[moshap-firebolt]

CREATE VECTOR INDEX creates an index for nearest-neighbor search over an embedding column in BigQuery, SQL Server and TiDB, but no dialect parsed it:

CREATE VECTOR INDEX idx ON t(embedding) OPTIONS(index_type = 'IVF', distance_type = 'COSINE')
-- Expected: an object type after CREATE, found: VECTOR

VECTOR becomes a keyword and a modifier on CREATE INDEX, like UNIQUE, so the statement reuses CreateIndex and its existing trailers (USING, INCLUDE, WITH, expression targets). CreateIndex gains vector, or_replace, options (BigQuery OPTIONS(...)), storing (BigQuery STORING(...)) and filegroup (SQL Server ON <filegroup>). A vector index accepts WITH (...) in every dialect, for SQL Server. Plain CREATE INDEX is unchanged.

Oracle's mandatory ORGANIZATION clause and MariaDB's bare M = ... option need their own grammar and are left for a follow-up.

Round-trip tests cover the shared form on all dialects in sqlparser_common.rs, plus BigQuery's OPTIONS/STORING, SQL Server's bracket names, WITH (...) and ON [PRIMARY], and TiDB's distance-function target with USING HNSW.

This PR was prepared with AI assistance (Claude).

Dialects: BigQuery, MsSql, MySQL, Generic

[LucaCappelletti94]

At this time, this PR edits would suggest that CREATE VECTOR INDEX is a BigQuery-specific syntax, but it is not, as it is supported also by:

• Oracle SQL
• Microsoft SQL
• Therefore also Generic Dialect

Therefore, move the tests to sqlparser common and ensure that all dialects that support the syntax successfully parse it. It may be acceptable to maintainers that dialects that do not support the syntax also parse it given the current preference for permissive parsing. I suggest you also review the possible syntax variations that these other dialects may have on this syntax, so as to cover more of its variants completely.

[moshap-firebolt]

@LucaCappelletti94 - done, added tests demonstrating different syntaxes. Note - didn't implement full blown Oracle's one.

[LucaCappelletti94]

I believe any Oracle support note should be removed at this time, given that its mandatory grammar is still unsupported.

Do recall to rebase on main to avoid this branch getting stale.

[LucaCappelletti94]

None of these inputs are valid Oracle statements, and the common test already covers permissive parsing of the shared prefix across every dialect. Since correct Oracle statements are not currently supported, it does not make sense to test for incorrect ones even though this implementation parses them without erroring out, which I am unsure whether that is desirable.

Suggested change

	#[test]
	fn parse_oracle_create_vector_index() {
	// Oracle's specialized clauses (ORGANIZATION / DISTANCE / WITH TARGET
	// ACCURACY / PARAMETERS) are not yet parsed; the forms it shares with the
	// common grammar — an expression target and an `INCLUDE` list — round-trip.
	oracle().verified_stmt("CREATE VECTOR INDEX g_idx ON galaxies(embedding)");
	oracle().verified_stmt("CREATE VECTOR INDEX g_idx ON galaxies(VEC_DISTANCE(embedding))");
	oracle().verified_stmt("CREATE VECTOR INDEX g_idx ON galaxies(embedding) INCLUDE (id)");
	}

[LucaCappelletti94]

VECTOR likely should be a keyword

Suggested change

	VECTOR,
	VERBOSE,

[LucaCappelletti94]

After adding the keyword, you can now replace this with:

Suggested change

	} else if matches!(
	&self.peek_token_ref().token,
	Token::Word(w) if w.keyword == Keyword::NoKeyword && w.value.eq_ignore_ascii_case("VECTOR")
	) {
	} else if self.parse_keyword(Keyword::VECTOR) {

[LucaCappelletti94]

The ON syntax from SQL Server CREATE VECTOR INDEX grammar still does not parse, here is a red test for it:

Suggested change

	);
	);
	ms().verified_stmt(
	"CREATE VECTOR INDEX vec_idx ON [dbo].[articles]([title_vector]) WITH (METRIC = 'cosine', TYPE = 'DiskANN', MAXDOP = 8) ON [PRIMARY]",
	);

[codecov-commenter]

Codecov Report

❌ Patch coverage is 67.44186% with 14 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.58%. Comparing base (50813ee) to head (cd6a0a8).
⚠️ Report is 5 commits behind head on main.

Files with missing lines	Patch %	Lines
src/ast/spans.rs	0.00%	6 Missing ⚠️
src/parser/mod.rs	80.76%	0 Missing and 5 partials ⚠️
src/ast/ddl.rs	72.72%	0 Missing and 3 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2437      +/-   ##
==========================================
+ Coverage   82.54%   82.58%   +0.04%     
==========================================
  Files          42       42              
  Lines       35553    35651      +98     
  Branches    35553    35651      +98     
==========================================
+ Hits        29346    29443      +97     
+ Misses       2672     2665       -7     
- Partials     3535     3543       +8     

☔ 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.