Skip to content

[Spark] Fix null expansion in MERGE with multiple clauses - #5613

Merged
OussamaSaoudi merged 1 commit into
delta-io:masterfrom
zhipengmao-db:zhipengmao-db/fix-null-expansion-multi-clause
Dec 2, 2025
Merged

[Spark] Fix null expansion in MERGE with multiple clauses#5613
OussamaSaoudi merged 1 commit into
delta-io:masterfrom
zhipengmao-db:zhipengmao-db/fix-null-expansion-multi-clause

Conversation

@zhipengmao-db

@zhipengmao-db zhipengmao-db commented Dec 2, 2025

Copy link
Copy Markdown
Contributor

Which Delta project/connector is this regarding?

  • Spark
  • Standalone
  • Flink
  • Kernel
  • Other (fill in here)

Description

This PR fixes null expansion issues in multi-clause MERGEs, when fields excluded in one MERGE clause are added to the final evolved schema by another MERGE clause, guarded by DELTA_MERGE_PRESERVE_NULL_SOURCE_STRUCTS.

Problem:
When using multiple UPDATE clauses with clauses in a single MERGE statement, NULL source structs were incorrectly expanded into non-NULL structs with all fields set to NULL.

Example:

Source table schema s: id: int, col struct<x: int, y: int>, extra: <val: int, val2: int>

Target table schema t: id: int, col struct<y: int, z: int>.

Source Table:

{id: 0, col: NULL, extra: NULL},
{id: 1, col: NULL, extra: NULL}

Target Table Before Query:

{id: 0, col: NULL},
{id: 1, col: NULL}

Query

MERGE WITH SCHEMA EVOLUTION INTO t
USING s
ON t.id = s.id
WHEN MATCHED AND id = 0 THEN UPDATE SET col = s.col
WHEN MATCHED AND id = 1 THEN UPDATE SET *

Target Table After Query - Current Behavior

{id: 0, col: NULL, extra: {val: NULL, val2: NULL},
{id: 1, col: NULL, extra: NULL}

Target Table After Query - Target Behavior

{id: 0, col: NULL, extra: NULL},
{id: 1, col: NULL, extra: NULL}

When a MERGE statement contains multiple UPDATE clauses with different clauses, e.g., one clause with UPDATE SET col = s.col and another with UPDATE *, we do not generate an action for extra for clause UPDATE SET col = s.col, while extra will be added to the final evolved schema by another clause UPDATE *.

We call generateUpdateOpsForNewTargetFields in PreprocessTableMerge to handle this case, which generates leaf-level operations for new target fields, causing null structs to be incorrectly expanded into non-null structs with null fields.

Solution:
Skip PreprocessTableMerge.generateUpdateOpsForNewTargetFields when the null expansion fix is enabled, since generateUpdateExpressions already handles evolved schema fields that do not have corresponding actions correctly by assigning them to NULL or default target expression.

How was this patch tested?

MergeIntoStructEvolutionNullnessMultiClauseTests.

Does this PR introduce any user-facing changes?

No. The change is protected by a flag disabled by default.

@zhipengmao-db zhipengmao-db changed the title [Spark] [Spark] Fix null expansion in MERGE with multiple clauses Dec 2, 2025

@tomvanbussel tomvanbussel left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks for fixing this long standing bug! LGTM.

@zhipengmao-db
zhipengmao-db force-pushed the zhipengmao-db/fix-null-expansion-multi-clause branch from 5203c7a to 2aa219d Compare December 2, 2025 19:13
@OussamaSaoudi
OussamaSaoudi merged commit 780dcb2 into delta-io:master Dec 2, 2025
20 checks passed
harperjiang pushed a commit to harperjiang/delta that referenced this pull request Dec 8, 2025
)

<!--
Thanks for sending a pull request!  Here are some tips for you:
1. If this is your first time, please read our contributor guidelines:
https://github.com/delta-io/delta/blob/master/CONTRIBUTING.md
2. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP]
Your PR title ...'.
  3. Be sure to keep the PR description updated to reflect all changes.
  4. Please write your PR title to summarize what this PR proposes.
5. If possible, provide a concise example to reproduce the issue for a
faster review.
6. If applicable, include the corresponding issue number in the PR title
and link it in the body.
-->

