Parse SETTINGS clause for ClickHouse table-valued functions (apache#1358)

Author: Jesse-Bakker

src/ast/mod.rs

@@ -50,8 +50,8 @@ pub use self::query::{
     OffsetRows, OrderBy, OrderByExpr, PivotValueSource, Query, RenameSelectItem,
     RepetitionQuantifier, ReplaceSelectElement, ReplaceSelectItem, RowsPerMatch, Select,
     SelectInto, SelectItem, SetExpr, SetOperator, SetQuantifier, Setting, SymbolDefinition, Table,
-    TableAlias, TableFactor, TableVersion, TableWithJoins, Top, TopQuantity, ValueTableMode,
-    Values, WildcardAdditionalOptions, With, WithFill,
+    TableAlias, TableFactor, TableFunctionArgs, TableVersion, TableWithJoins, Top, TopQuantity,
+    ValueTableMode, Values, WildcardAdditionalOptions, With, WithFill,
 };
 pub use self::value::{
     escape_double_quote_string, escape_quoted_string, DateTimeField, DollarQuotedString,

src/ast/query.rs

@@ -899,6 +899,19 @@ impl fmt::Display for ExprWithAlias {
     }
 }
 
+/// Arguments to a table-valued function
+#[derive(Debug, Clone, PartialEq, PartialOrd, Eq, Ord, Hash)]
+#[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
+#[cfg_attr(feature = "visitor", derive(Visit, VisitMut))]
+pub struct TableFunctionArgs {
+    pub args: Vec<FunctionArg>,
+    /// ClickHouse-specific SETTINGS clause.
+    /// For example,
+    /// `SELECT * FROM executable('generate_random.py', TabSeparated, 'id UInt32, random String', SETTINGS send_chunk_header = false, pool_size = 16)`
+    /// [`executable` table function](https://clickhouse.com/docs/en/engines/table-functions/executable)
+    pub settings: Option<Vec<Setting>>,
+}
+
 /// A table name or a parenthesized subquery with an optional alias
 #[derive(Debug, Clone, PartialEq, PartialOrd, Eq, Ord, Hash)]
 #[cfg_attr(feature = "serde", derive(Serialize, Deserialize))]
@@ -916,7 +929,7 @@ pub enum TableFactor {
         /// This field's value is `Some(v)`, where `v` is a (possibly empty)
         /// vector of arguments, in the case of a table-valued function call,
         /// whereas it's `None` in the case of a regular table name.
-        args: Option<Vec<FunctionArg>>,
+        args: Option<TableFunctionArgs>,
         /// MSSQL-specific `WITH (...)` hints such as NOLOCK.
         with_hints: Vec<Expr>,
         /// Optional version qualifier to facilitate table time-travel, as
@@ -1314,7 +1327,15 @@ impl fmt::Display for TableFactor {
                     write!(f, "PARTITION ({})", display_comma_separated(partitions))?;
                 }
                 if let Some(args) = args {
-                    write!(f, "({})", display_comma_separated(args))?;
+                    write!(f, "(")?;
+                    write!(f, "{}", display_comma_separated(&args.args))?;
+                    if let Some(ref settings) = args.settings {
+                        if !args.args.is_empty() {
+                            write!(f, ", ")?;
+                        }
+                        write!(f, "SETTINGS {}", display_comma_separated(settings))?;
+                    }
+                    write!(f, ")")?;
                 }
                 if *with_ordinality {
                     write!(f, " WITH ORDINALITY")?;

src/parser/mod.rs

@@ -3430,6 +3430,29 @@ impl<'a> Parser<'a> {
         Ok(values)
     }
 
+    /// Parse the comma of a comma-separated syntax element.
+    /// Returns true if there is a next element
+    fn is_parse_comma_separated_end(&mut self) -> bool {
+        if !self.consume_token(&Token::Comma) {
+            true
+        } else if self.options.trailing_commas {
+            let token = self.peek_token().token;
+            match token {
+                Token::Word(ref kw)
+                    if keywords::RESERVED_FOR_COLUMN_ALIAS.contains(&kw.keyword) =>
+                {
+                    true
+                }
+                Token::RParen | Token::SemiColon | Token::EOF | Token::RBracket | Token::RBrace => {
+                    true
+                }
+                _ => false,
+            }
+        } else {
+            false
+        }
+    }
+
     /// Parse a comma-separated list of 1+ items accepted by `F`
     pub fn parse_comma_separated<T, F>(&mut self, mut f: F) -> Result<Vec<T>, ParserError>
     where
@@ -3438,22 +3461,8 @@ impl<'a> Parser<'a> {
         let mut values = vec![];
         loop {
             values.push(f(self)?);
-            if !self.consume_token(&Token::Comma) {
+            if self.is_parse_comma_separated_end() {
                 break;
-            } else if self.options.trailing_commas {
-                match self.peek_token().token {
-                    Token::Word(kw)
-                        if keywords::RESERVED_FOR_COLUMN_ALIAS.contains(&kw.keyword) =>
-                    {
-                        break;
-                    }
-                    Token::RParen
-                    | Token::SemiColon
-                    | Token::EOF
-                    | Token::RBracket
-                    | Token::RBrace => break,
-                    _ => continue,
-                }
             }
         }
         Ok(values)
@@ -8104,19 +8113,7 @@ impl<'a> Parser<'a> {
                 vec![]
             };
 
