com.hazelcast.org.apache.calcite.sql.dialect.MssqlSqlDialect Maven / Gradle / Ivy
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to you under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package com.hazelcast.org.apache.calcite.sql.dialect;
import com.hazelcast.org.apache.calcite.avatica.util.TimeUnitRange;
import com.hazelcast.org.apache.calcite.config.NullCollation;
import com.hazelcast.org.apache.calcite.rel.type.RelDataTypeSystem;
import com.hazelcast.org.apache.calcite.sql.SqlAbstractDateTimeLiteral;
import com.hazelcast.org.apache.calcite.sql.SqlCall;
import com.hazelcast.org.apache.calcite.sql.SqlDialect;
import com.hazelcast.org.apache.calcite.sql.SqlFunction;
import com.hazelcast.org.apache.calcite.sql.SqlFunctionCategory;
import com.hazelcast.org.apache.calcite.sql.SqlIntervalLiteral;
import com.hazelcast.org.apache.calcite.sql.SqlIntervalQualifier;
import com.hazelcast.org.apache.calcite.sql.SqlKind;
import com.hazelcast.org.apache.calcite.sql.SqlLiteral;
import com.hazelcast.org.apache.calcite.sql.SqlNode;
import com.hazelcast.org.apache.calcite.sql.SqlNodeList;
import com.hazelcast.org.apache.calcite.sql.SqlUtil;
import com.hazelcast.org.apache.calcite.sql.SqlWriter;
import com.hazelcast.org.apache.calcite.sql.fun.SqlStdOperatorTable;
import com.hazelcast.org.apache.calcite.sql.parser.SqlParserPos;
import com.hazelcast.org.apache.calcite.sql.type.ReturnTypes;
/**
* A SqlDialect
implementation for the Microsoft SQL Server
* database.
*/
public class MssqlSqlDialect extends SqlDialect {
public static final Context DEFAULT_CONTEXT = SqlDialect.EMPTY_CONTEXT
.withDatabaseProduct(SqlDialect.DatabaseProduct.MSSQL)
.withIdentifierQuoteString("[")
.withCaseSensitive(false)
.withNullCollation(NullCollation.LOW);
public static final SqlDialect DEFAULT = new MssqlSqlDialect(DEFAULT_CONTEXT);
private static final SqlFunction MSSQL_SUBSTRING =
new SqlFunction("SUBSTRING", SqlKind.OTHER_FUNCTION,
ReturnTypes.ARG0_NULLABLE_VARYING, null, null,
SqlFunctionCategory.STRING);
/** Whether to generate "SELECT TOP(fetch)" rather than
* "SELECT ... FETCH NEXT fetch ROWS ONLY". */
private final boolean top;
/** Creates a MssqlSqlDialect. */
public MssqlSqlDialect(Context context) {
super(context);
// MSSQL 2008 (version 10) and earlier only supports TOP
// MSSQL 2012 (version 11) and higher supports OFFSET and FETCH
top = context.databaseMajorVersion() < 11;
}
/** {@inheritDoc}
*
* MSSQL does not support NULLS FIRST, so we emulate using CASE
* expressions. For example,
*
*
{@code ORDER BY x NULLS FIRST}
*
* becomes
*
*
* {@code ORDER BY CASE WHEN x IS NULL THEN 0 ELSE 1 END, x}
*
*/
@Override public SqlNode emulateNullDirection(SqlNode node,
boolean nullsFirst, boolean desc) {
// Default ordering preserved
if (nullCollation.isDefaultOrder(nullsFirst, desc)) {
return null;
}
// Grouping node should preserve grouping, no emulation needed
if (node.getKind() == SqlKind.GROUPING) {
return node;
}
// Emulate nulls first/last with case ordering
final SqlParserPos pos = SqlParserPos.ZERO;
final SqlNodeList whenList =
SqlNodeList.of(SqlStdOperatorTable.IS_NULL.createCall(pos, node));
final SqlNode oneLiteral = SqlLiteral.createExactNumeric("1", pos);
final SqlNode zeroLiteral = SqlLiteral.createExactNumeric("0", pos);
if (nullsFirst) {
// IS NULL THEN 0 ELSE 1 END
return SqlStdOperatorTable.CASE.createCall(null, pos,
null, whenList, SqlNodeList.of(zeroLiteral), oneLiteral);
} else {
// IS NULL THEN 1 ELSE 0 END
return SqlStdOperatorTable.CASE.createCall(null, pos,
null, whenList, SqlNodeList.of(oneLiteral), zeroLiteral);
}
}
@Override public void unparseOffsetFetch(SqlWriter writer, SqlNode offset,
SqlNode fetch) {
if (!top) {
super.unparseOffsetFetch(writer, offset, fetch);
}
}
@Override public void unparseTopN(SqlWriter writer, SqlNode offset,
SqlNode fetch) {
if (top) {
// Per Microsoft:
// "For backward compatibility, the parentheses are optional in SELECT
// statements. We recommend that you always use parentheses for TOP in
// SELECT statements. Doing so provides consistency with its required
// use in INSERT, UPDATE, MERGE, and DELETE statements."
