feat(datafusion): Add LIKE predicate pushdown for StartsWith patterns #2014
+262
−2
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
Contributor
liurenjie1024
left a comment
liurenjie1024
left a comment
There was a problem hiding this comment.
Thanks @viirya for this pr!
| assert_eq!(predicate, None); | ||
| } | ||
|
|
||
| #[test] |
Contributor
There was a problem hiding this comment.
We recently added support for sqllogictests, see https://github.com/liurenjie1024/iceberg-rust/blob/666a9fe1aaf1692583d6f44e4f7a1d52a688b217/crates/sqllogictest/testdata/schedules/df_test.toml#L19
It would be better if we also include such sql logic tests in spite of ut.
Member
Author
There was a problem hiding this comment.
Okay. Added test in sqllogictests.
| }) => { | ||
| // Only support simple prefix patterns (e.g., 'prefix%') | ||
| // Escape characters and case-insensitive matching are not supported yet | ||
| if escape_char.is_some() || *case_insensitive { |
Contributor
There was a problem hiding this comment.
IIRC, iceberg's starts with is case sensitive.
Member
Author
There was a problem hiding this comment.
I updated the document here.
Implement pushdown for LIKE predicates with simple prefix patterns (e.g., 'prefix%') to Iceberg's StartsWith operator. This optimization allows filtering to be performed at the storage layer, significantly improving query performance for prefix searches. Changes: - Add support for Expr::Like in predicate conversion - Convert LIKE 'prefix%' patterns to StartsWith operator - Convert NOT LIKE 'prefix%' patterns to NotStartsWith operator - Handle edge cases: empty prefix, unicode, special characters - Reject complex patterns that cannot be pushed down (wildcards in middle, underscore, ILIKE) - Add 8 comprehensive unit tests covering various scenarios Implementation details: - Only simple prefix patterns ending with % are converted - Patterns with % or _ wildcards in the prefix are not pushed down - Case-insensitive LIKE (ILIKE) is not supported for pushdown - Escape characters are not supported for pushdown - Works seamlessly with other predicates in AND/OR expressions Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
… clarify case sensitivity Address review comments on PR apache#2014: 1. Add comprehensive sqllogictest for LIKE predicate pushdown 2. Clarify that Iceberg's StartsWith operator is case-sensitive Changes: - Add like_predicate_pushdown.slt test file with comprehensive test coverage: * Verify LIKE 'prefix%' and NOT LIKE 'prefix%' predicates push down to IcebergTableScan * Test case-sensitive behavior (e.g., 'Al%' matches 'Alice' but not 'alice') * Test empty prefix, single character prefix * Test LIKE combined with other predicates (AND/OR expressions) - Add test to df_test.toml schedule - Update code comments to explicitly note that Iceberg's StartsWith is case-sensitive, explaining why ILIKE cannot be pushed down The sqllogictests verify that: - LIKE predicates are correctly converted to StartsWith/NotStartsWith - Case sensitivity is preserved (Iceberg StartsWith is case-sensitive) - Predicates are properly pushed down to the storage layer - Complex patterns that cannot be pushed down remain in FilterExec Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
viirya
force-pushed
the
feat/datafusion-string-predicate-pushdown
branch
from
99e1b6a to
09d52d6
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Implement pushdown for LIKE predicates with simple prefix patterns (e.g., 'prefix%') to Iceberg's StartsWith operator. This optimization allows filtering to be performed at the storage layer, significantly improving query performance for prefix searches.
Changes:
Implementation details:
Which issue does this PR close?
What changes are included in this PR?
Are these changes tested?