BigQuery: Add support for BEGIN (apache#1718)

Author: iffyio

src/ast/mod.rs

@@ -3072,6 +3072,28 @@ pub enum Statement {
         begin: bool,
         transaction: Option<BeginTransactionKind>,
         modifier: Option<TransactionModifier>,
+        /// List of statements belonging to the `BEGIN` block.
+        /// Example:
+        /// ```sql
+        /// BEGIN
+        ///     SELECT 1;
+        ///     SELECT 2;
+        /// END;
+        /// ```
+        statements: Vec<Statement>,
+        /// Statements of an exception clause.
+        /// Example:
+        /// ```sql
+        /// BEGIN
+        ///     SELECT 1;
+        /// EXCEPTION WHEN ERROR THEN
+        ///     SELECT 2;
+        ///     SELECT 3;
+        /// END;
+        /// <https://cloud.google.com/bigquery/docs/reference/standard-sql/procedural-language#beginexceptionend>
+        exception_statements: Option<Vec<Statement>>,
+        /// TRUE if the statement has an `END` keyword.
+        has_end_keyword: bool,
     },
     /// ```sql
     /// SET TRANSACTION ...
@@ -4815,6 +4837,9 @@ impl fmt::Display for Statement {
                 begin: syntax_begin,
                 transaction,
                 modifier,
+                statements,
+                exception_statements,
+                has_end_keyword,
             } => {
                 if *syntax_begin {
                     if let Some(modifier) = *modifier {
@@ -4831,6 +4856,24 @@ impl fmt::Display for Statement {
                 if !modes.is_empty() {
                     write!(f, " {}", display_comma_separated(modes))?;
                 }
+                if !statements.is_empty() {
+                    write!(f, " {}", display_separated(statements, "; "))?;
+                    // We manually insert semicolon for the last statement,
+                    // since display_separated doesn't handle that case.
+                    write!(f, ";")?;
+                }
+                if let Some(exception_statements) = exception_statements {
+                    write!(f, " EXCEPTION WHEN ERROR THEN")?;
+                    if !exception_statements.is_empty() {
+                        write!(f, " {}", display_separated(exception_statements, "; "))?;
+                        // We manually insert semicolon for the last statement,
+                        // since display_separated doesn't handle that case.
+                        write!(f, ";")?;
+                    }
+                }
+                if *has_end_keyword {
+                    write!(f, " END")?;
+                }
                 Ok(())
             }
             Statement::SetTransaction {

src/dialect/bigquery.rs

@@ -15,9 +15,10 @@
 // specific language governing permissions and limitations
 // under the License.
 
+use crate::ast::Statement;
 use crate::dialect::Dialect;
 use crate::keywords::Keyword;
-use crate::parser::Parser;
+use crate::parser::{Parser, ParserError};
 
 /// These keywords are disallowed as column identifiers. Such that
 /// `SELECT 5 AS <col> FROM T` is rejected by BigQuery.
@@ -44,7 +45,11 @@ const RESERVED_FOR_COLUMN_ALIAS: &[Keyword] = &[
 pub struct BigQueryDialect;
 
 impl Dialect for BigQueryDialect {
-    // See https://cloud.google.com/bigquery/docs/reference/standard-sql/lexical#identifiers
+    fn parse_statement(&self, parser: &mut Parser) -> Option<Result<Statement, ParserError>> {
+        self.maybe_parse_statement(parser)
+    }
+
+    /// See <https://cloud.google.com/bigquery/docs/reference/standard-sql/lexical#identifiers>
     fn is_delimited_identifier_start(&self, ch: char) -> bool {
         ch == '`'
     }
@@ -60,6 +65,9 @@ impl Dialect for BigQueryDialect {
 
     fn is_identifier_start(&self, ch: char) -> bool {
         ch.is_ascii_lowercase() || ch.is_ascii_uppercase() || ch == '_'
+            // BigQuery supports `@@foo.bar` variable syntax in its procedural language.
+            // https://cloud.google.com/bigquery/docs/reference/standard-sql/procedural-language#beginexceptionend
+            || ch == '@'
     }
 
     fn is_identifier_part(&self, ch: char) -> bool {
@@ -129,3 +137,48 @@ impl Dialect for BigQueryDialect {
         !RESERVED_FOR_COLUMN_ALIAS.contains(kw)
     }
 }
+
+impl BigQueryDialect {
+    fn maybe_parse_statement(&self, parser: &mut Parser) -> Option<Result<Statement, ParserError>> {
+        if parser.peek_keyword(Keyword::BEGIN) {
+            return Some(self.parse_begin(parser));
+        }
+        None
+    }
+
+    /// Parse a `BEGIN` statement.
+    /// <https://cloud.google.com/bigquery/docs/reference/standard-sql/procedural-language#beginexceptionend>
+    fn parse_begin(&self, parser: &mut Parser) -> Result<Statement, ParserError> {
+        parser.expect_keyword(Keyword::BEGIN)?;
+
+        let statements = parser.parse_statement_list(&[Keyword::EXCEPTION, Keyword::END])?;
+
+        let has_exception_when_clause = parser.parse_keywords(&[
+            Keyword::EXCEPTION,
+            Keyword::WHEN,
+            Keyword::ERROR,
+            Keyword::THEN,
+        ]);
+        let exception_statements = if has_exception_when_clause {
+            if !parser.peek_keyword(Keyword::END) {
+                Some(parser.parse_statement_list(&[Keyword::END])?)
+            } else {
+                Some(Default::default())
+            }
+        } else {
+            None
+        };
+
+        parser.expect_keyword(Keyword::END)?;
+
+        Ok(Statement::StartTransaction {
+            begin: true,
+            statements,
+            exception_statements,
+            has_end_keyword: true,
+            transaction: None,
+            modifier: None,
+            modes: Default::default(),
+        })
+    }
+}

