WIP: Update sqlparser to next release

Cargo.toml

@@ -177,3 +177,7 @@ large_futures = "warn"
 [workspace.lints.rust]
 unexpected_cfgs = { level = "warn", check-cfg = ["cfg(tarpaulin)"] }
 unused_qualifications = "deny"
+
+## Temp patch for sqlparser
+[patch.crates-io]
+sqlparser = { git = "https://github.com/apache/datafusion-sqlparser-rs.git", rev = "d0fcc06652ba9880622d0ef8b426c809cee752fe" }

datafusion-cli/Cargo.toml

@@ -67,3 +67,7 @@ assert_cmd = "2.0"
 ctor = "0.2.0"
 predicates = "3.0"
 rstest = "0.22"
+
+## Temp patch for sqlparser
+[patch.crates-io]
+sqlparser = { git = "https://github.com/apache/datafusion-sqlparser-rs.git", rev = "d0fcc06652ba9880622d0ef8b426c809cee752fe" }

datafusion/common/src/utils/mod.rs

@@ -887,10 +887,10 @@ pub fn get_available_parallelism() -> usize {
 
 #[cfg(test)]
 mod tests {
+    use super::*;
     use crate::ScalarValue::Null;
     use arrow::array::Float64Array;
-
-    use super::*;
+    use sqlparser::tokenizer::Span;
 
     #[test]
     fn test_bisect_linear_left_and_right() -> Result<()> {
@@ -1118,6 +1118,7 @@ mod tests {
             let expected_parsed = vec![Ident {
                 value: identifier.to_string(),
                 quote_style,
+                span: Span::empty(),
             }];
 
             assert_eq!(

datafusion/core/tests/user_defined/user_defined_scalar_functions.rs

@@ -27,10 +27,6 @@ use arrow_array::{
     Array, ArrayRef, Float32Array, Float64Array, Int32Array, RecordBatch, StringArray,
 };
 use arrow_schema::{DataType, Field, Schema};
-use parking_lot::Mutex;
-use regex::Regex;
-use sqlparser::ast::Ident;
-
 use datafusion::execution::context::{FunctionFactory, RegisterFunction, SessionState};
 use datafusion::prelude::*;
 use datafusion::{execution::registry::FunctionRegistry, test_util};
@@ -48,6 +44,10 @@ use datafusion_expr::{
     Volatility,
 };
 use datafusion_functions_nested::range::range_udf;
+use parking_lot::Mutex;
+use regex::Regex;
+use sqlparser::ast::Ident;
+use sqlparser::tokenizer::Span;
 
 /// test that casting happens on udfs.
 /// c11 is f32, but `custom_sqrt` requires f64. Casting happens but the logical plan and
@@ -1187,6 +1187,7 @@ async fn create_scalar_function_from_sql_statement_postgres_syntax() -> Result<(
             name: Some(Ident {
                 value: "name".into(),
                 quote_style: None,
+                span: Span::empty(),
             }),
             data_type: DataType::Utf8,
             default_expr: None,
@@ -1196,6 +1197,7 @@ async fn create_scalar_function_from_sql_statement_postgres_syntax() -> Result<(
             language: Some(Ident {
                 value: "plrust".into(),
                 quote_style: None,
+                span: Span::empty(),
             }),
             behavior: None,
             function_body: Some(lit(body)),

datafusion/expr/src/expr.rs

@@ -222,6 +222,8 @@ use sqlparser::ast::{
 /// // to 42 = 5 AND b = 6
 /// assert_eq!(rewritten.data, lit(42).eq(lit(5)).and(col("b").eq(lit(6))));
 #[derive(Clone, PartialEq, Eq, PartialOrd, Hash, Debug)]
+// TODO make the enum smaller with more boxing (looks like Wildcard is now bigger)
+#[allow(clippy::large_enum_variant)]
 pub enum Expr {
     /// An expression with a specific name.
     Alias(Alias),

datafusion/sql/src/expr/function.rs

@@ -169,6 +169,11 @@ impl FunctionArgs {
                         "Calling {name}: SEPARATOR not supported in function arguments: {sep}"
                     )
                 }
+                FunctionArgumentClause::JsonNullClause(jn) => {
+                    return not_impl_err!(
+                        "Calling {name}: JSON NULL clause not supported in function arguments: {jn}"
+                    )
+                }
             }
         }
 

datafusion/sql/src/expr/mod.rs

@@ -565,11 +565,11 @@ impl<S: ContextProvider> SqlToRel<'_, S> {
                 }
                 not_impl_err!("AnyOp not supported by ExprPlanner: {binary_expr:?}")
             }
-            SQLExpr::Wildcard => Ok(Expr::Wildcard {
+            SQLExpr::Wildcard(_token) => Ok(Expr::Wildcard {
                 qualifier: None,
                 options: WildcardOptions::default(),
             }),
-            SQLExpr::QualifiedWildcard(object_name) => Ok(Expr::Wildcard {
+            SQLExpr::QualifiedWildcard(object_name, _token) => Ok(Expr::Wildcard {
                 qualifier: Some(self.object_name_to_table_reference(object_name)?),
                 options: WildcardOptions::default(),
             }),

datafusion/sql/src/parser.rs

@@ -20,14 +20,15 @@
 use std::collections::VecDeque;
 use std::fmt;
 
