feat: implement per-note search and document pipeline

- Add SourceType::Note with extract_note_document() and ParentMetadataCache
- Migration 022: composite indexes for notes queries + author_id column
- Migration 024: table rebuild adding 'note' to CHECK constraints, defense triggers
- Migration 025: backfill existing non-system notes into dirty queue
- Add lore notes CLI command with 17 filter options (author, path, resolution, etc.)
- Support table/json/jsonl/csv output formats with field selection
- Wire note dirty tracking through discussion and MR discussion ingestion
- Fix test_migration_024_preserves_existing_data off-by-one (tested wrong migration)
- Fix upsert_document_inner returning false for label/path-only changes

src/cli/commands/generate_docs.rs

@@ -39,6 +39,7 @@ pub fn run_generate_docs(
         result.seeded += seed_dirty(&conn, SourceType::Issue, project_filter)?;
         result.seeded += seed_dirty(&conn, SourceType::MergeRequest, project_filter)?;
         result.seeded += seed_dirty(&conn, SourceType::Discussion, project_filter)?;
+        result.seeded += seed_dirty_notes(&conn, project_filter)?;
     }
 
     let regen =
@@ -67,6 +68,10 @@ fn seed_dirty(
         SourceType::Issue => "issues",
         SourceType::MergeRequest => "merge_requests",
         SourceType::Discussion => "discussions",
+        SourceType::Note => {
+            // NOTE-2E will implement seed_dirty_notes separately (needs is_system filter)
+            unreachable!("Note seeding handled by seed_dirty_notes, not seed_dirty")
+        }
     };
     let type_str = source_type.as_str();
     let now = chrono::Utc::now().timestamp_millis();
@@ -125,6 +130,55 @@ fn seed_dirty(
     Ok(total_seeded)
 }
 
+fn seed_dirty_notes(conn: &Connection, project_filter: Option<&str>) -> Result<usize> {
+    let now = chrono::Utc::now().timestamp_millis();
+    let mut total_seeded: usize = 0;
+    let mut last_id: i64 = 0;
+
+    loop {
+        let inserted = if let Some(project) = project_filter {
+            let project_id = resolve_project(conn, project)?;
+
+            conn.execute(
+                "INSERT INTO dirty_sources (source_type, source_id, queued_at, attempt_count, last_attempt_at, last_error, next_attempt_at)
+                 SELECT 'note', id, ?1, 0, NULL, NULL, NULL
+                 FROM notes WHERE id > ?2 AND project_id = ?3 AND is_system = 0 ORDER BY id LIMIT ?4
+                 ON CONFLICT(source_type, source_id) DO NOTHING",
+                rusqlite::params![now, last_id, project_id, FULL_MODE_CHUNK_SIZE],
+            )?
+        } else {
+            conn.execute(
+                "INSERT INTO dirty_sources (source_type, source_id, queued_at, attempt_count, last_attempt_at, last_error, next_attempt_at)
+                 SELECT 'note', id, ?1, 0, NULL, NULL, NULL
+                 FROM notes WHERE id > ?2 AND is_system = 0 ORDER BY id LIMIT ?3
+                 ON CONFLICT(source_type, source_id) DO NOTHING",
+                rusqlite::params![now, last_id, FULL_MODE_CHUNK_SIZE],
+            )?
+        };
+
+        if inserted == 0 {
+            break;
+        }
+
+        let max_id: i64 = conn.query_row(
+            "SELECT MAX(id) FROM (SELECT id FROM notes WHERE id > ?1 AND is_system = 0 ORDER BY id LIMIT ?2)",
+            rusqlite::params![last_id, FULL_MODE_CHUNK_SIZE],
+            |row| row.get(0),
+        )?;
+
+        total_seeded += inserted;
+        last_id = max_id;
+    }
+
+    info!(
+        source_type = "note",
+        seeded = total_seeded,
+        "Seeded dirty_sources"
+    );
+
+    Ok(total_seeded)
+}
+
 pub fn print_generate_docs(result: &GenerateDocsResult) {
     let mode = if result.full_mode {
         "full"
@@ -186,3 +240,81 @@ pub fn print_generate_docs_json(result: &GenerateDocsResult, elapsed_ms: u64) {
     };
     println!("{}", serde_json::to_string(&output).unwrap());
 }
+
+#[cfg(test)]
+mod tests {
+    use std::path::Path;
+
+    use crate::core::db::{create_connection, run_migrations};
+
+    use super::*;
+
+    fn setup_db() -> Connection {
+        let conn = create_connection(Path::new(":memory:")).unwrap();
+        run_migrations(&conn).unwrap();
+        conn.execute(
+            "INSERT INTO projects (id, gitlab_project_id, path_with_namespace, web_url) VALUES (1, 100, 'group/project', 'https://gitlab.com/group/project')",
+            [],
+        ).unwrap();
+        conn.execute(
+            "INSERT INTO issues (id, gitlab_id, project_id, iid, title, state, created_at, updated_at, last_seen_at) VALUES (1, 10, 1, 1, 'Test', 'opened', 1000, 2000, 3000)",
+            [],
+        ).unwrap();
+        conn.execute(
+            "INSERT INTO discussions (id, gitlab_discussion_id, project_id, issue_id, noteable_type, last_seen_at) VALUES (1, 'disc_1', 1, 1, 'Issue', 3000)",
+            [],
+        ).unwrap();
+        conn
+    }
+
+    fn insert_note(conn: &Connection, id: i64, gitlab_id: i64, is_system: bool) {
+        conn.execute(
+            "INSERT INTO notes (id, gitlab_id, discussion_id, project_id, author_username, body, created_at, updated_at, last_seen_at, is_system) VALUES (?1, ?2, 1, 1, 'alice', 'note body', 1000, 2000, 3000, ?3)",
+            rusqlite::params![id, gitlab_id, is_system as i32],
+        ).unwrap();
+    }
+
+    #[test]
+    fn test_full_seed_includes_notes() {
+        let conn = setup_db();
+        insert_note(&conn, 1, 101, false);
+        insert_note(&conn, 2, 102, false);
+        insert_note(&conn, 3, 103, false);
+        insert_note(&conn, 4, 104, true); // system note — should be excluded
+
+        let seeded = seed_dirty_notes(&conn, None).unwrap();
+        assert_eq!(seeded, 3);
+
+        let count: i64 = conn
+            .query_row(
+                "SELECT COUNT(*) FROM dirty_sources WHERE source_type = 'note'",
+                [],
+                |row| row.get(0),
+            )
+            .unwrap();
+        assert_eq!(count, 3);
+    }
+
+    #[test]
+    fn test_note_document_count_stable_after_second_generate_docs_full() {
+        let conn = setup_db();
+        insert_note(&conn, 1, 101, false);
+        insert_note(&conn, 2, 102, false);
+
+        let first = seed_dirty_notes(&conn, None).unwrap();
+        assert_eq!(first, 2);
+
+        // Second run should be idempotent (ON CONFLICT DO NOTHING)
+        let second = seed_dirty_notes(&conn, None).unwrap();
+        assert_eq!(second, 0);
+
+        let count: i64 = conn
+            .query_row(
+                "SELECT COUNT(*) FROM dirty_sources WHERE source_type = 'note'",
+                [],
+                |row| row.get(0),
+            )
+            .unwrap();
+        assert_eq!(count, 2);
+    }
+}