[fix](be) Fix ANN query vector extraction to handle all constant expression types by zhiqiang-hhhh · Pull Request #62637 · apache/doris

zhiqiang-hhhh · 2026-04-20T10:33:19Z

Proposed changes

Fix extract_query_vector in ann_topn_runtime.cpp to handle all constant expression types instead of only VArrayLiteral and VCastExpr.

What problem does this PR solve?

Issue Number: close #xxx

Problem Summary:

The extract_query_vector function previously checked whether the query vector argument was specifically a VArrayLiteral or VCastExpr via dynamic_pointer_cast. When FE produced other constant expression forms (e.g., after expression rewrites), ANN index queries would fail with InvalidArgument: Constant must be ArrayLiteral or CAST to array.

The fix removes the rigid type check and instead:

Calls get_const_col() to materialize the constant expression
Unwraps ColumnConst via check_and_get_column<ColumnConst> to get the underlying data column
Validates the resulting column shape (Nullable → Array → Float32)

This approach works for any constant expression that materializes to an array column, regardless of the expression node type.

Release note

Fix ANN vector index query failure when the query vector expression is not a direct array literal or CAST expression.

Check List (For Author)

Test: Regression test
Behavior changed: No
Does this need documentation: No

hello-stephen · 2026-04-20T10:33:25Z

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

What problem was fixed (it's best to include specific error reporting information). How it was fixed.
Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
What features were added. Why was this function added?
Which code was refactored and why was this part of the code refactored?
Which functions were optimized and what is the difference before and after the optimization?

zhiqiang-hhhh · 2026-04-21T04:01:06Z

run buildall

HappenLee

LGTM

HappenLee · 2026-04-21T04:12:13Z

better we add a regression test to test if it's valid @zhiqiang-hhhh

github-actions · 2026-04-21T04:12:54Z

PR approved by at least one committer and no changes requested.

github-actions · 2026-04-21T04:12:57Z

PR approved by anyone and no changes requested.

zhiqiang-hhhh · 2026-04-21T06:35:03Z

run buildall

zhiqiang-hhhh · 2026-04-23T02:32:59Z

run buildall

zhiqiang-hhhh · 2026-04-23T03:53:17Z

run buildall

zhiqiang-hhhh · 2026-04-23T06:34:24Z

run buildall

…ession types ### What problem does this PR solve? Problem Summary: The `extract_query_vector` function in `ann_topn_runtime.cpp` only accepted `VArrayLiteral` or `VCastExpr` as the query vector expression. This caused ANN index queries to fail when FE produced other constant expression types (e.g., `ColumnConst` wrapping). The fix removes the rigid type check and instead uses `get_const_col()` + `ColumnConst` unwrapping, which works for any constant expression that materializes to an array column. ### Release note Fix ANN vector index query failure when the query vector expression is not a direct array literal or CAST expression. ### Check List (For Author) - Test: Regression test - Behavior changed: No - Does this need documentation: No

### What problem does this PR solve? Problem Summary: Add BE unit tests and regression coverage for ANN query vectors produced by constant expressions such as `array_repeat` and `array_with_constant`, and update the NULL-array expectation to match the new extraction behavior. ### Release note None ### Check List (For Author) - Test: Unit Test and Regression test - Behavior changed: No - Does this need documentation: No

### What problem does this PR solve? Issue Number: None Related PR: None Problem Summary: Fix the ANN query vector unit test build failure caused by a stale MockConstVExpr override that no longer matches the current VExpr interface. ### Release note None ### Check List (For Author) - Test: No need to test (compile-only fix; full build not rerun in this session) - Behavior changed: No - Does this need documentation: No

### What problem does this PR solve? Issue Number: None Related PR: None Problem Summary: Fix the ANN extract_query_vector unit test compile failure in CI by updating the mock expression to match the current VExpr execution interface. ### Release note None ### Check List (For Author) - Test: Unit Test - Built extract_query_vector_test.cpp object and vector_search_test in be/ut_build_ASAN - Behavior changed: No - Does this need documentation: No

zhiqiang-hhhh · 2026-04-23T08:14:05Z

run buildall

### What problem does this PR solve? Issue Number: None Related PR: None Problem Summary: Fix the remaining BE UT compile failure in CI by matching MockConstVExpr to the current VExpr selector signature, where execute_column_impl takes const Selector*. ### Release note None ### Check List (For Author) - Test: Manual test - Verified latest TeamCity failure and aligned the mock signature with the VExpr API used in CI - Behavior changed: No - Does this need documentation: No

zhiqiang-hhhh · 2026-04-23T08:46:34Z

run buildall

hello-stephen · 2026-04-23T10:29:48Z

BE UT Coverage Report

Increment line coverage 100.00% (4/4) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	53.33% (20370/38194)
Line Coverage	36.88% (192013/520619)
Region Coverage	33.18% (149271/449848)
Branch Coverage	34.30% (65316/190404)

