derive(Clone, Copy)] #[derive(Clone, Copy)]
enum EventBusBackend { enum EventBusBackend {
Db, Db,

View File

@@ -181,6 +181,10 @@ fn api_routes(rate_limit: u64) -> Router<AppState> {
"/movies/{id}", "/movies/{id}",
routing::get(handlers::api::get_movie_detail), routing::get(handlers::api::get_movie_detail),
) )
.route(
"/movies/{id}/profile",
routing::get(handlers::api::get_movie_profile),
)
.route("/reviews", routing::post(handlers::api::post_review)) .route("/reviews", routing::post(handlers::api::post_review))
.route( .route(
"/reviews/{id}", "/reviews/{id}",

View File

@@ -147,6 +147,14 @@ impl domain::ports::DocumentParser for PanicDocumentParser {
} }
struct PanicImportProfile; struct PanicImportProfile;
struct PanicMovieProfile;
#[async_trait]
impl domain::ports::MovieProfileRepository for PanicMovieProfile {
async fn upsert(&self, _: &domain::models::MovieProfile) -> Result<(), DomainError> { panic!() }
async fn get_by_movie_id(&self, _: &domain::value_objects::MovieId) -> Result<Option<domain::models::MovieProfile>, DomainError> { Ok(None) }
async fn list_stale(&self) -> Result<Vec<(domain::value_objects::MovieId, String)>, DomainError> { Ok(vec![]) }
}
#[async_trait] #[async_trait]
impl domain::ports::ImportProfileRepository for PanicImportProfile { impl domain::ports::ImportProfileRepository for PanicImportProfile {
async fn save(&self, _: &domain::models::ImportProfile) -> Result<(), DomainError> { panic!() } async fn save(&self, _: &domain::models::ImportProfile) -> Result<(), DomainError> { panic!() }
@@ -198,6 +206,7 @@ async fn test_app() -> Router {
user_repository: Arc::new(NobodyUserRepo), user_repository: Arc::new(NobodyUserRepo),
import_session_repository: Arc::new(PanicImportSession), import_session_repository: Arc::new(PanicImportSession),
import_profile_repository: Arc::new(PanicImportProfile), import_profile_repository: Arc::new(PanicImportProfile),
movie_profile_repository: Arc::new(PanicMovieProfile),
config: AppConfig { config: AppConfig {
allow_registration: false, allow_registration: false,
base_url: "http://localhost:3000".to_string(), base_url: "http://localhost:3000".to_string(),

View File

@@ -26,6 +26,7 @@ poster-fetcher = { workspace = true }
image-storage = { workspace = true } image-storage = { workspace = true }
poster-sync = { workspace = true } poster-sync = { workspace = true }
export = { workspace = true } export = { workspace = true }
tmdb-enrichment = { workspace = true }
importer = { workspace = true } importer = { workspace = true }
nats = { workspace = true, optional = true } nats = { workspace = true, optional = true }
sqlx = { workspace = true } sqlx = { workspace = true }

