Java Code Examples for org.apache.kylin.metadata.model.FunctionDesc#isCountDistinct()
The following examples show how to use
org.apache.kylin.metadata.model.FunctionDesc#isCountDistinct() .
You can vote up the ones you like or vote down the ones you don't like,
and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example 1
Source File: QueryGenerator.java From kylin with Apache License 2.0 | 6 votes |
public static String createMeasureStatement(List<MeasureDesc> measureList) { StringBuilder sql = new StringBuilder(); for (MeasureDesc measureDesc : measureList) { FunctionDesc functionDesc = measureDesc.getFunction(); if (functionDesc.isSum() || functionDesc.isMax() || functionDesc.isMin()) { sql.append("," + functionDesc.getExpression() + "(" + functionDesc.getParameter().getValue() + ")\n"); break; } else if (functionDesc.isCountDistinct()) { sql.append(",COUNT" + "(DISTINCT " + functionDesc.getParameter().getValue() + ")\n"); break; } } return sql.toString(); }
Example 2
Source File: CubeCapabilityChecker.java From Kylin with Apache License 2.0 | 6 votes |
private static boolean isWeaklyMatchedWithAggregations(Collection<FunctionDesc> aggregations, Collection<TblColRef> metricColumns, CubeInstance cube) { CubeDesc cubeDesc = cube.getDescriptor(); Collection<FunctionDesc> cubeFuncs = cubeDesc.listAllFunctions(); boolean matched = true; for (FunctionDesc functionDesc : aggregations) { if (cubeFuncs.contains(functionDesc)) continue; // only inverted-index cube does not have count, and let calcite handle in this case if (functionDesc.isCount()) continue; if (functionDesc.isCountDistinct()) // calcite can not handle distinct count matched = false; TblColRef col = functionDesc.selectTblColRef(metricColumns, cubeDesc.getFactTable()); if (col == null || !cubeDesc.listDimensionColumnsIncludingDerived().contains(col)) { matched = false; } } return matched; }
Example 3
Source File: FunctionRule.java From Kylin with Apache License 2.0 | 6 votes |
private void validateReturnType(ValidateContext context, CubeDesc cube, FunctionDesc funcDesc) { String func = funcDesc.getExpression(); DataType rtype = funcDesc.getReturnDataType(); if (funcDesc.isCount()) { if (rtype.isIntegerFamily() == false) { context.addResult(ResultLevel.ERROR, "Return type for function " + func + " must be one of " + DataType.INTEGER_FAMILY); } } else if (funcDesc.isCountDistinct()) { if (rtype.isHLLC() == false && funcDesc.isHolisticCountDistinct() == false) { context.addResult(ResultLevel.ERROR, "Return type for function " + func + " must be hllc(10), hllc(12) etc."); } } else if (funcDesc.isMax() || funcDesc.isMin() || funcDesc.isSum()) { if (rtype.isNumberFamily() == false) { context.addResult(ResultLevel.ERROR, "Return type for function " + func + " must be one of " + DataType.NUMBER_FAMILY); } } else { if (StringUtils.equalsIgnoreCase(KylinConfig.getInstanceFromEnv().getProperty(KEY_IGNORE_UNKNOWN_FUNC, "false"), "false")) { context.addResult(ResultLevel.ERROR, "Unrecognized function: [" + func + "]"); } } }
Example 4
Source File: OLAPAggregateRel.java From Kylin with Apache License 2.0 | 5 votes |
private AggregateCall rewriteAggregateCall(AggregateCall aggCall, FunctionDesc func) { // rebuild parameters List<Integer> newArgList = new ArrayList<Integer>(1); String fieldName = func.getRewriteFieldName(); RelDataTypeField field = getChild().getRowType().getField(fieldName, true); newArgList.add(field.getIndex()); // rebuild function RelDataType fieldType = aggCall.getType(); Aggregation newAgg = aggCall.getAggregation(); if (func.isCountDistinct()) { newAgg = createHyperLogLogAggFunction(fieldType); } else if (func.isCount()) { newAgg = new