zhiqiang-hhhh · 2026-04-23T10:40:52Z

run cloud_p0

hello-stephen · 2026-04-23T11:11:01Z

BE Regression && UT Coverage Report

Increment line coverage 100.00% (4/4) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	71.22% (26661/37433)
Line Coverage	53.64% (278530/519230)
Region Coverage	46.93% (213705/455328)
Branch Coverage	50.27% (96837/192617)

hello-stephen · 2026-04-23T11:43:26Z

BE Regression && UT Coverage Report

Increment line coverage 100.00% (4/4) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	71.22% (26661/37433)
Line Coverage	53.64% (278530/519230)
Region Coverage	46.93% (213705/455328)
Branch Coverage	50.27% (96837/192617)

hello-stephen · 2026-04-23T11:50:39Z

BE Regression && UT Coverage Report

Increment line coverage 100.00% (4/4) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	71.22% (26660/37433)
Line Coverage	53.63% (278482/519230)
Region Coverage	46.92% (213620/455328)
Branch Coverage	50.27% (96827/192617)

HappenLee

LGTM

github-actions · 2026-04-23T12:04:07Z

PR approved by at least one committer and no changes requested.

…ession types (#62637) ## Proposed changes Fix `extract_query_vector` in `ann_topn_runtime.cpp` to handle all constant expression types instead of only `VArrayLiteral` and `VCastExpr`. ### What problem does this PR solve? Issue Number: close #xxx Problem Summary: The `extract_query_vector` function previously checked whether the query vector argument was specifically a `VArrayLiteral` or `VCastExpr` via `dynamic_pointer_cast`. When FE produced other constant expression forms (e.g., after expression rewrites), ANN index queries would fail with `InvalidArgument: Constant must be ArrayLiteral or CAST to array`. The fix removes the rigid type check and instead: 1. Calls `get_const_col()` to materialize the constant expression 2. Unwraps `ColumnConst` via `check_and_get_column<ColumnConst>` to get the underlying data column 3. Validates the resulting column shape (Nullable → Array → Float32) This approach works for any constant expression that materializes to an array column, regardless of the expression node type. ### Release note Fix ANN vector index query failure when the query vector expression is not a direct array literal or CAST expression. ### Check List (For Author) - Test: Regression test - Behavior changed: No - Does this need documentation: No

HappenLee previously approved these changes Apr 21, 2026

View reviewed changes

github-actions Bot added the approved Indicates a PR has been approved by one committer. label Apr 21, 2026

github-actions Bot added the reviewed label Apr 21, 2026

zhiqiang-hhhh dismissed HappenLee’s stale review via c3f3cfa April 21, 2026 06:34

github-actions Bot removed the approved Indicates a PR has been approved by one committer. label Apr 21, 2026

zhiqiang-hhhh added 5 commits April 23, 2026 16:13

format

c3ecfc6

zhiqiang-hhhh force-pushed the fix-ann-const-literal-extract branch from eb2cd14 to 66ba080 Compare April 23, 2026 08:13

HappenLee approved these changes Apr 23, 2026

View reviewed changes

github-actions Bot added the approved Indicates a PR has been approved by one committer. label Apr 23, 2026

yiguolei added the dev/4.1.x label Apr 28, 2026

yiguolei merged commit 430781b into apache:master Apr 29, 2026
30 of 32 checks passed

github-actions Bot mentioned this pull request Apr 29, 2026

branch-4.1: [fix](be) Fix ANN query vector extraction to handle all constant expression types #62637 #62926

Open

zhiqiang-hhhh deleted the fix-ann-const-literal-extract branch April 29, 2026 06:15

Conversation

zhiqiang-hhhh commented Apr 20, 2026

Proposed changes

What problem does this PR solve?

Release note

Check List (For Author)

Uh oh!

hello-stephen commented Apr 20, 2026

Uh oh!

zhiqiang-hhhh commented Apr 21, 2026

Uh oh!

HappenLee left a comment

Choose a reason for hiding this comment

Uh oh!

HappenLee commented Apr 21, 2026

Uh oh!

github-actions Bot commented Apr 21, 2026

Uh oh!

github-actions Bot commented Apr 21, 2026

Uh oh!

zhiqiang-hhhh commented Apr 21, 2026

Uh oh!

zhiqiang-hhhh commented Apr 23, 2026

Uh oh!

zhiqiang-hhhh commented Apr 23, 2026

Uh oh!

zhiqiang-hhhh commented Apr 23, 2026

Uh oh!

zhiqiang-hhhh commented Apr 23, 2026

Uh oh!

zhiqiang-hhhh commented Apr 23, 2026

Uh oh!

hello-stephen commented Apr 23, 2026

BE UT Coverage Report

Uh oh!

zhiqiang-hhhh commented Apr 23, 2026

Uh oh!

hello-stephen commented Apr 23, 2026

BE Regression && UT Coverage Report

Uh oh!

hello-stephen commented Apr 23, 2026

BE Regression && UT Coverage Report

Uh oh!

hello-stephen commented Apr 23, 2026

BE Regression && UT Coverage Report

Uh oh!

HappenLee left a comment

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Apr 23, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants