[moshap-firebolt]

Databricks SQL is Spark SQL plus extensions, but each dialect parsed pieces the other did not.

• Databricks did not parse a DIV b, CREATE TABLE ... USING fmt or the LONG type. It now does; the DIV infix is shared with Spark.
• Databricks read 1L as an identifier that starts with a digit, because its numeric-prefix rule ran before the L suffix. A new dialect hook, supports_long_suffix_before_numeric_prefix, lets an uppercase L that ends the number win; Databricks and Hive opt in, MySQL does not, and 1La is still an identifier. This matches the upstream draft; lowercase 1l stays an identifier under Databricks.
• Spark parsed exists(arr, x -> ...) as an EXISTS subquery. It is now a function, as under Databricks, unless the parentheses hold a query, parenthesized or not; NOT exists(...) negates the function.
• Spark did not read backslash escapes in string literals, so 'it\'s' was unterminated. Spark reads them; \% and \_ stay as written, as they already do for Databricks.

New tests in sqlparser_spark.rs and sqlparser_databricks.rs cover each. The full suite (--all-features), fmt and clippy are clean.

🤖 Generated with Claude Code

---

Note

Medium Risk
Changes core expression parsing, tokenization, and EXISTS disambiguation across Spark, Databricks, and Hive, which could affect downstream AST shape for edge cases.

Overview
Aligns Spark and Databricks parsing where each dialect previously handled the same syntax differently.

Databricks now parses DIV (via shared parse_div_infix with Spark), CREATE TABLE ... USING, LONG as BIGINT, and the collection exists(...) function hook. 1L is tokenized as a BIGINT literal instead of a digit-prefixed identifier when supports_long_suffix_before_numeric_prefix is enabled (Databricks and Hive; 1La stays an identifier).

Spark treats exists(arr, lambda) as a function call while EXISTS (SELECT …) remains a subquery, using a new supports_exists_function dialect flag and maybe_parse_exists_expr instead of Databricks-only EXISTS lookahead. Spark also enables backslash escapes in single-quoted strings (e.g. 'it\'s').

The tokenizer checks the L suffix before the numeric-prefix identifier rule; tests cover EXISTS, DIV, literals, DDL, and escapes across dialects.

^{Reviewed by Cursor Bugbot for commit 1b3d937. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Stale Bugbot comment from a previous run.

[cursor]

Stale Bugbot comment from a previous run.

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

Want higher recall? High effort reviews run extra passes and find more bugs. A team admin can switch effort levels in the Cursor dashboard.

^{Reviewed by Cursor Bugbot for commit 90bda1a. Configure here.}