HHH-17639 Make recursive CTE cycle detection emulation independent of…

… collation
hibernate · Jan 15, 2024 · 8250f13 · 8250f13
1 parent 899bf7b
commit 8250f13
Show file tree

Hide file tree

Showing 8 changed files with 93 additions and 8 deletions.
diff --git a/...dialects/src/main/java/org/hibernate/community/dialect/MariaDBLegacySqlAstTranslator.java b/...dialects/src/main/java/org/hibernate/community/dialect/MariaDBLegacySqlAstTranslator.java
@@ -250,4 +250,13 @@ public void visitCastTarget(CastTarget castTarget) {
 			super.visitCastTarget( castTarget );
 		}
 	}
+
+	@Override
+	protected void renderStringContainsExactlyPredicate(Expression haystack, Expression needle) {
+		// MariaDB can't cope with NUL characters in the position function, so we use a like predicate instead
+		haystack.accept( this );
+		appendSql( " like concat('%',replace(replace(replace(" );
+		needle.accept( this );
+		appendSql( ",'~','~~'),'?','~?'),'%','~%'),'%') escape '~'" );
+	}
 }
diff --git a/...y-dialects/src/main/java/org/hibernate/community/dialect/MySQLLegacySqlAstTranslator.java b/...y-dialects/src/main/java/org/hibernate/community/dialect/MySQLLegacySqlAstTranslator.java
@@ -258,4 +258,13 @@ public void visitCastTarget(CastTarget castTarget) {
 			super.visitCastTarget( castTarget );
 		}
 	}
+
+	@Override
+	protected void renderStringContainsExactlyPredicate(Expression haystack, Expression needle) {
+		// MySQL can't cope with NUL characters in the position function, so we use a like predicate instead
+		haystack.accept( this );
+		appendSql( " like concat('%',replace(replace(replace(" );
+		needle.accept( this );
+		appendSql( ",'~','~~'),'?','~?'),'%','~%'),'%') escape '~'" );
+	}
 }
diff --git a/...alects/src/main/java/org/hibernate/community/dialect/SQLServerLegacySqlAstTranslator.java b/...alects/src/main/java/org/hibernate/community/dialect/SQLServerLegacySqlAstTranslator.java
@@ -470,4 +470,15 @@ enum OffsetFetchClauseMode {
 		TOP_ONLY,
 		EMULATED;
 	}
+
+	@Override
+	protected void renderStringContainsExactlyPredicate(Expression haystack, Expression needle) {
+		// SQL Server ignores NUL characters in string on case-insensitive collations, so we force a binary collation.
+		// This is needed for the emulation of cycle detection in recursive queries
+		appendSql( "charindex(" );
+		needle.accept( this );
+		appendSql( " collate Latin1_General_100_BIN2," );
+		haystack.accept( this );
+		append( ")>0" );
+	}
 }
diff --git a/hibernate-core/src/main/java/org/hibernate/dialect/MariaDBSqlAstTranslator.java b/hibernate-core/src/main/java/org/hibernate/dialect/MariaDBSqlAstTranslator.java
@@ -240,4 +240,13 @@ public void visitCastTarget(CastTarget castTarget) {
 		}
 	}
 
