SQLite: Support multi-word type names

fuzz/fuzz_seeds/sqlite_2556_multiword_type_names.sql

@@ -0,0 +1,6 @@
+CREATE TABLE t (a UNSIGNED BIG INT, b NATIVE CHARACTER(70), c VARYING CHARACTER(10), d 'text', e foo bar NOT NULL DEFAULT 0, f foo GENERATED, g foo GENERATED ALWAYS AS (1), h foo PATH bar);
+SELECT CAST(a AS foo bar), CAST(b AS VARYING CHARACTER) FROM t;
+ALTER TABLE t ADD i foo bar;
+CREATE TABLE u (a NULL, b INT FOO, c TEXT FOO PRIMARY KEY, d ENUM FOO, e TIMESTAMP WITH FOO);
+SELECT CAST(1 AS INT FOO), CAST(1 AS ENUM FOO), CAST(1 AS TIMESTAMP WITH FOO);
+ALTER TABLE u ADD f NULL;

src/ast/data_type.rs

@@ -25,7 +25,9 @@ use serde::{Deserialize, Serialize};
 #[cfg(feature = "visitor")]
 use sqlparser_derive::{Visit, VisitMut};
 
-use crate::ast::{display_comma_separated, Expr, Ident, ObjectName, StructField, UnionField};
+use crate::ast::{
+    display_comma_separated, display_separated, Expr, Ident, ObjectName, StructField, UnionField,
+};
 
 use super::{value::escape_single_quote_string, ColumnDef};
 
@@ -466,6 +468,8 @@ pub enum DataType {
     VarBit(Option<u64>),
     /// Custom type with a modifier list, e.g. `GEOMETRY(POINT, 4326)`.
     Custom(ObjectName, Vec<TypeModifier>),
+    /// Multi-word type name whose parts are space-separated, as in SQLite's `typename` grammar.
+    CustomMultiWord(Vec<Ident>, Vec<TypeModifier>),
     /// Arrays.
     Array(ArrayElemTypeDef),
     /// Map, see [ClickHouse], [Hive].
@@ -770,6 +774,13 @@ impl fmt::Display for DataType {
                     write!(f, "{ty}({})", display_comma_separated(modifiers))
                 }
             }
+            DataType::CustomMultiWord(parts, modifiers) => {
+                write!(f, "{}", display_separated(parts, " "))?;
+                if !modifiers.is_empty() {
+                    write!(f, "({})", display_comma_separated(modifiers))?;
+                }
+                Ok(())
+            }
             DataType::Enum(vals, bits) => {
                 match bits {
                     Some(bits) => write!(f, "ENUM{bits}"),

src/dialect/mod.rs

@@ -586,6 +586,11 @@ pub trait Dialect: Debug + Any {
         false
     }
 
+    /// Whether type names accept space-separated words or strings, including built-in type keywords.
+    fn supports_multiword_type_names(&self) -> bool {
+        false
+    }
+
     /// Returns true if the dialects supports specifying null treatment
     /// as part of a window function's parameter list as opposed
     /// to after the parameter list.

src/dialect/sqlite.rs

@@ -171,6 +171,10 @@ impl Dialect for SQLiteDialect {
         true
     }
 
+    fn supports_multiword_type_names(&self) -> bool {
+        true
+    }
+
     fn supports_double_eq_assignment(&self) -> bool {
         true
     }

src/keywords.rs

@@ -566,6 +566,7 @@ define_keywords!(
     INVOKER,
     IO,
     IS,
+    ISNULL,
     ISODOW,
     ISOLATION,
     ISOWEEK,

