First batch of smart stuff

notes-infra/Cargo.toml

@@ -4,9 +4,10 @@ version = "0.1.0"
 edition = "2024"
 
 [features]
-default = ["sqlite"]
+default = ["sqlite", "smart-features"]
 sqlite = ["sqlx/sqlite", "tower-sessions-sqlx-store/sqlite"]
 postgres = ["sqlx/postgres", "tower-sessions-sqlx-store/postgres"]
+smart-features = ["dep:fastembed", "dep:qdrant-client"]
 
 [dependencies]
 notes-domain = { path = "../notes-domain" }
@@ -19,3 +20,6 @@ tracing = "0.1"
 uuid = { version = "1.19.0", features = ["v4", "serde"] }
 tower-sessions = "0.14.0"
 tower-sessions-sqlx-store = { version = "0.15.0", default-features = false }
+fastembed = { version = "5.4", optional = true }
+qdrant-client = { version = "1.16", optional = true }
+serde_json = "1.0"

notes-infra/src/embeddings/fastembed.rs

@@ -0,0 +1,48 @@
+use async_trait::async_trait;
+use fastembed::{EmbeddingModel, InitOptions, TextEmbedding};
+use notes_domain::errors::{DomainError, DomainResult};
+use notes_domain::ports::EmbeddingGenerator;
+use std::sync::{Arc, Mutex};
+
+pub struct FastEmbedAdapter {
+    model: Arc<Mutex<TextEmbedding>>,
+}
+
+impl FastEmbedAdapter {
+    pub fn new() -> DomainResult<Self> {
+        let mut options = InitOptions::default();
+        options.model_name = EmbeddingModel::AllMiniLML6V2;
+        options.show_download_progress = false;
+
+        let model = TextEmbedding::try_new(options).map_err(|e| {
+            DomainError::InfrastructureError(format!("Failed to init fastembed: {}", e))
+        })?;
+
+        Ok(Self {
+            model: Arc::new(Mutex::new(model)),
+        })
+    }
+}
+
+#[async_trait]
+impl EmbeddingGenerator for FastEmbedAdapter {
+    async fn generate_embedding(&self, text: &str) -> DomainResult<Vec<f32>> {
+        let model = self.model.clone();
+        let text = text.to_string();
+
+        let embeddings = tokio::task::spawn_blocking(move || {
+            let mut model = model.lock().map_err(|e| format!("Lock error: {}", e))?;
+            model
+                .embed(vec![text], None)
+                .map_err(|e| format!("Embed error: {}", e))
+        })
+        .await
+        .map_err(|e| DomainError::InfrastructureError(format!("Join error: {}", e)))?
+        .map_err(|e| DomainError::InfrastructureError(e))?;
+
+        embeddings
+            .into_iter()
+            .next()
+            .ok_or_else(|| DomainError::InfrastructureError("No embedding generated".to_string()))
+    }
+}

notes-infra/src/embeddings/mod.rs

@@ -0,0 +1 @@
+pub mod fastembed;

notes-infra/src/factory.rs

@@ -11,10 +11,66 @@ pub enum FactoryError {
     Database(#[from] sqlx::Error),
     #[error("Not implemented: {0}")]
     NotImplemented(String),
+    #[error("Infrastructure error: {0}")]
+    Infrastructure(#[from] notes_domain::DomainError),
 }
 
 pub type FactoryResult<T> = Result<T, FactoryError>;
 
+#[cfg(feature = "smart-features")]
+#[derive(Debug, Clone)]
+pub enum EmbeddingProvider {
+    FastEmbed,
+    // Ollama(String), // Url
+    // OpenAI(String), // ApiKey
+}
+
+#[cfg(feature = "smart-features")]
+#[derive(Debug, Clone)]
+pub enum VectorProvider {
+    Qdrant { url: String, collection: String },
+    // InMemory,
+}
+
+#[cfg(feature = "smart-features")]
+pub async fn build_embedding_generator(
+    provider: &EmbeddingProvider,
+) -> FactoryResult<Arc<dyn notes_domain::ports::EmbeddingGenerator>> {
+    match provider {
+        EmbeddingProvider::FastEmbed => {
+            let adapter = crate::embeddings::fastembed::FastEmbedAdapter::new()?;
+            Ok(Arc::new(adapter))
+        }
+    }
+}
+
+#[cfg(feature = "smart-features")]
+pub async fn build_vector_store(
+    provider: &VectorProvider,
+) -> FactoryResult<Arc<dyn notes_domain::ports::VectorStore>> {
+    match provider {
+        VectorProvider::Qdrant { url, collection } => {
+            let adapter = crate::vector::qdrant::QdrantVectorAdapter::new(url, collection)?;
+            adapter.create_collection_if_not_exists().await?;
+            Ok(Arc::new(adapter))
+        }
+    }
+}
+
+#[cfg(feature = "sqlite")]
+pub async fn build_link_repository(
+    pool: &DatabasePool,
+) -> FactoryResult<Arc<dyn notes_domain::ports::LinkRepository>> {
+    match pool {
+        DatabasePool::Sqlite(pool) => Ok(Arc::new(
+            crate::link_repository::SqliteLinkRepository::new(pool.clone()),
+        )),
+        _ => Err(FactoryError::NotImplemented(
+            "LinkRepostiory for non-sqlite".to_string(),
+        )),
+    }
+}
+
 pub async fn build_database_pool(db_config: &DatabaseConfig) -> FactoryResult<DatabasePool> {
     if db_config.url.starts_with("sqlite:") {
         #[cfg(feature = "sqlite")]

