Skip to content

Commit

Permalink
feat(rust,python,cli): add SQL support for timestamp precision modi…
Browse files Browse the repository at this point in the history 
…fier (#13936)
  • Loading branch information
@alexander-beedie
alexander-beedie authored Jan 24, 2024
1 parent af79f69 commit 7e06238
Show file tree
Hide file tree
Showing 7 changed files with 112 additions and 50 deletions.
12 changes: 6 additions & 6 deletions crates/polars-sql/src/context.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -194,9 +194,9 @@ impl SQLContext {
right,
} => self.process_union(left, right, set_quantifier, query),
SetExpr::SetOperation { op, .. } => {
polars_bail!(InvalidOperation: "{} operation not yet supported", op)
polars_bail!(InvalidOperation: "'{}' operation not yet supported", op)
},
op => polars_bail!(InvalidOperation: "{} operation not yet supported", op),
op => polars_bail!(InvalidOperation: "'{}' operation not yet supported", op),
}
}

Expand Down Expand Up @@ -232,7 +232,7 @@ impl SQLContext {
concatenated.map(|lf| lf.unique(None, UniqueKeepStrategy::Any))
},
#[allow(unreachable_patterns)]
_ => polars_bail!(InvalidOperation: "UNION {} is not yet supported", quantifier),
_ => polars_bail!(InvalidOperation: "'UNION {}' is not yet supported", quantifier),
}
}

Expand Down Expand Up @@ -610,11 +610,11 @@ impl SQLContext {
self.table_map.insert(alias.name.value.clone(), lf.clone());
Ok((alias.name.value.clone(), lf))
} else {
polars_bail!(ComputeError: "Derived tables must have aliases");
polars_bail!(ComputeError: "derived tables must have aliases");
}
},
// Support bare table, optional with alias for now
_ => polars_bail!(ComputeError: "not implemented"),
_ => polars_bail!(ComputeError: "not yet implemented: {}", relation),
}
}

