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

add support for unary and binary values in values list, update docs #1172

Merged
merged 3 commits into from
Oct 25, 2021
Merged

add support for unary and binary values in values list, update docs #1172

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
975d22c
add support for unary and binary values in values list
jimexist Oct 22, 2021
d233616
add explain
jimexist Oct 23, 2021
38c156b
fix clippy
jimexist Oct 23, 2021
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
3 changes: 2 additions & 1 deletion README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -159,6 +159,7 @@ DataFusion also includes a simple command-line interactive SQL utility. See the
- [x] Common math functions
- [x] cast
- [x] try_cast
- [x] [`VALUES` lists](https://www.postgresql.org/docs/current/queries-values.html)
- Postgres compatible String functions
- [x] ascii
- [x] bit_length
Expand Down Expand Up @@ -191,7 +192,7 @@ DataFusion also includes a simple command-line interactive SQL utility. See the
- Miscellaneous/Boolean functions
- [x] nullif
- Approximation functions
- [ ] approx_distinct
- [x] approx_distinct
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

❤️

- Common date/time functions
- [ ] Basic date functions
- [ ] Basic time functions
Expand Down
24 changes: 19 additions & 5 deletions datafusion/src/physical_plan/values.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -24,6 +24,7 @@ use crate::physical_plan::{
Partitioning, PhysicalExpr,
};
use crate::scalar::ScalarValue;
use arrow::array::new_null_array;
use arrow::datatypes::SchemaRef;
use arrow::record_batch::RecordBatch;
use async_trait::async_trait;
Expand All @@ -48,20 +49,33 @@ impl ValuesExec {
if data.is_empty() {
return Err(DataFusionError::Plan("Values list cannot be empty".into()));
}
// we have this empty batch as a placeholder to satisfy evaluation argument
let batch = RecordBatch::new_empty(schema.clone());
let n_row = data.len();
let n_col = schema.fields().len();
// we have this single row, null, typed batch as a placeholder to satisfy evaluation argument
let batch = RecordBatch::try_new(
schema.clone(),
schema
.fields()
.iter()
.map(|field| new_null_array(field.data_type(), 1))
.collect::<Vec<_>>(),
)?;
let arr = (0..n_col)
.map(|j| {
(0..n_row)
.map(|i| {
let r = data[i][j].evaluate(&batch);
match r {
Ok(ColumnarValue::Scalar(scalar)) => Ok(scalar),
Ok(ColumnarValue::Array(_)) => Err(DataFusionError::Plan(
"Cannot have array values in a values list".into(),
)),
Ok(ColumnarValue::Array(a)) if a.len() == 1 => {
ScalarValue::try_from_array(&a, 0)
}
Ok(ColumnarValue::Array(a)) => {
Err(DataFusionError::Plan(format!(
"Cannot have array values {:?} in a values list",
a
)))
}
Err(err) => Err(err),
}
})
Expand Down
170 changes: 99 additions & 71 deletions datafusion/src/sql/planner.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1069,24 +1069,107 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
}
}

fn parse_sql_binary_op(
&self,
left: &SQLExpr,
op: &BinaryOperator,
right: &SQLExpr,
schema: &DFSchema,
) -> Result<Expr> {
let operator = match *op {
BinaryOperator::Gt => Ok(Operator::Gt),
BinaryOperator::GtEq => Ok(Operator::GtEq),
BinaryOperator::Lt => Ok(Operator::Lt),
BinaryOperator::LtEq => Ok(Operator::LtEq),
BinaryOperator::Eq => Ok(Operator::Eq),
BinaryOperator::NotEq => Ok(Operator::NotEq),
BinaryOperator::Plus => Ok(Operator::Plus),
BinaryOperator::Minus => Ok(Operator::Minus),
BinaryOperator::Multiply => Ok(Operator::Multiply),
BinaryOperator::Divide => Ok(Operator::Divide),
BinaryOperator::Modulo => Ok(Operator::Modulo),
BinaryOperator::And => Ok(Operator::And),
BinaryOperator::Or => Ok(Operator::Or),
BinaryOperator::Like => Ok(Operator::Like),
BinaryOperator::NotLike => Ok(Operator::NotLike),
BinaryOperator::PGRegexMatch => Ok(Operator::RegexMatch),
BinaryOperator::PGRegexIMatch => Ok(Operator::RegexIMatch),
BinaryOperator::PGRegexNotMatch => Ok(Operator::RegexNotMatch),
BinaryOperator::PGRegexNotIMatch => Ok(Operator::RegexNotIMatch),
_ => Err(DataFusionError::NotImplemented(format!(
"Unsupported SQL binary operator {:?}",
op
))),
}?;

