feat(sql): HAVING — post-aggregation filter (SQLR-52)

[joaoh82]

Summary

WHERE is the pre-aggregate filter; HAVING is the post-aggregate one. This lifts the NotImplemented and closes the Phase 9e aggregates story (SQLR-52).

SELECT dept, COUNT(*) FROM emp GROUP BY dept HAVING COUNT(*) > 1;
SELECT dept, SUM(salary) AS total FROM emp GROUP BY dept HAVING total > 100000;
SELECT dept FROM emp GROUP BY dept HAVING COUNT(*) > 1 AND SUM(salary) > 100;

Design

Parser (src/sql/parser/select.rs):

• SelectQuery grows having: Option<Expr>, carried raw from sqlparser exactly like WHERE's selection.
• HAVING without GROUP BY is rejected with a typed NotImplemented ("use WHERE for row-level filters or restructure with a subquery"). SQLite's degenerate single-group form isn't worth the executor branch in v0.
• HAVING + JOIN stays covered by the existing GROUP-BY-over-JOIN rejection (SQLR-6 is the follow-up increment).

Executor (src/sql/executor.rs):

• lower_having_expr rewrites aggregate calls in the HAVING tree into identifiers naming their output slot (SUM(salary) → "SUM(salary)"), registering hidden trailing projection slots for aggregates and GROUP BY keys referenced only in HAVING — so SELECT dept FROM emp GROUP BY dept HAVING COUNT(*) > 1 computes the count without projecting it.
• A new GroupRowScope (third RowScope impl) resolves those identifiers against each group's output row through the shared expression evaluator — comparisons, AND/OR/NOT, arithmetic, IS NULL, LIKE, IN (list) all work in HAVING for free, with the same NULL-as-false collapse WHERE applies (design-decisions §13).
• filter_groups_by_having runs after aggregation and before DISTINCT / ORDER BY / LIMIT; hidden slots are stripped after filtering so they never leak into output width.

Tests

Engine suite 612 → 625 (+13), all green:

• COUNT(*) / SUM thresholds; aggregate alias (HAVING total > 100); lowercase call form
• Aggregate only in HAVING (hidden slot); GROUP BY key only in HAVING
• Compound AND predicate; composition with ORDER BY + LIMIT; all-groups-excluded
• NULL aggregate collapses to false (all-NULL SUM group dropped)
• HAVING without GROUP BY → typed error; out-of-scope column → typed error
• All four JOIN flavors + GROUP BY + HAVING → clean NotImplemented, no wrong results

Verification

• CI shape: cargo build / test / fmt --check / clippy (no new warnings) / doc — all pass
• REPL smoke (in-memory + file-backed with close/reopen round-trip) — correct rows, clean rejection message
• web/ npm run build — pass (three web files updated where they claimed HAVING was unsupported)

Docs

supported-sql.md (new "HAVING semantics" section + syntax block), sql-engine.md aggregation pipeline, roadmap.md shipped entry, README extras list, design-decisions.md §13 wording, web docs page / sql-ref / roadmap component.

Ref: SQLR-52

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
rust-sqlite	Ready	Preview, Comment	Jun 9, 2026 9:57pm