src/parser/mod.rs

@@ -4273,6 +4273,27 @@ impl<'a> Parser<'a> {
         self.parse_comma_separated(f)
     }
 
+    /// Parses 0 or more statements, each followed by a semicolon.
+    /// If the next token is any of `terminal_keywords` then no more
+    /// statements will be parsed.
+    pub(crate) fn parse_statement_list(
+        &mut self,
+        terminal_keywords: &[Keyword],
+    ) -> Result<Vec<Statement>, ParserError> {
+        let mut values = vec![];
+        loop {
+            if let Token::Word(w) = &self.peek_nth_token_ref(0).token {
+                if w.quote_style.is_none() && terminal_keywords.contains(&w.keyword) {
+                    break;
+                }
+            }
+
+            values.push(self.parse_statement()?);
+            self.expect_token(&Token::SemiColon)?;
+        }
+        Ok(values)
+    }
+
     /// Default implementation of a predicate that returns true if
     /// the specified keyword is reserved for column alias.
     /// See [Dialect::is_column_alias]
@@ -13783,6 +13804,9 @@ impl<'a> Parser<'a> {
             begin: false,
             transaction: Some(BeginTransactionKind::Transaction),
             modifier: None,
+            statements: vec![],
+            exception_statements: None,
+            has_end_keyword: false,
         })
     }
 
@@ -13812,6 +13836,9 @@ impl<'a> Parser<'a> {
             begin: true,
             transaction,
             modifier,
+            statements: vec![],
+            exception_statements: None,
+            has_end_keyword: false,
         })
     }
 

tests/sqlparser_bigquery.rs

@@ -236,6 +236,52 @@ fn parse_big_query_non_reserved_column_alias() {
     bigquery().verified_stmt(sql);
 }
 
+#[test]
+fn parse_at_at_identifier() {
+    bigquery().verified_stmt("SELECT @@error.stack_trace, @@error.message");
+}
+
+#[test]
+fn parse_begin() {
+    let sql = r#"BEGIN SELECT 1; EXCEPTION WHEN ERROR THEN SELECT 2; END"#;
+    let Statement::StartTransaction {
+        statements,
+        exception_statements,
+        has_end_keyword,
+        ..
+    } = bigquery().verified_stmt(sql)
+    else {
+        unreachable!();
+    };
+    assert_eq!(1, statements.len());
+    assert_eq!(1, exception_statements.unwrap().len());
+    assert!(has_end_keyword);
+
+    bigquery().verified_stmt(
+        "BEGIN SELECT 1; SELECT 2; EXCEPTION WHEN ERROR THEN SELECT 2; SELECT 4; END",
+    );
+    bigquery()
+        .verified_stmt("BEGIN SELECT 1; EXCEPTION WHEN ERROR THEN SELECT @@error.stack_trace; END");
+    bigquery().verified_stmt("BEGIN EXCEPTION WHEN ERROR THEN SELECT 2; END");
+    bigquery().verified_stmt("BEGIN SELECT 1; SELECT 2; EXCEPTION WHEN ERROR THEN END");
+    bigquery().verified_stmt("BEGIN EXCEPTION WHEN ERROR THEN END");
+    bigquery().verified_stmt("BEGIN SELECT 1; SELECT 2; END");
+    bigquery().verified_stmt("BEGIN END");
+
+    assert_eq!(
+        bigquery()
+            .parse_sql_statements("BEGIN SELECT 1; SELECT 2 END")
+            .unwrap_err(),
+        ParserError::ParserError("Expected: ;, found: END".to_string())
+    );
+    assert_eq!(
+        bigquery()
+            .parse_sql_statements("BEGIN SELECT 1; EXCEPTION WHEN ERROR THEN SELECT 2 END")
+            .unwrap_err(),
+        ParserError::ParserError("Expected: ;, found: END".to_string())
+    );
+}
+
 #[test]
 fn parse_delete_statement() {
     let sql = "DELETE \"table\" WHERE 1";

tests/sqlparser_common.rs

@@ -8343,7 +8343,12 @@ fn lateral_function() {
 
 #[test]
 fn parse_start_transaction() {
-    match verified_stmt("START TRANSACTION READ ONLY, READ WRITE, ISOLATION LEVEL SERIALIZABLE") {
+    let dialects = all_dialects_except(|d|
+        // BigQuery does not support this syntax
+        d.is::<BigQueryDialect>());
+    match dialects
+        .verified_stmt("START TRANSACTION READ ONLY, READ WRITE, ISOLATION LEVEL SERIALIZABLE")
+    {
         Statement::StartTransaction { modes, .. } => assert_eq!(
             modes,
             vec![
@@ -8357,7 +8362,7 @@ fn parse_start_transaction() {
 
     // For historical reasons, PostgreSQL allows the commas between the modes to
     // be omitted.
-    match one_statement_parses_to(
+    match dialects.one_statement_parses_to(
         "START TRANSACTION READ ONLY READ WRITE ISOLATION LEVEL SERIALIZABLE",
         "START TRANSACTION READ ONLY, READ WRITE, ISOLATION LEVEL SERIALIZABLE",
     ) {
@@ -8372,40 +8377,40 @@ fn parse_start_transaction() {
         _ => unreachable!(),
     }
 
-    verified_stmt("START TRANSACTION");
-    verified_stmt("BEGIN");
-    verified_stmt("BEGIN WORK");
-    verified_stmt("BEGIN TRANSACTION");
+    dialects.verified_stmt("START TRANSACTION");
+    dialects.verified_stmt("BEGIN");
+    dialects.verified_stmt("BEGIN WORK");
+    dialects.verified_stmt("BEGIN TRANSACTION");
 
-    verified_stmt("START TRANSACTION ISOLATION LEVEL READ UNCOMMITTED");
-    verified_stmt("START TRANSACTION ISOLATION LEVEL READ COMMITTED");
-    verified_stmt("START TRANSACTION ISOLATION LEVEL REPEATABLE READ");
-    verified_stmt("START TRANSACTION ISOLATION LEVEL SERIALIZABLE");
+    dialects.verified_stmt("START TRANSACTION ISOLATION LEVEL READ UNCOMMITTED");
+    dialects.verified_stmt("START TRANSACTION ISOLATION LEVEL READ COMMITTED");
+    dialects.verified_stmt("START TRANSACTION ISOLATION LEVEL REPEATABLE READ");
+    dialects.verified_stmt("START TRANSACTION ISOLATION LEVEL SERIALIZABLE");
 
     // Regression test for https://github.com/sqlparser-rs/sqlparser-rs/pull/139,
     // in which START TRANSACTION would fail to parse if followed by a statement
     // terminator.
     assert_eq!(
-        parse_sql_statements("START TRANSACTION; SELECT 1"),
+        dialects.parse_sql_statements("START TRANSACTION; SELECT 1"),
         Ok(vec![
             verified_stmt("START TRANSACTION"),
             verified_stmt("SELECT 1"),
         ])
     );
 
-    let res = parse_sql_statements("START TRANSACTION ISOLATION LEVEL BAD");
+    let res = dialects.parse_sql_statements("START TRANSACTION ISOLATION LEVEL BAD");
     assert_eq!(
         ParserError::ParserError("Expected: isolation level, found: BAD".to_string()),
         res.unwrap_err()
     );
 
-    let res = parse_sql_statements("START TRANSACTION BAD");
+    let res = dialects.parse_sql_statements("START TRANSACTION BAD");
     assert_eq!(
         ParserError::ParserError("Expected: end of statement, found: BAD".to_string()),
         res.unwrap_err()
     );
 
-    let res = parse_sql_statements("START TRANSACTION READ ONLY,");
+    let res = dialects.parse_sql_statements("START TRANSACTION READ ONLY,");
     assert_eq!(
         ParserError::ParserError("Expected: transaction mode, found: EOF".to_string()),
         res.unwrap_err()

tests/sqlparser_sqlite.rs

@@ -518,23 +518,6 @@ fn parse_start_transaction_with_modifier() {
     sqlite_and_generic().verified_stmt("BEGIN DEFERRED");
     sqlite_and_generic().verified_stmt("BEGIN IMMEDIATE");
     sqlite_and_generic().verified_stmt("BEGIN EXCLUSIVE");
-
-    let unsupported_dialects = all_dialects_except(|d| d.supports_start_transaction_modifier());
-    let res = unsupported_dialects.parse_sql_statements("BEGIN DEFERRED");
-    assert_eq!(
-        ParserError::ParserError("Expected: end of statement, found: DEFERRED".to_string()),
-        res.unwrap_err(),
-    );
-    let res = unsupported_dialects.parse_sql_statements("BEGIN IMMEDIATE");
-    assert_eq!(
-        ParserError::ParserError("Expected: end of statement, found: IMMEDIATE".to_string()),
-        res.unwrap_err(),
-    );
-    let res = unsupported_dialects.parse_sql_statements("BEGIN EXCLUSIVE");
-    assert_eq!(
-        ParserError::ParserError("Expected: end of statement, found: EXCLUSIVE".to_string()),
-        res.unwrap_err(),
-    );
 }
 
 #[test]