Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

feat: add ColumnarValue operations #331

Merged
merged 2 commits into from
Nov 4, 2024
Merged

feat: add ColumnarValue operations #331

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
d32a469
feat: add `ColumnarValue` operations
iajoiner Oct 31, 2024
9b7b7d5
fix: add panic doc to `scale_and_subtract_literal`
iajoiner Nov 4, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
120 changes: 115 additions & 5 deletions crates/proof-of-sql/src/sql/proof_exprs/comparison_util.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,15 +1,15 @@
use crate::{
base::{
database::Column,
database::{Column, ColumnarValue, LiteralValue},
if_rayon,
math::decimal::{DecimalError, Precision},
scalar::Scalar,
scalar::{Scalar, ScalarExt},
},
sql::parse::{type_check_binary_operation, ConversionError, ConversionResult},
};
use alloc::string::ToString;
use bumpalo::Bump;
use core::cmp;
use core::cmp::{max, Ordering};
use proof_of_sql_parser::intermediate_ast::BinaryOperator;
#[cfg(feature = "rayon")]
use rayon::iter::{IndexedParallelIterator, IntoParallelRefMutIterator, ParallelIterator};
Expand All @@ -31,6 +31,70 @@ fn unchecked_subtract_impl<'a, S: Scalar>(
Ok(result)
}

#[allow(
clippy::missing_panics_doc,
reason = "precision and scale are validated prior to calling this function, ensuring no panic occurs"
)]
/// Scale LHS and RHS to the same scale if at least one of them is decimal
/// and take the difference. This function is used for comparisons.
iajoiner marked this conversation as resolved.
Show resolved Hide resolved
#[allow(clippy::cast_sign_loss)]
pub fn scale_and_subtract_literal<S: Scalar>(
lhs: &LiteralValue<S>,
rhs: &LiteralValue<S>,
lhs_scale: i8,
rhs_scale: i8,
is_equal: bool,
) -> ConversionResult<S> {
let lhs_type = lhs.column_type();
let rhs_type = rhs.column_type();
let operator = if is_equal {
BinaryOperator::Equal
} else {
BinaryOperator::LessThanOrEqual
};
if !type_check_binary_operation(&lhs_type, &rhs_type, operator) {
return Err(ConversionError::DataTypeMismatch {
left_type: lhs_type.to_string(),
right_type: rhs_type.to_string(),
});
}
let max_scale = max(lhs_scale, rhs_scale);
let lhs_upscale = max_scale - lhs_scale;
let rhs_upscale = max_scale - rhs_scale;
// Only check precision overflow issues if at least one side is decimal
if max_scale != 0 {
let lhs_precision_value = lhs_type
.precision_value()
.expect("If scale is set, precision must be set");
let rhs_precision_value = rhs_type
.precision_value()
.expect("If scale is set, precision must be set");
let max_precision_value = max(
lhs_precision_value + (max_scale - lhs_scale) as u8,
rhs_precision_value + (max_scale - rhs_scale) as u8,
);
// Check if the precision is valid
let _max_precision = Precision::new(max_precision_value).map_err(|_| {
ConversionError::DecimalConversionError {
source: DecimalError::InvalidPrecision {
error: max_precision_value.to_string(),
},
}
})?;
}
match lhs_scale.cmp(&rhs_scale) {
Ordering::Less => {
let upscale_factor = S::pow10(rhs_upscale as u8);
Ok(lhs.to_scalar() * upscale_factor - rhs.to_scalar())
}
Ordering::Equal => Ok(lhs.to_scalar() - rhs.to_scalar()),
Ordering::Greater => {
let upscale_factor = S::pow10(lhs_upscale as u8);
Ok(lhs.to_scalar() - rhs.to_scalar() * upscale_factor)
}
}
}

