Some cleanup

.env.example

@@ -11,4 +11,3 @@ YOUTUBE_API_KEY=
 # Gemini API key (optional - for AI companion and exams)
 # Get from: https://aistudio.google.com/app/apikey
 GEMINI_API_KEY=
-

.github/workflows/release.yml

@@ -44,6 +44,16 @@ jobs:
         with:
           fetch-depth: 0
 
+      - name: Free Disk Space (Linux)
+        if: matrix.platform == 'linux'
+        run: |
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf /usr/local/lib/android
+          sudo rm -rf /opt/ghc
+          sudo rm -rf "/usr/local/share/boost"
+          sudo rm -rf "$AGENT_TOOLSDIRECTORY"
+        shell: bash
+
       - name: Configure Cargo git CLI
         run: |
           mkdir -p .cargo

Cargo.toml

@@ -87,11 +87,18 @@ keyring = { version = "3", features = ["windows-native"] }
 [target.'cfg(target_os = "linux")'.dependencies]
 keyring = { version = "3", features = ["linux-native"] }
 
+[profile.dev]
+debug = 1 # Reduce debug info to save disk space and prevent Bus Error on CI
+
+[profile.test]
+debug = 1 # Reduce debug info for tests as well
+
 [profile.release]
-opt-level = "z"   # Optimize for size
-lto = true        # Enable Link Time Optimization
-codegen-units = 1 # Allow for maximum cross-module optimization
-panic = "abort"   # Remove unwinding code for smaller binaries
+opt-level = "z"     # Optimize for size
+lto = "thin"        # Use Thin LTO to reduce memory/disk pressure during linking
+codegen-units = 16  # Increase units to parallelize and reduce resource usage
+panic = "abort"     # Remove unwinding code for smaller binaries
+strip = true        # Strip symbols to further reduce binary size
 
 [patch.crates-io]
 dioxus = { git = "https://github.com/k5602/dioxus", branch = "navigation_policy_hook", package = "dioxus" }

diesel.toml

@@ -4,6 +4,14 @@
 [print_schema]
 file = "src/schema.rs"
 custom_type_derives = ["diesel::query_builder::QueryId", "Clone"]
+filter = { except_tables = [
+    "search_index",
+    "search_index_config",
+    "search_index_content",
+    "search_index_data",
+    "search_index_docsize",
+    "search_index_idx",
+] }
 
 [migrations_directory]
 dir = "migrations"

migrations/2026-02-02-000000-0000_make_youtube_id_nullable/down.sql

@@ -0,0 +1,54 @@
+-- Revert videos.youtube_id to NOT NULL by rebuilding the table (SQLite)
+PRAGMA foreign_keys=off;
+
+CREATE TABLE videos_old (
+    id TEXT PRIMARY KEY NOT NULL,
+    module_id TEXT NOT NULL REFERENCES modules(id) ON DELETE CASCADE,
+    youtube_id TEXT NOT NULL,
+    title TEXT NOT NULL,
+    duration_secs INTEGER NOT NULL,
+    is_completed BOOLEAN NOT NULL DEFAULT FALSE,
+    sort_order INTEGER NOT NULL,
+    description TEXT,
+    transcript TEXT,
+    summary TEXT,
+    source_type TEXT NOT NULL DEFAULT 'youtube',
+    source_ref TEXT NOT NULL DEFAULT ''
+);
+
+INSERT INTO videos_old (
+    id,
+    module_id,
+    youtube_id,
+    title,
+    duration_secs,
+    is_completed,
+    sort_order,
+    description,
+    transcript,
+    summary,
+    source_type,
+    source_ref
+)
+SELECT
+    id,
+    module_id,
+    COALESCE(youtube_id, ''),
+    title,
+    duration_secs,
+    is_completed,
+    sort_order,
+    description,
+    transcript,
+    summary,
+    source_type,
+    source_ref
+FROM videos;
+
+DROP TABLE videos;
+ALTER TABLE videos_old RENAME TO videos;
+
+CREATE INDEX idx_videos_module_id ON videos(module_id);
+CREATE INDEX idx_videos_youtube_id ON videos(youtube_id);
+
+PRAGMA foreign_keys=on;