#### Which Delta project/connector is this regarding?
<!--
Please add the component selected below to the beginning of the pull
request title
For example: [Spark] Title of my pull request
-->

- [x] Spark
- [ ] Standalone
- [ ] Flink
- [ ] Kernel
- [ ] Other (fill in here)

## Description
This PR fixes null expansion issues in multi-clause MERGEs, when fields
excluded in one MERGE clause are added to the final evolved schema by
another MERGE clause, guarded by
`DELTA_MERGE_PRESERVE_NULL_SOURCE_STRUCTS`.

**Problem**:
When using multiple UPDATE clauses with clauses in a single MERGE
statement, NULL source structs were incorrectly expanded into non-NULL
structs with all fields set to NULL.

 Example:

**Source table schema** s: `id: int, col struct<x: int, y: int>, extra:
<val: int, val2: int>`

**Target table schema** t: `id: int, col struct<y: int, z: int>`.

**Source Table**: 
```
{id: 0, col: NULL, extra: NULL},
{id: 1, col: NULL, extra: NULL}
```

**Target Table Before Query**: 
```
{id: 0, col: NULL},
{id: 1, col: NULL}
```

**Query**
```
MERGE WITH SCHEMA EVOLUTION INTO t
USING s
ON t.id = s.id
WHEN MATCHED AND id = 0 THEN UPDATE SET col = s.col
WHEN MATCHED AND id = 1 THEN UPDATE SET *
```

**Target Table After Query - Current Behavior**

```
{id: 0, col: NULL, extra: {val: NULL, val2: NULL},
{id: 1, col: NULL, extra: NULL}
```

**Target Table After Query - Target Behavior**

```
{id: 0, col: NULL, extra: NULL},
{id: 1, col: NULL, extra: NULL}
```

When a MERGE statement contains multiple UPDATE clauses with different
clauses, e.g., one clause with `UPDATE SET col = s.col` and another with
`UPDATE *`, we do not generate an action for `extra` for clause `UPDATE
SET col = s.col`, while `extra` will be added to the final evolved
schema by another clause `UPDATE *`.

We call `generateUpdateOpsForNewTargetFields` in `PreprocessTableMerge`
to handle this case, which generates leaf-level operations for new
target fields, causing null structs to be incorrectly expanded into
non-null structs with null fields.


**Solution**:
Skip `PreprocessTableMerge.generateUpdateOpsForNewTargetFields` when the
null expansion fix is enabled, since `generateUpdateExpressions` already
handles evolved schema fields that do not have corresponding actions
correctly by assigning them to NULL or default target expression.

<!--
- Describe what this PR changes.
- Describe why we need the change.
 
If this PR resolves an issue be sure to include "Resolves #XXX" to
correctly link and close the issue upon merge.
-->

## How was this patch tested?
`MergeIntoStructEvolutionNullnessMultiClauseTests`.

<!--
If tests were added, say they were added here. Please make sure to test
the changes thoroughly including negative and positive cases if
possible.
If the changes were tested in any way other than unit tests, please
clarify how you tested step by step (ideally copy and paste-able, so
that other reviewers can test and check, and descendants can verify in
the future).
If the changes were not tested, please explain why.
-->

## Does this PR introduce _any_ user-facing changes?

<!--
If yes, please clarify the previous behavior and the change this PR
proposes - provide the console output, description and/or an example to
show the behavior difference if possible.
If possible, please also clarify if this is a user-facing change
compared to the released Delta Lake versions or within the unreleased
branches such as master.
If no, write 'No'.
-->
No. The change is protected by a flag disabled by default.
huashi-st pushed a commit to huashi-st/delta that referenced this pull request Apr 24, 2026
)

