SQLite: Support IN with a table name or table-valued function

src/ast/mod.rs

@@ -883,6 +883,15 @@ pub enum InTarget {
     /// `UNNEST(array_expression)`, see
     /// [BigQuery](https://cloud.google.com/bigquery/docs/reference/standard-sql/operators#in_operators)
     Unnest(Box<Expr>),
+    /// A table name or table-valued function, read as `(SELECT * FROM name)`, see
+    /// [SQLite](https://www.sqlite.org/lang_expr.html#in_op)
+    Table {
+        #[cfg_attr(feature = "visitor", visit(with = "visit_relation"))]
+        /// Table or table-valued function name, visited as a relation.
+        name: ObjectName,
+        /// Table-valued function arguments, `None` for a plain table name.
+        args: Option<Vec<FunctionArg>>,
+    },
 }
 
 impl fmt::Display for InTarget {
@@ -891,6 +900,13 @@ impl fmt::Display for InTarget {
             InTarget::List(list) => write!(f, "({})", display_comma_separated(list)),
             InTarget::Subquery(subquery) => write!(f, "({subquery})"),
             InTarget::Unnest(array_expr) => write!(f, "UNNEST({array_expr})"),
+            InTarget::Table { name, args } => {
+                write!(f, "{name}")?;
+                if let Some(args) = args {
+                    write!(f, "({})", display_comma_separated(args))?;
+                }
+                Ok(())
+            }
         }
     }
 }

src/ast/spans.rs

@@ -1536,6 +1536,11 @@ impl Spanned for Expr {
                 ),
                 InTarget::Subquery(subquery) => expr.span().union(&subquery.span()),
                 InTarget::Unnest(array_expr) => expr.span().union(&array_expr.span()),
+                InTarget::Table { name, args } => union_spans(
+                    [expr.span(), name.span()]
+                        .into_iter()
+                        .chain(args.iter().flatten().map(|arg| arg.span())),
+                ),
             },
             Expr::Between {
                 expr,
@@ -3258,4 +3263,20 @@ SETTINGS update_parallel_mode = 'auto'"#;
             assert_eq!(test.get_source(expr.span()), expected, "{sql}");
         }
     }
+
+    #[test]
+    fn test_in_table_expression_spans() {
+        let dialect = &crate::dialect::SQLiteDialect {};
+        for (sql, expected) in [
+            ("  needle IN main.items", "needle IN main.items"),
+            (
+                "  needle IN main.f(first, last)",
+                "needle IN main.f(first, last",
+            ),
+        ] {
+            let mut test = SpanTest::new(dialect, sql);
+            let expr = test.0.parse_expr().unwrap();
+            assert_eq!(test.get_source(expr.span()), expected, "{sql}");
+        }
+    }
 }

src/dialect/generic.rs

@@ -173,6 +173,10 @@ impl Dialect for GenericDialect {
         true
     }
 
+    fn supports_in_table_expression(&self) -> bool {
+        true
+    }
+
     fn supports_comment_on(&self) -> bool {
         true
     }

src/dialect/mod.rs

@@ -476,6 +476,14 @@ pub trait Dialect: Debug + Any {
         false
     }
 
+    /// Returns true if the dialect supports a table name or table-valued function
+    /// as the right-hand side of `IN`, e.g. `x IN t`, `x IN schema.t`, or `x IN f(args)`.
+    ///
+    /// [SQLite](https://www.sqlite.org/lang_expr.html#in_op)
+    fn supports_in_table_expression(&self) -> bool {
+        false
+    }
+
     /// Returns true if the dialect supports `BEGIN {DEFERRED | IMMEDIATE | EXCLUSIVE | TRY | CATCH} [TRANSACTION]` statements
     fn supports_start_transaction_modifier(&self) -> bool {
         false