-            let settings = if dialect_of!(self is ClickHouseDialect|GenericDialect)
-                && self.parse_keyword(Keyword::SETTINGS)
-            {
-                let key_values = self.parse_comma_separated(|p| {
-                    let key = p.parse_identifier(false)?;
-                    p.expect_token(&Token::Eq)?;
-                    let value = p.parse_value()?;
-                    Ok(Setting { key, value })
-                })?;
-                Some(key_values)
-            } else {
-                None
-            };
+            let settings = self.parse_settings()?;
 
             let fetch = if self.parse_keyword(Keyword::FETCH) {
                 Some(self.parse_fetch()?)
@@ -8163,6 +8160,23 @@ impl<'a> Parser<'a> {
         }
     }
 
+    fn parse_settings(&mut self) -> Result<Option<Vec<Setting>>, ParserError> {
+        let settings = if dialect_of!(self is ClickHouseDialect|GenericDialect)
+            && self.parse_keyword(Keyword::SETTINGS)
+        {
+            let key_values = self.parse_comma_separated(|p| {
+                let key = p.parse_identifier(false)?;
+                p.expect_token(&Token::Eq)?;
+                let value = p.parse_value()?;
+                Ok(Setting { key, value })
+            })?;
+            Some(key_values)
+        } else {
+            None
+        };
+        Ok(settings)
+    }
+
     /// Parse a mssql `FOR [XML | JSON | BROWSE]` clause
     pub fn parse_for_clause(&mut self) -> Result<Option<ForClause>, ParserError> {
         if self.parse_keyword(Keyword::XML) {
@@ -9382,9 +9396,9 @@ impl<'a> Parser<'a> {
             // Parse potential version qualifier
             let version = self.parse_table_version()?;
 
-            // Postgres, MSSQL: table-valued functions:
+            // Postgres, MSSQL, ClickHouse: table-valued functions:
             let args = if self.consume_token(&Token::LParen) {
-                Some(self.parse_optional_args()?)
+                Some(self.parse_table_function_args()?)
             } else {
                 None
             };
@@ -10327,6 +10341,27 @@ impl<'a> Parser<'a> {
         }
     }
 
+    fn parse_table_function_args(&mut self) -> Result<TableFunctionArgs, ParserError> {
+        if self.consume_token(&Token::RParen) {
+            return Ok(TableFunctionArgs {
+                args: vec![],
+                settings: None,
+            });
+        }
+        let mut args = vec![];
+        let settings = loop {
+            if let Some(settings) = self.parse_settings()? {
+                break Some(settings);
+            }
+            args.push(self.parse_function_args()?);
+            if self.is_parse_comma_separated_end() {
+                break None;
+            }
+        };
+        self.expect_token(&Token::RParen)?;
+        Ok(TableFunctionArgs { args, settings })
+    }
+
     /// Parses a potentially empty list of arguments to a window function
     /// (including the closing parenthesis).
     ///

tests/sqlparser_clickhouse.rs

@@ -1151,6 +1151,83 @@ fn parse_create_table_on_commit_and_as_query() {
     }
 }
 
+#[test]
+fn parse_select_table_function_settings() {
+    fn check_settings(sql: &str, expected: &TableFunctionArgs) {
+        match clickhouse_and_generic().verified_stmt(sql) {
+            Statement::Query(q) => {
+                let from = &q.body.as_select().unwrap().from;
+                assert_eq!(from.len(), 1);
+                assert_eq!(from[0].joins, vec![]);
+                match &from[0].relation {
+                    Table { args, .. } => {
+                        let args = args.as_ref().unwrap();
+                        assert_eq!(args, expected);
+                    }
+                    _ => unreachable!(),
+                }
+            }
+            _ => unreachable!(),
+        }
+    }
+    check_settings(
+        "SELECT * FROM table_function(arg, SETTINGS s0 = 3, s1 = 's')",
+        &TableFunctionArgs {
+            args: vec![FunctionArg::Unnamed(FunctionArgExpr::Expr(
+                Expr::Identifier("arg".into()),
+            ))],
+
+            settings: Some(vec![
+                Setting {
+                    key: "s0".into(),
+                    value: Value::Number("3".parse().unwrap(), false),
+                },
+                Setting {
+                    key: "s1".into(),
+                    value: Value::SingleQuotedString("s".into()),
+                },
+            ]),
+        },
+    );
+    check_settings(
+        r#"SELECT * FROM table_function(arg)"#,
+        &TableFunctionArgs {
+            args: vec![FunctionArg::Unnamed(FunctionArgExpr::Expr(
+                Expr::Identifier("arg".into()),
+            ))],
+            settings: None,
+        },
+    );
+    check_settings(
+        "SELECT * FROM table_function(SETTINGS s0 = 3, s1 = 's')",
+        &TableFunctionArgs {
+            args: vec![],
+            settings: Some(vec![
+                Setting {
+                    key: "s0".into(),
+                    value: Value::Number("3".parse().unwrap(), false),
+                },
+                Setting {
+                    key: "s1".into(),
+                    value: Value::SingleQuotedString("s".into()),
+                },
+            ]),
+        },
+    );
+    let invalid_cases = vec![
+        "SELECT * FROM t(SETTINGS a)",
+        "SELECT * FROM t(SETTINGS a=)",
+        "SELECT * FROM t(SETTINGS a=1, b)",
+        "SELECT * FROM t(SETTINGS a=1, b=)",
+        "SELECT * FROM t(SETTINGS a=1, b=c)",
+    ];
+    for sql in invalid_cases {
+        clickhouse_and_generic()
+            .parse_sql_statements(sql)
+            .expect_err("Expected: SETTINGS key = value, found: ");
+    }
+}
+
 fn clickhouse() -> TestedDialects {
     TestedDialects {
         dialects: vec![Box::new(ClickHouseDialect {})],