Indexing perf work

.gitignore

@@ -16,6 +16,7 @@ TestConfig.toml
 **/*.sqlite-wal
 polaris.log
 polaris.pid
+profile.json
 /thumbnails
 
 # Release process artifacts (usually runs on CI)

src/app/collection/index.rs

@@ -2,20 +2,20 @@ use std::{
 	borrow::BorrowMut,
 	collections::{HashMap, HashSet},
 	hash::{DefaultHasher, Hash, Hasher},
-	sync::Arc,
+	sync::{Arc, RwLock},
 };
 
 use log::{error, info};
 use rand::{rngs::ThreadRng, seq::IteratorRandom};
 use serde::{Deserialize, Serialize};
-use tokio::sync::RwLock;
+use tokio::task::spawn_blocking;
 
 use crate::{app::collection, db::DB};
 
 #[derive(Clone)]
 pub struct IndexManager {
 	db: DB,
-	index: Arc<RwLock<Index>>,
+	index: Arc<RwLock<Index>>, // Not a tokio RwLock as we want to do CPU-bound work with Index
 }
 
 impl IndexManager {
@@ -31,8 +31,15 @@ impl IndexManager {
 	}
 
 	pub(super) async fn replace_index(&mut self, new_index: Index) {
-		let mut lock = self.index.write().await;
-		*lock = new_index;
+		spawn_blocking({
+			let index_manager = self.clone();
+			move || {
+				let mut lock = index_manager.index.write().unwrap();
+				*lock = new_index;
+			}
+		})
+		.await
+		.unwrap()
 	}
 
 	pub(super) async fn persist_index(&mut self, index: &Index) -> Result<(), collection::Error> {
@@ -66,53 +73,93 @@ impl IndexManager {
 		Ok(true)
 	}
 
+	pub(super) async fn get_songs(&self) -> Vec<collection::Song> {
+		spawn_blocking({
+			let index_manager = self.clone();
+			move || {
+				let index = index_manager.index.read().unwrap();
+				index.songs.values().cloned().collect::<Vec<_>>()
+			}
+		})
+		.await
+		.unwrap()
+	}
+
 	pub async fn get_artist(
 		&self,
 		artist_key: &ArtistKey,
 	) -> Result<collection::Artist, collection::Error> {
-		let index = self.index.read().await;
-		let artist_id = artist_key.into();
-		index
-			.get_artist(artist_id)
-			.ok_or_else(|| collection::Error::ArtistNotFound)
+		spawn_blocking({
+			let index_manager = self.clone();
+			let artist_id = artist_key.into();
+			move || {
+				let index = index_manager.index.read().unwrap();
+				index
+					.get_artist(artist_id)
+					.ok_or_else(|| collection::Error::ArtistNotFound)
+			}
+		})
+		.await
+		.unwrap()
 	}
 
 	pub async fn get_album(
 		&self,
 		album_key: &AlbumKey,
 	) -> Result<collection::Album, collection::Error> {
-		let index = self.index.read().await;
-		let album_id = album_key.into();
-		index
-			.get_album(album_id)
-			.ok_or_else(|| collection::Error::AlbumNotFound)
+		spawn_blocking({
+			let index_manager = self.clone();
+			let album_id = album_key.into();
+			move || {
+				let index = index_manager.index.read().unwrap();
+				index
+					.get_album(album_id)
+					.ok_or_else(|| collection::Error::AlbumNotFound)
+			}
+		})
+		.await
+		.unwrap()
 	}
 
 	pub async fn get_random_albums(
 		&self,
 		count: usize,
 	) -> Result<Vec<collection::Album>, collection::Error> {
-		let index = self.index.read().await;
-		Ok(index
-			.albums
-			.keys()
-			.choose_multiple(&mut ThreadRng::default(), count)
-			.into_iter()
-			.filter_map(|k| index.get_album(*k))
-			.collect())
+		spawn_blocking({
+			let index_manager = self.clone();
+			move || {
+				let index = index_manager.index.read().unwrap();
+				Ok(index
+					.albums
+					.keys()
+					.choose_multiple(&mut ThreadRng::default(), count)
+					.into_iter()
+					.filter_map(|k| index.get_album(*k))
+					.collect())
+			}
+		})
+		.await
+		.unwrap()
 	}
 