Expand Down Expand Up @@ -774,7 +774,7 @@ impl SQLContext {
cols(schema.iter_names())
},
e => polars_bail!(
ComputeError: "Invalid wildcard expression: {:?}",
ComputeError: "invalid wildcard expression: {:?}",
e
),
};
Expand Down
42 changes: 21 additions & 21 deletions crates/polars-sql/src/functions.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -773,10 +773,10 @@ impl SQLFunctionVisitor<'_> {
polars_bail!(InvalidOperation: "Round does not (yet) support negative 'decimals': {}", function.args[1])
}
},
_ => polars_bail!(InvalidOperation: "Invalid 'decimals' for Round: {}", function.args[1]),
_ => polars_bail!(InvalidOperation: "invalid 'decimals' for Round: {}", function.args[1]),
}))
}),
_ => polars_bail!(InvalidOperation:"Invalid number of arguments for Round: {}", function.args.len()),
_ => polars_bail!(InvalidOperation: "invalid number of arguments for Round: {}", function.args.len()),
},
Sign => self.visit_unary(Expr::sign),
Sqrt => self.visit_unary(Expr::sqrt),
Expand Down Expand Up @@ -811,7 +811,7 @@ impl SQLFunctionVisitor<'_> {
3 => self.try_visit_ternary(|cond: Expr, expr1: Expr, expr2: Expr| {
Ok(when(cond).then(expr1).otherwise(expr2))
}),
_ => polars_bail!(InvalidOperation: "Invalid number of arguments for If: {}", function.args.len()
_ => polars_bail!(InvalidOperation: "invalid number of arguments for If: {}", function.args.len()
),
},
IfNull => match function.args.len() {
Expand All @@ -826,13 +826,13 @@ impl SQLFunctionVisitor<'_> {
Date => match function.args.len() {
1 => self.visit_unary(|e| e.str().to_date(StrptimeOptions::default())),
2 => self.visit_binary(|e, fmt| e.str().to_date(fmt)),
_ => polars_bail!(InvalidOperation: "Invalid number of arguments for Date: {}", function.args.len()),
_ => polars_bail!(InvalidOperation: "invalid number of arguments for Date: {}", function.args.len()),
},
DatePart => self.try_visit_binary(|e, part| {
match part {
Expr::Literal(LiteralValue::String(p)) => parse_date_part(e, &p),
_ => {
polars_bail!(InvalidOperation: "Invalid 'part' for DatePart: {}", function.args[1]);
polars_bail!(InvalidOperation: "invalid 'part' for DatePart: {}", function.args[1]);
}
}
}),
Expand All @@ -842,12 +842,12 @@ impl SQLFunctionVisitor<'_> {
// ----
BitLength => self.visit_unary(|e| e.str().len_bytes() * lit(8)),
Concat => if function.args.is_empty() {
polars_bail!(InvalidOperation: "Invalid number of arguments for Concat: 0");
polars_bail!(InvalidOperation: "invalid number of arguments for Concat: 0");
} else {
self.visit_variadic(|exprs: &[Expr]| concat_str(exprs, "", true))
},
ConcatWS => if function.args.len() < 2 {
polars_bail!(InvalidOperation: "Invalid number of arguments for ConcatWS: {}", function.args.len());
polars_bail!(InvalidOperation: "invalid number of arguments for ConcatWS: {}", function.args.len());
} else {
self.try_visit_variadic(|exprs: &[Expr]| {
match &exprs[0] {
Expand All @@ -867,7 +867,7 @@ impl SQLFunctionVisitor<'_> {
let len = if n > 0 { lit(n) } else { (e.clone().str().len_chars() + lit(n)).clip_min(lit(0)) };
e.str().slice(lit(0), len)
},
Expr::Literal(_) => polars_bail!(InvalidOperation: "Invalid 'n_chars' for Left: {}", function.args[1]),
Expr::Literal(_) => polars_bail!(InvalidOperation: "invalid 'n_chars' for Left: {}", function.args[1]),
_ => {
when(length.clone().gt_eq(lit(0)))
.then(e.clone().str().slice(lit(0), length.clone().abs()))
Expand All @@ -880,7 +880,7 @@ impl SQLFunctionVisitor<'_> {
LTrim => match function.args.len() {
1 => self.visit_unary(|e| e.str().strip_chars_start(lit(Null))),
2 => self.visit_binary(|e, s| e.str().strip_chars_start(s)),
_ => polars_bail!(InvalidOperation: "Invalid number of arguments for LTrim: {}", function.args.len()),
_ => polars_bail!(InvalidOperation: "invalid number of arguments for LTrim: {}", function.args.len()),
},
OctetLength => self.visit_unary(|e| e.str().len_bytes()),
StrPos => {
Expand All @@ -894,23 +894,23 @@ impl SQLFunctionVisitor<'_> {
match (pat, flags) {
(Expr::Literal(LiteralValue::String(s)), Expr::Literal(LiteralValue::String(f))) => {
if f.is_empty() {
polars_bail!(InvalidOperation: "Invalid/empty 'flags' for RegexpLike: {}", function.args[2]);