SqlSumEmptyIsZeroAggFunction(fieldType); } // rebuild aggregate call AggregateCall newAggCall = new AggregateCall(newAgg, false, newArgList, fieldType, newAgg.getName()); // To make sure specified type matches the inferReturnType, or otherwise // there will be assertion failure in optiq // The problem is BIGINT != BIGINT NOT NULL // Details see https://github.scm.corp.ebay.com/Kylin/Kylin/issues/323 SqlAggFunction aggFunction = (SqlAggFunction) newAggCall.getAggregation(); AggCallBinding callBinding = newAggCall.createBinding(this); RelDataType inferReturnType = aggFunction.inferReturnType(callBinding); return new AggregateCall(newAgg, false, newArgList, inferReturnType, newAgg.getName()); }
Example 5
Source File: RowValueDecoder.java From Kylin with Apache License 2.0 | 5 votes |
public boolean hasMemHungryCountDistinct() { for (int i = projectionIndex.nextSetBit(0); i >= 0; i = projectionIndex.nextSetBit(i + 1)) { FunctionDesc func = measures[i].getFunction(); if (func.isCountDistinct() && !func.isHolisticCountDistinct()) { return true; } } return false; }
Example 6
Source File: EndpointTupleIterator.java From Kylin with Apache License 2.0 | 5 votes |
/** * measure comes from query engine, does not contain enough information * * @param measures * @param columns */ private void rewriteMeasureParameters(List<FunctionDesc> measures, List<TblColRef> columns) { for (FunctionDesc functionDesc : measures) { if (functionDesc.isCount()) { functionDesc.setReturnType("bigint"); functionDesc.setReturnDataType(DataType.getInstance(functionDesc.getReturnType())); } else { boolean updated = false; for (TblColRef column : columns) { if (column.isSameAs(factTableName, functionDesc.getParameter().getValue())) { if (functionDesc.isCountDistinct()) { //TODO: default precision might need be configurable String iiDefaultHLLC = "hllc10"; functionDesc.setReturnType(iiDefaultHLLC); functionDesc.setReturnDataType(DataType.getInstance(iiDefaultHLLC)); } else { functionDesc.setReturnType(column.getColumn().getType().toString()); functionDesc.setReturnDataType(DataType.getInstance(functionDesc.getReturnType())); } functionDesc.getParameter().setColRefs(ImmutableList.of(column)); updated = true; break; } } if (!updated) { throw new RuntimeException("Func " + functionDesc + " is not related to any column in fact table " + factTableName); } } } }
Example 7
Source File: EndpointAggregators.java From Kylin with Apache License 2.0 | 5 votes |
public static EndpointAggregators fromFunctions(TableRecordInfo tableInfo, List<FunctionDesc> metrics) { String[] funcNames = new String[metrics.size()]; String[] dataTypes = new String[metrics.size()]; MetricInfo[] metricInfos = new MetricInfo[metrics.size()]; for (int i = 0; i < metrics.size(); i++) { FunctionDesc functionDesc = metrics.get(i); //TODO: what if funcionDesc's type is different from tablDesc? cause scale difference funcNames[i] = functionDesc.getExpression(); dataTypes[i] = functionDesc.getReturnType(); if (functionDesc.isCount()) { metricInfos[i] = new MetricInfo(MetricType.Count); } else if (functionDesc.isDimensionAsMetric()) { metricInfos[i] = new MetricInfo(MetricType.DimensionAsMetric); } else { int index = tableInfo.findFactTableColumn(functionDesc.getParameter().getValue()); if (index < 0) { throw new IllegalStateException("Column " + functionDesc.getParameter().getValue() + " is not found in II"); } if (functionDesc.isCountDistinct()) { metricInfos[i] = new MetricInfo(MetricType.DistinctCount, index, functionDesc.getReturnDataType().getPrecision()); } else { metricInfos[i] = new MetricInfo(MetricType.Normal, index); } } } return new EndpointAggregators(funcNames, dataTypes, metricInfos, tableInfo.getDigest()); }