parse COLLATE operator

src/Ast/Expr/Collate.php

@@ -0,0 +1,25 @@
+<?php
+
+declare(strict_types=1);
+
+namespace MariaStan\Ast\Expr;
+
+use MariaStan\Ast\BaseNode;
+use MariaStan\Parser\Position;
+
+final class Collate extends BaseNode implements Expr
+{
+	public function __construct(
+		Position $startPosition,
+		Position $endPosition,
+		public readonly Expr $expression,
+		public readonly string $collation,
+	) {
+		parent::__construct($startPosition, $endPosition);
+	}
+
+	public static function getExprType(): ExprTypeEnum
+	{
+		return ExprTypeEnum::COLLATE;
+	}
+}

src/Ast/Expr/ExprTypeEnum.php

@@ -68,4 +68,7 @@ enum ExprTypeEnum: string
 
 	/** @see ColumnDefault */
 	case COLUMN_DEFAULT = 'COLUMN_DEFAULT';
+
+	/** @see Collate */
+	case COLLATE = 'COLLATE';
 }

src/Ast/Expr/SpecialOpTypeEnum.php

@@ -9,6 +9,9 @@ enum SpecialOpTypeEnum: string
 	/** @see Between */
 	case BETWEEN = 'BETWEEN';
 
+	/** @see Collate */
+	case COLLATE = 'COLLATE';
+
 	/** @see Is */
 	case IS = 'IS';
 

src/Parser/Exception/ShouldNotHappenException.php

@@ -0,0 +1,9 @@
+<?php
+
+declare(strict_types=1);
+
+namespace MariaStan\Parser\Exception;
+
+class ShouldNotHappenException extends ParserException
+{
+}

src/Parser/MariaDbParserState.php

@@ -22,6 +22,7 @@
 use MariaStan\Ast\Expr\CastType\FloatCastType;
 use MariaStan\Ast\Expr\CastType\IntegerCastType;
 use MariaStan\Ast\Expr\CastType\TimeCastType;
+use MariaStan\Ast\Expr\Collate;
 use MariaStan\Ast\Expr\Column;
 use MariaStan\Ast\Expr\ColumnDefault;
 use MariaStan\Ast\Expr\Exists;
@@ -86,6 +87,7 @@
 use MariaStan\Database\FunctionInfo\FunctionInfoRegistry;
 use MariaStan\Parser\Exception\MissingSubqueryAliasException;
 use MariaStan\Parser\Exception\ParserException;
+use MariaStan\Parser\Exception\ShouldNotHappenException;
 use MariaStan\Parser\Exception\UnexpectedTokenException;
 use MariaStan\Parser\Exception\UnsupportedQueryException;
 
@@ -1091,6 +1093,8 @@ private function parseExpression(int $precedence = 0): Expr
 				$exp = $this->parseRestOfIsOperator($exp);
 			} elseif ($operator === SpecialOpTypeEnum::IN) {
 				$exp = $this->parseRestOfInOperator($exp);
+			} elseif ($operator === SpecialOpTypeEnum::COLLATE) {
+				$exp = $this->parseRestOfCollateOperator($exp);
 			} else {
 				// INTERVAL is handled as a unary operator
 				assert($operator !== SpecialOpTypeEnum::INTERVAL);
@@ -1222,6 +1226,19 @@ private function parseRestOfInOperator(Expr $left): In
 		);
 	}
 
