优化分钟历史窗口读取并移除滚动双口径

This commit is contained in:
boris
2026-08-27 13:14:41 +08:00
parent 45cafa5c96
commit ed126a3630
2 changed files with 239 additions and 21 deletions
+60 -21
View File
@@ -1174,6 +1174,7 @@ pub struct DataSet {
candidate_symbol_ids_by_date: Arc<BTreeMap<NaiveDate, Vec<u32>>>,
corporate_actions_by_date: Arc<BTreeMap<NaiveDate, Vec<CorporateAction>>>,
execution_quotes_by_date: Arc<HashMap<NaiveDate, HashMap<String, Vec<IntradayExecutionQuote>>>>,
execution_quote_dates: Arc<Vec<NaiveDate>>,
order_book_depth_index: Arc<HashMap<(NaiveDate, String), Vec<IntradayOrderBookDepthLevel>>>,
benchmark_by_date: Arc<BTreeMap<NaiveDate, BenchmarkSnapshot>>,
market_series_by_symbol: Arc<AHashMap<String, Arc<SymbolPriceSeries>>>,
@@ -1441,6 +1442,8 @@ impl DataSet {
}
let corporate_actions_by_date = group_by_date(corporate_actions, |item| item.date);
let execution_quotes_by_date = build_execution_quote_index(execution_quotes);
let mut execution_quote_dates = execution_quotes_by_date.keys().copied().collect::<Vec<_>>();
execution_quote_dates.sort_unstable();
let order_book_depth_index = build_order_book_depth_index(order_book_depth);
let benchmark_by_date = benchmarks
@@ -1464,6 +1467,7 @@ impl DataSet {
candidate_symbol_ids_by_date: Arc::new(candidate_symbol_ids_by_date),
corporate_actions_by_date: Arc::new(corporate_actions_by_date),
execution_quotes_by_date: Arc::new(execution_quotes_by_date),
execution_quote_dates: Arc::new(execution_quote_dates),
order_book_depth_index: Arc::new(order_book_depth_index),
benchmark_by_date: Arc::new(benchmark_by_date),
market_series_by_symbol: Arc::new(market_series_by_symbol),
@@ -1648,9 +1652,14 @@ impl DataSet {
.push(quote);
}
let mut added = 0usize;
let mut new_dates = Vec::new();
let execution_quotes_by_date = Arc::make_mut(&mut self.execution_quotes_by_date);
for (date, rows_by_symbol) in grouped {
let date_is_new = !execution_quotes_by_date.contains_key(&date);
let target_by_symbol = execution_quotes_by_date.entry(date).or_default();
if date_is_new {
new_dates.push(date);
}
for (symbol, mut incoming) in rows_by_symbol {
incoming.sort_by_key(|quote| quote.timestamp);
incoming.dedup_by(|left, right| left.timestamp == right.timestamp);
@@ -1688,6 +1697,14 @@ impl DataSet {
*target = merged;
}
}
if !new_dates.is_empty() {
let dates = Arc::make_mut(&mut self.execution_quote_dates);
for date in new_dates {
if let Err(index) = dates.binary_search(&date) {
dates.insert(index, date);
}
}
}
added
}
@@ -1744,10 +1761,18 @@ impl DataSet {
}
pub fn remove_execution_quotes_on_date(&mut self, date: NaiveDate) -> usize {
Arc::make_mut(&mut self.execution_quotes_by_date)
.remove(&date)
.map(|rows_by_symbol| rows_by_symbol.into_values().map(|rows| rows.len()).sum())
.unwrap_or_default()
let removed = Arc::make_mut(&mut self.execution_quotes_by_date).remove(&date);
let Some(rows_by_symbol) = removed else {
return 0;
};
let dates = Arc::make_mut(&mut self.execution_quote_dates);
if let Ok(index) = dates.binary_search(&date) {
dates.remove(index);
}
rows_by_symbol
.into_values()
.map(|rows| rows.len())
.sum()
}
pub fn snapshot_components(&self) -> DataSetSnapshotComponents {
@@ -1901,16 +1926,29 @@ impl DataSet {
if bar_count == 0 {
return Vec::new();
}
let mut quotes = self
.execution_quotes_by_date
.values()
.filter_map(|rows_by_symbol| rows_by_symbol.get(symbol))
.flat_map(|rows| rows.iter())
.filter(|quote| intraday_quote_visible(quote, date, active_datetime, include_now))
.cloned()
.collect::<Vec<_>>();
quotes.sort_by_key(|quote| quote.timestamp);
take_last(quotes, bar_count)
let end = self
.execution_quote_dates
.partition_point(|quote_date| *quote_date <= date);
let mut quotes = Vec::with_capacity(bar_count);
'dates: for quote_date in self.execution_quote_dates[..end].iter().rev() {
let Some(rows) = self
.execution_quotes_by_date
.get(quote_date)
.and_then(|rows_by_symbol| rows_by_symbol.get(symbol))
else {
continue;
};
for quote in rows.iter().rev() {
if intraday_quote_visible(quote, date, active_datetime, include_now) {
quotes.push(quote.clone());
if quotes.len() == bar_count {
break 'dates;
}
}
}
}
quotes.reverse();
quotes
}
pub fn trading_dates(&self, start: NaiveDate, end: NaiveDate) -> Vec<NaiveDate> {
@@ -3396,13 +3434,6 @@ fn normalize_history_frequency(frequency: &str) -> Option<String> {
}
}
fn take_last<T>(mut rows: Vec<T>, count: usize) -> Vec<T> {
if rows.len() <= count {
return rows;
}
rows.split_off(rows.len() - count)
}
fn group_by_date<T, F>(rows: Vec<T>, mut date_of: F) -> BTreeMap<NaiveDate, Vec<T>>
where
F: FnMut(&T) -> NaiveDate,
@@ -3864,6 +3895,10 @@ mod tests {
&data.execution_quotes_by_date,
&run_data.execution_quotes_by_date
));
assert!(Arc::ptr_eq(
&data.execution_quote_dates,
&run_data.execution_quote_dates
));
run_data.add_execution_quotes(vec![IntradayExecutionQuote {
date,
@@ -3886,6 +3921,10 @@ mod tests {
&data.execution_quotes_by_date,
&run_data.execution_quotes_by_date
));
assert!(!Arc::ptr_eq(
&data.execution_quote_dates,
&run_data.execution_quote_dates
));
}
#[test]