parser: clean & set transaction historical syntax (#491)

Various renames to prepare for the codegen'd stuff. AST node names need to match their SyntaxKind names.

Also added a test case and fixed parsing historical syntax for set transaction that I found when looking through the pg gram.y

Also delete a dead test helper

Author: sbdchd

crates/squawk_parser/src/grammar.rs

@@ -1654,10 +1654,10 @@ fn json_key_value(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
 }
 
 fn named_arg(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
-    assert!(p.at(FAT_ARROW) || p.at(COLONEQ));
+    assert!(p.at(FAT_ARROW) || p.at(COLON_EQ));
     let m = lhs.precede(p);
-    if p.at(COLONEQ) {
-        p.bump(COLONEQ);
+    if p.at(COLON_EQ) {
+        p.bump(COLON_EQ);
     } else {
         p.bump(FAT_ARROW);
     }
@@ -1668,9 +1668,9 @@ fn named_arg(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
 }
 
 fn cast_expr(p: &mut Parser<'_>, lhs: CompletedMarker) -> CompletedMarker {
-    assert!(p.at(COLON2));
+    assert!(p.at(COLON_COLON));
     let m = lhs.precede(p);
-    p.bump(COLON2);
+    p.bump(COLON_COLON);
     type_name(p);
     m.complete(p, CAST_EXPR)
 }
@@ -2089,9 +2089,9 @@ fn current_op(p: &Parser<'_>, r: &Restrictions) -> (u8, SyntaxKind, Associativit
         MINUS if p.next_not_joined_op(0) => (8, MINUS, Left), // symbol
         // Later on we return a NAMED_ARG for this instead of BIN_EXPR
         // :=
-        COLON if p.at(COLONEQ) => (5, COLONEQ, Right), // symbol
+        COLON if p.at(COLON_EQ) => (5, COLON_EQ, Right), // symbol
         // ::
-        COLON if p.at(COLON2) => (15, COLON2, Left), // symbol
+        COLON if p.at(COLON_COLON) => (15, COLON_COLON, Left), // symbol
         // Only used in json_object, like json_object('a' value 1) instead of json_object('a': 1)
         // value
         VALUE_KW if r.json_field_arg_allowed => (7, VALUE_KW, Right),
@@ -2159,11 +2159,11 @@ fn expr_bp(p: &mut Parser<'_>, bp: u8, r: &Restrictions) -> Option<CompletedMark
             break;
         }
         match op {
-            COLON2 => {
+            COLON_COLON => {
                 lhs = cast_expr(p, lhs);
                 continue;
             }
-            FAT_ARROW | COLONEQ => {
+            FAT_ARROW | COLON_EQ => {
                 lhs = named_arg(p, lhs);
                 continue;
             }
@@ -2740,15 +2740,15 @@ fn data_source(p: &mut Parser<'_>) {
 
 // USING data_source ON join_condition
 fn merge_using_clause(p: &mut Parser<'_>) {
-    let m1 = p.start();
+    let m = p.start();
     p.expect(USING_KW);
     data_source(p);
     p.expect(ON_KW);
     // join_condition
     if expr(p).is_none() {
         p.error("expected an expression");
     }
-    m1.complete(p, USING_CLAUSE);
+    m.complete(p, USING_CLAUSE);
 }
 
 // where from_item can be one of:
@@ -3119,8 +3119,7 @@ fn opt_with_params(p: &mut Parser<'_>) -> Option<CompletedMarker> {
 // [ INCLUDE ( column_name [, ... ] ) ]
 // [ WITH ( storage_parameter [= value] [, ... ] ) ]
 // [ USING INDEX TABLESPACE tablespace_name ]
-#[must_use]
-fn opt_index_parameters(p: &mut Parser<'_>) -> bool {
+fn opt_index_parameters(p: &mut Parser<'_>) {
     opt_include_columns(p);
     opt_with_params(p);
     if p.at(USING_KW) {
@@ -3131,7 +3130,6 @@ fn opt_index_parameters(p: &mut Parser<'_>) -> bool {
         name_ref(p);
         m.complete(p, CONSTRAINT_INDEX_TABLESPACE);
     }
-    true
 }
 
 // referential_action in a FOREIGN KEY/REFERENCES constraint is:
@@ -3293,18 +3291,14 @@ fn opt_constraint_inner(p: &mut Parser<'_>) -> Option<SyntaxKind> {
                 p.eat(NOT_KW);
                 p.expect(DISTINCT_KW);
             }
-            if !opt_index_parameters(p) {
-                p.error("expected index parameters");
-            }
+            opt_index_parameters(p);
             UNIQUE_CONSTRAINT
         }
         // PRIMARY KEY index_parameters
         PRIMARY_KW => {
             p.bump(PRIMARY_KW);
             p.expect(KEY_KW);
-            if !opt_index_parameters(p) {
-                p.error("expected index parameters");
-            }
+            opt_index_parameters(p);
             PRIMARY_KEY_CONSTRAINT
         }
         // REFERENCES reftable [ ( refcolumn ) ] [ MATCH FULL | MATCH PARTIAL | MATCH SIMPLE ]
@@ -3506,9 +3500,7 @@ fn table_constraint(p: &mut Parser<'_>) -> CompletedMarker {
                     p.eat(DISTINCT_KW);
                 }
                 column_list(p);
-                if !opt_index_parameters(p) {
-                    p.error("expected index parameters");
-                }
+                opt_index_parameters(p);
             }
             UNIQUE_CONSTRAINT
         }
@@ -3523,9 +3515,7 @@ fn table_constraint(p: &mut Parser<'_>) -> CompletedMarker {
             // ( column_name [, ... ] ) index_parameters
             } else {
                 column_list(p);
-                if !opt_index_parameters(p) {
-                    p.error("expected index parameters");
-                }
+                opt_index_parameters(p);
             }
             PRIMARY_KEY_CONSTRAINT
         }
@@ -3553,9 +3543,7 @@ fn table_constraint(p: &mut Parser<'_>) -> CompletedMarker {
             }
             p.expect(R_PAREN);
             m.complete(p, CONSTRAINT_EXCLUSIONS);
-            if !opt_index_parameters(p) {
-                p.error("expected index parameters");
-            }
+            opt_index_parameters(p);
             if p.at(WHERE_KW) {
                 let m = p.start();
                 p.bump(WHERE_KW);
@@ -3637,7 +3625,7 @@ fn opt_initally_constraint_option(p: &mut Parser<'_>) -> Option<CompletedMarker>
         (INITIALLY_KW, DEFERRED_KW) => {
             p.bump(INITIALLY_KW);
             p.bump(DEFERRED_KW);
-            INITALLY_DEFERRED_CONSTRAINT_OPTION
+            INITIALLY_DEFERRED_CONSTRAINT_OPTION
         }
         (INITIALLY_KW, IMMEDIATE_KW) => {
             p.bump(INITIALLY_KW);
@@ -3665,7 +3653,7 @@ fn opt_constraint_options(p: &mut Parser<'_>) {
         }
         (Some(deferrable), Some(initially)) => {
             if deferrable.kind() == NOT_DEFERRABLE_CONSTRAINT_OPTION
-                && initially.kind() == INITALLY_DEFERRED_CONSTRAINT_OPTION
+                && initially.kind() == INITIALLY_DEFERRED_CONSTRAINT_OPTION
             {
                 p.error("constraint declared INITIALLY DEFERRED must be DEFERRABLE");
             }
@@ -4176,9 +4164,8 @@ fn opt_target_el(p: &mut Parser) -> Option<CompletedMarker> {
         return None;
     } else if p.at(STAR) && !p.nth_at_ts(1, OPERATOR_FIRST) {
         p.bump(STAR);
-        true
     } else if expr(p).is_some() {
-        opt_as_col_label(p) || p.at(COMMA)
+        opt_as_col_label(p);
     } else {
         m.abandon(p);
         p.error(format!(
@@ -4562,7 +4549,7 @@ fn commit_stmt(p: &mut Parser<'_>) -> CompletedMarker {
             p.expect(CHAIN_KW);
         }
     }
-    m.complete(p, COMMIT_STMT)
+    m.complete(p, COMMIT)
 }
 
 const TRANSACTION_MODE_FIRST: TokenSet =
@@ -4635,7 +4622,7 @@ fn begin_stmt(p: &mut Parser<'_>) -> CompletedMarker {
         p.expect(TRANSACTION_KW);
         opt_transaction_mode_list(p);
     }
-    m.complete(p, BEGIN_STMT)
+    m.complete(p, BEGIN)
 }
 
