Handle empty branches during grammr parsing

Author: JanJakes

custom-parser/grammar-factoring/MySQLParser-factored-versioned.json

@@ -10184,6 +10184,9 @@
         "bnf": [
             [
                 "%xaConvert00"
+            ],
+            [
+                "\u03b5"
             ]
         ]
     },
@@ -11570,6 +11573,9 @@
         "bnf": [
             [
                 "%slaveConnectionOptions00"
+            ],
+            [
+                "\u03b5"
             ]
         ]
     },
@@ -12435,11 +12441,47 @@
             ]
         ]
     },
+    {
+        "name": "createUser",
+        "bnf": [
+            [
+                "CREATE_SYMBOL",
+                "USER_SYMBOL",
+                "%createUser02",
+                "createUserList",
+                "defaultRoleClause",
+                "createUserTail"
+            ]
+        ]
+    },
+    {
+        "versions": "serverVersion >= 50706",
+        "name": "%createUser0200",
+        "bnf": [
+            [
+                "ifNotExists"
+            ]
+        ]
+    },
+    {
+        "name": "%createUser02",
+        "bnf": [
+            [
+                "%createUser0200"
+            ],
+            [
+                "\u03b5"
+            ]
+        ]
+    },
     {
         "name": "createUserTail",
         "bnf": [
             [
                 "%createUserTail00"
+            ],
+            [
+                "\u03b5"
             ]
         ]
     },
@@ -12496,6 +12538,9 @@
         "bnf": [
             [
                 "%defaultRoleClause00"
+            ],
+            [
+                "\u03b5"
             ]
         ]
     },
@@ -15603,6 +15648,9 @@
         "bnf": [
             [
                 "%nonBlocking00"
+            ],
+            [
+                "\u03b5"
             ]
         ]
     },
@@ -24202,6 +24250,9 @@
         "bnf": [
             [
                 "%createTableOption21100"
+            ],
+            [
+                "\u03b5"
             ]
         ]
     },
@@ -30457,30 +30508,6 @@
             ]
         ]
     },
-    {
-        "name": "ifNotExists_zero_or_one",
-        "bnf": [
-            [
-                "ifNotExists"
-            ],
-            [
-                "\u03b5"
-            ]
-        ]
-    },
-    {
-        "name": "createUser",
-        "bnf": [
-            [
-                "CREATE_SYMBOL",
-                "USER_SYMBOL",
-                "ifNotExists_zero_or_one",
-                "createUserList",
-                "defaultRoleClause",
-                "createUserTail"
-            ]
-        ]
-    },
     {
         "name": "castType",
         "bnf": [

custom-parser/grammar-factoring/MySQLParser.g4

@@ -1765,10 +1765,10 @@ userFunction:
     USER_SYMBOL parentheses
 ;
 
-//createUser:
-//    CREATE_SYMBOL USER_SYMBOL ({serverVersion >= 50706}? ifNotExists | /* empty */) createUserList defaultRoleClause
-//        createUserTail
-//;
+createUser:
+    CREATE_SYMBOL USER_SYMBOL ({serverVersion >= 50706}? ifNotExists | /* empty */) createUserList defaultRoleClause
+        createUserTail
+;
 
 createUserTail:
     {serverVersion >= 50706}? requireClause? connectOptions? accountLockPasswordExpireOptions*
@@ -5036,13 +5036,6 @@ roleOrLabelKeyword:
  */
 alterOrderList: qualifiedIdentifier direction? (COMMA_SYMBOL qualifiedIdentifier direction?)*;
 
-/*
- * Fix CREATE USER statement.
- * The original grammar contains "({serverVersion >= 50706}? ifNotExists | / * empty * /)",
- * but the optionality of "| / * empty * /" part was lost during the conversion.
- */
-createUser: CREATE_SYMBOL USER_SYMBOL ifNotExists? createUserList defaultRoleClause createUserTail;
-
 /*
  * Fix CAST(2024 AS YEAR).
  * The original grammar was missing the YEAR_SYMBOL in the "castType" rule.

custom-parser/grammar-factoring/convert-grammar.php

@@ -20,13 +20,23 @@ function process_rule(string $rule) {
 
 	$parens_regex = PARENS_REGEX;
 
+	// Match empty branches in the original grammar. The equal to "ε", making the parent optional.
+	// This matches a "|" not followed by any rule, e.g. (A | B |) or (A | | B), etc.
+	$empty_branch_regex = '\|(?=\s*(?:\||\)|$))';
+
 	// extract rule branches (split by | not inside parentheses)
-	preg_match_all("/((?:[^()|]|$parens_regex)+)/", $rule, $matches);
+	preg_match_all("/((?:[^()|]|$parens_regex)+|$empty_branch_regex)/", $rule, $matches);
 	$branches = $matches[0];
 	$subrules = [];
 	foreach ($branches as $branch) {
 		$branch = trim($branch);
 
+		// empty branch equals to "ε"
+		if ($branch === '|') {
+			$subrules[] = ["ε"];
+			continue;
+		}
+
 		// extract version specifiers (like "{serverVersion >= 80000}?")
 		$versions = null;
 		if (preg_match('/^\{(.+?)}\?\s+(.*)$/s', $branch, $matches)) {