src/dialect/sqlite.rs

@@ -160,4 +160,75 @@ impl Dialect for SQLiteDialect {
     fn supports_unterminated_block_comments(&self) -> bool {
         true
     }
+
+    fn supports_in_table_expression(&self) -> bool {
+        true
+    }
+
+    /// The keywords SQLite does not accept as a name, as opposed to the
+    /// keywords its grammar lets fall back to an identifier.
+    /// See <https://www.sqlite.org/lang_keywords.html> and the `%fallback ID`
+    /// list in [parse.y](https://sqlite.org/src/file?name=src/parse.y)
+    fn is_table_factor(&self, kw: &Keyword, _parser: &mut Parser) -> bool {
+        !matches!(
+            kw,
+            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::DEFAULT
+                | Keyword::DEFERRABLE
+                | Keyword::DELETE
+                | Keyword::DISTINCT
+                | Keyword::DROP
+                | Keyword::ELSE
+                | Keyword::ESCAPE
+                | Keyword::EXCEPT
+                | Keyword::EXISTS
+                | Keyword::FOREIGN
+                | Keyword::FROM
+                | Keyword::GROUP
+                | Keyword::HAVING
+                | Keyword::IN
+                | Keyword::INDEX
+                | Keyword::INSERT
+                | Keyword::INTERSECT
+                | Keyword::INTO
+                | Keyword::IS
+                | Keyword::JOIN
+                | Keyword::LIMIT
+                | Keyword::NOT
+                | Keyword::NOTHING
+                | Keyword::NOTNULL
+                | Keyword::NULL
+                | Keyword::ON
+                | Keyword::OR
+                | Keyword::ORDER
+                | Keyword::PRIMARY
+                | Keyword::REFERENCES
+                | Keyword::RETURNING
+                | Keyword::SELECT
+                | Keyword::SET
+                | Keyword::TABLE
+                | Keyword::THEN
+                | Keyword::TO
+                | Keyword::TRANSACTION
+                | Keyword::UNION
+                | Keyword::UNIQUE
+                | Keyword::UPDATE
+                | Keyword::USING
+                | Keyword::VALUES
+                | Keyword::WHEN
+                | Keyword::WHERE
+        )
+    }
 }

src/parser/mod.rs

@@ -4515,6 +4515,19 @@ impl<'a> Parser<'a> {
             let array_expr = self.parse_expr()?;
             self.expect_token(&Token::RParen)?;
             InTarget::Unnest(Box::new(array_expr))
+        } else if !global
+            && self.dialect.supports_in_table_expression()
+            && self.next_word_is_table_name()
+        {
+            let name = self.parse_object_name(false)?;
+            let args = if self.consume_token(&Token::LParen) {
+                let args = self.parse_comma_separated0(Self::parse_function_args, Token::RParen)?;
+                self.expect_token(&Token::RParen)?;
+                Some(args)
+            } else {
+                None
+            };
+            InTarget::Table { name, args }
         } else if self.dialect.supports_in_unparenthesized_expr()
             && self.peek_token_ref().token != Token::LParen
         {
@@ -4532,6 +4545,20 @@ impl<'a> Parser<'a> {
         })
     }
 
+    /// Returns true if the next token is a word the dialect accepts as a table
+    /// name. The word is consumed while asking, so `is_table_factor` peeks past
+    /// it as it does at the trailing comma check.
+    fn next_word_is_table_name(&mut self) -> bool {
+        let Token::Word(word) = &self.peek_token_ref().token else {
+            return false;
+        };
+        let keyword = word.keyword;
+        self.advance_token();
+        let is_name = self.dialect.is_table_factor(&keyword, self);
+        self.prev_token();
+        is_name
+    }
+
     /// Parses the parenthesized subquery or value list of an `IN` predicate.
     fn parse_in_parenthesized_target(&mut self) -> Result<InTarget, ParserError> {
         self.expect_token(&Token::LParen)?;