+	/** @throws ParserException */
+	private function parseRestOfCollateOperator(Expr $left): Collate
+	{
+		$token = $this->expectAnyOfTokens(TokenTypeEnum::LITERAL_STRING, TokenTypeEnum::IDENTIFIER);
+		$collation = match ($token->type) {
+			TokenTypeEnum::LITERAL_STRING => $this->cleanStringLiteral($token->content),
+			TokenTypeEnum::IDENTIFIER => $this->cleanIdentifier($token->content),
+			default => throw new ShouldNotHappenException(),
+		};
+
+		return new Collate($left->getStartPosition(), $token->getEndPosition(), $left, $collation);
+	}
+
 	/** @throws ParserException */
 	private function parseTimeUnit(): TimeUnitEnum
 	{
@@ -1266,6 +1283,7 @@ private function findBinaryOrSpecialOpFromToken(Token $token): BinaryOpTypeEnum|
 			TokenTypeEnum::BETWEEN => SpecialOpTypeEnum::BETWEEN,
 			TokenTypeEnum::IS => SpecialOpTypeEnum::IS,
 			TokenTypeEnum::LIKE => SpecialOpTypeEnum::LIKE,
+			TokenTypeEnum::COLLATE => SpecialOpTypeEnum::COLLATE,
 			default => null,
 		};
 	}
@@ -1288,8 +1306,7 @@ private function getOperatorPrecedence(BinaryOpTypeEnum|UnaryOpTypeEnum|SpecialO
 		// https://mariadb.com/kb/en/operator-precedence/
 		return match ($op) {
 			SpecialOpTypeEnum::INTERVAL => 17,
-			// COLLATE => 16
-			UnaryOpTypeEnum::BINARY => 16,
+			UnaryOpTypeEnum::BINARY, SpecialOpTypeEnum::COLLATE => 16,
 			UnaryOpTypeEnum::LOGIC_NOT => 15,
 			UnaryOpTypeEnum::PLUS, UnaryOpTypeEnum::MINUS, UnaryOpTypeEnum::BITWISE_NOT => 14,
 			// || as string concat => 13

src/Util/MariaDbErrorCodes.php

@@ -52,6 +52,9 @@ class MariaDbErrorCodes
 	// 1264	22003	ER_WARN_DATA_OUT_OF_RANGE	Out of range value for column '%s' at row %ld
 	public const ER_WARN_DATA_OUT_OF_RANGE = 1264;
 
+	// 1273	HY000	ER_UNKNOWN_COLLATION	Unknown collation: '%s'
+	public const ER_UNKNOWN_COLLATION = 1273;
+
 	// 1292	22007	ER_TRUNCATED_WRONG_VALUE	Truncated incorrect %s value: '%s'
 	public const ER_TRUNCATED_WRONG_VALUE = 1292;
 

tests/Parser/MariaDbParserTest.php

@@ -126,6 +126,8 @@ public function testParseInvalid(string $name, string $code, string $expectedOut
 					MariaDbErrorCodes::ER_UNKNOWN_DATA_TYPE,
 					// Some of these functions have custom syntax and in those cases it can be enforced on AST level.
 					MariaDbErrorCodes::ER_WRONG_PARAMCOUNT_TO_NATIVE_FCT,
+					// E.g. 'x' COLLATE CONCAT(...) returns this error instead of parser error.
+					MariaDbErrorCodes::ER_UNKNOWN_COLLATION,
 				],
 				true,
 			)

tests/code/Parser/MariaDbLexer/select.test

@@ -175,4 +175,12 @@ SELECT 000123
 -----
 SELECT: [SELECT]
 INT_LITERAL: [000123]
+END_OF_INPUT: []
+-----
+SELECT 'aaa' COLLATE 'utf8mb4_unicode_ci'
+-----
+SELECT: [SELECT]
+STRING_LITERAL: ['aaa']
+COLLATE: [COLLATE]
+STRING_LITERAL: ['utf8mb4_unicode_ci']
 END_OF_INPUT: []

tests/code/Parser/MariaDbParser/invalid/collate.test

@@ -0,0 +1,15 @@
+COLLATE
+-----
+SELECT 'a' COLLATE CONCAT('utf8mb4', '_unicode_ci')
+-----
+MariaStan\Parser\Exception\UnexpectedTokenException
+Expected END_OF_INPUT, got '(' after: SELECT 'a' COLLATE CONCAT
+#####
+1273: Unknown collation: 'CONCAT'
+-----
+SELECT 'a' COLLATE
+-----
+MariaStan\Parser\Exception\UnexpectedTokenException
+Expected one of STRING_LITERAL, IDENTIFIER, got END_OF_INPUT after: SELECT 'a' COLLATE
+#####
+1064: You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near '' at line 1