feat(tesseract): Support use_original_sql_pre_aggregations in rollup

[ovr]

No description provided.

[claude]

Claude finished @ovr's task in 1m 2s —— View job

---

Code Review

No prior review threads. Verdict: LGTM — 0 high, 0 medium, 2 low.

Details (click to expand)

• Read core Rust diff (query_properties, query_properties_compiler, top_level_planner, bridge)
• Read JS side (BaseQuery.js, PreAggregations.ts) for context
• Read integration tests
• Verify naming/threading consistency

Summary

This PR threads the existing JS-side useOriginalSqlPreAggregationsInPreAggregation flag into the Tesseract (Rust) planner. The behavior mirrors BaseQuery.js:2639 exactly: when set, a rollup-build query also collects originalSql pre-aggregations and substitutes them in, so a rollup is sourced from the cube's originalSql table instead of the raw cube SQL.

The change is small, localized, and well-tested:

• Bridge struct, compiler, builder, and PartialEq are all updated symmetrically — no field forgotten.
• The new logic in top_level_planner.rs is just an OR-extension of the existing !is_pre_aggregation_query() gate — minimal blast radius.
• Two new integration tests (positive + negative) directly assert the SQL substitution effect, plus a separate lambda_union test for an adjacent fix.
• Test fixtures (MockBaseQueryOptions, YamlBaseQueryOptions, TestContext) are updated; bridge coverage test gets the new field added.

Code quality

• The doc comment on query_properties.rs:170-172 is well-scoped and explains the why, not the what — good.
• The inline comment in top_level_planner.rs:56-58 is similarly useful for explaining the OR condition.
• Naming matches the JS source-of-truth (useOriginalSqlPreAggregationsInPreAggregation → use_original_sql_pre_aggregations_in_pre_aggregation); long but precedent-consistent.

Minor observations (low)

1. top_level_planner.rs:53 — the blank line between the let templates = … statement and the PhysicalPlanBuilder::new line was removed in this diff. Not wrong, but it was an unrelated formatting change; small nit.

2. Test naming test_rollup_lambda_cross_cube_union_aliases in sql_generation.rs is unrelated to the PR title (it's for a lambda_union matcher concern, not useOriginalSqlPreAggregations). Worth either splitting into a separate PR or noting in the PR description, since the title says only "Support use_original_sql_pre_aggregations in rollup." Not blocking.

Security / performance

• No security surface touched (boolean propagation only).
• No new query paths; the same OriginalSqlCollector is reused. No perf concern.

Tests

• Coverage looks solid: positive case asserts foo__main appears and foo_table does not; negative case asserts the inverse. Both assertions on both directions — robust against accidental no-op regressions.

[claude]

Claude finished @ovr's task in 43s —— View job

---

Code Review (re-run after sync)

Verdict: LGTM — 0 high, 0 medium, 1 low.

• Diff vs origin/master re-read
• Verified chore: drop test removed the unrelated test_rollup_lambda_cross_cube_union_aliases flagged previously
• Verified chore: fix build added the new field to object-bridges-coverage.test.ts
• Re-checked Rust threading (bridge → compiler → properties → planner) for symmetry

Details (click to expand)

Summary

This PR threads the existing JS-side useOriginalSqlPreAggregationsInPreAggregation flag into the Tesseract (Rust) planner. Behavior mirrors BaseQuery.js:2639: when set, a rollup-build query also collects originalSql pre-aggregations and substitutes them in, so a rollup is sourced from the cube's originalSql table instead of the raw cube SQL.

Since the previous review:

• The unrelated test_rollup_lambda_cross_cube_union_aliases test was removed (152fc43), so the diff now matches the PR title scope.
• object-bridges-coverage.test.ts was updated (1f2f2b5) to include the new field in the bridge coverage list — required for that test to pass.

The change is small, localized, well-tested:

• Bridge struct, compiler, builder, PartialEq, and the MockBaseQueryOptions/YamlBaseQueryOptions/TestContext fixtures are all updated symmetrically — no field forgotten.
• The new logic in top_level_planner.rs is just an OR-extension of the existing !is_pre_aggregation_query() gate — minimal blast radius.
• Two integration tests (positive + negative) directly assert the SQL substitution effect.

Code quality

• Doc comment on query_properties.rs:170-172 explains the why, not the what — good.
• Inline comment in top_level_planner.rs:56-58 similarly useful for explaining the OR condition.
• Naming consistent with the JS source-of-truth (useOriginalSqlPreAggregationsInPreAggregation → use_original_sql_pre_aggregations_in_pre_aggregation).

Minor observation (low)

1. top_level_planner.rs:53 — the blank line between the let templates = … statement and PhysicalPlanBuilder::new was removed. Unrelated formatting nit; not blocking.

Security / performance

• No security surface touched (boolean propagation only).
• No new query paths; same OriginalSqlCollector is reused. No perf concern.

Tests

• Coverage looks solid: positive case asserts foo__main appears and foo_table does not; negative case asserts the inverse. Both directions covered — robust against accidental no-op regressions.

• [Branch](https://github.com/cube-js/cube/tree/feat/tesseract-use-original-pre-agg)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 79.19%. Comparing base (e317395) to head (152fc43).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #11146      +/-   ##
==========================================
- Coverage   83.68%   79.19%   -4.50%     
==========================================
  Files         257      474     +217     
  Lines       79384    96895   +17511     
  Branches        0     3563    +3563     
==========================================
+ Hits        66434    76732   +10298     
- Misses      12950    19645    +6695     
- Partials        0      518     +518     

Flag	Coverage Δ
cube-backend	58.80% <ø> (?)
cubesql	83.68% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.