Ok(Expr::BinaryExpr {
left: Box::new(self.sql_expr_to_logical_expr(left, schema)?),
op: operator,
right: Box::new(self.sql_expr_to_logical_expr(right, schema)?),
})
}

fn parse_sql_unary_op(
&self,
op: &UnaryOperator,
expr: &SQLExpr,
schema: &DFSchema,
) -> Result<Expr> {
match op {
UnaryOperator::Not => Ok(Expr::Not(Box::new(
self.sql_expr_to_logical_expr(expr, schema)?,
))),
UnaryOperator::Plus => Ok(self.sql_expr_to_logical_expr(expr, schema)?),
UnaryOperator::Minus => {
match expr {
// optimization: if it's a number literal, we apply the negative operator
// here directly to calculate the new literal.
SQLExpr::Value(Value::Number(n,_)) => match n.parse::<i64>() {
Ok(n) => Ok(lit(-n)),
Err(_) => Ok(lit(-n
.parse::<f64>()
.map_err(|_e| {
DataFusionError::Internal(format!(
"negative operator can be only applied to integer and float operands, got: {}",
n))
})?)),
},
// not a literal, apply negative operator on expression
_ => Ok(Expr::Negative(Box::new(self.sql_expr_to_logical_expr(expr, schema)?))),
}
}
_ => Err(DataFusionError::NotImplemented(format!(
"Unsupported SQL unary operator {:?}",
op
))),
}
}

fn sql_values_to_plan(&self, values: &SQLValues) -> Result<LogicalPlan> {
// values should not be based on any other schema
let schema = DFSchema::empty();
let values = values
.0
.iter()
.map(|row| {
row.iter()
.map(|v| match v {
SQLExpr::Value(Value::Number(n, _)) => match n.parse::<i64>() {
Ok(n) => Ok(lit(n)),
Err(_) => Ok(lit(n.parse::<f64>().unwrap())),
},
SQLExpr::Value(Value::Number(n, _)) => parse_sql_number(n),
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wonder if we could just call sql_to_logical_expr here (rather than repeating part of the parsing) so that any expression in the values list could be supported

That would allow things like CASE .. WHEN .. END type expressions in the VALUES lists as well

Although if we went with that approach, we probably would have to then do a post creation check for unsupported expressions (like Column, Aggregate, etc)

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

i'd like to keep this list small for now unless really needed in future.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

the key assumption here is that for all the expression in the values list they should be evaluated to a scalar or singleton array, but there's no way to test for that during planning time?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

i'd like to keep this list small for now unless really needed in future.

Makes sense to me

the key assumption here is that for all the expression in the values list they should be evaluated to a scalar or singleton array, but there's no way to test for that during planning time?

I think in general most Exprs have the property that they can be compiled into a PhysicalExpr and that PhysicalExpr can be evaluated to produce a ArrayRef. There are some exceptions like Expr::Aggregrate etc that don't compile directly to a PhysicalExpr but are special cased

