You are viewing a plain text version of this content. The canonical link for it is here.
Posted to github@arrow.apache.org by "mustafasrepo (via GitHub)" <gi...@apache.org> on 2023/02/24 07:09:36 UTC
[GitHub] [arrow-datafusion] mustafasrepo opened a new pull request, #5384: Bug fix: Window frame range value outside the type range
mustafasrepo opened a new pull request, #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384
# Which issue does this PR close?
<!--
We generally require a GitHub issue to be filed for all bug fixes and enhancements and this helps us generate change logs for our releases. You can link an issue to this PR using the GitHub syntax. For example `Closes #123` indicates that this PR will close issue #123.
-->
Closes [#5346](https://github.com/apache/arrow-datafusion/issues/5346).
# Rationale for this change
<!--
Why are you proposing this change? If this is already explained clearly in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand your changes and offer better suggestions for fixes.
-->
# What changes are included in this PR?
<!--
There is no need to duplicate the description in the issue here but it is sometimes worth providing a summary of the individual changes in this PR.
-->
If we cannot cast range value to the target type. We now check for If range value can be successfully casted to largest type of the family (for instance for `Int` types largest type is `Int64`, for `Uint` types it is `Uint64`). If we can accomplish so this range value is treated as unbounded. See #5346 for context.
# Are these changes tested?
<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code
If tests are not included in your PR, please explain why (for example, are they covered by existing tests)?
-->
Yes
# Are there any user-facing changes?
<!--
If there are user-facing changes then we may require documentation to be updated before approving the PR.
-->
<!--
If there are any breaking changes to public APIs, please add the `api change` label.
-->
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] mustafasrepo commented on a diff in pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "mustafasrepo (via GitHub)" <gi...@apache.org>.
mustafasrepo commented on code in PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384#discussion_r1116950002
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
+/// Downstream code uses this signal to treat these values as *unbounded*.
+fn coerce_scalar_range_aware(
+ target_type: &DataType,
+ value: &ScalarValue,
+) -> Result<ScalarValue> {
+ coerce_scalar(target_type, value).or_else(|err| {
+ // If type coercion fails, check if the largest type in family works:
+ if let Some(largest_type) = get_widest_type_in_family(target_type) {
+ coerce_scalar(largest_type, value).map_or_else(
+ |_| {
+ Err(DataFusionError::NotImplemented(format!(
+ "Cannot cast {:?} to {:?}",
+ value, target_type
+ )))
Review Comment:
Indeed you are right
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] mustafasrepo commented on a diff in pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "mustafasrepo (via GitHub)" <gi...@apache.org>.
mustafasrepo commented on code in PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384#discussion_r1117404087
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
+/// Downstream code uses this signal to treat these values as *unbounded*.
+fn coerce_scalar_range_aware(
+ target_type: &DataType,
+ value: &ScalarValue,
+) -> Result<ScalarValue> {
+ coerce_scalar(target_type, value).or_else(|err| {
+ // If type coercion fails, check if the largest type in family works:
+ if let Some(largest_type) = get_widest_type_in_family(target_type) {
+ coerce_scalar(largest_type, value).map_or_else(
+ |_| {
+ Err(DataFusionError::Execution(format!(
+ "Cannot cast {:?} to {:?}",
+ value, target_type
+ )))
+ },
+ |_| ScalarValue::try_from(target_type),
Review Comment:
Consider the case when target type is `Int8` and range value used in the query is 10000 like following: `OVER(ORDER BY c1 RANGE BETWEEN 10000 PRECEDING AND 10000 FOLLOWING)`). 10000 cannot be casted to `Int8`. In this case we try to cast value to `Int64` type. If it succeeds it means that 10000 couldn't be casted to `Int8` because of overflow in the first case. It means that we can treat the `OVER(ORDER BY c1 RANGE BETWEEN 10000 PRECEDING AND 10000 FOLLOWING)` as `OVER(ORDER BY c1 RANGE BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING)` without loss of generality, since we know that range would cover the whole table. In this case we are not interested in the casted version (We ignore the result. Then return `NULL` to convert range to `UNBOUNDED`. By the way we also use `NULL` for the types in the form `ScalarValue::Int8(None)` not just for the types `ScalarValue::Null` in case that terminology is misleading.). We just need to know if value can be casted to large type. If large
type casting fails also it means that some strange range entered not compatible with the original column like following: `OVER(ORDER BY c1 RANGE BETWEEN '3 day' PRECEDING AND '1 hour' FOLLOWING)`. In that case we return error.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] ursabot commented on pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "ursabot (via GitHub)" <gi...@apache.org>.