<!--
Thanks for sending a pull request!  Here are some tips for you:
1. If this is your first time, please read our contributor guidelines:
https://github.com/delta-io/delta/blob/master/CONTRIBUTING.md
2. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP]
Your PR title ...'.
  3. Be sure to keep the PR description updated to reflect all changes.
  4. Please write your PR title to summarize what this PR proposes.
5. If possible, provide a concise example to reproduce the issue for a
faster review.
6. If applicable, include the corresponding issue number in the PR title
and link it in the body.
-->

#### Which Delta project/connector is this regarding?
<!--
Please add the component selected below to the beginning of the pull
request title
For example: [Spark] Title of my pull request
-->

- [x] Spark
- [ ] Standalone
- [ ] Flink
- [ ] Kernel
- [ ] Other (fill in here)

## Description
This PR fixes null expansion issues in multi-clause MERGEs, when fields
excluded in one MERGE clause are added to the final evolved schema by
another MERGE clause, guarded by
`DELTA_MERGE_PRESERVE_NULL_SOURCE_STRUCTS`.

**Problem**:
When using multiple UPDATE clauses with clauses in a single MERGE
statement, NULL source structs were incorrectly expanded into non-NULL
structs with all fields set to NULL.

 Example:

**Source table schema** s: `id: int, col struct<x: int, y: int>, extra:
<val: int, val2: int>`

**Target table schema** t: `id: int, col struct<y: int, z: int>`.

**Source Table**: 
```
{id: 0, col: NULL, extra: NULL},
{id: 1, col: NULL, extra: NULL}
```

**Target Table Before Query**: 
```
{id: 0, col: NULL},
{id: 1, col: NULL}
```

**Query**
```
MERGE WITH SCHEMA EVOLUTION INTO t
USING s
ON t.id = s.id
WHEN MATCHED AND id = 0 THEN UPDATE SET col = s.col
WHEN MATCHED AND id = 1 THEN UPDATE SET *
```

**Target Table After Query - Current Behavior**

```
{id: 0, col: NULL, extra: {val: NULL, val2: NULL},
{id: 1, col: NULL, extra: NULL}
```

**Target Table After Query - Target Behavior**

```
{id: 0, col: NULL, extra: NULL},
{id: 1, col: NULL, extra: NULL}
```

When a MERGE statement contains multiple UPDATE clauses with different
clauses, e.g., one clause with `UPDATE SET col = s.col` and another with
`UPDATE *`, we do not generate an action for `extra` for clause `UPDATE
SET col = s.col`, while `extra` will be added to the final evolved
schema by another clause `UPDATE *`.

We call `generateUpdateOpsForNewTargetFields` in `PreprocessTableMerge`
to handle this case, which generates leaf-level operations for new
target fields, causing null structs to be incorrectly expanded into
non-null structs with null fields.


**Solution**:
Skip `PreprocessTableMerge.generateUpdateOpsForNewTargetFields` when the
null expansion fix is enabled, since `generateUpdateExpressions` already
handles evolved schema fields that do not have corresponding actions
correctly by assigning them to NULL or default target expression.

<!--
- Describe what this PR changes.
- Describe why we need the change.
 
If this PR resolves an issue be sure to include "Resolves #XXX" to
correctly link and close the issue upon merge.
-->

## How was this patch tested?
`MergeIntoStructEvolutionNullnessMultiClauseTests`.

<!--
If tests were added, say they were added here. Please make sure to test
the changes thoroughly including negative and positive cases if
possible.
If the changes were tested in any way other than unit tests, please
clarify how you tested step by step (ideally copy and paste-able, so
that other reviewers can test and check, and descendants can verify in
the future).
If the changes were not tested, please explain why.
-->

## Does this PR introduce _any_ user-facing changes?

<!--
If yes, please clarify the previous behavior and the change this PR
proposes - provide the console output, description and/or an example to
show the behavior difference if possible.
If possible, please also clarify if this is a user-facing change
compared to the released Delta Lake versions or within the unreleased
branches such as master.
If no, write 'No'.
-->
No. The change is protected by a flag disabled by default.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants