fix: support spark type-hinted number literals

CHANGELOG.md

@@ -4,6 +4,11 @@ All notable changes to this project will be documented in this file.
 
 ## [Unreleased]
 
+### Formatting Changes and Bug Fixes
+
+- sqlfmt no longer adds a space between the function name and parens for `filter()` and `isnull()` (but it also permits `filter ()` and `isnull ()` to support dialects where those are operators, not function names) ([#641](https://github.com/tconbeer/sqlfmt/issues/641) - thank you [@williamscs](https://github.com/williamscs) and [@hongtron](https://github.com/hongtron)!).
+- sqlfmt now supports Spark type-hinted numeric literals like `32y` and `+3.2e6bd` and will not introduce a space between the digits and their type suffix ([#640](https://github.com/tconbeer/sqlfmt/issues/640) - thank you [@ShaneMazur](https://github.com/ShaneMazur)!).
+
 ## [0.23.3] - 2024-11-12
 
 ### Bug Fixes

src/sqlfmt/rules/core.py

@@ -65,11 +65,22 @@
         action=actions.raise_sqlfmt_bracket_error,
     ),
     Rule(
+        # see https://spark.apache.org/docs/latest/sql-ref-literals.html#integral-literal-syntax
+        name="spark_int_literals",
+        priority=349,
+        pattern=group(
+            r"(\+|-)?\d+(l|s|y)",
+        ),
+        action=actions.handle_number,
+    ),
+    Rule(
+        # the (bd|d|f) groups add support for Spark fractional literals
+        # https://spark.apache.org/docs/latest/sql-ref-literals.html#fractional-literals-syntax
         name="number",
         priority=350,
         pattern=group(
-            r"(\+|-)?\d+(\.\d*)?(e(\+|-)?\d+)?",
-            r"(\+|-)?\.\d+(e(\+|-)?\d+)?",
+            r"(\+|-)?\d+(\.\d*)?(e(\+|-)?\d+)?(bd|d|f)?",
+            r"(\+|-)?\.\d+(e(\+|-)?\d+)?(bd|d|f)?",
         ),
         action=actions.handle_number,
     ),

tests/data/unformatted/132_spark_number_literals.sql

@@ -0,0 +1,54 @@
+-- source: https://spark.apache.org/docs/latest/sql-ref-literals.html#integral-literal-syntax
+-- see: https://github.com/tconbeer/sqlfmt/issues/640
+SELECT -2147483648 AS col;
+SELECT 9223372036854775807l AS col;
+SELECT -32Y AS col;
+SELECT 482S AS col;
+SELECT 12.578 AS col;
+SELECT -0.1234567 AS col;
+SELECT -.1234567 AS col;
+SELECT 123. AS col;
+SELECT 123.BD AS col;
+SELECT 5E2 AS col;
+SELECT 5D AS col;
+SELECT -5BD AS col;
+SELECT 12.578e-2d AS col;
+SELECT -.1234567E+2BD AS col;
+SELECT +3.e+3 AS col;
+SELECT -3.E-3D AS col;
+)))))__SQLFMT_OUTPUT__(((((
+-- source:
+-- https://spark.apache.org/docs/latest/sql-ref-literals.html#integral-literal-syntax
+-- see: https://github.com/tconbeer/sqlfmt/issues/640
+select -2147483648 as col
+;
+select 9223372036854775807l as col
+;
+select -32Y as col
+;
+select 482S as col
+;
+select 12.578 as col
+;
+select -0.1234567 as col
+;
+select -.1234567 as col
+;
+select 123. as col
+;
+select 123.BD as col
+;
+select 5E2 as col
+;
+select 5D as col
+;
+select -5BD as col
+;
+select 12.578e-2d as col
+;
+select -.1234567E+2BD as col
+;
+select +3.e+3 as col
+;
+select -3.E-3D as col
+;

tests/functional_tests/test_general_formatting.py

@@ -51,6 +51,7 @@
         "unformatted/129_duckdb_joins.sql",
         "unformatted/130_athena_data_types.sql",
         "unformatted/131_assignment_statement.sql",
+        "unformatted/132_spark_number_literals.sql",
         "unformatted/133_for_else.sql",
         "unformatted/200_base_model.sql",
         "unformatted/201_basic_snapshot.sql",

tests/unit_tests/test_rule.py

@@ -57,13 +57,23 @@ def get_rule(ruleset: List[Rule], rule_name: str) -> Rule:
         (MAIN, "statement_start", "case"),
         (MAIN, "statement_end", "END"),
         (CORE, "star", "*"),
+        (CORE, "spark_int_literals", "145y"),
+        (CORE, "spark_int_literals", "-145s"),
+        (CORE, "spark_int_literals", "5l"),
         (CORE, "number", "145.8"),
         (CORE, "number", "-.58"),
         (CORE, "number", "+145.8"),
         (CORE, "number", "+.58"),
         (CORE, "number", "1e9"),
         (CORE, "number", "1e-9"),
         (CORE, "number", "1.55e-9"),
+        (CORE, "number", "145.8bd"),
+        (CORE, "number", "-.58d"),
+        (CORE, "number", "+145.8f"),
+        (CORE, "number", "+.58bd"),
+        (CORE, "number", "1e9d"),
+        (CORE, "number", "1e-9f"),
+        (CORE, "number", "1.55e-9bd"),
         (CORE, "bracket_open", "["),
         (CORE, "bracket_close", ")"),
         (CORE, "double_colon", "::"),