 // Sconst
@@ -9179,7 +9166,7 @@ fn drop_aggregate_stmt(p: &mut Parser<'_>) -> CompletedMarker {
         aggregate(p);
     }
     opt_cascade_or_restrict(p);
-    m.complete(p, DROP_AGGREGATE_STMT)
+    m.complete(p, DROP_AGGREGATE)
 }
 
 fn source_type_as_target_type(p: &mut Parser<'_>) {
@@ -10554,6 +10541,18 @@ fn set_session_auth_stmt(p: &mut Parser<'_>) -> CompletedMarker {
     m.complete(p, SET_SESSION_AUTH_STMT)
 }
 
+fn transaction_mode_list(p: &mut Parser<'_>) {
+    // TODO: generalize
+    // transaction_mode [, ...]
+    while !p.at(EOF) && p.at_ts(TRANSACTION_MODE_FIRST) {
+        if !opt_transaction_mode(p) {
+            p.error("expected transaction mode");
+        }
+        // historical pg syntax doesn't require commas
+        p.eat(COMMA);
+    }
+}
+
 // SET TRANSACTION transaction_mode [, ...]
 // SET TRANSACTION SNAPSHOT snapshot_id
 // SET SESSION CHARACTERISTICS AS TRANSACTION transaction_mode [, ...]
@@ -10570,32 +10569,14 @@ fn set_transaction_stmt(p: &mut Parser<'_>) -> CompletedMarker {
         p.expect(CHARACTERISTICS_KW);
         p.expect(AS_KW);
         p.expect(TRANSACTION_KW);
-        // TODO: generalize
-        // transaction_mode [, ...]
-        while !p.at(EOF) {
-            if !opt_transaction_mode(p) {
-                p.error("expected transaction mode");
-            }
-            if !p.eat(COMMA) {
-                break;
-            }
-        }
+        transaction_mode_list(p);
     } else {
         p.expect(TRANSACTION_KW);
         // [ SNAPSHOT snapshot_id ]
         if p.eat(SNAPSHOT_KW) {
             string_literal(p);
         } else {
-            // TODO: generalize
-            // transaction_mode [, ...]
-            while !p.at(EOF) {
-                if !opt_transaction_mode(p) {
-                    break;
-                }
-                if !p.eat(COMMA) {
-                    break;
-                }
-            }
+            transaction_mode_list(p);
         }
     }
     m.complete(p, SET_TRANSACTION_STMT)
@@ -12017,7 +11998,7 @@ fn create_index_stmt(p: &mut Parser<'_>) -> CompletedMarker {
     }
     // [ WHERE predicate ]
     opt_where_clause(p);
-    m.complete(p, CREATE_INDEX_STMT)
+    m.complete(p, CREATE_INDEX)
 }
 
 // (
@@ -12060,7 +12041,7 @@ fn opt_param_mode(p: &mut Parser<'_>) -> Option<CompletedMarker> {
         IN_KW => {
             p.bump(IN_KW);
             if p.eat(OUT_KW) {
-                PARAM_INOUT
+                PARAM_IN_OUT
             } else {
                 PARAM_IN
             }
@@ -12071,7 +12052,7 @@ fn opt_param_mode(p: &mut Parser<'_>) -> Option<CompletedMarker> {
         }
         INOUT_KW => {
             p.bump(INOUT_KW);
-            PARAM_INOUT
+            PARAM_IN_OUT
         }
         _ => {
             m.abandon(p);

crates/squawk_parser/src/lib.rs

@@ -210,8 +210,8 @@ impl<'t> Parser<'t> {
             return false;
         }
         let n_raw_tokens = match kind {
-            SyntaxKind::COLON2
-            | SyntaxKind::COLONEQ
+            SyntaxKind::COLON_COLON
+            | SyntaxKind::COLON_EQ
             | SyntaxKind::NEQ
             | SyntaxKind::NEQB
             | SyntaxKind::LTEQ
@@ -508,14 +508,14 @@ impl<'t> Parser<'t> {
                 TrivaBetween::NotAllowed,
             ),
             // :=
-            SyntaxKind::COLONEQ => self.at_composite2(
+            SyntaxKind::COLON_EQ => self.at_composite2(
                 n,
                 SyntaxKind::COLON,
                 SyntaxKind::EQ,
                 TrivaBetween::NotAllowed,
             ),
             // ::
-            SyntaxKind::COLON2 => self.at_composite2(
+            SyntaxKind::COLON_COLON => self.at_composite2(
                 n,
                 SyntaxKind::COLON,
                 SyntaxKind::COLON,