ursabot commented on PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384#issuecomment-1444304327
Benchmark runs are scheduled for baseline = d6ef4637447e5e7e74d9dfc92a2e7ad27adb491b and contender = 49237a21a707944aa0218c346f1c03e8d2cf478e. 49237a21a707944aa0218c346f1c03e8d2cf478e is a master commit associated with this PR. Results will be available as each benchmark for each run completes.
Conbench compare runs links:
[Skipped :warning: Benchmarking of arrow-datafusion-commits is not supported on ec2-t3-xlarge-us-east-2] [ec2-t3-xlarge-us-east-2](https://conbench.ursa.dev/compare/runs/71cab6f31970453cb6d0eaed302202ed...0dca7d1910584c3cad2b802e87e71166/)
[Skipped :warning: Benchmarking of arrow-datafusion-commits is not supported on test-mac-arm] [test-mac-arm](https://conbench.ursa.dev/compare/runs/51a8056dfdbd4f8bb0c2b5049755f8b6...a98c2131c02e497f84c8a71c72da2a75/)
[Skipped :warning: Benchmarking of arrow-datafusion-commits is not supported on ursa-i9-9960x] [ursa-i9-9960x](https://conbench.ursa.dev/compare/runs/3624134dda83455c98cffa41a48d9034...0565bf4e76754088a14757bd3148a96e/)
[Skipped :warning: Benchmarking of arrow-datafusion-commits is not supported on ursa-thinkcentre-m75q] [ursa-thinkcentre-m75q](https://conbench.ursa.dev/compare/runs/ad7b09dff9e74a6d89f8d52a3f59b2e9...989a9d7b667845bfb8981f93362d03db/)
Buildkite builds:
Supported benchmarks:
ec2-t3-xlarge-us-east-2: Supported benchmark langs: Python, R. Runs only benchmarks with cloud = True
test-mac-arm: Supported benchmark langs: C++, Python, R
ursa-i9-9960x: Supported benchmark langs: Python, R, JavaScript
ursa-thinkcentre-m75q: Supported benchmark langs: C++, Java
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] waynexia commented on a diff in pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "waynexia (via GitHub)" <gi...@apache.org>.
waynexia commented on code in PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384#discussion_r1116851452
##########
datafusion/expr/src/type_coercion.rs:
##########
@@ -62,17 +62,18 @@ pub fn is_numeric(dt: &DataType) -> bool {
)
}
-/// Determine if a DataType is Timestamp or not
+/// Determine whether the given data type `dt` is a `Timestamp`.
pub fn is_timestamp(dt: &DataType) -> bool {
matches!(dt, DataType::Timestamp(_, _))
}
-/// Determine if a DataType is Date or not
+/// Determine whether the given data type `dt` is a `Date`.
pub fn is_date(dt: &DataType) -> bool {
matches!(dt, DataType::Date32 | DataType::Date64)
}
-pub fn is_uft8(dt: &DataType) -> bool {
+/// Determine whether the given data type `dt` is a `Utf8`.
+pub fn is_utf8(dt: &DataType) -> bool {
matches!(dt, DataType::Utf8)
}
Review Comment:
Not related to this PR, but I think we missed `DataType::LargeUtf8` in #5234. cc @jackwener
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
+/// Downstream code uses this signal to treat these values as *unbounded*.