#[allow(
clippy::missing_panics_doc,
reason = "precision and scale are validated prior to calling this function, ensuring no panic occurs"
Expand Down Expand Up @@ -67,7 +131,7 @@ pub(crate) fn scale_and_subtract<'a, S: Scalar>(
right_type: rhs_type.to_string(),
});
}
let max_scale = cmp::max(lhs_scale, rhs_scale);
let max_scale = max(lhs_scale, rhs_scale);
let lhs_upscale = max_scale - lhs_scale;
let rhs_upscale = max_scale - rhs_scale;
// Only check precision overflow issues if at least one side is decimal
Expand All @@ -78,7 +142,7 @@ pub(crate) fn scale_and_subtract<'a, S: Scalar>(
let rhs_precision_value = rhs_type
.precision_value()
.expect("If scale is set, precision must be set");
let max_precision_value = cmp::max(
let max_precision_value = max(
lhs_precision_value + (max_scale - lhs_scale) as u8,
rhs_precision_value + (max_scale - rhs_scale) as u8,
);
Expand All @@ -98,3 +162,49 @@ pub(crate) fn scale_and_subtract<'a, S: Scalar>(
lhs_len,
)
}

#[allow(clippy::cast_sign_loss)]
#[allow(dead_code)]
/// Scale LHS and RHS to the same scale if at least one of them is decimal
/// and take the difference. This function is used for comparisons.
pub(crate) fn scale_and_subtract_columnar_value<'a, S: Scalar>(
alloc: &'a Bump,
lhs: ColumnarValue<'a, S>,
rhs: ColumnarValue<'a, S>,
lhs_scale: i8,
rhs_scale: i8,
is_equal: bool,
) -> ConversionResult<ColumnarValue<'a, S>> {
match (lhs, rhs) {
(ColumnarValue::Column(lhs), ColumnarValue::Column(rhs)) => {
Ok(ColumnarValue::Column(Column::Scalar(scale_and_subtract(
alloc, lhs, rhs, lhs_scale, rhs_scale, is_equal,
)?)))
}
(ColumnarValue::Literal(lhs), ColumnarValue::Column(rhs)) => {
Ok(ColumnarValue::Column(Column::Scalar(scale_and_subtract(
alloc,
Column::from_literal_with_length(&lhs, rhs.len(), alloc),
rhs,
lhs_scale,
rhs_scale,
is_equal,
)?)))
}
(ColumnarValue::Column(lhs), ColumnarValue::Literal(rhs)) => {
Ok(ColumnarValue::Column(Column::Scalar(scale_and_subtract(
alloc,
lhs,
Column::from_literal_with_length(&rhs, lhs.len(), alloc),
lhs_scale,
rhs_scale,
is_equal,
)?)))
}
(ColumnarValue::Literal(lhs), ColumnarValue::Literal(rhs)) => {
Ok(ColumnarValue::Literal(LiteralValue::Scalar(
scale_and_subtract_literal(&lhs, &rhs, lhs_scale, rhs_scale, is_equal)?,
)))
}
}
}
117 changes: 115 additions & 2 deletions crates/proof-of-sql/src/sql/proof_exprs/numerical_util.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,8 +1,36 @@
use crate::base::{
database::Column,
database::{Column, ColumnarValue, LiteralValue},
scalar::{Scalar, ScalarExt},
};
use bumpalo::Bump;
use core::cmp::Ordering;

#[allow(clippy::cast_sign_loss)]
/// Add or subtract two literals together.
pub(crate) fn add_subtract_literals<S: Scalar>(
lhs: &LiteralValue<S>,
rhs: &LiteralValue<S>,
lhs_scale: i8,
rhs_scale: i8,
is_subtract: bool,
) -> S {
let (lhs_scaled, rhs_scaled) = match lhs_scale.cmp(&rhs_scale) {
Ordering::Less => {
let scaling_factor = S::pow10((rhs_scale - lhs_scale) as u8);
(lhs.to_scalar() * scaling_factor, rhs.to_scalar())
}
Ordering::Equal => (lhs.to_scalar(), rhs.to_scalar()),
Ordering::Greater => {
let scaling_factor = S::pow10((lhs_scale - rhs_scale) as u8);
(lhs.to_scalar(), rhs.to_scalar() * scaling_factor)
}
};
if is_subtract {
lhs_scaled - rhs_scaled
} else {
lhs_scaled + rhs_scaled
}
}