//
// Note that "fetch" is ignored.
writer.keyword("TOP");
writer.keyword("(");
fetch.unparse(writer, -1, -1);
writer.keyword(")");
}
}
@Override public void unparseDateTimeLiteral(SqlWriter writer,
SqlAbstractDateTimeLiteral literal, int leftPrec, int rightPrec) {
writer.literal("'" + literal.toFormattedString() + "'");
}
@Override public void unparseCall(SqlWriter writer, SqlCall call,
int leftPrec, int rightPrec) {
if (call.getOperator() == SqlStdOperatorTable.SUBSTRING) {
if (call.operandCount() != 3) {
throw new IllegalArgumentException("MSSQL SUBSTRING requires FROM and FOR arguments");
}
SqlUtil.unparseFunctionSyntax(MSSQL_SUBSTRING, writer, call);
} else {
switch (call.getKind()) {
case FLOOR:
if (call.operandCount() != 2) {
super.unparseCall(writer, call, leftPrec, rightPrec);
return;
}
unparseFloor(writer, call);
break;
default:
super.unparseCall(writer, call, leftPrec, rightPrec);
}
}
}
@Override public boolean supportsCharSet() {
return false;
}
@Override public boolean supportsGroupByWithRollup() {
return true;
}
@Override public boolean supportsGroupByWithCube() {
return true;
}
/**
* Unparses datetime floor for Microsoft SQL Server.
* There is no TRUNC function, so simulate this using calls to CONVERT.
*
* @param writer Writer
* @param call Call
*/
private void unparseFloor(SqlWriter writer, SqlCall call) {
SqlLiteral node = call.operand(1);
TimeUnitRange unit = (TimeUnitRange) node.getValue();
switch (unit) {
case YEAR:
unparseFloorWithUnit(writer, call, 4, "-01-01");
break;
case MONTH:
unparseFloorWithUnit(writer, call, 7, "-01");
break;
case WEEK:
writer.print("CONVERT(DATETIME, CONVERT(VARCHAR(10), "
+ "DATEADD(day, - (6 + DATEPART(weekday, ");
call.operand(0).unparse(writer, 0, 0);
writer.print(")) % 7, ");
call.operand(0).unparse(writer, 0, 0);
writer.print("), 126))");
break;
case DAY:
unparseFloorWithUnit(writer, call, 10, "");
break;
case HOUR:
unparseFloorWithUnit(writer, call, 13, ":00:00");
break;
case MINUTE:
unparseFloorWithUnit(writer, call, 16, ":00");
break;
case SECOND:
unparseFloorWithUnit(writer, call, 19, ":00");
break;
default:
throw new IllegalArgumentException("MSSQL does not support FLOOR for time unit: "
+ unit);
}
}
@Override public void unparseSqlDatetimeArithmetic(SqlWriter writer,
SqlCall call, SqlKind sqlKind, int leftPrec, int rightPrec) {
final SqlWriter.Frame frame = writer.startFunCall("DATEADD");
SqlNode operand = call.operand(1);
if (operand instanceof SqlIntervalLiteral) {
//There is no DATESUB method available, so change the sign.
unparseSqlIntervalLiteralMssql(
writer, (SqlIntervalLiteral) operand, sqlKind == SqlKind.MINUS ? -1 : 1);
} else {
operand.unparse(writer, leftPrec, rightPrec);
}
writer.sep(",", true);
call.operand(0).unparse(writer, leftPrec, rightPrec);
writer.endList(frame);
}
@Override public void unparseSqlIntervalQualifier(SqlWriter writer,
SqlIntervalQualifier qualifier, RelDataTypeSystem typeSystem) {
switch (qualifier.timeUnitRange) {
case YEAR:
case QUARTER:
case MONTH:
case WEEK:
case DAY:
case HOUR:
case MINUTE:
case SECOND:
case MILLISECOND:
case MICROSECOND:
final String timeUnit = qualifier.timeUnitRange.startUnit.name();
writer.keyword(timeUnit);
break;
default:
throw new AssertionError("Unsupported type: " + qualifier.timeUnitRange);
}
if (null != qualifier.timeUnitRange.endUnit) {
throw new AssertionError("End unit is not supported now: "
+ qualifier.timeUnitRange.endUnit);
}
}
@Override public void unparseSqlIntervalLiteral(
SqlWriter writer, SqlIntervalLiteral literal, int leftPrec, int rightPrec) {
unparseSqlIntervalLiteralMssql(writer, literal, 1);
}
private void unparseSqlIntervalLiteralMssql(
SqlWriter writer, SqlIntervalLiteral literal, int sign) {
final SqlIntervalLiteral.IntervalValue interval =
(SqlIntervalLiteral.IntervalValue) literal.getValue();
unparseSqlIntervalQualifier(writer, interval.getIntervalQualifier(),
RelDataTypeSystem.DEFAULT);
writer.sep(",", true);
if (interval.getSign() * sign == -1) {
writer.print("-");
}
writer.literal(literal.getValue().toString());
}
private void unparseFloorWithUnit(SqlWriter writer, SqlCall call, int charLen,
String offset) {
writer.print("CONVERT");
SqlWriter.Frame frame = writer.startList("(", ")");
writer.print("DATETIME, CONVERT(VARCHAR(" + charLen + "), ");
call.operand(0).unparse(writer, 0, 0);
writer.print(", 126)");
if (offset.length() > 0) {
writer.print("+'" + offset + "'");
}
writer.endList(frame);
}
}