SQLExpr::Value(Value::SingleQuotedString(ref s)) => {
Ok(lit(s.clone()))
}
SQLExpr::Value(Value::Null) => {
Ok(Expr::Literal(ScalarValue::Utf8(None)))
}
SQLExpr::Value(Value::Boolean(n)) => Ok(lit(*n)),
SQLExpr::UnaryOp { ref op, ref expr } => {
self.parse_sql_unary_op(op, expr, &schema)
}
SQLExpr::BinaryOp {
ref left,
ref op,
ref right,
} => self.parse_sql_binary_op(left, op, right, &schema),
other => Err(DataFusionError::NotImplemented(format!(
"Unsupported value {:?} in a values list expression",
other
Expand All @@ -1100,14 +1183,9 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {

fn sql_expr_to_logical_expr(&self, sql: &SQLExpr, schema: &DFSchema) -> Result<Expr> {
match sql {
SQLExpr::Value(Value::Number(n, _)) => match n.parse::<i64>() {
Ok(n) => Ok(lit(n)),
Err(_) => Ok(lit(n.parse::<f64>().unwrap())),
},
SQLExpr::Value(Value::Number(n, _)) => parse_sql_number(n),
SQLExpr::Value(Value::SingleQuotedString(ref s)) => Ok(lit(s.clone())),

SQLExpr::Value(Value::Boolean(n)) => Ok(lit(*n)),

SQLExpr::Value(Value::Null) => Ok(Expr::Literal(ScalarValue::Utf8(None))),
SQLExpr::Extract { field, expr } => Ok(Expr::ScalarFunction {
fun: functions::BuiltinScalarFunction::DatePart,
Expand Down Expand Up @@ -1244,34 +1322,9 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
right: Box::new(self.sql_expr_to_logical_expr(right, schema)?),
}),

SQLExpr::UnaryOp { ref op, ref expr } => match op {
UnaryOperator::Not => Ok(Expr::Not(Box::new(
self.sql_expr_to_logical_expr(expr, schema)?,
))),
UnaryOperator::Plus => Ok(self.sql_expr_to_logical_expr(expr, schema)?),
UnaryOperator::Minus => {
match expr.as_ref() {
// optimization: if it's a number literal, we apply the negative operator
// here directly to calculate the new literal.
SQLExpr::Value(Value::Number(n,_)) => match n.parse::<i64>() {
Ok(n) => Ok(lit(-n)),
Err(_) => Ok(lit(-n
.parse::<f64>()
.map_err(|_e| {
DataFusionError::Internal(format!(
"negative operator can be only applied to integer and float operands, got: {}",
n))
})?)),
},
// not a literal, apply negative operator on expression
_ => Ok(Expr::Negative(Box::new(self.sql_expr_to_logical_expr(expr, schema)?))),
}
}
_ => Err(DataFusionError::NotImplemented(format!(
"Unsupported SQL unary operator {:?}",
op
))),
},
SQLExpr::UnaryOp { ref op, ref expr } => {
self.parse_sql_unary_op(op, expr, schema)
}

SQLExpr::Between {
ref expr,
Expand Down Expand Up @@ -1306,39 +1359,7 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
ref left,
ref op,
ref right,
} => {
let operator = match *op {
BinaryOperator::Gt => Ok(Operator::Gt),
BinaryOperator::GtEq => Ok(Operator::GtEq),
BinaryOperator::Lt => Ok(Operator::Lt),
BinaryOperator::LtEq => Ok(Operator::LtEq),
BinaryOperator::Eq => Ok(Operator::Eq),
BinaryOperator::NotEq => Ok(Operator::NotEq),
BinaryOperator::Plus => Ok(Operator::Plus),
BinaryOperator::Minus => Ok(Operator::Minus),
BinaryOperator::Multiply => Ok(Operator::Multiply),
BinaryOperator::Divide => Ok(Operator::Divide),
BinaryOperator::Modulo => Ok(Operator::Modulo),
BinaryOperator::And => Ok(Operator::And),
BinaryOperator::Or => Ok(Operator::Or),
BinaryOperator::Like => Ok(Operator::Like),
BinaryOperator::NotLike => Ok(Operator::NotLike),
BinaryOperator::PGRegexMatch => Ok(Operator::RegexMatch),
BinaryOperator::PGRegexIMatch => Ok(Operator::RegexIMatch),
BinaryOperator::PGRegexNotMatch => Ok(Operator::RegexNotMatch),
BinaryOperator::PGRegexNotIMatch => Ok(Operator::RegexNotIMatch),
_ => Err(DataFusionError::NotImplemented(format!(
"Unsupported SQL binary operator {:?}",
op
))),
}?;