migrations/2026-02-02-000000-0000_make_youtube_id_nullable/up.sql

@@ -0,0 +1,54 @@
+-- Make videos.youtube_id nullable by rebuilding the table
+PRAGMA foreign_keys=off;
+
+CREATE TABLE videos_new (
+    id TEXT PRIMARY KEY NOT NULL,
+    module_id TEXT NOT NULL REFERENCES modules(id) ON DELETE CASCADE,
+    youtube_id TEXT,
+    title TEXT NOT NULL,
+    duration_secs INTEGER NOT NULL,
+    is_completed BOOLEAN NOT NULL DEFAULT FALSE,
+    sort_order INTEGER NOT NULL,
+    description TEXT,
+    transcript TEXT,
+    summary TEXT,
+    source_type TEXT NOT NULL DEFAULT 'youtube',
+    source_ref TEXT NOT NULL DEFAULT ''
+);
+
+INSERT INTO videos_new (
+    id,
+    module_id,
+    youtube_id,
+    title,
+    duration_secs,
+    is_completed,
+    sort_order,
+    description,
+    transcript,
+    summary,
+    source_type,
+    source_ref
+)
+SELECT
+    id,
+    module_id,
+    youtube_id,
+    title,
+    duration_secs,
+    is_completed,
+    sort_order,
+    description,
+    transcript,
+    summary,
+    source_type,
+    source_ref
+FROM videos;
+
+DROP TABLE videos;
+ALTER TABLE videos_new RENAME TO videos;
+
+CREATE INDEX idx_videos_module_id ON videos(module_id);
+CREATE INDEX idx_videos_youtube_id ON videos(youtube_id);
+
+PRAGMA foreign_keys=on;

src/application/use_cases/ingest_local.rs

@@ -13,7 +13,7 @@ use crate::domain::{
         CourseRepository, LocalMediaScanner, ModuleRepository, RawLocalMediaMetadata,
         SearchRepository, VideoRepository,
     },
-    services::{BoundaryDetector, SubtitleCleaner, TitleSanitizer},
+    services::{BoundaryDetector, SubtitleCleaner, TitleSanitizer, title_number_sequence},
     value_objects::{CourseId, ModuleId, PlaylistUrl, VideoId, VideoSource},
 };
 
@@ -217,6 +217,20 @@ fn group_by_folder(
         grouped.entry(folder).or_default().push(item.clone());
     }
 
+    for entries in grouped.values_mut() {
+        entries.sort_by(|a, b| {
+            let a_key = title_number_sequence(&a.title);
+            let b_key = title_number_sequence(&b.title);
+
+            match (a_key, b_key) {
+                (Some(a_seq), Some(b_seq)) => a_seq.cmp(&b_seq).then_with(|| a.title.cmp(&b.title)),
+                (Some(_), None) => std::cmp::Ordering::Less,
+                (None, Some(_)) => std::cmp::Ordering::Greater,
+                (None, None) => a.title.cmp(&b.title),
+            }
+        });
+    }
+
     grouped
 }
 
@@ -245,7 +259,8 @@ fn split_root_group_if_needed(
     }
 
     let detector = BoundaryDetector::new();
-    let groups = detector.group_into_modules(items.len());
+    let raw_titles: Vec<&str> = items.iter().map(|item| item.title.as_str()).collect();
+    let groups = detector.group_by_titles(&raw_titles);
     if groups.len() <= 1 {
         return grouped.clone();
     }

src/application/use_cases/ingest_playlist.rs

@@ -105,9 +105,10 @@ where
         // 3. Sanitize titles
         let sanitized_titles: Vec<String> =
             raw_videos.iter().map(|v| self.sanitizer.sanitize(&v.title)).collect();
+        let raw_titles: Vec<&str> = raw_videos.iter().map(|v| v.title.as_str()).collect();
 
-        // 4. Group videos into modules (simple batch grouping)
-        let module_groups = self.boundary_detector.group_into_modules(raw_videos.len());
+        // 4. Group videos into modules (title-aware with batch fallback)
+        let module_groups = self.boundary_detector.group_by_titles(&raw_titles);
 
         // 5. Create course
         let course_name = input