 	pub async fn get_recent_albums(
 		&self,
 		count: usize,
 	) -> Result<Vec<collection::Album>, collection::Error> {
-		let index = self.index.read().await;
-		Ok(index
-			.recent_albums
-			.iter()
-			.take(count)
-			.filter_map(|k| index.get_album(*k))
-			.collect())
+		spawn_blocking({
+			let index_manager = self.clone();
+			move || {
+				let index = index_manager.index.read().unwrap();
+				Ok(index
+					.recent_albums
+					.iter()
+					.take(count)
+					.filter_map(|k| index.get_album(*k))
+					.collect())
+			}
+		})
+		.await
+		.unwrap()
 	}
 }
 

src/app/collection/inserter.rs

@@ -64,6 +64,9 @@ where
 	}
 
 	pub async fn flush(&mut self) {
+		if self.new_entries.is_empty() {
+			return;
+		}
 		let Ok(connection) = self.db.connect().await else {
 			error!("Could not acquire connection to insert new entries in database");
 			return;

src/app/collection/scanner.rs

@@ -43,7 +43,7 @@ impl Scanner {
 		let num_threads = std::env::var_os(key)
 			.map(|v| v.to_string_lossy().to_string())
 			.and_then(|v| usize::from_str(&v).ok())
-			.unwrap_or_else(|| min(num_cpus::get(), 4));
+			.unwrap_or_else(|| min(num_cpus::get(), 8));
 		info!("Browsing collection using {} threads", num_threads);
 
 		let directories_output = self.directories_output.clone();

src/app/collection/updater.rs

@@ -78,9 +78,6 @@ impl Updater {
 		let start = Instant::now();
 		info!("Beginning collection scan");
 
-		let cleaner = Cleaner::new(self.db.clone(), self.vfs_manager.clone());
-		cleaner.clean().await?;
-
 		let album_art_pattern = self
 			.settings_manager
 			.get_index_album_art_pattern()
@@ -97,7 +94,6 @@ impl Updater {
 			album_art_pattern,
 		);
 
-		let mut song_inserter = Inserter::<Song>::new(self.db.clone());
 		let mut directory_inserter = Inserter::<Directory>::new(self.db.clone());
 
 		let directory_task = tokio::spawn(async move {
@@ -132,13 +128,11 @@ impl Updater {
 					0 => break,
 					_ => {
 						for song in buffer.drain(0..) {
-							index_builder.add_song(song.clone());
-							song_inserter.insert(song).await;
+							index_builder.add_song(song);
 						}
 					}
 				}
 			}
-			song_inserter.flush().await;
 			index_builder.build()
 		});
 
@@ -151,6 +145,33 @@ impl Updater {
 			start.elapsed().as_millis() as f32 / 1000.0
 		);
 
+		let start = Instant::now();
+		info!("Beginning collection DB update");
+
+		tokio::task::spawn({
+			let db = self.db.clone();
+			let vfs_manager = self.vfs_manager.clone();
+			let index_manager = self.index_manager.clone();
+			async move {
+				let cleaner = Cleaner::new(db.clone(), vfs_manager);
+				if let Err(e) = cleaner.clean().await {
+					error!("Error while cleaning up database: {}", e);
+				}
+
+				let mut song_inserter = Inserter::<Song>::new(db.clone());
+				for song in index_manager.get_songs().await {
+					song_inserter.insert(song).await;
+				}
+				song_inserter.flush().await;
+			}
+		})
+		.await?;
+
+		info!(
+			"Collection DB update took {} seconds",
+			start.elapsed().as_millis() as f32 / 1000.0
+		);
+
 		Ok(())
 	}
 }