Fix parquet hybrid scan dictionary query bugs

[pmattione-nvidia]

For a predicate with more than MAX_INLINE_LITERALS (2) literals, query_dictionaries decided a column chunk had no dictionary page by testing its cuco set for zero slots. The problem is that make_valid_extent rounds every capacity up to at least one bucket, so that test never fired and the chunk was instead probed against a set that was never built — every literal came back absent and the row group was wrongly pruned, dropping rows. Detect the empty dictionary from the chunk's decoded value count instead, which is what the few-literals kernel already does.

Also add a missing synchronization between initializing and further writing to results, as threads initializing could stomp on threads filling it below.

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[coderabbitai]

📝 Walkthrough

Summary by CodeRabbit

• Bug Fixes

  • Improved dictionary-based filtering for empty dictionaries.
  • Ensured reliable evaluation of fixed-width literal filters during parallel processing.
  • Corrected filtering behavior for OR expressions with multiple literals, including row groups with and without dictionary pages.

• Tests

  • Added coverage for retaining or pruning row groups based on dictionary contents.

Walkthrough

The change updates Parquet dictionary filtering to skip empty dictionaries safely and synchronize predicate initialization before decoding. New tests cover three-literal OR filters with absent and present dictionary values.

Changes

Dictionary filtering

Layer / File(s)	Summary
Dictionary query and decoding updates cpp/src/io/parquet/experimental/dictionary_page_filter.cu	query_dictionaries computes value counts before hash-set probing and skips empty dictionaries. evaluate_few_fixed_width_literals adds synchronization before dictionary decoding.
Multi-literal OR coverage cpp/tests/io/experimental/hybrid_scan_filters_test.cpp	Tests cover three-literal OR filters where queried values are absent from or partially present in a row-group dictionary.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Suggested labels: bug, cuIO

Suggested reviewers: vuule, bdice, mhaseeb123

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Title check	✅ Passed	The title clearly identifies the main change: fixing Parquet hybrid scan dictionary query bugs.
Description check	✅ Passed	The description directly explains the dictionary detection bug, synchronization fix, regression tests, and affected predicates.

✨ Finishing Touches 💡 1

🛠️ Fix failing CI checks 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@cpp/tests/io/experimental/hybrid_scan_filters_test.cpp`:
- Around line 1891-1939: Add a test case alongside the existing multi-literal
filter cases in filter_row_groups_with_dictionaries using a non-ASCII UTF-8
value stored in a dictionary and the same value as a cudf::string_scalar
literal. Build the filter expression and reader options consistently with the
surrounding tests, then assert the expected row groups to exercise hashing and
equality for UTF-8 dictionary strings.

🪄 Autofix

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: aa7d5704-45ba-432b-b40a-43f1146a765e

📥 Commits

Reviewing files that changed from the base of the PR and between 463fba2 and a5133a2.

📒 Files selected for processing (2)

• cpp/src/io/parquet/experimental/dictionary_page_filter.cu
• cpp/tests/io/experimental/hybrid_scan_filters_test.cpp