Improve Parquet case-insensitive column selection semantics

[qbacpey]

Description

Close #21864

Follow-up to #21700. That PR repeated normalize_column_path(...) at every insertion/lookup of the column-name hash containers. As suggested, this PR encapsulates that policy in a small pair of hash/equality functors that carry the case_sensitive_names policy.

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.

[qbacpey]

Benchmark — no statistically significant overhead

No statistically significant change was observed; deltas stay near zero, CIs include zero, and the spread matches same-code run-to-run noise.

comparison	light — signed mean (95% CI)	heavy — signed mean (95% CI)	within noise
branch vs main (change)	−0.4% ([−4.1, +3.3])	−0.5% ([−2.1, +1.2])	16/16
same code twice (control)	−1.3% ([−2.9, +0.3])	−0.3% ([−1.6, +0.9])	16/16

Light filter

Filter references 1 column (1 name lookup)

Heavy filter (every column referenced)

Filter references every column (O(num_cols) lookups)

Wrap-up

main (blue) and branch (orange) bars overlap at every column count in both regimes. No measurable overhead

[coderabbitai]

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

• @coderabbitai resume to resume automatic reviews.
• @coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

• ▶️ Resume reviews
• 🔍 Trigger review

📝 Walkthrough

Walkthrough

Adds case-sensitivity-aware column-path utilities and factories, replaces ad-hoc normalization with transparent container-based maps/sets in Parquet expression and hybrid-scan code, wires the new source into the build, and adds an NVBench verifying AST filter name-to-column resolution.

Changes

Case-Sensitive Column Path Utilities and Reader Integration

Layer / File(s)	Summary
Column-path helpers and build integration cpp/src/io/parquet/column_path_helpers.hpp, cpp/src/io/parquet/column_path_helpers.cpp, cpp/CMakeLists.txt, cpp/src/io/parquet/reader_impl_helpers.cpp, cpp/src/io/parquet/reader_impl_helpers.hpp	Adds normalize_column_path/are_column_paths_equal, transparent column_path_hash/column_path_equal, column_path_set/column_path_map aliases, and make_* factories; wires new column_path_helpers.cpp into the cudf target and removes local normalizer declarations.
Expression-transform integration cpp/src/io/parquet/expression_transform_helpers.hpp, cpp/src/io/parquet/expression_transform_helpers.cpp	Refactors names_from_expression and named_to_reference_converter to use column_path_set/column_path_map<size_type>, removes stored _case_sensitive_names, and switches lookups/skip-checks to use raw expression keys with the transparent containers.
Hybrid-scan filter integration cpp/src/io/parquet/experimental/hybrid_scan_helpers.cpp	Replaces per-name normalization logic in select_payload_columns and named_to_reference_converter with make_column_path_set(case_sensitive_names)-backed sets and direct membership/map lookups using schema names.
Filter name-resolution benchmark cpp/benchmarks/io/parquet/parquet_reader_metadata.cpp	Adds BM_parquet_filter_name_resolution NVBench that creates deterministic Parquet schemas and AST predicates (case-sensitive/insensitive, light/heavy) and times cudf::io::chunked_parquet_reader construction to validate name resolution.

---

Estimated code review effort

🎯 4 (Complex) | ⏱️ ~45 minutes

---

Possibly related PRs

• rapidsai/cudf#22616: Touches experimental/hybrid_scan_helpers.cpp and modifies filter/payload column matching and named-to-reference mapping; closely related to the hybrid-scan/name-resolution changes in this PR.

---

Suggested reviewers

• davidwendt
• PointKernel
• lamarrr
• mythrocks

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 36.84% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title 'Improve Parquet case-insensitive column selection semantics' accurately and concisely describes the primary change: refactoring column-name matching to encapsulate normalization in hash/equality functors rather than repeating normalize calls throughout the codebase.
Description check	✅ Passed	The description is related to the changeset, explaining that it follows up on PR #21700 by encapsulating normalization policy in hash/equality functors, which aligns with the code changes that introduce column_path_helpers with these functors.
Linked Issues check	✅ Passed	The PR successfully implements the objectives from #21864: defines hash/equality functors (column_path_hash, column_path_equal) with case-sensitivity policies, creates container aliases (column_path_set, column_path_map) using these functors, and removes scattered normalize calls throughout the codebase.
Out of Scope Changes check	✅ Passed	All changes are in scope. They focus on centralizing normalization logic via column_path_helpers (new file with helper functions and functors), updating internal implementations to use these helpers, adding a benchmark for the new functionality, and removing outdated normalization functions. No extraneous changes detected.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[PointKernel]

Very clean code. Only some non-blocking nits.

Thanks!

[coderabbitai]

🧹 Nitpick comments (1)

cpp/src/io/parquet/column_path_helpers.cpp (1)

32-35: ⚡ Quick win

Simplify the comparison lambda for clarity.

The std::equal_to<>{} call is unnecessarily verbose. Direct == comparison is more idiomatic and clearer.

♻️ Simplified comparison

   return std::equal(
     lhs.begin(), lhs.end(), rhs.begin(), [](unsigned char lhs_char, unsigned char rhs_char) {
-      return std::equal_to<>{}(std::tolower(lhs_char), std::tolower(rhs_char));
+      return std::tolower(lhs_char) == std::tolower(rhs_char);
     });

🤖 Prompt for 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.

In `@cpp/src/io/parquet/column_path_helpers.cpp` around lines 32 - 35, The lambda
inside the std::equal call can be simplified: replace the verbose
std::equal_to<>{}(std::tolower(lhs_char), std::tolower(rhs_char)) expression in
the lambda (the one taking lhs_char and rhs_char) with a direct == comparison of
the two std::tolower results; update the lambda used in the std::equal(...)
invocation in column_path_helpers.cpp accordingly for clarity.

🤖 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.

Nitpick comments:
In `@cpp/src/io/parquet/column_path_helpers.cpp`:
- Around line 32-35: The lambda inside the std::equal call can be simplified:
replace the verbose std::equal_to<>{}(std::tolower(lhs_char),
std::tolower(rhs_char)) expression in the lambda (the one taking lhs_char and
rhs_char) with a direct == comparison of the two std::tolower results; update
the lambda used in the std::equal(...) invocation in column_path_helpers.cpp
accordingly for clarity.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 4d0880da-19dd-4dcb-ba6d-e4499b47a5a5

📥 Commits

Reviewing files that changed from the base of the PR and between d2993e8 and c3ac600.

📒 Files selected for processing (9)

• cpp/CMakeLists.txt
• cpp/benchmarks/io/parquet/parquet_reader_metadata.cpp
• cpp/src/io/parquet/column_path_helpers.cpp
• cpp/src/io/parquet/column_path_helpers.hpp
• cpp/src/io/parquet/experimental/hybrid_scan_helpers.cpp
• cpp/src/io/parquet/expression_transform_helpers.cpp
• cpp/src/io/parquet/expression_transform_helpers.hpp
• cpp/src/io/parquet/reader_impl_helpers.cpp
• cpp/src/io/parquet/reader_impl_helpers.hpp

🚧 Files skipped from review as they are similar to previous changes (5)

• cpp/CMakeLists.txt
• cpp/src/io/parquet/expression_transform_helpers.cpp
• cpp/src/io/parquet/column_path_helpers.hpp
• cpp/src/io/parquet/experimental/hybrid_scan_helpers.cpp
• cpp/src/io/parquet/reader_impl_helpers.hpp

[mhaseeb123]

Approving with minor optional nits. Great work 🚀

[qbacpey]

/merge