Implement SUBSTRING(col [FROM <expr>] [FOR <expr>]) syntax (apache#293)

Author: Dandandan

src/ast/mod.rs

@@ -201,10 +201,17 @@ pub enum Expr {
         expr: Box<Expr>,
         data_type: DataType,
     },
+    /// EXTRACT(DateTimeField FROM <expr>)
     Extract {
         field: DateTimeField,
         expr: Box<Expr>,
     },
+    /// SUBSTRING(<expr> [FROM <expr>] [FOR <expr>])
+    Substring {
+        expr: Box<Expr>,
+        substring_from: Option<Box<Expr>>,
+        substring_for: Option<Box<Expr>>,
+    },
     /// `expr COLLATE collation`
     Collate {
         expr: Box<Expr>,
@@ -333,6 +340,21 @@ impl fmt::Display for Expr {
             Expr::Exists(s) => write!(f, "EXISTS ({})", s),
             Expr::Subquery(s) => write!(f, "({})", s),
             Expr::ListAgg(listagg) => write!(f, "{}", listagg),
+            Expr::Substring {
+                expr,
+                substring_from,
+                substring_for,
+            } => {
+                write!(f, "SUBSTRING({}", expr)?;
+                if let Some(from_part) = substring_from {
+                    write!(f, " FROM {}", from_part)?;
+                }
+                if let Some(from_part) = substring_for {
+                    write!(f, " FOR {}", from_part)?;
+                }
+
+                write!(f, ")")
+            }
         }
     }
 }

src/parser.rs

@@ -350,6 +350,7 @@ impl<'a> Parser<'a> {
                 Keyword::CAST => self.parse_cast_expr(),
                 Keyword::EXISTS => self.parse_exists_expr(),
                 Keyword::EXTRACT => self.parse_extract_expr(),
+                Keyword::SUBSTRING => self.parse_substring_expr(),
                 Keyword::INTERVAL => self.parse_literal_interval(),
                 Keyword::LISTAGG => self.parse_listagg_expr(),
                 Keyword::NOT => Ok(Expr::UnaryOp {
@@ -606,6 +607,27 @@ impl<'a> Parser<'a> {
         })
     }
 
+    pub fn parse_substring_expr(&mut self) -> Result<Expr, ParserError> {
+        // PARSE SUBSTRING (EXPR [FROM 1] [FOR 3])
+        self.expect_token(&Token::LParen)?;
+        let expr = self.parse_expr()?;
+        let mut from_expr = None;
+        let mut to_expr = None;
+        if self.parse_keyword(Keyword::FROM) {
+            from_expr = Some(self.parse_expr()?);
+        }
+        if self.parse_keyword(Keyword::FOR) {
+            to_expr = Some(self.parse_expr()?);
+        }
+        self.expect_token(&Token::RParen)?;
+
+        Ok(Expr::Substring {
+            expr: Box::new(expr),
+            substring_from: from_expr.map(Box::new),
+            substring_for: to_expr.map(Box::new),
+        })
+    }
+
     /// Parse a SQL LISTAGG expression, e.g. `LISTAGG(...) WITHIN GROUP (ORDER BY ...)`.
     pub fn parse_listagg_expr(&mut self) -> Result<Expr, ParserError> {
         self.expect_token(&Token::LParen)?;

tests/sqlparser_common.rs

@@ -2598,6 +2598,23 @@ fn parse_scalar_subqueries() {
     );
 }
 
+#[test]
+fn parse_substring() {
+    one_statement_parses_to("SELECT SUBSTRING('1')", "SELECT SUBSTRING('1')");
+
+    one_statement_parses_to(
+        "SELECT SUBSTRING('1' FROM 1)",
+        "SELECT SUBSTRING('1' FROM 1)",
+    );
+
+    one_statement_parses_to(
+        "SELECT SUBSTRING('1' FROM 1 FOR 3)",
+        "SELECT SUBSTRING('1' FROM 1 FOR 3)",
+    );
+
+    one_statement_parses_to("SELECT SUBSTRING('1' FOR 3)", "SELECT SUBSTRING('1' FOR 3)");
+}
+
 #[test]
 fn parse_exists_subquery() {
     let expected_inner = verified_query("SELECT 1");

tests/sqlparser_regression.rs

@@ -25,10 +25,9 @@ macro_rules! tpch_tests {
         #[test]
         fn $name() {
             let dialect = GenericDialect {};
-
             let res = Parser::parse_sql(&dialect, QUERIES[$value -1]);
-            // Ignore 6.sql and 22.sql
-            if $value != 6 && $value != 22 {
+            // Ignore 6.sql
+            if $value != 6 {
                 assert!(res.is_ok());
             }
         }