优化策略选股的索引查找路径

This commit is contained in:
boris
2026-08-29 05:44:06 +08:00
parent ff2844884d
commit 257088d7d9
2 changed files with 21 additions and 96 deletions
+13 -88
View File
@@ -1335,13 +1335,6 @@ pub struct DataSet {
futures_params_by_symbol: Arc<HashMap<String, Vec<FuturesTradingParameter>>>, futures_params_by_symbol: Arc<HashMap<String, Vec<FuturesTradingParameter>>>,
} }
#[derive(Debug, Clone, Copy)]
pub(crate) struct SymbolSnapshotRefs<'a> {
pub market: Option<&'a DailyMarketSnapshot>,
pub factor: Option<&'a DailyFactorSnapshot>,
pub candidate: Option<&'a CandidateEligibility>,
}
#[derive(Debug, Clone, Copy)] #[derive(Debug, Clone, Copy)]
pub(crate) struct StandardRollingMeans { pub(crate) struct StandardRollingMeans {
pub close: [Option<f64>; 7], pub close: [Option<f64>; 7],
@@ -1963,68 +1956,6 @@ impl DataSet {
) )
} }
pub(crate) fn symbol_snapshots_by_id(
&self,
date: NaiveDate,
symbol_id: u32,
) -> SymbolSnapshotRefs<'_> {
let market_rows = self.market_by_date.get(&date).map(Vec::as_slice);
let market_symbol_ids = self
.market_symbol_ids_by_date
.get(&date)
.map(Vec::as_slice);
let market_index = dense_row_position(
&self.market_row_positions_by_date,
date,
symbol_id,
)
.or_else(|| {
market_rows
.zip(market_symbol_ids)
.and_then(|(rows, symbol_ids)| symbol_id_index(rows.len(), symbol_ids, symbol_id))
});
let market = market_index.and_then(|index| market_rows?.get(index));
let factor = self.factor_by_date.get(&date).and_then(|rows| {
dense_row_position(&self.factor_row_positions_by_date, date, symbol_id)
.and_then(|index| rows.get(index))
.or_else(|| {
self.factor_symbol_ids_by_date
.get(&date)
.and_then(|symbol_ids| {
find_by_symbol_id_with_preferred_index(
rows,
symbol_ids,
symbol_id,
market_index,
)
})
})
});
let candidate = self.candidate_by_date.get(&date).and_then(|rows| {
dense_row_position(&self.candidate_row_positions_by_date, date, symbol_id)
.and_then(|index| rows.get(index))
.or_else(|| {
self.candidate_symbol_ids_by_date
.get(&date)
.and_then(|symbol_ids| {
find_by_symbol_id_with_preferred_index(
rows,
symbol_ids,
symbol_id,
market_index,
)
})
})
});
SymbolSnapshotRefs {
market,
factor,
candidate,
}
}
pub(crate) fn market_standard_rolling_means_by_symbol_id( pub(crate) fn market_standard_rolling_means_by_symbol_id(
&self, &self,
date: NaiveDate, date: NaiveDate,
@@ -4155,13 +4086,6 @@ fn find_by_symbol_id<'a, T>(rows: &'a [T], symbol_ids: &[u32], symbol_id: u32) -
find_by_symbol_id_with_preferred_index(rows, symbol_ids, symbol_id, None) find_by_symbol_id_with_preferred_index(rows, symbol_ids, symbol_id, None)
} }
fn symbol_id_index(rows_len: usize, symbol_ids: &[u32], symbol_id: u32) -> Option<usize> {
if rows_len != symbol_ids.len() {
return None;
}
symbol_ids.binary_search(&symbol_id).ok()
}
fn find_by_symbol_id_with_preferred_index<'a, T>( fn find_by_symbol_id_with_preferred_index<'a, T>(
rows: &'a [T], rows: &'a [T],
symbol_ids: &[u32], symbol_ids: &[u32],
@@ -4786,7 +4710,7 @@ mod tests {
} }
#[test] #[test]
fn combined_symbol_snapshot_lookup_uses_alignment_and_falls_back_for_sparse_rows() { fn direct_symbol_id_snapshot_lookups_preserve_alignment_for_sparse_rows() {
let date = NaiveDate::parse_from_str("2025-01-02", "%Y-%m-%d").unwrap(); let date = NaiveDate::parse_from_str("2025-01-02", "%Y-%m-%d").unwrap();