notes-infra/src/lib.rs

@@ -15,20 +15,28 @@
 //! - [`db::run_migrations`] - Run database migrations
 
 pub mod db;
+#[cfg(feature = "smart-features")]
+pub mod embeddings;
 pub mod factory;
 #[cfg(feature = "sqlite")]
+pub mod link_repository;
+#[cfg(feature = "sqlite")]
 pub mod note_repository;
 pub mod session_store;
 #[cfg(feature = "sqlite")]
 pub mod tag_repository;
 #[cfg(feature = "sqlite")]
 pub mod user_repository;
+#[cfg(feature = "smart-features")]
+pub mod vector;
 
 // Re-export for convenience
 #[cfg(feature = "sqlite")]
 pub use db::create_pool;
 pub use db::{DatabaseConfig, run_migrations};
 #[cfg(feature = "sqlite")]
+pub use link_repository::SqliteLinkRepository;
+#[cfg(feature = "sqlite")]
 pub use note_repository::SqliteNoteRepository;
 #[cfg(feature = "sqlite")]
 pub use tag_repository::SqliteTagRepository;

notes-infra/src/link_repository.rs

@@ -0,0 +1,111 @@
+use async_trait::async_trait;
+use sqlx::SqlitePool;
+use uuid::Uuid;
+
+use notes_domain::entities::NoteLink;
+use notes_domain::errors::{DomainError, DomainResult};
+use notes_domain::ports::LinkRepository;
+
+pub struct SqliteLinkRepository {
+    pool: SqlitePool,
+}
+
+impl SqliteLinkRepository {
+    pub fn new(pool: SqlitePool) -> Self {
+        Self { pool }
+    }
+}
+
+#[async_trait]
+impl LinkRepository for SqliteLinkRepository {
+    async fn save_links(&self, links: &[NoteLink]) -> DomainResult<()> {
+        let mut tx = self
+            .pool
+            .begin()
+            .await
+            .map_err(|e| DomainError::RepositoryError(e.to_string()))?;
+
+        for link in links {
+            let source = link.source_note_id.to_string();
+            let target = link.target_note_id.to_string();
+            let created_at = link.created_at.to_rfc3339();
+
+            sqlx::query(
+                r#"
+                INSERT INTO note_links (source_note_id, target_note_id, score, created_at)
+                VALUES (?, ?, ?, ?)
+                ON CONFLICT(source_note_id, target_note_id) DO UPDATE SET
+                    score = excluded.score,
+                    created_at = excluded.created_at
+                "#,
+            )
+            .bind(source)
+            .bind(target)
+            .bind(link.score)
+            .bind(created_at)
+            .execute(&mut *tx)
+            .await
+            .map_err(|e| DomainError::RepositoryError(e.to_string()))?;
+        }
+
+        tx.commit()
+            .await
+            .map_err(|e| DomainError::RepositoryError(e.to_string()))?;
+
+        Ok(())
+    }
+
+    async fn delete_links_for_source(&self, source_note_id: Uuid) -> DomainResult<()> {
+        let source_str = source_note_id.to_string();
+        sqlx::query("DELETE FROM note_links WHERE source_note_id = ?")
+            .bind(source_str)
+            .execute(&self.pool)
+            .await
+            .map_err(|e| DomainError::RepositoryError(e.to_string()))?;
+
+        Ok(())
+    }
+
+    async fn get_links_for_note(&self, source_note_id: Uuid) -> DomainResult<Vec<NoteLink>> {
+        let source_str = source_note_id.to_string();
+
+        // We select links where the note is the source
+        // TODO: Should we also include links where the note is the target?
+        // For now, let's stick to outgoing links as defined by the service logic.
+        // Actually, semantic similarity is symmetric, but we only save (A -> B) if we process A.
+        // Ideally we should look for both directions or enforce symmetry.
+        // Given current implementation saves A->B when A is processed, if B is processed it saves B->A.
+        // So just querying source_note_id is fine if we assume all notes are processed.
+
+        let links = sqlx::query_as::<_, SqliteNoteLink>(
+            "SELECT * FROM note_links WHERE source_note_id = ? ORDER BY score DESC",
+        )
+        .bind(source_str)
+        .fetch_all(&self.pool)
+        .await
+        .map_err(|e| DomainError::RepositoryError(e.to_string()))?;
+
+        Ok(links.into_iter().map(NoteLink::from).collect())
+    }
+}
+
+#[derive(sqlx::FromRow)]
+struct SqliteNoteLink {
+    source_note_id: String,
+    target_note_id: String,
+    score: f32,
+    created_at: String, // Stored as ISO string
+}
+
+impl From<SqliteNoteLink> for NoteLink {
+    fn from(row: SqliteNoteLink) -> Self {
+        Self {
+            source_note_id: Uuid::parse_str(&row.source_note_id).unwrap_or_default(),
+            target_note_id: Uuid::parse_str(&row.target_note_id).unwrap_or_default(),
+            score: row.score,
+            created_at: chrono::DateTime::parse_from_rfc3339(&row.created_at)
+                .unwrap_or_default()
+                .with_timezone(&chrono::Utc),
+        }
+    }
+}