polars_bail!(InvalidOperation: "invalid/empty 'flags' for RegexpLike: {}", function.args[2]);
};
lit(format!("(?{}){}", f, s))
},
_ => {
polars_bail!(InvalidOperation: "Invalid arguments for RegexpLike: {}, {}", function.args[1], function.args[2]);
polars_bail!(InvalidOperation: "invalid arguments for RegexpLike: {}, {}", function.args[1], function.args[2]);
},
},
true))
}),
_ => polars_bail!(InvalidOperation:"Invalid number of arguments for RegexpLike: {}",function.args.len()),
_ => polars_bail!(InvalidOperation: "invalid number of arguments for RegexpLike: {}",function.args.len()),
},
Replace => match function.args.len() {
3 => self.try_visit_ternary(|e, old, new| {
Ok(e.str().replace_all(old, new, true))
}),
_ => polars_bail!(InvalidOperation: "Invalid number of arguments for Replace: {}", function.args.len()),
_ => polars_bail!(InvalidOperation: "invalid number of arguments for Replace: {}", function.args.len()),
},
Reverse => self.visit_unary(|e| e.str().reverse()),
Right => self.try_visit_binary(|e, length| {
Expand All @@ -921,7 +921,7 @@ impl SQLFunctionVisitor<'_> {
let offset = if n < 0 { lit(n.abs()) } else { e.clone().str().len_chars().cast(DataType::Int32) - lit(n) };
e.str().slice(offset, lit(Null))
},
Expr::Literal(_) => polars_bail!(InvalidOperation: "Invalid 'n_chars' for Right: {}", function.args[1]),
Expr::Literal(_) => polars_bail!(InvalidOperation: "invalid 'n_chars' for Right: {}", function.args[1]),
_ => {
when(length.clone().lt(lit(0)))
.then(e.clone().str().slice(length.clone().abs(), lit(Null)))
Expand All @@ -932,7 +932,7 @@ impl SQLFunctionVisitor<'_> {
RTrim => match function.args.len() {
1 => self.visit_unary(|e| e.str().strip_chars_end(lit(Null))),
2 => self.visit_binary(|e, s| e.str().strip_chars_end(s)),
_ => polars_bail!(InvalidOperation: "Invalid number of arguments for RTrim: {}", function.args.len()),
_ => polars_bail!(InvalidOperation: "invalid number of arguments for RTrim: {}", function.args.len()),
},
StartsWith => self.visit_binary(|e, s| e.str().starts_with(s)),
Substring => match function.args.len() {
Expand All @@ -942,7 +942,7 @@ impl SQLFunctionVisitor<'_> {
Expr::Literal(Null) => lit(Null),
Expr::Literal(LiteralValue::Int64(n)) if n <= 0 => e,
Expr::Literal(LiteralValue::Int64(n)) => e.str().slice(lit(n - 1), lit(Null)),
Expr::Literal(_) => polars_bail!(InvalidOperation: "Invalid 'start' for Substring: {}", function.args[1]),
Expr::Literal(_) => polars_bail!(InvalidOperation: "invalid 'start' for Substring: {}", function.args[1]),
_ => start.clone() + lit(1),
})
}),
Expand All @@ -956,9 +956,9 @@ impl SQLFunctionVisitor<'_> {
(Expr::Literal(LiteralValue::Int64(n)), _) => {
e.str().slice(lit(0), (length.clone() + lit(n - 1)).clip_min(lit(0)))
},
(Expr::Literal(_), _) => polars_bail!(InvalidOperation: "Invalid 'start' for Substring: {}", function.args[1]),
(Expr::Literal(_), _) => polars_bail!(InvalidOperation: "invalid 'start' for Substring: {}", function.args[1]),
(_, Expr::Literal(LiteralValue::Float64(_))) => {
polars_bail!(InvalidOperation: "Invalid 'length' for Substring: {}", function.args[1])
polars_bail!(InvalidOperation: "invalid 'length' for Substring: {}", function.args[1])
},
_ => {
let adjusted_start = start.clone() - lit(1);
Expand All @@ -968,7 +968,7 @@ impl SQLFunctionVisitor<'_> {
}
})
}),
_ => polars_bail!(InvalidOperation: "Invalid number of arguments for Substring: {}", function.args.len()),
_ => polars_bail!(InvalidOperation: "invalid number of arguments for Substring: {}", function.args.len()),
}
Upper => self.visit_unary(|e| e.str().to_uppercase()),
// ----
Expand Down Expand Up @@ -1006,10 +1006,10 @@ impl SQLFunctionVisitor<'_> {
e.list().eval(col("").fill_null(lit(v)), false).list().join(sep, false)
})
},
_ => polars_bail!(InvalidOperation: "Invalid null value for ArrayToString: {}", function.args[2]),
_ => polars_bail!(InvalidOperation: "invalid null value for ArrayToString: {}", function.args[2]),
}
}),
_ => polars_bail!(InvalidOperation: "Invalid number of arguments for ArrayToString: {}", function.args.len()),
_ => polars_bail!(InvalidOperation: "invalid number of arguments for ArrayToString: {}", function.args.len()),
}
ArrayUnique => self.visit_unary(|e| e.list().unique()),