+fn coerce_scalar_range_aware(
+ target_type: &DataType,
+ value: &ScalarValue,
+) -> Result<ScalarValue> {
+ coerce_scalar(target_type, value).or_else(|err| {
+ // If type coercion fails, check if the largest type in family works:
+ if let Some(largest_type) = get_widest_type_in_family(target_type) {
+ coerce_scalar(largest_type, value).map_or_else(
+ |_| {
+ Err(DataFusionError::NotImplemented(format!(
+ "Cannot cast {:?} to {:?}",
+ value, target_type
+ )))
Review Comment:
If the largest type doesn't work then this coercion is impossible rather than unimplemented?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] avantgardnerio commented on a diff in pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "avantgardnerio (via GitHub)" <gi...@apache.org>.
avantgardnerio commented on code in PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384#discussion_r1117510855
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
+/// Downstream code uses this signal to treat these values as *unbounded*.
+fn coerce_scalar_range_aware(
+ target_type: &DataType,
+ value: &ScalarValue,
+) -> Result<ScalarValue> {
+ coerce_scalar(target_type, value).or_else(|err| {
+ // If type coercion fails, check if the largest type in family works:
+ if let Some(largest_type) = get_widest_type_in_family(target_type) {
+ coerce_scalar(largest_type, value).map_or_else(
+ |_| {
+ Err(DataFusionError::Execution(format!(
+ "Cannot cast {:?} to {:?}",
+ value, target_type
+ )))
+ },
+ |_| ScalarValue::try_from(target_type),
Review Comment:
Thanks for the clarification!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] avantgardnerio commented on a diff in pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "avantgardnerio (via GitHub)" <gi...@apache.org>.
avantgardnerio commented on code in PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384#discussion_r1117290276
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
+/// Downstream code uses this signal to treat these values as *unbounded*.
+fn coerce_scalar_range_aware(
+ target_type: &DataType,
+ value: &ScalarValue,
+) -> Result<ScalarValue> {
+ coerce_scalar(target_type, value).or_else(|err| {
+ // If type coercion fails, check if the largest type in family works:
+ if let Some(largest_type) = get_widest_type_in_family(target_type) {
+ coerce_scalar(largest_type, value).map_or_else(
+ |_| {
+ Err(DataFusionError::Execution(format!(
+ "Cannot cast {:?} to {:?}",
+ value, target_type
+ )))
+ },
+ |_| ScalarValue::try_from(target_type),
Review Comment:
I don't understand - this is ignoring the result of `coerce_scalar()` and re-parsing it as `ScalarValue::try_from(target_type)`?
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
Review Comment:
Where is the `Null` introduced? `ScalarValue::try_from(target_type)`?
Rust has `Option` and `Result` to indicate an operation was unsuccessful, is there a reason to use a valid value (Null) which callers must be aware of, vs returning one of the above to indicate an error? I could see how callers that did not read this documentation might continue on with the `Null` introducing subtle errors later in the call stack.
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
Review Comment:
Oh, it looks like this is because that's the existing behavior of `WindowFrameBound`?:
```
pub enum WindowFrameBound {
Preceding(ScalarValue),
CurrentRow,
Following(ScalarValue),
}
```
##########
datafusion/expr/src/type_coercion.rs:
##########
@@ -62,17 +62,18 @@ pub fn is_numeric(dt: &DataType) -> bool {
)
}
-/// Determine if a DataType is Timestamp or not
+/// Determine whether the given data type `dt` is a `Timestamp`.
pub fn is_timestamp(dt: &DataType) -> bool {
matches!(dt, DataType::Timestamp(_, _))
}
-/// Determine if a DataType is Date or not
+/// Determine whether the given data type `dt` is a `Date`.
pub fn is_date(dt: &DataType) -> bool {
matches!(dt, DataType::Date32 | DataType::Date64)
}
-pub fn is_uft8(dt: &DataType) -> bool {
+/// Determine whether the given data type `dt` is a `Utf8`.
+pub fn is_utf8(dt: &DataType) -> bool {
Review Comment:
Thanks for fixing this.
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
Review Comment:
The issue was not introduced by this PR, but it seems like it would be a good idea to update `try_from_string()`...
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] mustafasrepo commented on a diff in pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "mustafasrepo (via GitHub)" <gi...@apache.org>.
mustafasrepo commented on code in PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384#discussion_r1116950002
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
+/// Downstream code uses this signal to treat these values as *unbounded*.