notes-infra/src/vector/mod.rs

@@ -0,0 +1 @@
+pub mod qdrant;

notes-infra/src/vector/qdrant.rs

@@ -0,0 +1,101 @@
+use async_trait::async_trait;
+use notes_domain::errors::{DomainError, DomainResult};
+use notes_domain::ports::VectorStore;
+use qdrant_client::Qdrant;
+use qdrant_client::qdrant::{
+    CreateCollectionBuilder, Distance, PointStruct, SearchPointsBuilder, UpsertPointsBuilder,
+    Value, VectorParamsBuilder,
+};
+use std::collections::HashMap;
+use std::sync::Arc;
+use uuid::Uuid;
+
+pub struct QdrantVectorAdapter {
+    client: Arc<Qdrant>,
+    collection_name: String,
+}
+
+impl QdrantVectorAdapter {
+    pub fn new(url: &str, collection_name: &str) -> DomainResult<Self> {
+        let client = Qdrant::from_url(url).build().map_err(|e| {
+            DomainError::InfrastructureError(format!("Failed to create Qdrant client: {}", e))
+        })?;
+
+        Ok(Self {
+            client: Arc::new(client),
+            collection_name: collection_name.to_string(),
+        })
+    }
+
+    pub async fn create_collection_if_not_exists(&self) -> DomainResult<()> {
+        if !self
+            .client
+            .collection_exists(&self.collection_name)
+            .await
+            .map_err(|e| {
+                DomainError::InfrastructureError(format!(
+                    "Failed to check collection existence: {}",
+                    e
+                ))
+            })?
+        {
+            self.client
+                .create_collection(
+                    CreateCollectionBuilder::new(self.collection_name.clone())
+                        .vectors_config(VectorParamsBuilder::new(384, Distance::Cosine)),
+                )
+                .await
+                .map_err(|e| {
+                    DomainError::InfrastructureError(format!("Failed to create collection: {}", e))
+                })?;
+        }
+        Ok(())
+    }
+}
+
+#[async_trait]
+impl VectorStore for QdrantVectorAdapter {
+    async fn upsert(&self, id: Uuid, vector: &[f32]) -> DomainResult<()> {
+        let payload: HashMap<String, Value> = HashMap::new();
+
+        let point = PointStruct::new(id.to_string(), vector.to_vec(), payload);
+
+        let upsert_points = UpsertPointsBuilder::new(self.collection_name.clone(), vec![point]);
+
+        self.client
+            .upsert_points(upsert_points)
+            .await
+            .map_err(|e| DomainError::InfrastructureError(format!("Qdrant upsert error: {}", e)))?;
+
+        Ok(())
+    }
+
+    async fn find_similar(&self, vector: &[f32], limit: usize) -> DomainResult<Vec<(Uuid, f32)>> {
+        let search_points =
+            SearchPointsBuilder::new(self.collection_name.clone(), vector.to_vec(), limit as u64)
+                .with_payload(true);
+
+        let search_result = self
+            .client
+            .search_points(search_points)
+            .await
+            .map_err(|e| DomainError::InfrastructureError(format!("Qdrant search error: {}", e)))?;
+
+        let results = search_result
+            .result
+            .into_iter()
+            .filter_map(|point| {
+                let id = point.id?;
+                let uuid_str = match id.point_id_options? {
+                    qdrant_client::qdrant::point_id::PointIdOptions::Uuid(u) => u,
+                    _ => return None,
+                };
+
+                let uuid = Uuid::parse_str(&uuid_str).ok()?;
+                Some((uuid, point.score))
+            })
+            .collect();
+
+        Ok(results)
+    }
+}