#[allow(
clippy::missing_panics_doc,
Expand Down Expand Up @@ -36,9 +64,62 @@ pub(crate) fn add_subtract_columns<'a, S: Scalar>(
result
}

/// Add or subtract two [`ColumnarValues`] together.
#[allow(dead_code)]
pub(crate) fn add_subtract_columnar_values<'a, S: Scalar>(
lhs: ColumnarValue<'a, S>,
rhs: ColumnarValue<'a, S>,
lhs_scale: i8,
rhs_scale: i8,
alloc: &'a Bump,
is_subtract: bool,
) -> ColumnarValue<'a, S> {
match (lhs, rhs) {
(ColumnarValue::Column(lhs), ColumnarValue::Column(rhs)) => {
ColumnarValue::Column(Column::Scalar(add_subtract_columns(
lhs,
rhs,
lhs_scale,
rhs_scale,
alloc,
is_subtract,
)))
}
(ColumnarValue::Literal(lhs), ColumnarValue::Column(rhs)) => {
ColumnarValue::Column(Column::Scalar(add_subtract_columns(
Column::from_literal_with_length(&lhs, rhs.len(), alloc),
rhs,
lhs_scale,
rhs_scale,
alloc,
is_subtract,
)))
}
(ColumnarValue::Column(lhs), ColumnarValue::Literal(rhs)) => {
ColumnarValue::Column(Column::Scalar(add_subtract_columns(
lhs,
Column::from_literal_with_length(&rhs, lhs.len(), alloc),
lhs_scale,
rhs_scale,
alloc,
is_subtract,
)))
}
(ColumnarValue::Literal(lhs), ColumnarValue::Literal(rhs)) => {
ColumnarValue::Literal(LiteralValue::Scalar(add_subtract_literals(
&lhs,
&rhs,
lhs_scale,
rhs_scale,
is_subtract,
)))
}
}
}

/// Multiply two columns together.
/// # Panics
/// Panics if: The lengths of `lhs` and `rhs` are not equal.`lhs.scalar_at(i)` or `rhs.scalar_at(i)` returns `None`, which occurs if the column does not have, a scalar at the given index `i`.
/// Panics if: `lhs` and `rhs` are not of the same length.
pub(crate) fn multiply_columns<'a, S: Scalar>(
lhs: &Column<'a, S>,
rhs: &Column<'a, S>,
Expand All @@ -55,6 +136,38 @@ pub(crate) fn multiply_columns<'a, S: Scalar>(
})
}

#[allow(dead_code)]
/// Multiply two [`ColumnarValues`] together.
/// # Panics
/// Panics if: `lhs` and `rhs` are not of the same length.
pub(crate) fn multiply_columnar_values<'a, S: Scalar>(
lhs: &ColumnarValue<'a, S>,
rhs: &ColumnarValue<'a, S>,
alloc: &'a Bump,
) -> ColumnarValue<'a, S> {
match (lhs, rhs) {
(ColumnarValue::Column(lhs), ColumnarValue::Column(rhs)) => {
ColumnarValue::Column(Column::Scalar(multiply_columns(lhs, rhs, alloc)))
}
(ColumnarValue::Literal(lhs), ColumnarValue::Column(rhs)) => {
let lhs_scalar = lhs.to_scalar();
let result =
alloc.alloc_slice_fill_with(rhs.len(), |i| lhs_scalar * rhs.scalar_at(i).unwrap());
ColumnarValue::Column(Column::Scalar(result))
}
(ColumnarValue::Column(lhs), ColumnarValue::Literal(rhs)) => {
let rhs_scalar = rhs.to_scalar();
let result =
alloc.alloc_slice_fill_with(lhs.len(), |i| lhs.scalar_at(i).unwrap() * rhs_scalar);
ColumnarValue::Column(Column::Scalar(result))
}
(ColumnarValue::Literal(lhs), ColumnarValue::Literal(rhs)) => {
let result = lhs.to_scalar() * rhs.to_scalar();
ColumnarValue::Literal(LiteralValue::Scalar(result))
}
}
}

#[allow(
clippy::missing_panics_doc,
reason = "scaling factor is guaranteed to not be negative based on input validation prior to calling this function"
Expand Down
Loading