+	@Override
+	protected void renderStringContainsExactlyPredicate(Expression haystack, Expression needle) {
+		// MariaDB can't cope with NUL characters in the position function, so we use a like predicate instead
+		haystack.accept( this );
+		appendSql( " like concat('%',replace(replace(replace(" );
+		needle.accept( this );
+		appendSql( ",'~','~~'),'?','~?'),'%','~%'),'%') escape '~'" );
+	}
+
 }
diff --git a/hibernate-core/src/main/java/org/hibernate/dialect/MySQLSqlAstTranslator.java b/hibernate-core/src/main/java/org/hibernate/dialect/MySQLSqlAstTranslator.java
@@ -294,4 +294,13 @@ public void visitCastTarget(CastTarget castTarget) {
 			super.visitCastTarget( castTarget );
 		}
 	}
+
+	@Override
+	protected void renderStringContainsExactlyPredicate(Expression haystack, Expression needle) {
+		// MySQL can't cope with NUL characters in the position function, so we use a like predicate instead
+		haystack.accept( this );
+		appendSql( " like concat('%',replace(replace(replace(" );
+		needle.accept( this );
+		appendSql( ",'~','~~'),'?','~?'),'%','~%'),'%') escape '~'" );
+	}
 }
diff --git a/hibernate-core/src/main/java/org/hibernate/dialect/SQLServerSqlAstTranslator.java b/hibernate-core/src/main/java/org/hibernate/dialect/SQLServerSqlAstTranslator.java
@@ -14,6 +14,8 @@
 import org.hibernate.metamodel.mapping.JdbcMappingContainer;
 import org.hibernate.query.sqm.ComparisonOperator;
 import org.hibernate.query.sqm.FetchClauseType;
+import org.hibernate.query.sqm.function.AbstractSqmSelfRenderingFunctionDescriptor;
+import org.hibernate.query.sqm.function.SelfRenderingFunctionSqlAstExpression;
 import org.hibernate.sql.ast.Clause;
 import org.hibernate.sql.ast.SqlAstJoinType;
 import org.hibernate.sql.ast.spi.SqlSelection;
@@ -451,4 +453,15 @@ protected void renderMergeStatement(OptionalTableUpdate optionalTableUpdate) {
 		super.renderMergeStatement( optionalTableUpdate );
 		appendSql( ";" );
 	}
+
+	@Override
+	protected void renderStringContainsExactlyPredicate(Expression haystack, Expression needle) {
+		// SQL Server ignores NUL characters in string on case-insensitive collations, so we force a binary collation.
+		// This is needed for the emulation of cycle detection in recursive queries
+		appendSql( "charindex(" );
+		needle.accept( this );
+		appendSql( " collate Latin1_General_100_BIN2," );
+		haystack.accept( this );
+		append( ")>0" );
+	}
 }
diff --git a/hibernate-core/src/main/java/org/hibernate/dialect/TiDBSqlAstTranslator.java b/hibernate-core/src/main/java/org/hibernate/dialect/TiDBSqlAstTranslator.java
@@ -201,4 +201,13 @@ public void visitCastTarget(CastTarget castTarget) {
 			super.visitCastTarget( castTarget );
 		}
 	}
+
+	@Override
+	protected void renderStringContainsExactlyPredicate(Expression haystack, Expression needle) {
+		// TiDB can't cope with NUL characters in the position function, so we use a like predicate instead
+		haystack.accept( this );
+		appendSql( " like concat('%',replace(replace(replace(" );
+		needle.accept( this );
+		appendSql( ",'~','~~'),'?','~?'),'%','~%'),'%') escape '~'" );
+	}
 }
diff --git a/hibernate-core/src/main/java/org/hibernate/sql/ast/spi/AbstractSqlAstTranslator.java b/hibernate-core/src/main/java/org/hibernate/sql/ast/spi/AbstractSqlAstTranslator.java
@@ -2660,7 +2660,6 @@ private void emulateCycleClauseWithString(SelectClause selectClause) {
 					null,
 					stringType
 			);
-			arguments.add( new QueryLiteral<>( "%", stringType ) );
 			for ( CteColumn cycleColumn : currentCteStatement.getCycleColumns() ) {
 				final int selectionIndex = currentCteStatement.getCteTable()
 						.getCteColumns()
@@ -2683,14 +2682,20 @@ private void emulateCycleClauseWithString(SelectClause selectClause) {
 				arguments.add( nullSeparator );
 			}
 			arguments.add( nullSeparator );
-			arguments.add( new QueryLiteral<>( "%", stringType ) );
 
 			if ( !supportsRecursiveCycleClause() ) {
 				// Cycle mark
 				appendSql( "case when " );
-				visitColumnReference( cyclePathColumnReference );
-				appendSql( " like " );
-				concat.render( this, arguments, stringType, this );
+				renderStringContainsExactlyPredicate(
+						cyclePathColumnReference,
+						new SelfRenderingFunctionSqlAstExpression(
+								"concat",
+								concat,
+								arguments,
+								stringType,
+								stringType
+						)
+				);
 				appendSql( " then " );
 				currentCteStatement.getCycleValue().accept( this );
 				appendSql( " else " );
@@ -2699,9 +2704,8 @@ private void emulateCycleClauseWithString(SelectClause selectClause) {
 				appendSql( COMMA_SEPARATOR );
 			}
 
-			// Remove the wildcard literals
-			arguments.remove( arguments.size() - 1 );
-			arguments.set( 0, cyclePathColumnReference );
+			// Add the previous path
+			arguments.add( 0, cyclePathColumnReference );
 			// Cycle path
 			concat.render( this, arguments, stringType, this );
 		}
@@ -2750,6 +2754,18 @@ private void emulateCycleClauseWithString(SelectClause selectClause) {
 		}
 	}
 
+	protected void renderStringContainsExactlyPredicate(Expression haystack, Expression needle) {
+		final AbstractSqmSelfRenderingFunctionDescriptor position = findSelfRenderingFunction( "position", 2 );
+		new SelfRenderingFunctionSqlAstExpression(
+				"position",
+				position,
+				List.of( needle, haystack ),
+				getStringType(),
+				getStringType()
+		).accept( this );
+		append( ">0" );
+	}
+
 	/**
 	 * Wraps the given expression so that it produces a string, which should have the same ordering as the original value.
 	 * Here are the mappings for various data types: