[task #8987]add_to_date_function

Signed-off-by: tangruilin <[email protected]>

Author: Tangruilin

datafusion/common/src/scalar.rs

@@ -52,6 +52,7 @@ use arrow::{
     },
 };
 use arrow_array::cast::as_list_array;
+use arrow_array::types::Date32Type;
 
 /// A dynamically typed, nullable single value, (the single-valued counter-part
 /// to arrow's [`Array`])
@@ -3239,6 +3240,12 @@ impl ScalarType<i64> for TimestampNanosecondType {
     }
 }
 
+impl ScalarType<i32> for Date32Type {
+    fn scalar(r: Option<i32>) -> ScalarValue {
+        ScalarValue::Date32(r)
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use std::cmp::Ordering;

datafusion/expr/src/built_in_function.rs

@@ -287,6 +287,8 @@ pub enum BuiltinScalarFunction {
     ToTimestampSeconds,
     /// from_unixtime
     FromUnixtime,
+    /// to_date
+    ToDate,
     ///now
     Now,
     ///current_date
@@ -490,6 +492,7 @@ impl BuiltinScalarFunction {
             BuiltinScalarFunction::RegexpMatch => Volatility::Immutable,
             BuiltinScalarFunction::Struct => Volatility::Immutable,
             BuiltinScalarFunction::FromUnixtime => Volatility::Immutable,
+            BuiltinScalarFunction::ToDate => Volatility::Immutable,
             BuiltinScalarFunction::ArrowTypeof => Volatility::Immutable,
             BuiltinScalarFunction::OverLay => Volatility::Immutable,
             BuiltinScalarFunction::Levenshtein => Volatility::Immutable,
@@ -806,6 +809,7 @@ impl BuiltinScalarFunction {
             BuiltinScalarFunction::ToTimestampMicros => Ok(Timestamp(Microsecond, None)),
             BuiltinScalarFunction::ToTimestampSeconds => Ok(Timestamp(Second, None)),
             BuiltinScalarFunction::FromUnixtime => Ok(Timestamp(Second, None)),
+            BuiltinScalarFunction::ToDate => Ok(Date32),
             BuiltinScalarFunction::Now => {
                 Ok(Timestamp(Nanosecond, Some("+00:00".into())))
             }
@@ -1055,6 +1059,7 @@ impl BuiltinScalarFunction {
             BuiltinScalarFunction::FromUnixtime => {
                 Signature::uniform(1, vec![Int64], self.volatility())
             }
+            BuiltinScalarFunction::ToDate => Signature::variadic_any(self.volatility()),
             BuiltinScalarFunction::Digest => Signature::one_of(
                 vec![
                     Exact(vec![Utf8, Utf8]),
@@ -1499,6 +1504,7 @@ impl BuiltinScalarFunction {
             BuiltinScalarFunction::ToTimestampSeconds => &["to_timestamp_seconds"],
             BuiltinScalarFunction::ToTimestampNanos => &["to_timestamp_nanos"],
             BuiltinScalarFunction::FromUnixtime => &["from_unixtime"],
+            BuiltinScalarFunction::ToDate => &["to_date"],
 
             // hashing functions
             BuiltinScalarFunction::Digest => &["digest"],

datafusion/expr/src/expr_fn.rs

@@ -891,6 +891,11 @@ nary_scalar_expr!(
 scalar_expr!(DatePart, date_part, part date, "extracts a subfield from the date");
 scalar_expr!(DateTrunc, date_trunc, part date, "truncates the date to a specified level of precision");
 scalar_expr!(DateBin, date_bin, stride source origin, "coerces an arbitrary timestamp to the start of the nearest specified interval");
+nary_scalar_expr!(
+    ToDate,
+    to_date,
+    "converts string to date according to the given format"
+);
 nary_scalar_expr!(
     ToTimestamp,
     to_timestamp,

datafusion/physical-expr/src/datetime_expressions.rs

@@ -53,8 +53,8 @@ use datafusion_common::cast::{
     as_timestamp_nanosecond_array, as_timestamp_second_array,
 };
 use datafusion_common::{
-    exec_err, internal_err, not_impl_err, DataFusionError, Result, ScalarType,
-    ScalarValue,
+    exec_err, internal_datafusion_err, internal_err, not_impl_err, DataFusionError,
+    Result, ScalarType, ScalarValue,
 };
 use datafusion_expr::ColumnarValue;
 
@@ -397,6 +397,39 @@ fn string_to_timestamp_nanos_shim(s: &str) -> Result<i64> {
     string_to_timestamp_nanos(s).map_err(|e| e.into())
 }
 
+fn to_date_impl(args: &[ColumnarValue], name: &str) -> Result<ColumnarValue> {
+    match args.len() {
+        1 => handle::<Date32Type, _, Date32Type>(
+            args,
+            |s| {
+                string_to_timestamp_nanos_shim(s)
+                    .map(|n| n / (1_000_000 * 24 * 60 * 60 * 1_000))
+                    .and_then(|v| {
+                        v.try_into().map_err(|_| {
+                            internal_datafusion_err!("Unable to cast to Date32 for converting from i64 to i32 failed")
+                        })
+                    })
+            },
+            name,
+        ),
+        n if n >= 2 => handle_multiple::<Date32Type, _, Date32Type, _>(
+            args,
+            |s, format| {
+                string_to_timestamp_nanos_formatted(s, format)
+                    .map(|n| n / (1_000_000 * 24 * 60 * 60 * 1_000))
+                    .and_then(|v| {
+                        v.try_into().map_err(|_| {
+                            internal_datafusion_err!("Unable to cast to Date32 for converting from i64 to i32 failed")
+                        })
+                    })
+            },
+            |n| n,
+            name,
+        ),
+        _ => internal_err!("Unsupported 0 argument count for function {name}"),
+    }
+}
+
 fn to_timestamp_impl<T: ArrowTimestampType + ScalarType<i64>>(
     args: &[ColumnarValue],
     name: &str,
@@ -424,6 +457,11 @@ fn to_timestamp_impl<T: ArrowTimestampType + ScalarType<i64>>(
     }
 }
 