src/parser/mod.rs

@@ -1854,7 +1854,11 @@ impl<'a> Parser<'a> {
             _ => None,
         };
         let opt_expr = self.maybe_parse(|parser| {
-            match parser.parse_data_type()? {
+            let (data_type, trailing_bracket) = parser.parse_data_type_inner()?;
+            if trailing_bracket.0 {
+                return parser_err!("dummy", loc);
+            }
+            match data_type {
                 DataType::Interval { .. } => parser.parse_interval(),
                 // Custom names need dialect consent, else `NOT 'a'` would read as a typed string.
                 DataType::Custom(name, modifiers)
@@ -1870,7 +1874,7 @@ impl<'a> Parser<'a> {
                         uses_odbc_syntax: false,
                     }))
                 }
-                DataType::Custom(..) => parser_err!("dummy", loc),
+                DataType::Custom(..) | DataType::CustomMultiWord(..) => parser_err!("dummy", loc),
                 // MySQL supports using the `BINARY` keyword as a cast to binary type.
                 DataType::Binary(..) if self.dialect.supports_binary_kw_as_cast() => {
                     Ok(Expr::Cast {
@@ -9986,6 +9990,7 @@ impl<'a> Parser<'a> {
                     Keyword::CONSTRAINT
                         | Keyword::PRIMARY
                         | Keyword::NOT
+                        | Keyword::NULL
                         | Keyword::UNIQUE
                         | Keyword::CHECK
                         | Keyword::DEFAULT
@@ -9994,7 +9999,9 @@ impl<'a> Parser<'a> {
                         | Keyword::GENERATED
                         | Keyword::AS
                 ),
-                _ => true, // e.g. comma immediately after column name
+                // A single-quoted string is a valid type name in SQLite's ids grammar.
+                Token::SingleQuotedString(_) => false,
+                _ => true,
             }
         } else {
             false
@@ -13222,10 +13229,46 @@ impl<'a> Parser<'a> {
         Ok(ty)
     }
 
-    #[cfg_attr(feature = "recursive-protection", recursive::recursive)]
     fn parse_data_type_helper(
         &mut self,
     ) -> Result<(DataType, MatchedTrailingBracket), ParserError> {
+        if !self.dialect.supports_multiword_type_names() {
+            return self.parse_data_type_inner();
+        }
+
+        let index = self.index;
+        let dedicated_error = match self.try_parse(|parser| parser.parse_data_type_inner()) {
+            Ok(data_type) => {
+                if self.peek_multiword_type_name_part().is_none() {
+                    return Ok(data_type);
+                }
+                self.index = index;
+                None
+            }
+            Err(ParserError::RecursionLimitExceeded) => {
+                return Err(ParserError::RecursionLimitExceeded);
+            }
+            Err(error) => Some(error),
+        };
+
+        self.advance_token();
+        let token = self.get_current_token();
+        let first = match &token.token {
+            Token::Word(word) => word.clone().into_ident(token.span),
+            Token::SingleQuotedString(value) => Ident::with_quote('\'', value.clone()),
+            _ => return self.expected_ref("a data type name", token),
+        };
+        if self.peek_multiword_type_name_part().is_none() {
+            if let Some(error) = dedicated_error {
+                return Err(error);
+            }
+        }
+        self.parse_multiword_type_name(first)
+            .map_err(|error| dedicated_error.unwrap_or(error))
+    }
+
+    #[cfg_attr(feature = "recursive-protection", recursive::recursive)]
+    fn parse_data_type_inner(&mut self) -> Result<(DataType, MatchedTrailingBracket), ParserError> {
         let _guard = self.recursion_counter.try_decrease()?;
 
         let dialect = self.dialect;
@@ -13679,13 +13722,21 @@ impl<'a> Parser<'a> {
                     }
                 }
                 Keyword::SIGNED => {
+                    if self.dialect.supports_multiword_type_names() {
+                        let first = w.clone().into_ident(next_token.span);
+                        return self.parse_multiword_type_name(first);
+                    }
                     if self.parse_keyword(Keyword::INTEGER) {
                         Ok(DataType::SignedInteger)
                     } else {
                         Ok(DataType::Signed)
                     }
                 }
                 Keyword::UNSIGNED => {
+                    if self.dialect.supports_multiword_type_names() {
+                        let first = w.clone().into_ident(next_token.span);
+                        return self.parse_multiword_type_name(first);
+                    }
                     if self.parse_keyword(Keyword::INTEGER) {
                         Ok(DataType::UnsignedInteger)
                     } else {
@@ -13699,6 +13750,10 @@ impl<'a> Parser<'a> {
                     Ok(DataType::TsQuery)
                 }
                 _ => {
+                    if self.dialect.supports_multiword_type_names() {
+                        let first = w.clone().into_ident(next_token.span);
+                        return self.parse_multiword_type_name(first);
+                    }
                     self.prev_token();
                     let type_name = self.parse_object_name(false)?;
                     if let Some(modifiers) = self.parse_optional_type_modifiers()? {
@@ -13708,6 +13763,10 @@ impl<'a> Parser<'a> {
                     }
                 }
             },
+            Token::SingleQuotedString(value) if self.dialect.supports_multiword_type_names() => {
+                let first = Ident::with_quote('\'', value.clone());
+                return self.parse_multiword_type_name(first);
+            }
             _ => self.expected_at("a data type name", next_token_index),
         }?;
 