View File

@@ -6,7 +6,7 @@ use export::ExportAdapter;
use importer::ImporterDocumentParser; use importer::ImporterDocumentParser;
use tracing_subscriber::{layer::SubscriberExt, util::SubscriberInitExt}; use tracing_subscriber::{layer::SubscriberExt, util::SubscriberInitExt};
use domain::ports::{DiaryExporter, DocumentParser, EventHandler}; use domain::ports::{DiaryExporter, DocumentParser, EventHandler, PeriodicJob};
#[cfg(not(any(feature = "sqlite", feature = "postgres")))] #[cfg(not(any(feature = "sqlite", feature = "postgres")))]
compile_error!("At least one database backend must be enabled. Use --features sqlite or --features postgres"); compile_error!("At least one database backend must be enabled. Use --features sqlite or --features postgres");
@@ -25,17 +25,17 @@ async fn main() -> anyhow::Result<()> {
let poster_fetcher = poster_fetcher::create()?; let poster_fetcher = poster_fetcher::create()?;
let image_storage = image_storage::create()?; let image_storage = image_storage::create()?;
let (movie_repository, review_repository, diary_repository, stats_repository, user_repository, import_session_repository, import_profile_repository, db_pool) = let (movie_repository, review_repository, diary_repository, stats_repository, user_repository, import_session_repository, import_profile_repository, movie_profile_repository, db_pool) =
match backend.as_str() { match backend.as_str() {
#[cfg(feature = "postgres")] #[cfg(feature = "postgres")]
"postgres" => { "postgres" => {
let (pool, m, r, d, s, u, is, ip) = postgres::wire(&database_url).await?; let (pool, m, r, d, s, u, is, ip, mp) = postgres::wire(&database_url).await?;
(m, r, d, s, u, is, ip, DbPool::Postgres(pool)) (m, r, d, s, u, is, ip, mp, DbPool::Postgres(pool))
} }
#[cfg(feature = "sqlite")] #[cfg(feature = "sqlite")]
_ => { _ => {
let (pool, m, r, d, s, u, is, ip) = sqlite::wire(&database_url).await?; let (pool, m, r, d, s, u, is, ip, mp) = sqlite::wire(&database_url).await?;
(m, r, d, s, u, is, ip, DbPool::Sqlite(pool)) (m, r, d, s, u, is, ip, mp, DbPool::Sqlite(pool))
} }
#[cfg(not(feature = "sqlite"))] #[cfg(not(feature = "sqlite"))]
_ => anyhow::bail!("DATABASE_BACKEND={backend} is not supported by this build"), _ => anyhow::bail!("DATABASE_BACKEND={backend} is not supported by this build"),
@@ -63,6 +63,8 @@ async fn main() -> anyhow::Result<()> {
} }
}; };
let profile_repo = movie_profile_repository;
// Clone what federation handler needs before ctx and app_config are consumed. // Clone what federation handler needs before ctx and app_config are consumed.
#[cfg(feature = "federation")] #[cfg(feature = "federation")]
let (fed_movie_repo, fed_review_repo, fed_diary_repo, fed_user_repo, base_url, allow_registration) = ( let (fed_movie_repo, fed_review_repo, fed_diary_repo, fed_user_repo, base_url, allow_registration) = (
@@ -90,19 +92,37 @@ async fn main() -> anyhow::Result<()> {
user_repository, user_repository,
import_session_repository, import_session_repository,
import_profile_repository, import_profile_repository,
movie_profile_repository: Arc::clone(&profile_repo) as _,
config: app_config, config: app_config,
}; };
// Spawn periodic import session cleanup (hourly) let enrichment_handler: Option<Arc<dyn EventHandler>> =
{ match tmdb_enrichment::TmdbEnrichmentClient::from_env() {
let cleanup_ctx = ctx.clone(); Ok(client) => {
tracing::info!("TMDb enrichment enabled");
Some(Arc::new(tmdb_enrichment::EnrichmentHandler {
enrichment_client: Arc::new(client),
profile_repo: Arc::clone(&profile_repo),
}))
}
Err(e) => {
tracing::warn!("TMDb enrichment disabled: {e}");
None
}
};
let periodic_jobs: Vec<Arc<dyn PeriodicJob>> = vec![
Arc::new(application::jobs::ImportSessionCleanupJob::new(ctx.clone())),
Arc::new(application::jobs::EnrichmentStalenessJob::new(ctx.clone())),
];
for job in periodic_jobs {
tokio::spawn(async move { tokio::spawn(async move {
let mut interval = tokio::time::interval(std::time::Duration::from_secs(3600)); let mut tick = tokio::time::interval(job.interval());
loop { loop {
interval.tick().await; tick.tick().await;
match application::use_cases::cleanup_expired_import_sessions::execute(&cleanup_ctx).await { if let Err(e) = job.run().await {
Ok(n) => tracing::info!("import session cleanup: removed {} expired sessions", n), tracing::error!("periodic job failed: {e}");
Err(e) => tracing::error!("import session cleanup failed: {:?}", e),
} }
} }
}); });
@@ -121,8 +141,14 @@ async fn main() -> anyhow::Result<()> {
Arc::clone(&ctx.image_storage), Arc::clone(&ctx.image_storage),
)) as Arc<dyn EventHandler>; )) as Arc<dyn EventHandler>;
let enrichment = enrichment_handler;
#[cfg(not(feature = "federation"))] #[cfg(not(feature = "federation"))]
{ vec![poster, cleanup] } {
let mut h: Vec<Arc<dyn EventHandler>> = vec![poster, cleanup];
if let Some(e) = enrichment { h.push(e); }
h
}
#[cfg(feature = "federation")] #[cfg(feature = "federation")]
{ {
@@ -145,7 +171,9 @@ async fn main() -> anyhow::Result<()> {
).await?.event_handler; ).await?.event_handler;
tracing::info!("federation event handler registered"); tracing::info!("federation event handler registered");
vec![poster, cleanup, ap] let mut h: Vec<Arc<dyn EventHandler>> = vec![poster, cleanup, ap];
if let Some(e) = enrichment { h.push(e); }
h
} }
}; };