crates/squawk_parser/src/syntax_kind.rs

@@ -56,9 +56,9 @@ pub enum SyntaxKind {
     /// `:`
     COLON,
     /// `::`
-    COLON2,
+    COLON_COLON,
     /// `:=`
-    COLONEQ,
+    COLON_EQ,
     /// `=`
     EQ,
     /// `=>`
@@ -1212,7 +1212,7 @@ pub enum SyntaxKind {
     ANALYZE_STMT,
     CLUSTER_STMT,
     COMMENT_STMT,
-    COMMIT_STMT,
+    COMMIT,
     CREATE_EXTENSION_STMT,
     CREATE_ACCESS_METHOD_STMT,
     CREATE_AGGREGATE_STMT,
@@ -1246,14 +1246,14 @@ pub enum SyntaxKind {
     CREATE_TEXT_SEARCH_PARSER_STMT,
     CREATE_TEXT_SEARCH_TEMPLATE_STMT,
     CREATE_TRANSFORM_STMT,
-    CREATE_INDEX_STMT,
+    CREATE_INDEX,
     CREATE_TYPE_STMT,
     CREATE_TRIGGER_STMT,
     CREATE_FUNCTION_STMT,
     PARAM,
     PARAM_IN,
     PARAM_OUT,
-    PARAM_INOUT,
+    PARAM_IN_OUT,
     PARAM_VARIADIC,
     BEGIN_FUNC_OPTION,
     RETURN_FUNC_OPTION,
@@ -1278,7 +1278,7 @@ pub enum SyntaxKind {
     OR_REPLACE,
     DROP_INDEX_STMT,
     DROP_TRIGGER_STMT,
-    BEGIN_STMT,
+    BEGIN,
     SHOW_STMT,
     SET_STMT,
     PREPARE_TRANSACTION_STMT,
@@ -1339,7 +1339,7 @@ pub enum SyntaxKind {
     DROP_CONVERSION_STMT,
     DROP_COLLATION_STMT,
     DROP_CAST_STMT,
-    DROP_AGGREGATE_STMT,
+    DROP_AGGREGATE,
     DROP_ACCESS_METHOD_STMT,
     DROP_USER_MAPPING_STMT,
     IMPORT_FOREIGN_SCHEMA,
@@ -1407,7 +1407,7 @@ pub enum SyntaxKind {
     CONSTRAINT_EXCLUSIONS,
     DEFERRABLE_CONSTRAINT_OPTION,
     NOT_DEFERRABLE_CONSTRAINT_OPTION,
-    INITALLY_DEFERRED_CONSTRAINT_OPTION,
+    INITIALLY_DEFERRED_CONSTRAINT_OPTION,
     INITIALLY_IMMEDIATE_CONSTRAINT_OPTION,
     CONSTRAINT_OPTION_LIST,
     SEQUENCE_OPTION_LIST,

crates/squawk_parser/tests/data/ok/set_transaction.sql

@@ -3,5 +3,8 @@ SET TRANSACTION SNAPSHOT '00000003-0000001B-1';
 
 SET SESSION CHARACTERISTICS AS TRANSACTION ISOLATION LEVEL READ COMMITTED, read write;
 
-SET TRANSACTION ISOLATION LEVEL SERIALIZABLE, READ WRITE, NOT DEFERRABLE
+SET TRANSACTION ISOLATION LEVEL SERIALIZABLE, READ WRITE, NOT DEFERRABLE;
 
+
+-- no commas is postgres historical according to gram.y
+SET TRANSACTION ISOLATION LEVEL SERIALIZABLE READ WRITE NOT DEFERRABLE;

crates/squawk_parser/tests/snapshots/tests__alter_aggregate_ok.snap

@@ -1,7 +1,6 @@
 ---
 source: crates/squawk_parser/tests/tests.rs
 input_file: crates/squawk_parser/tests/data/ok/alter_aggregate.sql
-snapshot_kind: text
 ---
 SOURCE_FILE
   COMMENT "-- star"

crates/squawk_parser/tests/snapshots/tests__alter_collation_ok.snap

@@ -1,7 +1,6 @@
 ---
 source: crates/squawk_parser/tests/tests.rs
 input_file: crates/squawk_parser/tests/data/ok/alter_collation.sql
-snapshot_kind: text
 ---
 SOURCE_FILE
   COMMENT "-- refresh"