+fn coerce_scalar_range_aware(
+ target_type: &DataType,
+ value: &ScalarValue,
+) -> Result<ScalarValue> {
+ coerce_scalar(target_type, value).or_else(|err| {
+ // If type coercion fails, check if the largest type in family works:
+ if let Some(largest_type) = get_widest_type_in_family(target_type) {
+ coerce_scalar(largest_type, value).map_or_else(
+ |_| {
+ Err(DataFusionError::NotImplemented(format!(
+ "Cannot cast {:?} to {:?}",
+ value, target_type
+ )))
Review Comment:
Indeed you are right. Fixed it. Thanks.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] mustafasrepo commented on a diff in pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "mustafasrepo (via GitHub)" <gi...@apache.org>.
mustafasrepo commented on code in PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384#discussion_r1116950002
##########
datafusion/optimizer/src/type_coercion.rs:
##########
@@ -426,95 +426,128 @@ impl ExprRewriter for TypeCoercionRewriter {
}
}
-/// Casts the ScalarValue `value` to coerced type.
-// When coerced type is `Interval` we use `parse_interval` since `try_from_string` not
-// supports conversion from string to Interval
-fn convert_to_coerced_type(
- coerced_type: &DataType,
- value: &ScalarValue,
-) -> Result<ScalarValue> {
+/// Casts the given `value` to `target_type`. Note that this function
+/// only considers `Null` or `Utf8` values.
+fn coerce_scalar(target_type: &DataType, value: &ScalarValue) -> Result<ScalarValue> {
match value {
- // In here we do casting either for NULL types or
- // ScalarValue::Utf8(Some(val)). The other types are already casted.
- // The reason is that we convert the sqlparser result
- // to the Utf8 for all possible cases. Hence the types other than Utf8
- // are already casted to appropriate type. Therefore they can be returned directly.
+ // Coerce Utf8 values:
ScalarValue::Utf8(Some(val)) => {
- // we need special handling for Interval types
- if let DataType::Interval(..) = coerced_type {
+ // When `target_type` is `Interval`, we use `parse_interval` since
+ // `try_from_string` does not support `String` to `Interval` coercions.
+ if let DataType::Interval(..) = target_type {
parse_interval("millisecond", val)
} else {
- ScalarValue::try_from_string(val.clone(), coerced_type)
+ ScalarValue::try_from_string(val.clone(), target_type)
}
}
s => {
if s.is_null() {
- ScalarValue::try_from(coerced_type)
+ // Coerce `Null` values:
+ ScalarValue::try_from(target_type)
} else {
+ // Values except `Utf8`/`Null` variants already have the right type
+ // (casted before) since we convert `sqlparser` outputs to `Utf8`
+ // for all possible cases. Therefore, we return a clone here.
Ok(s.clone())
}
}
}
}
+/// This function coerces `value` to `target_type` in a range-aware fashion.
+/// If the coercion is successful, we return an `Ok` value with the result.
+/// If the coercion fails because `target_type` is not wide enough (i.e. we
+/// can not coerce to `target_type`, but we can to a wider type in the same
+/// family), we return a `Null` value of this type to signal this situation.
+/// Downstream code uses this signal to treat these values as *unbounded*.
+fn coerce_scalar_range_aware(
+ target_type: &DataType,
+ value: &ScalarValue,
+) -> Result<ScalarValue> {
+ coerce_scalar(target_type, value).or_else(|err| {
+ // If type coercion fails, check if the largest type in family works:
+ if let Some(largest_type) = get_widest_type_in_family(target_type) {
+ coerce_scalar(largest_type, value).map_or_else(
+ |_| {
+ Err(DataFusionError::NotImplemented(format!(
+ "Cannot cast {:?} to {:?}",
+ value, target_type
+ )))
Review Comment:
Indeed you are right. Fixed it.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [arrow-datafusion] avantgardnerio merged pull request #5384: Bug fix: Window frame range value outside the type range
Posted by "avantgardnerio (via GitHub)" <gi...@apache.org>.
avantgardnerio merged PR #5384:
URL: https://github.com/apache/arrow-datafusion/pull/5384
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: github-unsubscribe@arrow.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org