feat: Streamline SQL INTERVAL handling and improve related error messages, update sqlparser-rs lib

Cargo.toml

@@ -72,7 +72,7 @@ serde_json = "1"
 simd-json = { version = "0.13", features = ["known-key"] }
 simdutf8 = "0.1.4"
 smartstring = "1"
-sqlparser = "0.39"
+sqlparser = "0.45"
 stacker = "0.1"
 streaming-iterator = "0.1.9"
 strength_reduce = "0.2"

crates/polars-sql/src/context.rs

@@ -119,7 +119,7 @@ impl SQLContext {
             .parse_statements()
             .map_err(to_compute_err)?;
 
-        polars_ensure!(ast.len() == 1, SQLInterface: "one (and only one) statement at a time please");
+        polars_ensure!(ast.len() == 1, SQLInterface: "one (and only one) statement can be parsed at a time");
         let res = self.execute_statement(ast.first().unwrap())?;
 
         // Ensure the result uses the proper arenas.
@@ -175,13 +175,11 @@ impl SQLContext {
 
     pub(crate) fn execute_query(&mut self, query: &Query) -> PolarsResult<LazyFrame> {
         self.register_ctes(query)?;
-
         self.execute_query_no_ctes(query)
     }
 
     pub(crate) fn execute_query_no_ctes(&mut self, query: &Query) -> PolarsResult<LazyFrame> {
         let lf = self.process_set_expr(&query.body, query)?;
-
         self.process_limit_offset(lf, &query.limit, &query.offset)
     }
 

crates/polars-sql/src/keywords.rs

@@ -39,6 +39,7 @@ pub fn all_keywords() -> Vec<&'static str> {
         keywords::IN,
         keywords::INNER,
         keywords::INT,
+        keywords::INTERVAL,
         keywords::JOIN,
         keywords::LEFT,
         keywords::LIMIT,

crates/polars-sql/src/sql_expr.rs

@@ -417,20 +417,11 @@ impl SQLExprVisitor<'_> {
             || interval.leading_precision.is_some()
             || interval.fractional_seconds_precision.is_some()
         {
-            polars_bail!(SQLInterface: "interval with explicit leading field or precision is not supported: {:?}", interval)
+            polars_bail!(SQLSyntax: "unsupported interval syntax: '{}'", interval)
         }
-        let mut negative = false;
         let s = match &*interval.value {
-            SQLExpr::UnaryOp {
-                op: UnaryOperator::Minus,
-                expr,
-            } if matches!(**expr, SQLExpr::Value(SQLValue::SingleQuotedString(_))) => {
-                if let SQLExpr::Value(SQLValue::SingleQuotedString(ref s)) = **expr {
-                    negative = true;
-                    Some(s)
-                } else {
-                    unreachable!()
-                }
+            SQLExpr::UnaryOp { .. } => {
+                polars_bail!(SQLSyntax: "unary ops are not valid on interval strings; found {}", interval.value)
             },
             SQLExpr::Value(SQLValue::SingleQuotedString(s)) => Some(s),
             _ => None,
@@ -439,10 +430,7 @@ impl SQLExprVisitor<'_> {
             Some(s) if s.contains('-') => {
                 polars_bail!(SQLInterface: "minus signs are not yet supported in interval strings; found '{}'", s)
             },
-            Some(s) => {
-                let d = Duration::parse_interval(s);
-                Ok(lit(if negative { -d } else { d }))
-            },
+            Some(s) => Ok(lit(Duration::parse_interval(s))),
             None => polars_bail!(SQLSyntax: "invalid interval {:?}", interval),
         }
     }
@@ -1193,7 +1181,7 @@ fn parse_extract(expr: Expr, field: &DateTimeField) -> PolarsResult<Expr> {
         DateTimeField::Year => expr.dt().year(),
         DateTimeField::Quarter => expr.dt().quarter(),
         DateTimeField::Month => expr.dt().month(),
-        DateTimeField::Week => expr.dt().week(),
+        DateTimeField::Week(_) => expr.dt().week(),
         DateTimeField::IsoWeek => expr.dt().week(),
         DateTimeField::DayOfYear | DateTimeField::Doy => expr.dt().ordinal_day(),
         DateTimeField::DayOfWeek | DateTimeField::Dow => {

py-polars/tests/unit/sql/test_literals.py

@@ -103,7 +103,7 @@ def test_intervals() -> None:
               INTERVAL '100ms 100us' AS i2,
               -- long form with/without commas (case-insensitive)
               INTERVAL '1 week, 2 hours, 3 minutes, 4 seconds' AS i3,
-              INTERVAL '1 Quarter 2 Months 987 Microseconds' AS i4,
+              INTERVAL '1 QUARTER 2 Months 987 microseconds' AS i4,
             FROM df
             """
         )
@@ -124,3 +124,9 @@ def test_intervals() -> None:
             match="minus signs are not yet supported in interval strings; found '-7d'",
         ):
             ctx.execute("SELECT INTERVAL '-7d' AS one_week_ago FROM df")
+
+        with pytest.raises(
+            SQLSyntaxError,
+            match="unary ops are not valid on interval strings; found -'7d'",
+        ):
+            ctx.execute("SELECT INTERVAL -'7d' AS one_week_ago FROM df")