feat(sql): close out Epic 14 — analytical-SQL parser regression audit (Story 14.6)#105
Merged
Conversation
Translate the ANSI NULLS FIRST / NULLS LAST clause to Elasticsearch's sort.missing parameter (_first / _last). Add NullOrdering AST + parser support (case-insensitive), and apply .missing(...) on the FieldSort builder across the ES7/8/9 bridge template and the hand-maintained ES6 bridge. Reject NULLS ordering on aggregation / GROUP BY ORDER BY (ES terms aggregations have no missing parameter) rather than silently dropping it. Covered by ParserSpec round-trip/reject cases, SQLQuerySpec JSON-emission assertions, and per-version integration specs (es6/jest, es6/rest, es7/rest, es8/java, es9/java). Docs updated in dql_statements.md. Closed Issue #99 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
N-ary GREATEST(e1, e2, …) / LEAST(e1, e2, …) with ANSI null handling: NULL args are ignored; result is NULL only when every arg is NULL. Numeric scope. Emitted as a right-folded nested-ternary Painless script field over Math.max / Math.min. - Nullability-aware guard: non-nullable args (literals) are not wrapped in `== null` (Painless rejects `<primitive> == null`). - validate() accepts numeric and unresolved (SQLAny) args, rejecting only definitively non-numeric types. - Help JSON, docs (dql_statements.md), parser/bridge/integration tests. Closed Issue #100 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
ANSI ranking windows over the existing top_hits enrichment path: - ROW_NUMBER / RANK / DENSE_RANK with required ORDER BY inside OVER, optional PARTITION BY - rank ordinals computed Scala-side in searchWithWindowEnrichment and injected per base-query row via (partitionKey, _id) lookup - top-N push-down via inline LIMIT N inside OVER -> top_hits.size = N (default cap = index.max_inner_result_window, 100) Review fixes: outer LIMIT no longer shrinks the window, LIMIT 0/negative guarded, dotted ORDER BY field fallback, null enrichment under SELECT alias. Closed Issue #101 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Accept and translate the six ANSI statistical aggregates (STDDEV, STDDEV_POP, STDDEV_SAMP, VARIANCE, VAR_POP, VAR_SAMP), all mapping to Elasticsearch's extended_stats aggregation via one ExtendedStatsAgg case class parameterised by a 6-variant ExtendedStatsKind ADT. - STDDEV = STDDEV_SAMP, VARIANCE = VAR_SAMP (ANSI default = sample). - Sample variants project the _sampling keys (ES 7.7+); population variants project the un-suffixed keys (ES 6+). On ES < 7.7 sample variants log a warning and return null. Gated by ElasticsearchVersion.supportsStdDevVariance. - Result key carried on ClientAggregation.aggResultField, set at conversion time and projected in extractMetrics; Stats branch gated on isEmpty to avoid fallthrough. - OVER (PARTITION BY ...) supported via the aggregation window pipeline. - Help JSON (6 entries + _index), docs (dql_statements, functions_aggregate), bridge + es6/bridge SQLQuerySpec JSON validation, testkit integration test. Closed Issue #102 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Maps both ANSI ordered-set percentile aggregates to the Elasticsearch `percentiles` aggregation (TDigest). One `PercentileAgg` case class; five syntax forms (WITHIN GROUP, OVER PARTITION BY/ORDER BY, top-level GROUP BY, (col, p) shorthand, bare whole-result-set), case-insensitive, p in [0,1] enforced at compile time. Multiple percentile calls on the same value column/partition coalesce into a single ES agg (sourceAgg delegate, split back per SQL alias at extraction). Nested `values.<key>` projection with numeric-proximity fallback for drifted fractional keys; locale-independent percent label. PERCENTILE_DISC is continuous-backed (documented). es6/bridge mirrored by hand. Closed Issue #103 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
… (Story 14.6) Gating verification story for Epic 14 (Analytical SQL, R1). Stories 14.1–14.5 added 18 new SQL keywords to the parser; this story verifies they introduce no regression and reconciles the docs/help indexes. - Cross-compile green on Scala 2.12.20 + 2.13.16 - ParserSpec, ElasticConversionSpec, es8 + es6 bridge SQLQuerySpec all green - Keyword-collision audit clean: no fixture rename required (rank/row_number bare-identifier uses exist only in dead block-comment code; live fixtures alias to rnum/rk/drk) - Case-insensitivity confirmed via (?i) TokenRegex + new lowercased-keyword ParserSpec test covering all 18 Epic-14 keywords - WindowFunctionSpec live-green on all 4 ES majors (es6/es7/es8 default JDK, es9 under JDK 17) - Docs reconciled in both repos: backfilled functions_conditional.md (GREATEST/LEAST) and functions_aggregate.md (ROW_NUMBER/RANK/DENSE_RANK) - Help index reconciled: indexed real SAFE_CAST help file + added the REGEXP_LIKE help file that resolves a dangling string/_index.json ref - scalafmtAll clean Epic 14 stories: - #99 Story 14.1 — ORDER BY ... NULLS FIRST / NULLS LAST - #100 Story 14.2 — GREATEST / LEAST conditional functions - #101 Story 14.3 — ROW_NUMBER / RANK / DENSE_RANK ranking windows - #102 Story 14.4 — STDDEV / VARIANCE statistical aggregates - #103 Story 14.5 — PERCENTILE_CONT / PERCENTILE_DISC percentiles - #104 Story 14.6 — parser regression audit (this commit) Closed Issue #104 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Gating verification story for Epic 14 (Analytical SQL, R1). Stories 14.1–14.5
added 18 new SQL keywords to the parser; this story verifies they introduce
no regression and reconciles the docs/help indexes.
bare-identifier uses exist only in dead block-comment code; live fixtures
alias to rnum/rk/drk)
ParserSpec test covering all 18 Epic-14 keywords
es9 under JDK 17)
(GREATEST/LEAST) and functions_aggregate.md (ROW_NUMBER/RANK/DENSE_RANK)
REGEXP_LIKE help file that resolves a dangling string/_index.json ref
Epic 14 stories:
Closed Issue #104