Explode => self.visit_unary(|e| e.explode()),
Expand Down
49 changes: 35 additions & 14 deletions crates/polars-sql/src/sql_expr.rs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -14,8 +14,8 @@ use sqlparser::ast::ExactNumberInfo;
use sqlparser::ast::{
ArrayAgg, ArrayElemTypeDef, BinaryOperator as SQLBinaryOperator, BinaryOperator, CastFormat,
DataType as SQLDataType, DateTimeField, Expr as SQLExpr, Function as SQLFunction, Ident,
JoinConstraint, OrderByExpr, Query as Subquery, SelectItem, TrimWhereField, UnaryOperator,
Value as SQLValue,
JoinConstraint, OrderByExpr, Query as Subquery, SelectItem, TimezoneInfo, TrimWhereField,
UnaryOperator, Value as SQLValue,
};
use sqlparser::dialect::GenericDialect;
use sqlparser::parser::{Parser, ParserOptions};
Expand Down Expand Up @@ -62,11 +62,32 @@ pub(crate) fn map_sql_polars_datatype(data_type: &SQLDataType) -> PolarsResult<D
SQLDataType::Int2(_) => DataType::Int16,
SQLDataType::Int4(_) => DataType::Int32,
SQLDataType::Int8(_) => DataType::Int64,
SQLDataType::Interval => DataType::Duration(TimeUnit::Milliseconds),
SQLDataType::Interval => DataType::Duration(TimeUnit::Microseconds),
SQLDataType::Real => DataType::Float32,
SQLDataType::SmallInt(_) => DataType::Int16,
SQLDataType::Time { .. } => DataType::Time,
SQLDataType::Timestamp { .. } => DataType::Datetime(TimeUnit::Milliseconds, None),
SQLDataType::Time(_, tz) => match tz {
TimezoneInfo::None => DataType::Time,
_ => {
polars_bail!(ComputeError: "`time` with timezone is not supported; found tz={}", tz)
},
},
SQLDataType::Timestamp(prec, tz) => {
let tu = match prec {
None => TimeUnit::Microseconds,
Some(3) => TimeUnit::Milliseconds,
Some(6) => TimeUnit::Microseconds,
Some(9) => TimeUnit::Nanoseconds,
Some(n) => {
polars_bail!(ComputeError: "unsupported `timestamp` precision; expected 3, 6 or 9, found prec={}", n)
},
};
match tz {
TimezoneInfo::None => DataType::Datetime(tu, None),
_ => {
polars_bail!(ComputeError: "`timestamp` with timezone is not (yet) supported; found tz={}", tz)
},
}
},
SQLDataType::TinyInt(_) => DataType::Int8,
SQLDataType::UnsignedBigInt(_) => DataType::UInt64,
SQLDataType::UnsignedInt(_) | SQLDataType::UnsignedInteger(_) => DataType::UInt32,
Expand Down Expand Up @@ -270,7 +291,7 @@ impl SQLExprVisitor<'_> {
}
},
_ => polars_bail!(
ComputeError: "Invalid identifier {:?}",
ComputeError: "invalid identifier {:?}",
idents
),
}
Expand Down Expand Up @@ -351,23 +372,23 @@ impl SQLExprVisitor<'_> {
// ----
SQLBinaryOperator::PGRegexMatch => match right {
Expr::Literal(LiteralValue::String(_)) => left.str().contains(right, true),
_ => polars_bail!(ComputeError: "Invalid pattern for '~' operator: {:?}", right),
_ => polars_bail!(ComputeError: "invalid pattern for '~' operator: {:?}", right),
},
SQLBinaryOperator::PGRegexNotMatch => match right {
Expr::Literal(LiteralValue::String(_)) => left.str().contains(right, true).not(),
_ => polars_bail!(ComputeError: "Invalid pattern for '!~' operator: {:?}", right),
_ => polars_bail!(ComputeError: "invalid pattern for '!~' operator: {:?}", right),
},
SQLBinaryOperator::PGRegexIMatch => match right {
Expr::Literal(LiteralValue::String(pat)) => {
left.str().contains(lit(format!("(?i){}", pat)), true)
},
_ => polars_bail!(ComputeError: "Invalid pattern for '~*' operator: {:?}", right),
_ => polars_bail!(ComputeError: "invalid pattern for '~*' operator: {:?}", right),
},
SQLBinaryOperator::PGRegexNotIMatch => match right {
Expr::Literal(LiteralValue::String(pat)) => {
left.str().contains(lit(format!("(?i){}", pat)), true).not()
},
_ => polars_bail!(ComputeError: "Invalid pattern for '!~*' operator: {:?}", right),
_ => polars_bail!(ComputeError: "invalid pattern for '!~*' operator: {:?}", right),
},
other => polars_bail!(ComputeError: "SQL operator {:?} is not yet supported", other),
})
Expand All @@ -388,7 +409,7 @@ impl SQLExprVisitor<'_> {
(UnaryOperator::Plus, _) => lit(0) + expr,
(UnaryOperator::Minus, _) => lit(0) - expr,
(UnaryOperator::Not, _) => expr.not(),
other => polars_bail!(InvalidOperation: "Unary operator {:?} is not supported", other),
other => polars_bail!(InvalidOperation: "unary operator {:?} is not supported", other),
})
}