@@ -14890,6 +14949,115 @@ impl<'a> Parser<'a> {
         }
     }
 
+    fn peek_multiword_type_name_part(&self) -> Option<&TokenWithSpan> {
+        let peeked = self.peek_token_ref();
+        let is_part = match &peeked.token {
+            Token::Word(w) if w.keyword == Keyword::GENERATED => !matches!(
+                &self.peek_nth_token_ref(1).token,
+                Token::Word(w) if w.keyword == Keyword::ALWAYS
+            ),
+            Token::Word(w) => !matches!(
+                w.keyword,
+                Keyword::ADD
+                    | Keyword::ALL
+                    | Keyword::ALTER
+                    | Keyword::AND
+                    | Keyword::AS
+                    | Keyword::AUTOINCREMENT
+                    | Keyword::BETWEEN
+                    | Keyword::CASE
+                    | Keyword::CHECK
+                    | Keyword::COLLATE
+                    | Keyword::COMMIT
+                    | Keyword::CONSTRAINT
+                    | Keyword::CREATE
+                    | Keyword::CROSS
+                    | Keyword::DEFAULT
+                    | Keyword::DELETE
+                    | Keyword::DEFERRABLE
+                    | Keyword::DISTINCT
+                    | Keyword::DROP
+                    | Keyword::ELSE
+                    | Keyword::ESCAPE
+                    | Keyword::EXCEPT
+                    | Keyword::EXISTS
+                    | Keyword::FOREIGN
+                    | Keyword::FROM
+                    | Keyword::FULL
+                    | Keyword::GROUP
+                    | Keyword::HAVING
+                    | Keyword::IN
+                    | Keyword::INDEX
+                    | Keyword::INDEXED
+                    | Keyword::INNER
+                    | Keyword::INSERT
+                    | Keyword::INTERSECT
+                    | Keyword::INTO
+                    | Keyword::IS
+                    | Keyword::ISNULL
+                    | Keyword::JOIN
+                    | Keyword::LEFT
+                    | Keyword::LIMIT
+                    | Keyword::NATURAL
+                    | Keyword::NOT
+                    | Keyword::NULL
+                    | Keyword::NOTHING
+                    | Keyword::NOTNULL
+                    | Keyword::ON
+                    | Keyword::OR
+                    | Keyword::ORDER
+                    | Keyword::OUTER
+                    | Keyword::PRIMARY
+                    | Keyword::REFERENCES
+                    | Keyword::RETURNING
+                    | Keyword::RIGHT
+                    | Keyword::SELECT
+                    | Keyword::SET
+                    | Keyword::TABLE
+                    | Keyword::THEN
+                    | Keyword::TO
+                    | Keyword::TRANSACTION
+                    | Keyword::UNIQUE
+                    | Keyword::UNION
+                    | Keyword::UPDATE
+                    | Keyword::USING
+                    | Keyword::VALUES
+                    | Keyword::WHEN
+                    | Keyword::WHERE
+            ),
+            Token::SingleQuotedString(_) => true,
+            _ => false,
+        };
+        is_part.then_some(peeked)
+    }
+
+    /// Parse a space-separated multi-word typename per SQLite's `typename` grammar.
+    fn parse_multiword_type_name(
+        &mut self,
+        first: Ident,
+    ) -> Result<(DataType, MatchedTrailingBracket), ParserError> {
+        let mut parts = Vec::with_capacity(4);
+        parts.push(first);
+        while let Some(peeked) = self.peek_multiword_type_name_part() {
+            let next_part = match &peeked.token {
+                Token::Word(word) => word.clone().into_ident(peeked.span),
+                Token::SingleQuotedString(value) => Ident::with_quote('\'', value.clone()),
+                _ => break,
+            };
+            self.advance_token();
+            parts.push(next_part);
+        }
+        let modifiers = self.parse_optional_type_modifiers()?.unwrap_or_default();
+        if parts.len() == 1 {
+            Ok((
+                DataType::Custom(ObjectName::from(vec![parts.remove(0)]), modifiers),
+                false.into(),
+            ))
+        } else {
+            Ok((DataType::CustomMultiWord(parts, modifiers), false.into()))
+        }
+    }
+
     fn parse_structured_object_type_def(&mut self) -> Result<Vec<ColumnDef>, ParserError> {
         self.expect_token(&Token::LParen)?;
         if self.consume_token(&Token::RParen) {