Ok(Expr::BinaryExpr {
left: Box::new(self.sql_expr_to_logical_expr(left, schema)?),
op: operator,
right: Box::new(self.sql_expr_to_logical_expr(right, schema)?),
})
}
} => self.parse_sql_binary_op(left, op, right, schema),

SQLExpr::Trim { expr, trim_where } => {
let (fun, where_expr) = match trim_where {
Expand Down Expand Up @@ -3630,3 +3651,10 @@ mod tests {
quick_test(sql, expected);
}
}

fn parse_sql_number(n: &str) -> Result<Expr> {
match n.parse::<i64>() {
Ok(n) => Ok(lit(n)),
Err(_) => Ok(lit(n.parse::<f64>().unwrap())),
}
}
38 changes: 38 additions & 0 deletions datafusion/tests/sql.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -491,6 +491,30 @@ async fn select_values_list() -> Result<()> {
];
assert_batches_eq!(expected, &actual);
}
{
let sql = "VALUES (-1)";
let actual = execute_to_batches(&mut ctx, sql).await;
let expected = vec![
"+---------+",
"| column1 |",
"+---------+",
"| -1 |",
"+---------+",
];
assert_batches_eq!(expected, &actual);
}
{
let sql = "VALUES (2+1,2-1,2>1)";
let actual = execute_to_batches(&mut ctx, sql).await;
let expected = vec![
"+---------+---------+---------+",
"| column1 | column2 | column3 |",
"+---------+---------+---------+",
"| 3 | 1 | true |",
"+---------+---------+---------+",
];
assert_batches_eq!(expected, &actual);
}
{
let sql = "VALUES";
let plan = ctx.create_logical_plan(sql);
Expand Down Expand Up @@ -647,6 +671,20 @@ async fn select_values_list() -> Result<()> {
];
assert_batches_eq!(expected, &actual);
}
{
let sql = "EXPLAIN VALUES (1, 'a', -1, 1.1),(NULL, 'b', -3, 0.5)";
let actual = execute_to_batches(&mut ctx, sql).await;
let expected = vec![
"+---------------+-----------------------------------------------------------------------------------------------------------+",
"| plan_type | plan |",
"+---------------+-----------------------------------------------------------------------------------------------------------+",
"| logical_plan | Values: (Int64(1), Utf8(\"a\"), Int64(-1), Float64(1.1)), (Int64(NULL), Utf8(\"b\"), Int64(-3), Float64(0.5)) |",
"| physical_plan | ValuesExec |",
"| | |",
"+---------------+-----------------------------------------------------------------------------------------------------------+",
];
assert_batches_eq!(expected, &actual);
}
Ok(())
}

Expand Down
19 changes: 19 additions & 0 deletions integration-tests/sqls/values_list.sql
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,19 @@
-- Licensed to the Apache Software Foundation (ASF) under one
-- or more contributor license agreements. See the NOTICE file
-- distributed with this work for additional information
-- regarding copyright ownership. The ASF licenses this file
-- to you under the Apache License, Version 2.0 (the
-- "License"); you may not use this file except in compliance
-- with the License. You may obtain a copy of the License at

-- http://www.apache.org/licenses/LICENSE-2.0

-- Unless required by applicable law or agreed to in writing, software
-- distributed under the License is distributed on an "AS IS" BASIS,
-- WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-- See the License for the specific language governing permissions and
-- limitations under the License.

SELECT * FROM
(VALUES (1,2.0,-3,1+1),(10,20.0,-30,2+2))
AS tbl(int_col, float_col, negative_col, summation);
2 changes: 1 addition & 1 deletion integration-tests/test_psql_parity.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -77,7 +77,7 @@ def generate_csv_from_psql(fname: str):

class TestPsqlParity:
def test_tests_count(self):
assert len(test_files) == 15, "tests are missed"
assert len(test_files) == 16, "tests are missed"

@pytest.mark.parametrize("fname", test_files)
def test_sql_file(self, fname):
Expand Down