+use sqlparser::tokenizer::TokenWithSpan;
 use sqlparser::{
     ast::{
         ColumnDef, ColumnOptionDef, Expr, ObjectName, OrderByExpr, Query,
         Statement as SQLStatement, TableConstraint, Value,
     },
     dialect::{keywords::Keyword, Dialect, GenericDialect},
     parser::{Parser, ParserError},
-    tokenizer::{Token, TokenWithLocation, Tokenizer, Word},
+    tokenizer::{Token, Tokenizer, Word},
 };
 
 // Use `Parser::expected` instead, if possible
@@ -337,7 +338,7 @@ impl<'a> DFParser<'a> {
     fn expected<T>(
         &self,
         expected: &str,
-        found: TokenWithLocation,
+        found: TokenWithSpan,
     ) -> Result<T, ParserError> {
         parser_err!(format!("Expected {expected}, found: {found}"))
     }
@@ -875,6 +876,7 @@ mod tests {
     use super::*;
     use sqlparser::ast::Expr::Identifier;
     use sqlparser::ast::{BinaryOperator, DataType, Expr, Ident};
+    use sqlparser::tokenizer::Span;
 
     fn expect_parse_ok(sql: &str, expected: Statement) -> Result<(), ParserError> {
         let statements = DFParser::parse_sql(sql)?;
@@ -910,6 +912,7 @@ mod tests {
             name: Ident {
                 value: name.into(),
                 quote_style: None,
+                span: Span::empty(),
             },
             data_type,
             collation: None,
@@ -1218,6 +1221,7 @@ mod tests {
                     expr: Identifier(Ident {
                         value: "c1".to_owned(),
                         quote_style: None,
+                        span: Span::empty(),
                     }),
                     asc,
                     nulls_first,
@@ -1249,6 +1253,7 @@ mod tests {
                     expr: Identifier(Ident {
                         value: "c1".to_owned(),
                         quote_style: None,
+                        span: Span::empty(),
                     }),
                     asc: Some(true),
                     nulls_first: None,
@@ -1258,6 +1263,7 @@ mod tests {
                     expr: Identifier(Ident {
                         value: "c2".to_owned(),
                         quote_style: None,
+                        span: Span::empty(),
                     }),
                     asc: Some(false),
                     nulls_first: Some(true),
@@ -1289,11 +1295,13 @@ mod tests {
                     left: Box::new(Identifier(Ident {
                         value: "c1".to_owned(),
                         quote_style: None,
+                        span: Span::empty(),
                     })),
                     op: BinaryOperator::Minus,
                     right: Box::new(Identifier(Ident {
                         value: "c2".to_owned(),
                         quote_style: None,
+                        span: Span::empty(),
                     })),
                 },
                 asc: Some(true),
@@ -1334,11 +1342,13 @@ mod tests {
                     left: Box::new(Identifier(Ident {
                         value: "c1".to_owned(),
                         quote_style: None,
+                        span: Span::empty(),
                     })),
                     op: BinaryOperator::Minus,
                     right: Box::new(Identifier(Ident {
                         value: "c2".to_owned(),
                         quote_style: None,
+                        span: Span::empty(),
                     })),
                 },
                 asc: Some(true),

datafusion/sql/src/planner.rs

@@ -339,7 +339,8 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
         plan: LogicalPlan,
         alias: TableAlias,
     ) -> Result<LogicalPlan> {
-        let plan = self.apply_expr_alias(plan, alias.columns)?;
+        let idents = alias.columns.into_iter().map(|c| c.name).collect();
+        let plan = self.apply_expr_alias(plan, idents)?;
 
         LogicalPlanBuilder::from(plan)
             .alias(TableReference::bare(
@@ -542,7 +543,7 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
             | SQLDataType::Regclass
             | SQLDataType::Custom(_, _)
             | SQLDataType::Array(_)
-            | SQLDataType::Enum(_)
+            | SQLDataType::Enum(_, _)
             | SQLDataType::Set(_)
             | SQLDataType::MediumInt(_)
             | SQLDataType::UnsignedMediumInt(_)
@@ -586,6 +587,15 @@ impl<'a, S: ContextProvider> SqlToRel<'a, S> {
             | SQLDataType::Nullable(_)
             | SQLDataType::LowCardinality(_)
             | SQLDataType::Trigger
+            // MySQL datatypes
+            | SQLDataType::TinyBlob
+            | SQLDataType::MediumBlob
+            | SQLDataType::LongBlob
+            | SQLDataType::TinyText
+            | SQLDataType::MediumText
+            | SQLDataType::LongText
+            | SQLDataType::Bit(_)
+            |SQLDataType::BitVarying(_)
             => not_impl_err!(
                 "Unsupported SQL type {sql_type:?}"
             ),

datafusion/sql/src/select.rs

@@ -655,6 +655,7 @@ impl<S: ContextProvider> SqlToRel<'_, S> {
             opt_rename,
             opt_replace: _opt_replace,
             opt_ilike: _opt_ilike,
+            wildcard_token: _wildcard_token,
         } = options;
 
         if opt_rename.is_some() {