fix: allow duplicate field names in table join, fix output with duplicated names #1023
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
houqp
commented
Sep 19, 2021
| "| 3 | 7 | 9 | | 7 | |", | ||
| "| 1 | 4 | 7 | | | |", | ||
| "| 2 | 5 | 8 | | | |", | ||
| "| 3 | 7 | 9 | | | |", |
There was a problem hiding this comment.
these test fixes demonstrate the incorrect join behavior that this PR also fixes.
houqp
commented
Sep 19, 2021
| @@ -188,7 +182,8 @@ impl HashJoinExec { | |||
| let right_schema = right.schema(); | |||
| check_join_is_valid(&left_schema, &right_schema, &on)?; | |||
|
|
|||
| let schema = Arc::new(build_join_schema(&left_schema, &right_schema, join_type)); | |||
| let (schema, column_indices) = | |||
| build_join_schema(&left_schema, &right_schema, join_type); | |||
There was a problem hiding this comment.
column indices are now created only once when we create the HashJoinExec node.
Dandandan
reviewed
Sep 20, 2021
Dandandan
approved these changes
Sep 20, 2021
Dandandan
reviewed
Sep 20, 2021
Dandandan
reviewed
Sep 20, 2021
houqp
commented
Sep 20, 2021
| @@ -229,38 +225,6 @@ impl HashJoinExec { | |||
| pub fn partition_mode(&self) -> &PartitionMode { | |||
| &self.mode | |||
| } | |||
|
|
|||
| /// Calculates column indices and left/right placement on input / output schemas and jointype | |||
| fn column_indices_from_schema(&self) -> ArrowResult<Vec<ColumnIndex>> { | |||
houqp
commented
Sep 20, 2021
| left: &Schema, | ||
| right: &Schema, | ||
| join_type: &JoinType, | ||
| ) -> (Schema, Vec<ColumnIndex>) { |
There was a problem hiding this comment.
this is the main change, we are now returning column index with schema in this function.
houqp
commented
Sep 20, 2021
| use std::sync::Arc; | ||
|
|
||
| use crate::logical_plan::JoinType; |
There was a problem hiding this comment.
moved join specific code into join_utils.rs
|
@Dandandan pushed refactor based on your suggestions :) PTAL. |
Dandandan
approved these changes
Sep 20, 2021
alamb
changed the title
alamb
approved these changes
Sep 20, 2021
| @@ -1375,7 +1344,7 @@ mod tests { | |||
| "| 1 | 4 | 7 | 10 | 4 | 70 |", | |||
| "| 2 | 5 | 8 | 20 | 5 | 80 |", | |||
| "| 2 | 5 | 8 | 20 | 5 | 80 |", | |||
| "| 3 | 7 | 9 | | 7 | |", | |||
| "| 3 | 7 | 9 | | | |", | |||
There was a problem hiding this comment.
I double checked this and the answer after this PR appears to be correct (there is no value 7 in b1 on the right input 👍
|
|
||
| let expected = vec
|
Thanks so much @houqp |
houqp
added
api change
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.
Rationale for this change
The current join constraint check is too strict, which causes false positives for valid join queries. Our current outer/full/right/left join logic is also not correct because it generates row values for both of the join columns from only one side of the join.
What changes are included in this PR?
check_join_set_is_validbuild_join_schemaso it can support column with duplicated names. This should also result in minor performance improvement because we are not rebuilding the same column index when executing join on every single partition anymore.Are there any user-facing changes?
no