SNOW-1748140: Modify schema_expression to be structured type aware. #2659
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
sfc-gh-jrose
added
the
NO-CHANGELOG-UPDATES
sfc-gh-jrose
requested review from
sfc-gh-jdu,
sfc-gh-yixie and
sfc-gh-yuwang
sfc-gh-aalam
approved these changes
Nov 22, 2024
sfc-gh-jdu
reviewed
Nov 22, 2024
| if data_type.structured: | ||
| key = schema_expression(data_type.key_type, is_nullable) | ||
| value = schema_expression(data_type.value_type, is_nullable) | ||
| return f"object_construct_keep_null({key}, {value}) :: {convert_sp_to_sf_type(data_type)}" |
There was a problem hiding this comment.
should we determine whether keeping null values based on is_nullable?
There was a problem hiding this comment.
If we don't keep nulls and either the key or value gets evaluated to a NULL :: type statement then that field would be dropped from the schema altogether. For this reason I think we always want nulls.
| assert table.union(table).schema == expected_schema | ||
| # Functions used in schema generation don't respect nested nullability so compare query string instead | ||
| non_null_union = non_null_table.union(non_null_table) | ||
| assert non_null_union._plan.schema_query == ( |
There was a problem hiding this comment.
can you also create a test case for nested array and object? like to_array(... to_array(...))
8 tasks
sfc-gh-jdu
approved these changes
Nov 25, 2024
sfc-gh-jrose
deleted the
jrose_snow_1748140_schema_expression_for_struct_types
branch
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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.
Which Jira issue is this PR addressing? Make sure that there is an accompanying issue to your PR.
Fixes SNOW-1748140
Fill out the following pre-review checklist:
Please describe how your code solves the related issue.
This pull request includes several changes to improve the handling of structured data types in the
schema_expressionfunction to support structured types.Schema expression has different case for when a field is nullable and when it is not. This change modifies the nullable case to skip the semi-structured branches of the if statement and instead use the default case so that
convert_sp_to_sf_typedoes the work to specify the schema instead.For non-nullable fields it relies on recursive calls to schema_expression in order to get good default values for the various nestsed data types and then casts a relevant data structured to the correct schema. Note that non-nullable columns do not have nullability respected in their child fields which is a limitation of the way we infer schemas using the dummy query approach.
While writing some test cases for this I also found that the nullability of child fields was not set correctly when parsing metadata so I added that in type_utils as well.
I've opened two new bugs as a result of this change:
SNOW-1819531 - Large query breakdown does not appear to work correctly with the structured types.
SNOW-1819428 - When calling create_dataframe with a schema that contains a StructType column, the child fields do not have their nullability respected. I suspect this due to a similar limitation in how we generate schema strings.