[SPARK-52791][PS] Fix error when inferring a UDT with a null first element #51475
Conversation
This file contains hidden or 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
petern48
changed the title
ueshin
reviewed
Jul 22, 2025
Comment on lines
365
to
367
| first_idx = pser.first_valid_index() | ||
| if first_idx is not None and hasattr(pser.loc[first_idx], "__UDT__"): | ||
| return pser.loc[first_idx].__UDT__ |
There was a problem hiding this comment.
Good catch! but using .loc sounds dangerous, e.g.:
>>> pser = pd.Series([None,None,3], index=[1,0,1])
>>> i = pser.first_valid_index()
>>> pser.loc[i]
1 NaN
1 3.0
dtype: float64How about using notnull()?
Suggested change
| first_idx = pser.first_valid_index() | |
| if first_idx is not None and hasattr(pser.loc[first_idx], "__UDT__"): | |
| return pser.loc[first_idx].__UDT__ | |
| notnull = pser[pser.notnull()] | |
| if hasattr(notnull.iloc[0], "__UDT__"): | |
| return notnull.iloc[0].__UDT__ |
There was a problem hiding this comment.
Nice! Accepted the change and all tests passing
Co-authored-by: Takuya UESHIN <[email protected]>
HyukjinKwon
approved these changes
Jul 23, 2025
|
Merged to master, branch-4.0 and branch-3.5. |
|
@HyukjinKwon Can we backport this to 3.5 too? |
|
👌 |
HyukjinKwon
pushed a commit
that referenced
this pull request
Jul 23, 2025
…ement
I modified the udt condition to check the first non-null element instead of the first element (which might be null).
```
import pyspark.pandas as ps
from pyspark.ml.linalg import SparseVector
sparse_values = {0: 0.1, 1: 1.1}
ps_series = ps.Series([None, SparseVector(1, \{0: 1.2}), SparseVector(1, \{0: 3})])
```
Error:
```
pyarrow.lib.ArrowInvalid: Could not convert SparseVector(1, {0: 1.2}) with type SparseVector: did not recognize Python value type when inferring an Arrow data type
```
This should work as normal, but it fails because the first element is None
Yes, previously it would error, but now it works properly. This is a behavior change from all previous spark versions, and should probably be backported.
Added a test
No
Closes #51475 from petern48/fix_infer_spark_type.
Authored-by: Peter Nguyen <[email protected]>
Signed-off-by: Hyukjin Kwon <[email protected]>
(cherry picked from commit 5182eb4)
Signed-off-by: Hyukjin Kwon <[email protected]>
HyukjinKwon
pushed a commit
that referenced
this pull request
Jul 23, 2025
…ement
I modified the udt condition to check the first non-null element instead of the first element (which might be null).
```
import pyspark.pandas as ps
from pyspark.ml.linalg import SparseVector
sparse_values = {0: 0.1, 1: 1.1}
ps_series = ps.Series([None, SparseVector(1, \{0: 1.2}), SparseVector(1, \{0: 3})])
```
Error:
```
pyarrow.lib.ArrowInvalid: Could not convert SparseVector(1, {0: 1.2}) with type SparseVector: did not recognize Python value type when inferring an Arrow data type
```
This should work as normal, but it fails because the first element is None
Yes, previously it would error, but now it works properly. This is a behavior change from all previous spark versions, and should probably be backported.
Added a test
No
Closes #51475 from petern48/fix_infer_spark_type.
Authored-by: Peter Nguyen <[email protected]>
Signed-off-by: Hyukjin Kwon <[email protected]>
(cherry picked from commit 5182eb4)
Signed-off-by: Hyukjin Kwon <[email protected]>
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?
I modified the udt condition to check the first non-null element instead of the first element (which might be null).
Why are the changes needed?
Error:
This should work as normal, but it fails because the first element is None
Does this PR introduce any user-facing change?
Yes, previously it would error, but now it works properly. This is a behavior change from all previous spark versions, and should probably be backported.
How was this patch tested?
Added a test
Was this patch authored or co-authored using generative AI tooling?
No