let instrument = |symbol: &str| Instrument { let instrument = |symbol: &str| Instrument {
symbol: symbol.to_string(), symbol: symbol.to_string(),
@@ -4848,29 +4772,30 @@ mod tests {
for symbol in ["000001.SZ", "600000.SH"] { for symbol in ["000001.SZ", "600000.SH"] {
let symbol_id = data.symbol_id(symbol).unwrap(); let symbol_id = data.symbol_id(symbol).unwrap();
let combined = data.symbol_snapshots_by_id(date, symbol_id);
assert_eq!( assert_eq!(
combined.market.map(|row| row.symbol.as_str()),
data.market_by_symbol_id(date, symbol_id) data.market_by_symbol_id(date, symbol_id)
.map(|row| row.symbol.as_str()) .map(|row| row.symbol.as_str()),
Some(symbol)
); );
assert_eq!( assert_eq!(
combined.factor.map(|row| row.symbol.as_str()),
data.factor_by_symbol_id(date, symbol_id) data.factor_by_symbol_id(date, symbol_id)
.map(|row| row.symbol.as_str()) .map(|row| row.symbol.as_str()),
Some(symbol)
); );
assert_eq!( assert_eq!(
combined.candidate.map(|row| row.symbol.as_str()),
data.candidate_by_symbol_id(date, symbol_id) data.candidate_by_symbol_id(date, symbol_id)
.map(|row| row.symbol.as_str()) .map(|row| row.symbol.as_str()),
Some(symbol)
); );
} }
let signal_id = data.symbol_id("000300.SH").unwrap(); let signal_id = data.symbol_id("000300.SH").unwrap();
let signal = data.symbol_snapshots_by_id(date, signal_id); assert_eq!(
assert_eq!(signal.market.map(|row| row.symbol.as_str()), Some("000300.SH")); data.market_by_symbol_id(date, signal_id).map(|row| row.symbol.as_str()),
assert!(signal.factor.is_none()); Some("000300.SH")
assert!(signal.candidate.is_none()); );
assert!(data.factor_by_symbol_id(date, signal_id).is_none());
assert!(data.candidate_by_symbol_id(date, signal_id).is_none());
} }
#[test] #[test]
@@ -3900,28 +3900,28 @@ impl PlatformExprStrategy {
return Ok(Arc::clone(state)); return Ok(Arc::clone(state));
} }
let execution_snapshots = ctx.data.symbol_snapshots_by_id(date, symbol_id); let market = ctx.data.market_by_symbol_id(date, symbol_id).ok_or_else(|| {
let market = execution_snapshots.market.ok_or_else(|| {
BacktestError::Data(crate::data::DataSetError::MissingSnapshot { BacktestError::Data(crate::data::DataSetError::MissingSnapshot {
kind: "market", kind: "market",
date, date,
symbol: symbol.to_string(), symbol: symbol.to_string(),
}) })
})?; })?;
let candidate = execution_snapshots.candidate.ok_or_else(|| { let candidate = ctx.data.candidate_by_symbol_id(date, symbol_id).ok_or_else(|| {
BacktestError::Data(crate::data::DataSetError::MissingSnapshot { BacktestError::Data(crate::data::DataSetError::MissingSnapshot {
kind: "candidate", kind: "candidate",
date, date,
symbol: symbol.to_string(), symbol: symbol.to_string(),
}) })
})?; })?;
let factor_snapshots = if factor_date == date { let feature_market = if factor_date == date {
execution_snapshots market
} else { } else {
ctx.data.symbol_snapshots_by_id(factor_date, symbol_id) ctx.data
.market_by_symbol_id(factor_date, symbol_id)
.unwrap_or(market)
}; };
let feature_market = factor_snapshots.market.unwrap_or(market); let factor = ctx.data.factor_by_symbol_id(factor_date, symbol_id).ok_or_else(|| {
let factor = factor_snapshots.factor.ok_or_else(|| {
BacktestError::Data(crate::data::DataSetError::MissingSnapshot { BacktestError::Data(crate::data::DataSetError::MissingSnapshot {
kind: "factor", kind: "factor",
date: factor_date, date: factor_date,