perf: intern stock-state symbols

crates/fidc-core/src/data.rs

@@ -1327,6 +1327,7 @@ pub struct DataSet {
     market_series_end_positions_by_symbol_id: Arc<Option<SymbolSeriesEndPositionIndex>>,
     benchmark_series_cache: Arc<BenchmarkPriceSeries>,
     symbol_id_by_code: Arc<AHashMap<String, u32>>,
+    symbol_by_id: Arc<Vec<Arc<str>>>,
     eligible_universe_by_date: Arc<OnceLock<BTreeMap<NaiveDate, Vec<EligibleUniverseSnapshot>>>>,
     benchmark_code: String,
     futures_params_by_symbol: Arc<HashMap<String, Vec<FuturesTradingParameter>>>,
@@ -1771,6 +1772,10 @@ impl DataSet {
             &factor_by_date,
             &candidate_by_date,
         );
+        let mut symbol_by_id = vec![Arc::<str>::from(""); symbol_id_by_code.len()];
+        for (symbol, symbol_id) in &symbol_id_by_code {
+            symbol_by_id[*symbol_id as usize] = Arc::<str>::from(symbol.as_str());
+        }
         let mut instruments_by_symbol_id = vec![None; symbol_id_by_code.len()];
         for (symbol, instrument) in &instruments {
             if let Some(symbol_id) = symbol_id_by_code.get(symbol).copied() {
@@ -1859,6 +1864,7 @@ impl DataSet {
             ),
             benchmark_series_cache: Arc::new(benchmark_series_cache),
             symbol_id_by_code: Arc::new(symbol_id_by_code),
+            symbol_by_id: Arc::new(symbol_by_id),
             eligible_universe_by_date: Arc::new(OnceLock::new()),
             benchmark_code,
             futures_params_by_symbol: Arc::new(futures_params_by_symbol),
@@ -1912,6 +1918,11 @@ impl DataSet {
         self.symbol_id_by_code.get(symbol).copied()
     }
 
+    pub(crate) fn shared_symbol_by_id(&self, symbol_id: u32) -> Option<Arc<str>> {
+        let symbol = self.symbol_by_id.get(symbol_id as usize)?;
+        (!symbol.is_empty()).then(|| Arc::clone(symbol))
+    }
+
     pub(crate) fn symbol_count(&self) -> usize {
         self.symbol_id_by_code.len()
     }
@@ -5047,6 +5058,10 @@ mod tests {
                     .map(|row| row.symbol.as_str()),
                 Some(symbol)
             );
+            let first_shared = data.shared_symbol_by_id(symbol_id).expect("shared symbol");
+            let second_shared = data.shared_symbol_by_id(symbol_id).expect("shared symbol");
+            assert_eq!(first_shared.as_ref(), symbol);
+            assert!(Arc::ptr_eq(&first_shared, &second_shared));
             assert_eq!(
                 data.market_by_symbol_id(date, symbol_id)
                     .map(|row| row.symbol.as_str()),