[SPARK-38176][SQL] ANSI mode: allow implicitly casting String to other simple types #35478
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
cloud-fan
reviewed
Feb 10, 2022
docs/sql-ref-ansi-compliance.md
Outdated
| | Double | Double | | ||
| | Date | Date -> Timestamp | | ||
| | Timestamp | Timestamp | | ||
| | String | String, Long -> Double, Date -> Timestamp, Boolean, Interval, Binary | |
There was a problem hiding this comment.
I find this line a bit hard to understand. Shall we add some explanation below?
cloud-fan
reviewed
Feb 10, 2022
| Some(DoubleType) | ||
|
|
||
| // If the target type is any Decimal type, convert the String type literal as Double type. | ||
| case (StringType, DecimalType) if isInputFoldable => | ||
| // If the target type is any Decimal type, convert the String type literal as the default |
There was a problem hiding this comment.
nit: not literal anymore
There was a problem hiding this comment.
please update other similar comments in this file.
cloud-fan
reviewed
Feb 10, 2022
sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/analysis/AnsiTypeCoercion.scala
Outdated
Show resolved
Hide resolved
…ysis/AnsiTypeCoercion.scala Co-authored-by: Wenchen Fan <cloud0fan@gmail.com>
cloud-fan
approved these changes
Feb 11, 2022
|
Merging to master for unblocking https://issues.apache.org/jira/browse/SPARK-38154 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What changes were proposed in this pull request?
Compared to the default behavior, the current ANSI type coercion rules don't allow the following cases:
str_col > date'2022-01-01'This PR is to remove the limitation. After changes, the String type can be implicit cast as Long/Double/Date/Timestamp/Boolean/Binary.
Note that Byte/Short/Int is not on the precedent list of String:
str_col > 1will becomecast(str_col as long) > 1L. So that we can avoid string parsing error if the string is out of the range of Byte/Short/Int in comparison/arithmetic/union operations.The design applies to Float/Decimal (especially Decimal), for SQL operators containing Float/Decimal and String, the type coercion system will convert both as Double.
Why are the changes needed?
The purpose of the current limitation is to prevent potential String parsing errors under ANSI mode. However, after doing research among real-world Spark SQL queries, I find that many users are actually using String as Date/Timestamp/Numeric/etc in their queries. For example, the purpose of query
where str_col > date'2022-01-01'is quite obvious, but users have to rewrite it aswhere cast(str_col as date) > date'2022-01-01'under ANSI mode.To make the migration to ANSI mode easier, I suggest removing this limitation. Let's treat it as an extension in our SQL dialect.
Does this PR introduce any user-facing change?
Yes, allow implicitly casting String to other simple types under ANSI mode
How was this patch tested?
Unit tests