+/// to_date SQL function
+pub fn to_date(args: &[ColumnarValue]) -> Result<ColumnarValue> {
+    to_date_impl(args, "to_date")
+}
+
 /// to_timestamp SQL function
 ///
 /// Note: `to_timestamp` returns `Timestamp(Nanosecond)` though its arguments are interpreted as **seconds**.
@@ -1343,6 +1381,36 @@ fn validate_to_timestamp_data_types(
     None
 }
 
+/// to_date SQL function implementation
+pub fn to_date_invoke(args: &[ColumnarValue]) -> Result<ColumnarValue> {
+    if args.is_empty() {
+        return exec_err!(
+            "to_date function requires 1 or more arguments, got {}",
+            args.len()
+        );
+    }
+
+    // validate that any args after the first one are Utf8
+    if args.len() > 1 {
+        if let Some(value) = validate_to_timestamp_data_types(args, "to_date") {
+            return value;
+        }
+    }
+
+    match args[0].data_type() {
+        DataType::Int32
+        | DataType::Int64
+        | DataType::Null
+        | DataType::Float64
+        | DataType::Date32
+        | DataType::Date64 => cast_column(&args[0], &DataType::Date32, None),
+        DataType::Utf8 => to_date(args),
+        other => {
+            internal_err!("Unsupported data type {:?} for function to_date", other)
+        }
+    }
+}
+
 /// to_timestamp() SQL function implementation
 pub fn to_timestamp_invoke(args: &[ColumnarValue]) -> Result<ColumnarValue> {
     if args.is_empty() {

datafusion/physical-expr/src/functions.rs

@@ -577,6 +577,7 @@ pub fn create_physical_fun(
         BuiltinScalarFunction::FromUnixtime => {
             Arc::new(datetime_expressions::from_unixtime_invoke)
         }
+        BuiltinScalarFunction::ToDate => Arc::new(datetime_expressions::to_date_invoke),
         BuiltinScalarFunction::InitCap => Arc::new(|args| match args[0].data_type() {
             DataType::Utf8 => {
                 make_scalar_function_inner(string_expressions::initcap::<i32>)(args)

datafusion/proto/proto/datafusion.proto

@@ -671,6 +671,7 @@ enum ScalarFunction {
   InStr = 132;
   MakeDate = 133;
   ArrayReverse = 134;
+  ToDate = 135;
 }
 
 message ScalarFunctionNode {

datafusion/proto/src/generated/pbjson.rs

@@ -575,6 +575,7 @@ impl From<&protobuf::ScalarFunction> for BuiltinScalarFunction {
             ScalarFunction::Levenshtein => Self::Levenshtein,
             ScalarFunction::SubstrIndex => Self::SubstrIndex,
             ScalarFunction::FindInSet => Self::FindInSet,
+            ScalarFunction::ToDate => Self::ToDate,
         }
     }
 }
@@ -1826,6 +1827,16 @@ pub fn parse_expr(
                 ScalarFunction::StructFun => {
                     Ok(struct_fun(parse_expr(&args[0], registry)?))
                 }
+                ScalarFunction::ToDate => {
+                    let args: Vec<_> = args
+                        .iter()
+                        .map(|expr| parse_expr(expr, registry))
+                        .collect::<std::result::Result<_, _>>()?;
+                    Ok(Expr::ScalarFunction(expr::ScalarFunction::new(
+                        BuiltinScalarFunction::ToDate,
+                        args,
+                    )))
+                }
             }
         }
         ExprType::ScalarUdfExpr(protobuf::ScalarUdfExprNode { fun_name, args }) => {

datafusion/proto/src/generated/prost.rs

@@ -1569,6 +1569,7 @@ impl TryFrom<&BuiltinScalarFunction> for protobuf::ScalarFunction {
             BuiltinScalarFunction::Levenshtein => Self::Levenshtein,
             BuiltinScalarFunction::SubstrIndex => Self::SubstrIndex,
             BuiltinScalarFunction::FindInSet => Self::FindInSet,
+            BuiltinScalarFunction::ToDate => Self::ToDate,
         };
 
         Ok(scalar_function)

datafusion/proto/src/logical_plan/from_proto.rs

@@ -107,3 +107,27 @@ query ?
 SELECT '2023-01-01T00:00:00'::timestamp - DATE '2021-01-01';
 ----
 730 days 0 hours 0 mins 0.000000000 secs
+
+# to_date_test
+statement ok
+create table to_date_t1(ts bigint) as VALUES
+   (1235865600000),
+   (1235865660000),
+   (1238544000000);
+
+
+# query_cast_timestamp_millis
+query D
+SELECT to_date(ts / 100000000) FROM to_date_t1 LIMIT 3
+----
+2003-11-02
+2003-11-02
+2003-11-29
+
+query D
+SELECT to_date('01-14-2023 01:01:30+05:30', '%q', '%d-%m-%Y %H/%M/%S', '%+', '%m-%d-%Y %H:%M:%S%#z');
+----
+2023-01-13
+
+statement error DataFusion error: Internal error: to_date function unsupported data type at index 1: List
+SELECT to_date('2022-08-03T14:38:50+05:30', make_array('%s', '%q', '%d-%m-%Y %H:%M:%S%#z', '%+'));