use crate::binding::{map_memory_bytes, map_payload_bytes, Binding};
use crate::pipeline::runtime_checkpoint;
use crate::{
QueryMemoryAccount, QueryMemoryClass, QueryMemoryLease, QueryMemoryLedger, QueryRows,
QueryRowsBuilder, Row,
};
use hawdb_core::{HawDBError, ReadBudgetResource, Result, RuntimeTaskContext};
use std::num::NonZeroUsize;
#[derive(Clone, Copy)]
pub enum ConsumerMemoryMode {
Retained,
ReleasedAfterCall,
DeferredUntilValidated,
}
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum StreamDelivery {
Validated,
Incremental,
}
impl StreamDelivery {
pub fn consumer_memory_mode(self, bounded: bool) -> ConsumerMemoryMode {
match (self, bounded) {
(Self::Validated, true) => ConsumerMemoryMode::DeferredUntilValidated,
(Self::Validated, false) | (Self::Incremental, _) => {
ConsumerMemoryMode::ReleasedAfterCall
}
}
}
}
#[derive(Clone, Copy, Default)]
pub struct OutputLimits {
pub max_rows: Option<usize>,
pub max_payload_bytes: Option<usize>,
}
#[derive(Clone, Copy, Default)]
pub struct OutputMetrics {
pub rows: usize,
pub payload_bytes: usize,
}
pub struct QueryOutputAccumulator<'a> {
consumer: Option<&'a mut dyn FnMut(Row) -> Result<()>>,
owned_rows: Option<QueryRowsBuilder>,
memory_mode: ConsumerMemoryMode,
max_rows: Option<usize>,
max_payload_bytes: Option<usize>,
result_account: QueryMemoryAccount,
retained_result_lease: QueryMemoryLease,
deferred_rows: Vec<Row>,
metrics: OutputMetrics,
}
impl<'a> QueryOutputAccumulator<'a> {
pub fn new(
limits: OutputLimits,
result_memory_budget: NonZeroUsize,
memory_ledger: &QueryMemoryLedger,
memory_mode: ConsumerMemoryMode,
consumer: &'a mut dyn FnMut(Row) -> Result<()>,
) -> Result<Self> {
Self::new_with_budget_option(
limits,
result_memory_budget,
memory_ledger,
"QueryMemoryLedger::account(budget_bytes)",
memory_mode,
consumer,
)
}
pub fn new_with_budget_option(
limits: OutputLimits,
result_memory_budget: NonZeroUsize,
memory_ledger: &QueryMemoryLedger,
budget_option: &'static str,
memory_mode: ConsumerMemoryMode,
consumer: &'a mut dyn FnMut(Row) -> Result<()>,
) -> Result<Self> {
let result_account = memory_ledger.account_with_budget_option(
QueryMemoryClass::ResultMaterialization,
"query result",
result_memory_budget,
budget_option,
);
let retained_result_lease = result_account.reserve(0)?;
Ok(Self {
consumer: Some(consumer),
owned_rows: None,
memory_mode,
max_rows: limits.max_rows,
max_payload_bytes: limits.max_payload_bytes,
result_account,
retained_result_lease,
deferred_rows: Vec::new(),
metrics: OutputMetrics::default(),
})
}
pub fn new_owned(
limits: OutputLimits,
result_memory_budget: NonZeroUsize,
memory_ledger: &QueryMemoryLedger,
) -> Result<Self> {
Self::new_owned_with_budget_option(
limits,
result_memory_budget,
memory_ledger,
"QueryMemoryLedger::account(budget_bytes)",
)
}
pub fn new_owned_with_budget_option(
limits: OutputLimits,
result_memory_budget: NonZeroUsize,
memory_ledger: &QueryMemoryLedger,
budget_option: &'static str,
) -> Result<Self> {
let result_account = memory_ledger.account_with_budget_option(
QueryMemoryClass::ResultMaterialization,
"query result",
result_memory_budget,
budget_option,
);
let retained_result_lease =
result_account.reserve(QueryRowsBuilder::owner_memory_bytes())?;
Ok(Self {
consumer: None,
owned_rows: Some(QueryRowsBuilder::new()),
memory_mode: ConsumerMemoryMode::Retained,
max_rows: limits.max_rows,
max_payload_bytes: limits.max_payload_bytes,
result_account,
retained_result_lease,
deferred_rows: Vec::new(),
metrics: OutputMetrics::default(),
})
}
pub fn take_owned_rows(&mut self) -> Option<QueryRows> {
self.owned_rows.take().map(QueryRowsBuilder::finish)
}
pub fn emit(&mut self, binding: Binding) -> Result<()> {
if let Some(max_rows) = self.max_rows
&& self.metrics.rows >= max_rows
{
return Err(HawDBError::read_budget_exceeded(ReadBudgetResource::Rows, max_rows, format!(
"read query returned more than {max_rows} rows, exceeding max_read_result_rows {max_rows}"
)));
}
let row = binding.values;
let row_payload_bytes = map_payload_bytes(&row);
let next_payload_bytes = self.metrics.payload_bytes.saturating_add(row_payload_bytes);
if let Some(max_payload_bytes) = self.max_payload_bytes
&& next_payload_bytes > max_payload_bytes
{
return Err(HawDBError::read_budget_exceeded(ReadBudgetResource::PayloadBytes, max_payload_bytes, format!(
"read query payload would exceed max_payload_bytes {max_payload_bytes} (max_read_result_payload_bytes {max_payload_bytes}; next total {next_payload_bytes})"
)));
}
if let Some(rows) = &mut self.owned_rows {
rows.push_budgeted_named_row(
row,
&self.result_account,
&mut self.retained_result_lease,
)?;
self.metrics.rows = self.metrics.rows.saturating_add(1);
self.metrics.payload_bytes = next_payload_bytes;
return Ok(());
}
let row_memory_bytes = map_memory_bytes(&row);
let transient_result_lease = match self.memory_mode {
ConsumerMemoryMode::Retained | ConsumerMemoryMode::DeferredUntilValidated => {
self.retained_result_lease.grow(row_memory_bytes)?;
None
}
ConsumerMemoryMode::ReleasedAfterCall => {
Some(self.result_account.reserve(row_memory_bytes)?)
}
};
match self.memory_mode {
ConsumerMemoryMode::Retained | ConsumerMemoryMode::ReleasedAfterCall => {
(self.consumer.as_mut().expect("row consumer is present"))(row)?;
}
ConsumerMemoryMode::DeferredUntilValidated => self.deferred_rows.push(row),
}
drop(transient_result_lease);
self.metrics.rows = self.metrics.rows.saturating_add(1);
self.metrics.payload_bytes = next_payload_bytes;
Ok(())
}
pub fn metrics(&self) -> OutputMetrics {
self.metrics
}
pub fn finish_delivery(&mut self, task_context: Option<&RuntimeTaskContext>) -> Result<()> {
if !matches!(self.memory_mode, ConsumerMemoryMode::DeferredUntilValidated) {
return Ok(());
}
for row in self.deferred_rows.drain(..) {
runtime_checkpoint(task_context)?;
(self.consumer.as_mut().expect("row consumer is present"))(row)?;
runtime_checkpoint(task_context)?;
}
self.retained_result_lease.reset();
Ok(())
}
}
#[cfg(test)]
mod tests;