Expand Down Expand Up @@ -424,7 +445,7 @@ impl SQLExprVisitor<'_> {
BinaryOperator::LtEq => Ok(left.lt_eq(right.min())),
BinaryOperator::Eq => polars_bail!(ComputeError: "ALL cannot be used with ="),
BinaryOperator::NotEq => polars_bail!(ComputeError: "ALL cannot be used with !="),
_ => polars_bail!(ComputeError: "Invalid comparison operator"),
_ => polars_bail!(ComputeError: "invalid comparison operator"),
}
}

Expand All @@ -447,7 +468,7 @@ impl SQLExprVisitor<'_> {
BinaryOperator::LtEq => Ok(left.lt_eq(right.max())),
BinaryOperator::Eq => Ok(left.is_in(right)),
BinaryOperator::NotEq => Ok(left.is_in(right).not()),
_ => polars_bail!(ComputeError: "Invalid comparison operator"),
_ => polars_bail!(ComputeError: "invalid comparison operator"),
}
}

Expand Down Expand Up @@ -891,7 +912,7 @@ pub(super) fn process_join_constraint(
return Ok((using.clone(), using.clone()));
}
}
polars_bail!(InvalidOperation: "Unsupported SQL join constraint:\n{:?}", constraint);
polars_bail!(InvalidOperation: "unsupported SQL join constraint:\n{:?}", constraint);
}

/// parse a SQL expression to a polars expression
Expand Down
2 changes: 1 addition & 1 deletion py-polars/tests/unit/sql/test_numeric.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -109,7 +109,7 @@ def test_round_ndigits_errors() -> None:
df = pl.DataFrame({"n": [99.999]})
with pl.SQLContext(df=df, eager_execution=True) as ctx:
with pytest.raises(
InvalidOperationError, match="Invalid 'decimals' for Round: ??"
InvalidOperationError, match="invalid 'decimals' for Round: ??"
):
ctx.execute("SELECT ROUND(n,'??') AS n FROM df")
with pytest.raises(
Expand Down
Loading

0 comments on commit 7e06238

Please sign in to comment.