tests/sqlparser_common.rs

@@ -2621,8 +2621,10 @@ fn parse_in_error() {
     // <expr> IN <expr> is no valid, except in dialects that accept an
     // unparenthesized expression as the IN right-hand side (e.g. ClickHouse).
     let sql = "SELECT * FROM customers WHERE segment in segment";
-    let res =
-        all_dialects_except(|d| d.supports_in_unparenthesized_expr()).parse_sql_statements(sql);
+    let res = all_dialects_except(|d| {
+        d.supports_in_unparenthesized_expr() || d.supports_in_table_expression()
+    })
+    .parse_sql_statements(sql);
     assert_eq!(
         ParserError::ParserError("Expected: (, found: segment".to_string()),
         res.unwrap_err()
@@ -21022,3 +21024,127 @@ fn parse_window_spec_keyword_base_window_name() {
         "SELECT count(*) OVER (ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW) FROM t",
     );
 }
+
+#[test]
+fn parse_in_table_expression() {
+    let dialects = all_dialects_where(|d| d.supports_in_table_expression());
+    for sql in [
+        "SELECT a IN t FROM t",
+        "SELECT a NOT IN t FROM t",
+        "SELECT a IN main.t FROM t",
+        "SELECT a IN f() FROM t",
+        "SELECT a NOT IN f(1, 2) FROM t",
+    ] {
+        dialects.verified_stmt(sql);
+    }
+    match dialects.verified_expr("a IN main.f(1)") {
+        Expr::In {
+            target:
+                InTarget::Table {
+                    name,
+                    args: Some(args),
+                },
+            negated: false,
+            global: false,
+            ..
+        } => {
+            assert_eq!(name.to_string(), "main.f");
+            assert_eq!(args.len(), 1);
+        }
+        other => panic!("expected an IN table target, got {other:?}"),
+    }
+    dialects.verified_stmt("SELECT a IN (1, 2) FROM t");
+    dialects.verified_stmt("SELECT a IN (SELECT b FROM u) FROM t");
+}
+
+/// Malformed table and table-function targets of `IN` keep their precise
+/// parse errors.
+#[test]
+fn parse_in_table_expression_malformed_targets() {
+    let dialects = all_dialects_where(|d| d.supports_in_table_expression());
+    for (sql, expected) in [
+        // A dangling dot leaves the qualified name without its last part.
+        ("SELECT a IN t.", "Expected: identifier, found: EOF"),
+        // A trailing comma leaves the next argument without an expression.
+        ("SELECT a IN f(x,)", "Expected: an expression, found: )"),
+        // The argument list never closes its parenthesis.
+        ("SELECT a IN f(1, 2", "Expected: ), found: EOF"),
+    ] {
+        assert_eq!(
+            ParserError::ParserError(expected.to_string()),
+            dialects.parse_sql_statements(sql).unwrap_err(),
+            "{sql}"
+        );
+    }
+}
+
+/// Display of an `IN` table target propagates formatter errors from the
+/// name and the argument list writes.
+#[test]
+fn display_in_table_target_propagates_formatter_errors() {
+    use core::fmt::{self, Write};
+
+    struct RejectingWriter(char);
+
+    impl Write for RejectingWriter {
+        fn write_str(&mut self, value: &str) -> fmt::Result {
+            if value.contains(self.0) {
+                Err(fmt::Error)
+            } else {
+                Ok(())
+            }
+        }
+    }
+
+    let plain = InTarget::Table {
+        name: ObjectName::from(Ident::new("t")),
+        args: None,
+    };
+    let mut writer = RejectingWriter('t');
+    assert_eq!(write!(writer, "{plain}"), Err(fmt::Error));
+
+    let function = InTarget::Table {
+        name: ObjectName::from(Ident::new("t")),
+        args: Some(vec![FunctionArg::Unnamed(FunctionArgExpr::Expr(
+            Expr::value(number("1")),
+        ))]),
+    };
+    let mut writer = RejectingWriter('1');
+    assert_eq!(write!(writer, "{function}"), Err(fmt::Error));
+}
+
+#[test]
+#[cfg(feature = "visitor")]
+fn test_visit_relations_in_table_target() {
+    let dialects = all_dialects_where(|d| d.supports_in_table_expression());
+    for dialect in dialects.dialects {
+        let stmts = Parser::parse_sql(dialect.as_ref(), "SELECT a IN lookup FROM source").unwrap();
+        let mut relations = vec![];
+        let _ = sqlparser::ast::visit_relations(&stmts, |relation| {
+            relations.push(relation.to_string());
+            core::ops::ControlFlow::<()>::Continue(())
+        });
+        assert_eq!(relations, ["lookup", "source"], "{dialect:?}");
+    }
+}
+
+#[test]
+#[cfg(feature = "visitor")]
+fn test_visit_relations_mut_in_table_target() {
+    let dialects = all_dialects_where(|d| d.supports_in_table_expression());
+    for dialect in dialects.dialects {
+        let mut stmts =
+            Parser::parse_sql(dialect.as_ref(), "SELECT a IN lookup FROM source").unwrap();
+        let _ = sqlparser::ast::visit_relations_mut(&mut stmts, |relation| {
+            if relation.to_string() == "lookup" {
+                *relation = ObjectName::from(Ident::new("renamed"));
+            }
+            core::ops::ControlFlow::<()>::Continue(())
+        });
+        assert_eq!(
+            stmts[0].to_string(),
+            "SELECT a IN renamed FROM source",
+            "{dialect:?}"
+        );
+    }
+}

tests/sqlparser_sqlite.rs

@@ -1250,3 +1250,32 @@ fn parse_bitwise_and_or_share_precedence() {
         assert_eq!(sqlite().verified_expr(sql), expected, "{sql}");
     }
 }
+
+#[test]
+fn parse_in_table_keyword_names() {
+    // Keywords SQLite lets fall back to an identifier name a table after `IN`.
+    for sql in [
+        "SELECT 1 IN view",
+        "SELECT 1 IN offset",
+        "SELECT 1 IN key",
+        "SELECT 1 IN abort",
+    ] {
+        sqlite().verified_stmt(sql);
+    }
+    // Its reserved keywords do not, so the target falls back to a parenthesized list.
+    for (sql, found) in [
+        ("SELECT 1 IN FROM t", "FROM"),
+        ("SELECT 1 IN select", "select"),
+        ("SELECT 1 IN where", "where"),
+        ("SELECT 1 IN NULL", "NULL"),
+        ("SELECT 1 IN table", "table"),
+        ("SELECT a FROM t WHERE a IN AND b", "AND"),
+    ] {
+        assert_eq!(
+            sqlite().parse_sql_statements(sql).unwrap_err(),
+            ParserError::ParserError(format!("Expected: (, found: {found}")),
+        );
+    }
+    // A quoted keyword is a plain name.
+    sqlite().verified_stmt(r#"SELECT 1 IN "from""#);
+}