├── .DS_Store ├── .asf.yaml ├── .github ├── PULL_REQUEST_TEMPLATE.md └── workflows │ ├── build-test.yml │ └── maven-publish.yml ├── .gitignore ├── LICENSE ├── NOTICE ├── README.md ├── assembly ├── pom.xml └── src │ └── main │ └── assembly │ └── assembly.xml ├── bin ├── carbon-spark-sql ├── start-indexserver.sh └── stop-indexserver.sh ├── build ├── How-to-build-carbondata-notebook-docker-image-by-dockerfile.md ├── How-to-build-carbondata-notebook-docker-image-by-manual.md ├── README.md ├── carbondata-build-info.bat ├── carbondata-build-info.sh └── docker │ └── carbondata-notebook │ └── Dockerfile ├── common ├── pom.xml └── src │ ├── main │ └── java │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── common │ │ ├── CarbonIterator.java │ │ ├── Maps.java │ │ ├── Strings.java │ │ ├── annotations │ │ ├── InterfaceAudience.java │ │ └── InterfaceStability.java │ │ ├── constants │ │ └── LoggerAction.java │ │ ├── exceptions │ │ ├── DeprecatedFeatureException.java │ │ ├── MetadataProcessException.java │ │ ├── NoSuchStreamException.java │ │ ├── TableStatusLockException.java │ │ └── sql │ │ │ ├── CarbonSchemaException.java │ │ │ ├── InvalidLoadOptionException.java │ │ │ ├── MalformedCarbonCommandException.java │ │ │ ├── MalformedIndexCommandException.java │ │ │ ├── MalformedMVCommandException.java │ │ │ ├── NoSuchIndexException.java │ │ │ └── NoSuchMVException.java │ │ └── logging │ │ ├── LogService.java │ │ ├── LogServiceFactory.java │ │ └── impl │ │ ├── AuditExtendedRollingFileAppender.java │ │ ├── AuditLevel.java │ │ ├── ExtendedRollingFileAppender.java │ │ └── StatisticLevel.java │ └── test │ └── java │ ├── log4j.properties │ └── org │ └── apache │ └── carbondata │ └── common │ ├── StringsSuite.java │ └── logging │ ├── LogServiceFactoryTest_UT.java │ └── impl │ ├── AuditExtendedRollingFileAppenderTest_UT.java │ ├── AuditLevelTest_UT.java │ └── ExtendedRollingFileAppenderTest_UT.java ├── conf ├── carbon.properties.template └── dataload.properties.template ├── core ├── CARBON_CORELogResource.properties ├── pom.xml └── src │ ├── main │ └── java │ │ └── org │ │ └── apache │ │ └── carbondata │ │ ├── core │ │ ├── cache │ │ │ ├── Cache.java │ │ │ ├── CacheProvider.java │ │ │ ├── CacheType.java │ │ │ ├── Cacheable.java │ │ │ ├── CarbonLRUCache.java │ │ │ └── dictionary │ │ │ │ └── DictionaryByteArrayWrapper.java │ │ ├── constants │ │ │ ├── CarbonCommonConstants.java │ │ │ ├── CarbonCommonConstantsInternal.java │ │ │ ├── CarbonLoadOptionConstants.java │ │ │ ├── CarbonV3DataFormatConstants.java │ │ │ ├── CarbonVersionConstants.java │ │ │ └── SortScopeOptions.java │ │ ├── datastore │ │ │ ├── ColumnType.java │ │ │ ├── DataRefNode.java │ │ │ ├── FileReader.java │ │ │ ├── ReusableDataBuffer.java │ │ │ ├── TableSegmentUniqueIdentifier.java │ │ │ ├── TableSpec.java │ │ │ ├── block │ │ │ │ ├── AbstractIndex.java │ │ │ │ ├── Distributable.java │ │ │ │ ├── SegmentProperties.java │ │ │ │ ├── SegmentPropertiesAndSchemaHolder.java │ │ │ │ ├── TableBlockInfo.java │ │ │ │ └── TaskBlockInfo.java │ │ │ ├── blocklet │ │ │ │ ├── BlockletEncodedColumnPage.java │ │ │ │ └── EncodedBlocklet.java │ │ │ ├── chunk │ │ │ │ ├── AbstractRawColumnChunk.java │ │ │ │ ├── DimensionColumnPage.java │ │ │ │ ├── impl │ │ │ │ │ ├── AbstractDimensionColumnPage.java │ │ │ │ │ ├── DimensionRawColumnChunk.java │ │ │ │ │ ├── FixedLengthDimensionColumnPage.java │ │ │ │ │ ├── MeasureRawColumnChunk.java │ │ │ │ │ └── VariableLengthDimensionColumnPage.java │ │ │ │ ├── reader │ │ │ │ │ ├── CarbonDataReaderFactory.java │ │ │ │ │ ├── DimensionColumnChunkReader.java │ │ │ │ │ ├── MeasureColumnChunkReader.java │ │ │ │ │ ├── dimension │ │ │ │ │ │ ├── AbstractDimensionChunkReader.java │ │ │ │ │ │ └── v3 │ │ │ │ │ │ │ ├── DimensionChunkPageReaderV3.java │ │ │ │ │ │ │ └── DimensionChunkReaderV3.java │ │ │ │ │ └── measure │ │ │ │ │ │ ├── AbstractMeasureChunkReader.java │ │ │ │ │ │ └── v3 │ │ │ │ │ │ ├── MeasureChunkPageReaderV3.java │ │ │ │ │ │ └── MeasureChunkReaderV3.java │ │ │ │ └── store │ │ │ │ │ ├── ColumnPageWrapper.java │ │ │ │ │ ├── DimensionChunkStoreFactory.java │ │ │ │ │ ├── DimensionDataChunkStore.java │ │ │ │ │ └── impl │ │ │ │ │ ├── LocalDictDimensionDataChunkStore.java │ │ │ │ │ ├── safe │ │ │ │ │ ├── AbstractNonDictionaryVectorFiller.java │ │ │ │ │ ├── SafeAbstractDimensionDataChunkStore.java │ │ │ │ │ ├── SafeFixedLengthDimensionDataChunkStore.java │ │ │ │ │ ├── SafeVariableIntLengthDimensionDataChunkStore.java │ │ │ │ │ ├── SafeVariableLengthDimensionDataChunkStore.java │ │ │ │ │ └── SafeVariableShortLengthDimensionDataChunkStore.java │ │ │ │ │ └── unsafe │ │ │ │ │ ├── UnsafeAbstractDimensionDataChunkStore.java │ │ │ │ │ ├── UnsafeFixedLengthDimensionDataChunkStore.java │ │ │ │ │ ├── UnsafeVariableIntLengthDimensionDataChunkStore.java │ │ │ │ │ ├── UnsafeVariableLengthDimensionDataChunkStore.java │ │ │ │ │ └── UnsafeVariableShortLengthDimensionDataChunkStore.java │ │ │ ├── columnar │ │ │ │ ├── BlockIndexerStorage.java │ │ │ │ ├── ByteArrayBlockIndexerStorage.java │ │ │ │ ├── ByteArrayBlockIndexerStorageWithoutRowId.java │ │ │ │ ├── ByteArrayColumnWithRowId.java │ │ │ │ ├── DummyBlockIndexerStorage.java │ │ │ │ ├── ObjectArrayBlockIndexerStorage.java │ │ │ │ ├── ObjectColumnWithRowId.java │ │ │ │ └── UnBlockIndexer.java │ │ │ ├── compression │ │ │ │ ├── AbstractCompressor.java │ │ │ │ ├── Compressor.java │ │ │ │ ├── CompressorFactory.java │ │ │ │ ├── GzipCompressor.java │ │ │ │ ├── SnappyCompressor.java │ │ │ │ └── ZstdCompressor.java │ │ │ ├── exception │ │ │ │ ├── CarbonDataWriterException.java │ │ │ │ └── IndexBuilderException.java │ │ │ ├── filesystem │ │ │ │ ├── AbstractDFSCarbonFile.java │ │ │ │ ├── AlluxioCarbonFile.java │ │ │ │ ├── CarbonFile.java │ │ │ │ ├── CarbonFileFilter.java │ │ │ │ ├── HDFSCarbonFile.java │ │ │ │ ├── LocalCarbonFile.java │ │ │ │ ├── S3CarbonFile.java │ │ │ │ └── ViewFSCarbonFile.java │ │ │ ├── impl │ │ │ │ ├── DFSFileReaderImpl.java │ │ │ │ ├── DefaultFileTypeProvider.java │ │ │ │ ├── FileFactory.java │ │ │ │ ├── FileReaderImpl.java │ │ │ │ └── FileTypeInterface.java │ │ │ ├── page │ │ │ │ ├── ActualDataBasedFallbackEncoder.java │ │ │ │ ├── ColumnPage.java │ │ │ │ ├── ColumnPageValueConverter.java │ │ │ │ ├── ComplexColumnPage.java │ │ │ │ ├── DecimalColumnPage.java │ │ │ │ ├── DecoderBasedFallbackEncoder.java │ │ │ │ ├── EncodedTablePage.java │ │ │ │ ├── FallbackEncodedColumnPage.java │ │ │ │ ├── LVByteBufferColumnPage.java │ │ │ │ ├── LazyColumnPage.java │ │ │ │ ├── LocalDictColumnPage.java │ │ │ │ ├── SafeDecimalColumnPage.java │ │ │ │ ├── SafeFixLengthColumnPage.java │ │ │ │ ├── SafeVarLengthColumnPage.java │ │ │ │ ├── UnsafeDecimalColumnPage.java │ │ │ │ ├── UnsafeFixLengthColumnPage.java │ │ │ │ ├── UnsafeVarLengthColumnPage.java │ │ │ │ ├── UnsafeVarLengthColumnPageBase.java │ │ │ │ ├── VarLengthColumnPageBase.java │ │ │ │ ├── encoding │ │ │ │ │ ├── ColumnPageCodec.java │ │ │ │ │ ├── ColumnPageDecoder.java │ │ │ │ │ ├── ColumnPageEncoder.java │ │ │ │ │ ├── ColumnPageEncoderMeta.java │ │ │ │ │ ├── DefaultEncodingFactory.java │ │ │ │ │ ├── EncodedColumnPage.java │ │ │ │ │ ├── EncodingFactory.java │ │ │ │ │ ├── adaptive │ │ │ │ │ │ ├── AdaptiveCodec.java │ │ │ │ │ │ ├── AdaptiveDeltaFloatingCodec.java │ │ │ │ │ │ ├── AdaptiveDeltaIntegralCodec.java │ │ │ │ │ │ ├── AdaptiveFloatingCodec.java │ │ │ │ │ │ └── AdaptiveIntegralCodec.java │ │ │ │ │ ├── bool │ │ │ │ │ │ └── BooleanConvert.java │ │ │ │ │ ├── compress │ │ │ │ │ │ └── DirectCompressCodec.java │ │ │ │ │ ├── dimension │ │ │ │ │ │ └── legacy │ │ │ │ │ │ │ ├── ComplexDimensionIndexCodec.java │ │ │ │ │ │ │ ├── DirectDictDimensionIndexCodec.java │ │ │ │ │ │ │ ├── IndexStorageCodec.java │ │ │ │ │ │ │ ├── IndexStorageEncoder.java │ │ │ │ │ │ │ └── PlainDimensionIndexCodec.java │ │ │ │ │ └── rle │ │ │ │ │ │ ├── RLECodec.java │ │ │ │ │ │ └── RLEEncoderMeta.java │ │ │ │ └── statistics │ │ │ │ │ ├── ColumnPageStatsCollector.java │ │ │ │ │ ├── DummyStatsCollector.java │ │ │ │ │ ├── KeyPageStatsCollector.java │ │ │ │ │ ├── PrimitivePageStatsCollector.java │ │ │ │ │ ├── SimpleStatsResult.java │ │ │ │ │ ├── StringStatsCollector.java │ │ │ │ │ └── TablePageStatistics.java │ │ │ └── row │ │ │ │ ├── CarbonRow.java │ │ │ │ ├── ComplexColumnInfo.java │ │ │ │ └── WriteStepRowUtil.java │ │ ├── devapi │ │ │ └── BiDictionary.java │ │ ├── enums │ │ │ └── EscapeSequences.java │ │ ├── exception │ │ │ ├── CarbonFileException.java │ │ │ ├── ConcurrentOperationException.java │ │ │ └── InvalidConfigurationException.java │ │ ├── features │ │ │ └── TableOperation.java │ │ ├── fileoperations │ │ │ ├── AtomicFileOperationFactory.java │ │ │ ├── AtomicFileOperationS3Impl.java │ │ │ ├── AtomicFileOperations.java │ │ │ ├── AtomicFileOperationsImpl.java │ │ │ └── FileWriteOperation.java │ │ ├── index │ │ │ ├── AbstractIndexJob.java │ │ │ ├── IndexChooser.java │ │ │ ├── IndexFilter.java │ │ │ ├── IndexInputFormat.java │ │ │ ├── IndexInputSplit.java │ │ │ ├── IndexJob.java │ │ │ ├── IndexLevel.java │ │ │ ├── IndexMeta.java │ │ │ ├── IndexRegistry.java │ │ │ ├── IndexStoreManager.java │ │ │ ├── IndexUtil.java │ │ │ ├── Segment.java │ │ │ ├── SegmentIndexGroup.java │ │ │ ├── TableIndex.java │ │ │ ├── dev │ │ │ │ ├── BlockletSerializer.java │ │ │ │ ├── CacheableIndex.java │ │ │ │ ├── Index.java │ │ │ │ ├── IndexBuilder.java │ │ │ │ ├── IndexFactory.java │ │ │ │ ├── IndexModel.java │ │ │ │ ├── IndexWriter.java │ │ │ │ ├── cgindex │ │ │ │ │ ├── CoarseGrainIndex.java │ │ │ │ │ └── CoarseGrainIndexFactory.java │ │ │ │ ├── expr │ │ │ │ │ ├── AndIndexExprWrapper.java │ │ │ │ │ ├── IndexExprWrapper.java │ │ │ │ │ ├── IndexExprWrapperImpl.java │ │ │ │ │ ├── IndexInputSplitWrapper.java │ │ │ │ │ ├── IndexWrapperSimpleInfo.java │ │ │ │ │ └── OrIndexExprWrapper.java │ │ │ │ └── fgindex │ │ │ │ │ ├── FineGrainBlocklet.java │ │ │ │ │ ├── FineGrainIndex.java │ │ │ │ │ └── FineGrainIndexFactory.java │ │ │ ├── secondaryindex │ │ │ │ └── CarbonCostBasedOptimizer.java │ │ │ └── status │ │ │ │ └── IndexStatus.java │ │ ├── indexstore │ │ │ ├── AbstractMemoryDMStore.java │ │ │ ├── BlockMetaInfo.java │ │ │ ├── Blocklet.java │ │ │ ├── BlockletDetailInfo.java │ │ │ ├── BlockletDetailsFetcher.java │ │ │ ├── BlockletIndexStore.java │ │ │ ├── BlockletIndexWrapper.java │ │ │ ├── ExtendedBlocklet.java │ │ │ ├── ExtendedBlockletWrapper.java │ │ │ ├── ExtendedBlockletWrapperContainer.java │ │ │ ├── PartitionSpec.java │ │ │ ├── SafeMemoryDMStore.java │ │ │ ├── SegmentBlockIndexInfo.java │ │ │ ├── SegmentPropertiesFetcher.java │ │ │ ├── SegmentWrapper.java │ │ │ ├── SegmentWrapperContainer.java │ │ │ ├── TableBlockIndexUniqueIdentifier.java │ │ │ ├── TableBlockIndexUniqueIdentifierWrapper.java │ │ │ ├── UnsafeMemoryDMStore.java │ │ │ ├── blockletindex │ │ │ │ ├── BlockIndex.java │ │ │ │ ├── BlockletDataRefNode.java │ │ │ │ ├── BlockletIndex.java │ │ │ │ ├── BlockletIndexFactory.java │ │ │ │ ├── BlockletIndexInputSplit.java │ │ │ │ ├── BlockletIndexModel.java │ │ │ │ ├── BlockletIndexRowIndexes.java │ │ │ │ ├── IndexWrapper.java │ │ │ │ └── SegmentIndexFileStore.java │ │ │ ├── row │ │ │ │ ├── IndexRow.java │ │ │ │ ├── IndexRowImpl.java │ │ │ │ └── UnsafeIndexRow.java │ │ │ └── schema │ │ │ │ ├── CarbonRowSchema.java │ │ │ │ └── SchemaGenerator.java │ │ ├── keygenerator │ │ │ ├── KeyGenException.java │ │ │ ├── KeyGenerator.java │ │ │ ├── directdictionary │ │ │ │ ├── DirectDictionaryGenerator.java │ │ │ │ ├── DirectDictionaryKeyGeneratorFactory.java │ │ │ │ └── timestamp │ │ │ │ │ ├── AbstractDirectDictionaryGenerator.java │ │ │ │ │ ├── DateDirectDictionaryGenerator.java │ │ │ │ │ ├── TimeStampDirectDictionaryGenerator.java │ │ │ │ │ ├── TimeStampGranularityConstants.java │ │ │ │ │ └── TimeStampGranularityTypeValue.java │ │ │ ├── factory │ │ │ │ └── KeyGeneratorFactory.java │ │ │ └── mdkey │ │ │ │ ├── AbstractKeyGenerator.java │ │ │ │ ├── Bits.java │ │ │ │ └── MultiDimKeyVarLengthGenerator.java │ │ ├── localdictionary │ │ │ ├── PageLevelDictionary.java │ │ │ ├── dictionaryholder │ │ │ │ ├── DictionaryStore.java │ │ │ │ └── MapBasedDictionaryStore.java │ │ │ ├── exception │ │ │ │ └── DictionaryThresholdReachedException.java │ │ │ └── generator │ │ │ │ ├── ColumnLocalDictionaryGenerator.java │ │ │ │ └── LocalDictionaryGenerator.java │ │ ├── locks │ │ │ ├── AbstractCarbonLock.java │ │ │ ├── AlluxioFileLock.java │ │ │ ├── CarbonLockFactory.java │ │ │ ├── CarbonLockUtil.java │ │ │ ├── HdfsFileLock.java │ │ │ ├── ICarbonLock.java │ │ │ ├── LocalFileLock.java │ │ │ ├── LockUsage.java │ │ │ ├── S3FileLock.java │ │ │ ├── ZooKeeperLocking.java │ │ │ └── ZookeeperInit.java │ │ ├── memory │ │ │ ├── CarbonUnsafe.java │ │ │ ├── HeapMemoryAllocator.java │ │ │ ├── IntPointerBuffer.java │ │ │ ├── MemoryAllocator.java │ │ │ ├── MemoryBlock.java │ │ │ ├── MemoryException.java │ │ │ ├── MemoryLocation.java │ │ │ ├── MemoryType.java │ │ │ ├── UnsafeMemoryAllocator.java │ │ │ ├── UnsafeMemoryManager.java │ │ │ └── UnsafeSortMemoryManager.java │ │ ├── metadata │ │ │ ├── AbsoluteTableIdentifier.java │ │ │ ├── CarbonMetadata.java │ │ │ ├── CarbonTableIdentifier.java │ │ │ ├── ColumnIdentifier.java │ │ │ ├── ColumnarFormatVersion.java │ │ │ ├── DatabaseLocationProvider.java │ │ │ ├── SegmentFileStore.java │ │ │ ├── ValueEncoderMeta.java │ │ │ ├── blocklet │ │ │ │ ├── BlockletInfo.java │ │ │ │ ├── DataFileFooter.java │ │ │ │ └── index │ │ │ │ │ ├── BlockletBTreeIndex.java │ │ │ │ │ ├── BlockletIndex.java │ │ │ │ │ └── BlockletMinMaxIndex.java │ │ │ ├── converter │ │ │ │ ├── SchemaConverter.java │ │ │ │ └── ThriftWrapperSchemaConverterImpl.java │ │ │ ├── datatype │ │ │ │ ├── ArrayType.java │ │ │ │ ├── BinaryType.java │ │ │ │ ├── BooleanType.java │ │ │ │ ├── ByteArrayType.java │ │ │ │ ├── ByteType.java │ │ │ │ ├── DataType.java │ │ │ │ ├── DataTypeAdapter.java │ │ │ │ ├── DataTypeDeserializer.java │ │ │ │ ├── DataTypes.java │ │ │ │ ├── DateType.java │ │ │ │ ├── DecimalConverterFactory.java │ │ │ │ ├── DecimalType.java │ │ │ │ ├── DoubleType.java │ │ │ │ ├── Field.java │ │ │ │ ├── FloatType.java │ │ │ │ ├── IntType.java │ │ │ │ ├── LongType.java │ │ │ │ ├── MapType.java │ │ │ │ ├── NullType.java │ │ │ │ ├── ShortIntType.java │ │ │ │ ├── ShortType.java │ │ │ │ ├── StringType.java │ │ │ │ ├── StructField.java │ │ │ │ ├── StructType.java │ │ │ │ ├── TimestampType.java │ │ │ │ └── VarcharType.java │ │ │ ├── encoder │ │ │ │ └── Encoding.java │ │ │ ├── index │ │ │ │ ├── BlockIndexInfo.java │ │ │ │ └── IndexType.java │ │ │ └── schema │ │ │ │ ├── BucketingInfo.java │ │ │ │ ├── ColumnRangeInfo.java │ │ │ │ ├── PartitionInfo.java │ │ │ │ ├── SchemaEvolution.java │ │ │ │ ├── SchemaEvolutionEntry.java │ │ │ │ ├── SchemaReader.java │ │ │ │ ├── SortColumnRangeInfo.java │ │ │ │ ├── index │ │ │ │ └── IndexProperty.java │ │ │ │ ├── indextable │ │ │ │ ├── IndexMetadata.java │ │ │ │ └── IndexTableInfo.java │ │ │ │ ├── partition │ │ │ │ └── PartitionType.java │ │ │ │ └── table │ │ │ │ ├── CarbonTable.java │ │ │ │ ├── CarbonTableBuilder.java │ │ │ │ ├── IndexSchema.java │ │ │ │ ├── RelationIdentifier.java │ │ │ │ ├── TableInfo.java │ │ │ │ ├── TableSchema.java │ │ │ │ ├── TableSchemaBuilder.java │ │ │ │ ├── Writable.java │ │ │ │ ├── WritableUtil.java │ │ │ │ └── column │ │ │ │ ├── CarbonColumn.java │ │ │ │ ├── CarbonDimension.java │ │ │ │ ├── CarbonImplicitDimension.java │ │ │ │ ├── CarbonMeasure.java │ │ │ │ ├── ColumnSchema.java │ │ │ │ ├── ColumnUniqueIdGenerator.java │ │ │ │ └── ParentColumnTableRelation.java │ │ ├── mutate │ │ │ ├── CarbonUpdateUtil.java │ │ │ ├── CdcVO.java │ │ │ ├── DeleteDeltaBlockDetails.java │ │ │ ├── DeleteDeltaBlockletDetails.java │ │ │ ├── DeleteDeltaVo.java │ │ │ ├── FilePathMinMaxVO.java │ │ │ ├── SegmentUpdateDetails.java │ │ │ ├── TupleIdEnum.java │ │ │ ├── UpdateVO.java │ │ │ └── data │ │ │ │ ├── BlockMappingVO.java │ │ │ │ └── RowCountDetailsVO.java │ │ ├── preagg │ │ │ ├── DaysOfWeekEnum.java │ │ │ ├── TimeSeriesFunctionEnum.java │ │ │ └── TimeSeriesUDF.java │ │ ├── profiler │ │ │ ├── ExplainCollector.java │ │ │ └── TablePruningInfo.java │ │ ├── range │ │ │ ├── BlockMinMaxTree.java │ │ │ └── MinMaxNode.java │ │ ├── readcommitter │ │ │ ├── LatestFilesReadCommittedScope.java │ │ │ ├── ReadCommittedIndexFileSnapShot.java │ │ │ ├── ReadCommittedScope.java │ │ │ └── TableStatusReadCommittedScope.java │ │ ├── reader │ │ │ ├── CarbonDeleteDeltaFileReader.java │ │ │ ├── CarbonDeleteDeltaFileReaderImpl.java │ │ │ ├── CarbonDeleteFilesDataReader.java │ │ │ ├── CarbonDictionaryColumnMetaChunk.java │ │ │ ├── CarbonDictionaryReader.java │ │ │ ├── CarbonFooterReader.java │ │ │ ├── CarbonFooterReaderV3.java │ │ │ ├── CarbonHeaderReader.java │ │ │ ├── CarbonIndexFileReader.java │ │ │ └── ThriftReader.java │ │ ├── scan │ │ │ ├── collector │ │ │ │ ├── ResultCollectorFactory.java │ │ │ │ ├── ScannedResultCollector.java │ │ │ │ └── impl │ │ │ │ │ ├── AbstractScannedResultCollector.java │ │ │ │ │ ├── DictionaryBasedResultCollector.java │ │ │ │ │ ├── DictionaryBasedVectorResultCollector.java │ │ │ │ │ ├── RawBasedResultCollector.java │ │ │ │ │ ├── RestructureBasedDictionaryResultCollector.java │ │ │ │ │ ├── RestructureBasedRawResultCollector.java │ │ │ │ │ ├── RestructureBasedVectorResultCollector.java │ │ │ │ │ ├── RowIdBasedResultCollector.java │ │ │ │ │ ├── RowIdRawBasedResultCollector.java │ │ │ │ │ └── RowIdRestructureBasedRawResultCollector.java │ │ │ ├── complextypes │ │ │ │ ├── ArrayQueryType.java │ │ │ │ ├── ComplexQueryType.java │ │ │ │ ├── MapQueryType.java │ │ │ │ ├── PrimitiveQueryType.java │ │ │ │ └── StructQueryType.java │ │ │ ├── executor │ │ │ │ ├── QueryExecutor.java │ │ │ │ ├── QueryExecutorFactory.java │ │ │ │ ├── exception │ │ │ │ │ └── QueryExecutionException.java │ │ │ │ ├── impl │ │ │ │ │ ├── AbstractQueryExecutor.java │ │ │ │ │ ├── DetailQueryExecutor.java │ │ │ │ │ ├── QueryExecutorProperties.java │ │ │ │ │ └── VectorDetailQueryExecutor.java │ │ │ │ ├── infos │ │ │ │ │ ├── BlockExecutionInfo.java │ │ │ │ │ ├── DeleteDeltaInfo.java │ │ │ │ │ ├── DimensionInfo.java │ │ │ │ │ └── MeasureInfo.java │ │ │ │ └── util │ │ │ │ │ ├── QueryUtil.java │ │ │ │ │ └── RestructureUtil.java │ │ │ ├── expression │ │ │ │ ├── BinaryExpression.java │ │ │ │ ├── ColumnExpression.java │ │ │ │ ├── Expression.java │ │ │ │ ├── ExpressionResult.java │ │ │ │ ├── FilterModificationNode.java │ │ │ │ ├── LeafExpression.java │ │ │ │ ├── LiteralExpression.java │ │ │ │ ├── MatchExpression.java │ │ │ │ ├── RangeExpressionEvaluator.java │ │ │ │ ├── UnknownExpression.java │ │ │ │ ├── conditional │ │ │ │ │ ├── BinaryConditionalExpression.java │ │ │ │ │ ├── CDCBlockImplicitExpression.java │ │ │ │ │ ├── ConditionalExpression.java │ │ │ │ │ ├── EqualToExpression.java │ │ │ │ │ ├── GreaterThanEqualToExpression.java │ │ │ │ │ ├── GreaterThanExpression.java │ │ │ │ │ ├── ImplicitExpression.java │ │ │ │ │ ├── InExpression.java │ │ │ │ │ ├── LessThanEqualToExpression.java │ │ │ │ │ ├── LessThanExpression.java │ │ │ │ │ ├── ListExpression.java │ │ │ │ │ ├── NotEqualsExpression.java │ │ │ │ │ ├── NotInExpression.java │ │ │ │ │ └── StartsWithExpression.java │ │ │ │ ├── exception │ │ │ │ │ ├── FilterIllegalMemberException.java │ │ │ │ │ └── FilterUnsupportedException.java │ │ │ │ └── logical │ │ │ │ │ ├── AndExpression.java │ │ │ │ │ ├── BinaryLogicalExpression.java │ │ │ │ │ ├── FalseExpression.java │ │ │ │ │ ├── OrExpression.java │ │ │ │ │ ├── RangeExpression.java │ │ │ │ │ └── TrueExpression.java │ │ │ ├── filter │ │ │ │ ├── ColumnFilterInfo.java │ │ │ │ ├── FilterExecutorUtil.java │ │ │ │ ├── FilterExpressionProcessor.java │ │ │ │ ├── FilterProcessor.java │ │ │ │ ├── FilterUtil.java │ │ │ │ ├── GenericQueryType.java │ │ │ │ ├── executer │ │ │ │ │ ├── AndFilterExecutorImpl.java │ │ │ │ │ ├── BitSetUpdaterFactory.java │ │ │ │ │ ├── CDCBlockImplicitExecutorImpl.java │ │ │ │ │ ├── DimColumnExecutorFilterInfo.java │ │ │ │ │ ├── ExcludeFilterExecutorImpl.java │ │ │ │ │ ├── FalseFilterExecutor.java │ │ │ │ │ ├── FilterBitSetUpdater.java │ │ │ │ │ ├── FilterExecutor.java │ │ │ │ │ ├── ImplicitColumnFilterExecutor.java │ │ │ │ │ ├── ImplicitIncludeFilterExecutorImpl.java │ │ │ │ │ ├── IncludeFilterExecutorImpl.java │ │ │ │ │ ├── MeasureColumnExecutorFilterInfo.java │ │ │ │ │ ├── OrFilterExecutorImpl.java │ │ │ │ │ ├── RangeValueFilterExecutorImpl.java │ │ │ │ │ ├── RestructureEvaluatorImpl.java │ │ │ │ │ ├── RestructureExcludeFilterExecutorImpl.java │ │ │ │ │ ├── RestructureIncludeFilterExecutorImpl.java │ │ │ │ │ ├── RowLevelFilterExecutorImpl.java │ │ │ │ │ ├── RowLevelRangeGreaterThanEqualFilterExecutorImpl.java │ │ │ │ │ ├── RowLevelRangeGreaterThanFilterExecutorImpl.java │ │ │ │ │ ├── RowLevelRangeLessThanEqualFilterExecutorImpl.java │ │ │ │ │ ├── RowLevelRangeLessThanFilterExecutorImpl.java │ │ │ │ │ ├── RowLevelRangeTypeExecutorFactory.java │ │ │ │ │ └── TrueFilterExecutor.java │ │ │ │ ├── intf │ │ │ │ │ ├── ExpressionType.java │ │ │ │ │ ├── FilterExecutorType.java │ │ │ │ │ ├── FilterOptimizer.java │ │ │ │ │ ├── RowImpl.java │ │ │ │ │ └── RowIntf.java │ │ │ │ ├── optimizer │ │ │ │ │ └── RangeFilterOptimizer.java │ │ │ │ └── resolver │ │ │ │ │ ├── ConditionalFilterResolverImpl.java │ │ │ │ │ ├── FilterResolverIntf.java │ │ │ │ │ ├── LogicalFilterResolverImpl.java │ │ │ │ │ ├── RowLevelFilterResolverImpl.java │ │ │ │ │ ├── RowLevelRangeFilterResolverImpl.java │ │ │ │ │ ├── metadata │ │ │ │ │ └── FilterResolverMetadata.java │ │ │ │ │ └── resolverinfo │ │ │ │ │ ├── ColumnResolvedFilterInfo.java │ │ │ │ │ ├── DimColumnResolvedFilterInfo.java │ │ │ │ │ ├── FalseConditionalResolverImpl.java │ │ │ │ │ ├── MeasureColumnResolvedFilterInfo.java │ │ │ │ │ ├── TrueConditionalResolverImpl.java │ │ │ │ │ └── visitor │ │ │ │ │ ├── CustomTypeDictionaryVisitor.java │ │ │ │ │ ├── FilterInfoTypeVisitorFactory.java │ │ │ │ │ ├── ImplicitColumnVisitor.java │ │ │ │ │ ├── MeasureColumnVisitor.java │ │ │ │ │ ├── NoDictionaryTypeVisitor.java │ │ │ │ │ ├── RangeDirectDictionaryVisitor.java │ │ │ │ │ ├── RangeNoDictionaryTypeVisitor.java │ │ │ │ │ └── ResolvedFilterInfoVisitorIntf.java │ │ │ ├── model │ │ │ │ ├── ProjectionColumn.java │ │ │ │ ├── ProjectionDimension.java │ │ │ │ ├── ProjectionMeasure.java │ │ │ │ ├── QueryModel.java │ │ │ │ ├── QueryModelBuilder.java │ │ │ │ └── QueryProjection.java │ │ │ ├── processor │ │ │ │ ├── BlockletIterator.java │ │ │ │ ├── DataBlockIterator.java │ │ │ │ └── RawBlockletColumnChunks.java │ │ │ ├── result │ │ │ │ ├── BlockletScannedResult.java │ │ │ │ ├── RowBatch.java │ │ │ │ ├── impl │ │ │ │ │ ├── FilterQueryScannedResult.java │ │ │ │ │ └── NonFilterQueryScannedResult.java │ │ │ │ ├── iterator │ │ │ │ │ ├── AbstractDetailQueryResultIterator.java │ │ │ │ │ ├── ChunkRowIterator.java │ │ │ │ │ ├── ColumnDriftRawResultIterator.java │ │ │ │ │ ├── DetailQueryResultIterator.java │ │ │ │ │ ├── PartitionSplitterRawResultIterator.java │ │ │ │ │ ├── RawResultIterator.java │ │ │ │ │ └── VectorDetailQueryResultIterator.java │ │ │ │ └── vector │ │ │ │ │ ├── CarbonColumnVector.java │ │ │ │ │ ├── CarbonColumnarBatch.java │ │ │ │ │ ├── CarbonDictionary.java │ │ │ │ │ ├── ColumnVectorInfo.java │ │ │ │ │ ├── MeasureDataVectorProcessor.java │ │ │ │ │ └── impl │ │ │ │ │ ├── CarbonColumnVectorImpl.java │ │ │ │ │ ├── CarbonDictionaryImpl.java │ │ │ │ │ └── directread │ │ │ │ │ ├── AbstractCarbonColumnarVector.java │ │ │ │ │ ├── ColumnarVectorWrapperDirectFactory.java │ │ │ │ │ ├── ColumnarVectorWrapperDirectWithDeleteDelta.java │ │ │ │ │ ├── ColumnarVectorWrapperDirectWithDeleteDeltaAndInvertedIndex.java │ │ │ │ │ ├── ColumnarVectorWrapperDirectWithInvertedIndex.java │ │ │ │ │ ├── ConvertibleVector.java │ │ │ │ │ └── SequentialFill.java │ │ │ ├── scanner │ │ │ │ ├── BlockletScanner.java │ │ │ │ ├── LazyBlockletLoader.java │ │ │ │ ├── LazyPageLoader.java │ │ │ │ └── impl │ │ │ │ │ ├── BlockletFilterScanner.java │ │ │ │ │ └── BlockletFullScanner.java │ │ │ └── wrappers │ │ │ │ ├── ByteArrayWrapper.java │ │ │ │ └── IntArrayWrapper.java │ │ ├── segmentmeta │ │ │ ├── BlockColumnMetaDataInfo.java │ │ │ ├── SegmentColumnMetaDataInfo.java │ │ │ ├── SegmentMetaDataInfo.java │ │ │ └── SegmentMetaDataInfoStats.java │ │ ├── service │ │ │ ├── ColumnUniqueIdService.java │ │ │ └── impl │ │ │ │ └── ColumnUniqueIdGenerator.java │ │ ├── stats │ │ │ ├── DriverQueryStatisticsRecorderDummy.java │ │ │ ├── DriverQueryStatisticsRecorderImpl.java │ │ │ ├── QueryStatistic.java │ │ │ ├── QueryStatisticsConstants.java │ │ │ ├── QueryStatisticsModel.java │ │ │ ├── QueryStatisticsRecorder.java │ │ │ ├── QueryStatisticsRecorderDummy.java │ │ │ ├── QueryStatisticsRecorderImpl.java │ │ │ └── TaskStatistics.java │ │ ├── statusmanager │ │ │ ├── FileFormat.java │ │ │ ├── LoadMetadataDetails.java │ │ │ ├── SegmentRefreshInfo.java │ │ │ ├── SegmentStatus.java │ │ │ ├── SegmentStatusManager.java │ │ │ ├── SegmentUpdateStatusManager.java │ │ │ ├── StageInput.java │ │ │ └── StageInputCollector.java │ │ ├── stream │ │ │ ├── ExtendedByteArrayInputStream.java │ │ │ ├── ExtendedByteArrayOutputStream.java │ │ │ ├── ExtendedDataInputStream.java │ │ │ ├── StreamFile.java │ │ │ └── StreamPruner.java │ │ ├── util │ │ │ ├── AbstractDataFileFooterConverter.java │ │ │ ├── BitSetGroup.java │ │ │ ├── BlockletIndexUtil.java │ │ │ ├── ByteUtil.java │ │ │ ├── CarbonLoadStatisticsDummy.java │ │ │ ├── CarbonLoadStatisticsImpl.java │ │ │ ├── CarbonMetadataUtil.java │ │ │ ├── CarbonProperties.java │ │ │ ├── CarbonSessionInfo.java │ │ │ ├── CarbonTaskInfo.java │ │ │ ├── CarbonTestUtil.java │ │ │ ├── CarbonThreadFactory.java │ │ │ ├── CarbonTimeStatisticsFactory.java │ │ │ ├── CarbonUnsafeUtil.java │ │ │ ├── CarbonUtil.java │ │ │ ├── CleanFilesUtil.java │ │ │ ├── CustomIndex.java │ │ │ ├── DataFileFooterConverter.java │ │ │ ├── DataFileFooterConverterFactory.java │ │ │ ├── DataFileFooterConverterV3.java │ │ │ ├── DataLoadMetrics.java │ │ │ ├── DataTypeConverter.java │ │ │ ├── DataTypeConverterImpl.java │ │ │ ├── DataTypeUtil.java │ │ │ ├── DeleteLoadFolders.java │ │ │ ├── LoadStatistics.java │ │ │ ├── NonDictionaryUtil.java │ │ │ ├── ObjectSerializationUtil.java │ │ │ ├── ObjectSizeCalculator.java │ │ │ ├── ReUsableByteArrayDataOutputStream.java │ │ │ ├── SessionParams.java │ │ │ ├── TaskMetricsMap.java │ │ │ ├── ThreadLocalSessionInfo.java │ │ │ ├── ThreadLocalTaskInfo.java │ │ │ ├── TrashUtil.java │ │ │ ├── annotations │ │ │ │ └── CarbonProperty.java │ │ │ ├── comparator │ │ │ │ ├── BigDecimalSerializableComparator.java │ │ │ │ ├── BooleanSerializableComparator.java │ │ │ │ ├── ByteArraySerializableComparator.java │ │ │ │ ├── Comparator.java │ │ │ │ ├── DoubleSerializableComparator.java │ │ │ │ ├── FloatSerializableComparator.java │ │ │ │ ├── IntSerializableComparator.java │ │ │ │ ├── LongSerializableComparator.java │ │ │ │ ├── SerializableComparator.java │ │ │ │ ├── ShortSerializableComparator.java │ │ │ │ └── StringSerializableComparator.java │ │ │ └── path │ │ │ │ └── CarbonTablePath.java │ │ ├── view │ │ │ ├── MVCatalog.java │ │ │ ├── MVCatalogFactory.java │ │ │ ├── MVManager.java │ │ │ ├── MVProperty.java │ │ │ ├── MVProvider.java │ │ │ ├── MVSchema.java │ │ │ ├── MVStatus.java │ │ │ └── MVStatusDetail.java │ │ └── writer │ │ │ ├── CarbonDeleteDeltaWriter.java │ │ │ ├── CarbonDeleteDeltaWriterImpl.java │ │ │ ├── CarbonIndexFileMergeWriter.java │ │ │ ├── CarbonIndexFileWriter.java │ │ │ └── ThriftWriter.java │ │ ├── events │ │ ├── Event.java │ │ ├── OperationContext.java │ │ ├── OperationEventListener.java │ │ └── OperationListenerBus.java │ │ └── hadoop │ │ ├── CarbonInputSplit.java │ │ ├── CarbonInputSplitWrapper.java │ │ └── internal │ │ ├── ObjectArrayWritable.java │ │ └── index │ │ └── Block.java │ └── test │ ├── java │ └── org │ │ └── apache │ │ └── carbondata │ │ └── core │ │ ├── CarbonPropertiesValidationTest.java │ │ ├── cache │ │ ├── CarbonLRUCacheTest.java │ │ └── dictionary │ │ │ └── AbstractDictionaryCacheTest.java │ │ ├── carbon │ │ ├── AbsoluteTableIdentifierTest.java │ │ ├── CarbonTableIdentifierTest.java │ │ ├── ColumnIdentifierTest.java │ │ └── datastorage │ │ │ └── filesystem │ │ │ └── store │ │ │ └── impl │ │ │ ├── DFSFileReaderImplUnitTest.java │ │ │ ├── FileFactoryImplUnitTest.java │ │ │ ├── FileReaderImplUnitTest.java │ │ │ └── TestFileProvider.java │ │ ├── constants │ │ └── CarbondataVersionUnitTest.java │ │ ├── datastore │ │ ├── CompressdFileTest.java │ │ ├── block │ │ │ ├── SegmentPropertiesTest.java │ │ │ ├── SegmentPropertiesTestUtil.java │ │ │ └── TableBlockInfoTest.java │ │ ├── chunk │ │ │ └── impl │ │ │ │ └── FixedLengthDimensionDataChunkTest.java │ │ ├── filesystem │ │ │ ├── AlluxioCarbonFileTest.java │ │ │ ├── HDFSCarbonFileTest.java │ │ │ ├── LocalCarbonFileTest.java │ │ │ └── ViewFsCarbonFileTest.java │ │ └── page │ │ │ └── encoding │ │ │ ├── RLECodecTest.java │ │ │ └── TestEncodingFactory.java │ │ ├── indexstore │ │ └── blockletindex │ │ │ ├── TestBlockletIndex.java │ │ │ └── TestBlockletIndexFactory.java │ │ ├── keygenerator │ │ ├── directdictionary │ │ │ ├── DateDirectDictionaryGeneratorTest.java │ │ │ └── DirectDictionaryKeyGeneratorFactoryUnitTest.java │ │ └── mdkey │ │ │ ├── BitsUnitTest.java │ │ │ └── MultiDimKeyVarLengthGeneratorUnitTest.java │ │ ├── load │ │ └── LoadMetadataDetailsUnitTest.java │ │ ├── localdictionary │ │ ├── DictionaryByteArrayWrapperTest.java │ │ ├── TestDictionaryStore.java │ │ ├── TestLocalDictionaryGenerator.java │ │ └── TestPageLevelDictionary.java │ │ ├── locks │ │ └── CarbonLockFactoryTest.java │ │ ├── memory │ │ └── MemoryAllocatorUnitTest.java │ │ ├── metadata │ │ ├── CarbonMetadataTest.java │ │ ├── DatabaseLocationProviderTest.java │ │ ├── converter │ │ │ └── ThriftWrapperSchemaConverterImplTest.java │ │ └── schema │ │ │ └── table │ │ │ ├── CarbonTableBuilderSuite.java │ │ │ ├── CarbonTableTest.java │ │ │ ├── CarbonTableWithComplexTypesTest.java │ │ │ ├── ColumnSchemaTest.java │ │ │ ├── TableInfoTest.java │ │ │ ├── TableSchemaBuilderSuite.java │ │ │ └── TableSchemaTest.java │ │ ├── reader │ │ └── CarbonIndexFileReaderTest.java │ │ ├── scan │ │ ├── complextypes │ │ │ ├── ArrayQueryTypeTest.java │ │ │ ├── PrimitiveQueryTypeTest.java │ │ │ └── StructQueryTypeTest.java │ │ ├── executor │ │ │ └── util │ │ │ │ ├── QueryUtilTest.java │ │ │ │ └── RestructureUtilTest.java │ │ ├── expression │ │ │ ├── ColumnExpressionTest.java │ │ │ ├── ExpressionResultTest.java │ │ │ ├── LiteralExpressionTest.java │ │ │ ├── conditional │ │ │ │ ├── EqualToExpressionUnitTest.java │ │ │ │ ├── GreaterThanEqualToExpressionUnitTest.java │ │ │ │ ├── GreaterThanExpressionUnitTest.java │ │ │ │ ├── InExpressionUnitTest.java │ │ │ │ ├── LessThanEqualToExpressionUnitTest.java │ │ │ │ ├── LessThanExpressionUnitTest.java │ │ │ │ ├── ListExpressionUnitTest.java │ │ │ │ ├── NotEqualsExpressionUnitTest.java │ │ │ │ └── NotInExpressionUnitTest.java │ │ │ └── logical │ │ │ │ ├── AndExpressionTest.java │ │ │ │ ├── FalseExpressionTest.java │ │ │ │ ├── OrExpressionTest.java │ │ │ │ ├── RangeExpressionTest.java │ │ │ │ └── TrueExpressionTest.java │ │ ├── filter │ │ │ ├── FilterExpressionProcessorTest.java │ │ │ ├── FilterUtilTest.java │ │ │ └── executer │ │ │ │ ├── ExcludeFilterExecutorImplTest.java │ │ │ │ └── IncludeFilterExecutorImplTest.java │ │ ├── result │ │ │ └── RowBatchTest.java │ │ └── wrappers │ │ │ └── ByteArrayWrapperTest.java │ │ ├── stats │ │ ├── DriverQueryStatisticsRecorderImplTest.java │ │ └── QueryStasticsRecorderImplTest.java │ │ ├── util │ │ ├── ByteUtilTest.java │ │ ├── CarbonMetadataUtilTest.java │ │ ├── CarbonUtilTest.java │ │ ├── DataFileFooterConverterTest.java │ │ ├── DataTypeUtilTest.java │ │ ├── RangeFilterProcessorTest.java │ │ └── path │ │ │ └── CarbonFormatDirectoryStructureTest.java │ │ └── writer │ │ └── CarbonDictionaryWriterImplTest.java │ └── resources │ └── carbonTest.properties ├── dev ├── carbon-pr-readme.md ├── carbon_pr.py ├── findbugs-exclude.xml ├── java-code-format-template.xml ├── java.header ├── javastyle-config.xml └── javastyle-suppressions.xml ├── docs ├── addsegment-guide.md ├── alluxio-guide.md ├── carbon-as-spark-datasource-guide.md ├── clean-files.md ├── configuration-parameters.md ├── csdk-guide.md ├── ddl-of-carbondata.md ├── dml-of-carbondata.md ├── documentation.md ├── faq.md ├── file-structure-of-carbondata.md ├── flink-integration-guide.md ├── hive-guide.md ├── how-to-contribute-to-apache-carbondata.md ├── images │ ├── 2-1_1.png │ ├── 2-1_1_latest.PNG │ ├── 2-2_1.png │ ├── 2-3_1.png │ ├── 2-3_2.png │ ├── 2-3_3.png │ ├── 2-3_4.png │ ├── 2-4_1.png │ ├── 2-5_1.png │ ├── 2-5_2.png │ ├── 2-5_3.png │ ├── 2-6_1.png │ ├── CarbonData_logo.png │ ├── QRCode_WechatGroup.png │ ├── carbon_data_file_structure_new.png │ ├── carbon_data_format_new.png │ ├── carbondata-performance.png │ ├── carbondata-streamer-tool-pipeline.png │ ├── codegen.png │ ├── spatial-index-1.png │ ├── spatial-index-2.png │ ├── spatial-index-polygonlist.png │ ├── spatial-index-polylinelist.png │ ├── spatial-index-rangelist.png │ ├── using-carbondata-in-notebook-1.png │ ├── using-carbondata-in-notebook-2.png │ ├── using-carbondata-in-notebook-3.png │ ├── using-carbondata-in-notebook-visualization-0.png │ ├── using-carbondata-in-notebook-visualization-1.png │ ├── using-carbondata-in-notebook-visualization-2.png │ └── using-carbondata-in-notebook-visualization-3.png ├── index-developer-guide.md ├── index-server.md ├── index │ ├── bloomfilter-index-guide.md │ ├── index-management.md │ ├── lucene-index-guide.md │ └── secondary-index-guide.md ├── introduction.md ├── language-manual.md ├── mv-guide.md ├── notebook │ ├── carbondata_notebook.ipynb │ ├── carbondata_notebook_with_visualization.ipynb │ └── sample_data_simple.csv ├── performance-tuning.md ├── prestodb-guide.md ├── prestosql-guide.md ├── query-with-spark-sql-performance -tuning.md ├── quick-start-guide.md ├── release-guide.md ├── s3-guide.md ├── scd-and-cdc-guide.md ├── sdk-guide.md ├── segment-management-on-carbondata.md ├── spatial-index-guide.md ├── streaming-guide.md ├── supported-data-types-in-carbondata.md ├── usecases.md ├── using-carbondata-in-notebook.md ├── using-carbondata-to-visualization_in-notebook.md └── zh_cn │ ├── CarbonData与商业列存DB性能对比.md │ ├── CarbonData典型应用场景之明细数据查询：点查+过滤条件.md │ └── images │ └── SortColumns.png ├── examples ├── flink │ ├── pom.xml │ └── src │ │ └── main │ │ └── scala │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── examples │ │ └── FlinkExample.scala └── spark │ ├── pom.xml │ └── src │ ├── main │ ├── java │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ └── examples │ │ │ ├── sdk │ │ │ ├── CarbonReaderExample.java │ │ │ ├── SDKS3Example.java │ │ │ ├── SDKS3ReadExample.java │ │ │ └── SDKS3SchemaReadExample.java │ │ │ └── sql │ │ │ └── JavaCarbonSessionExample.java │ ├── resources │ │ ├── Test_Data1.csv │ │ ├── complexdata.csv │ │ ├── data.csv │ │ ├── data1.csv │ │ ├── dataSample.csv │ │ ├── dimSample.csv │ │ ├── factSample.csv │ │ ├── log4j.properties │ │ ├── sample.csv │ │ └── streamSample.csv │ └── scala │ │ └── org │ │ └── apache │ │ └── carbondata │ │ ├── benchmark │ │ ├── ConcurrentQueryBenchmark.scala │ │ ├── Query.scala │ │ ├── SCDType2Benchmark.scala │ │ └── SimpleQueryBenchmark.scala │ │ └── examples │ │ ├── AlluxioExample.scala │ │ ├── AlterTableExample.scala │ │ ├── CDCExample.scala │ │ ├── CarbonDataFrameExample.scala │ │ ├── CarbonSessionExample.scala │ │ ├── CarbonSortColumnsExample.scala │ │ ├── CaseClassDataFrameAPIExample.scala │ │ ├── CustomCompactionExample.scala │ │ ├── DataFrameComplexTypeExample.scala │ │ ├── DataManagementExample.scala │ │ ├── DataMergeIntoExample.scala │ │ ├── DataUPSERTExample.scala │ │ ├── DataUpdateDeleteExample.scala │ │ ├── DedupExample.scala │ │ ├── DirectSQLExample.scala │ │ ├── ExternalTableExample.scala │ │ ├── GeoTableExampleWithCarbonSession.scala │ │ ├── HadoopFileExample.scala │ │ ├── HiveExample.scala │ │ ├── LuceneIndexExample.scala │ │ ├── MVExample.scala │ │ ├── QuerySegmentExample.scala │ │ ├── S3CsvExample.scala │ │ ├── S3Example.scala │ │ ├── S3UsingSDkExample.scala │ │ ├── SparkSessionExample.scala │ │ ├── SparkStreamingExample.scala │ │ ├── StandardPartitionExample.scala │ │ ├── StreamSQLExample.scala │ │ ├── StreamingUsingBatchLoadExample.scala │ │ ├── StreamingWithRowParserExample.scala │ │ ├── StructuredStreamingExample.scala │ │ ├── TableLevelCompactionOptionExample.scala │ │ └── util │ │ └── ExampleUtils.scala │ └── test │ └── scala │ └── org │ └── apache │ └── carbondata │ └── examplesCI │ └── RunExamples.scala ├── format ├── pom.xml └── src │ └── main │ └── thrift │ ├── carbondata.thrift │ ├── carbondata_index.thrift │ ├── carbondata_index_merge.thrift │ ├── dictionary.thrift │ └── schema.thrift ├── geo ├── pom.xml └── src │ ├── main │ └── java │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── geo │ │ ├── GeoConstants.java │ │ ├── GeoHashIndex.java │ │ ├── GeoHashUtils.java │ │ ├── GeoOperationType.java │ │ ├── QuadTreeCls.java │ │ └── scan │ │ ├── expression │ │ ├── PolygonExpression.java │ │ ├── PolygonListExpression.java │ │ ├── PolygonRangeListExpression.java │ │ └── PolylineListExpression.java │ │ └── filter │ │ └── executor │ │ └── PolygonFilterExecutorImpl.java │ └── test │ └── java │ └── org │ └── apache │ └── carbondata │ └── geo │ ├── GeoHashUtilsTest.java │ └── QuadTreeClsTest.java ├── hadoop ├── pom.xml └── src │ ├── main │ └── java │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── hadoop │ │ ├── AbstractRecordReader.java │ │ ├── CarbonMultiBlockSplit.java │ │ ├── CarbonProjection.java │ │ ├── CarbonRecordReader.java │ │ ├── InputMetricsStats.java │ │ ├── api │ │ ├── CarbonFileInputFormat.java │ │ ├── CarbonInputFormat.java │ │ ├── CarbonOutputCommitter.java │ │ ├── CarbonTableInputFormat.java │ │ └── CarbonTableOutputFormat.java │ │ ├── readsupport │ │ ├── CarbonReadSupport.java │ │ └── impl │ │ │ └── CarbonRowReadSupport.java │ │ ├── stream │ │ ├── CarbonStreamInputFormat.java │ │ ├── CarbonStreamUtils.java │ │ ├── StreamBlockletReader.java │ │ └── StreamRecordReader.java │ │ ├── testutil │ │ └── StoreCreator.java │ │ └── util │ │ ├── CarbonInputFormatUtil.java │ │ ├── CarbonInputSplitTaskInfo.java │ │ └── CarbonVectorizedRecordReader.java │ └── test │ ├── java │ └── org │ │ └── apache │ │ └── carbondata │ │ └── hadoop │ │ ├── ft │ │ ├── CarbonTableInputFormatTest.java │ │ └── CarbonTableOutputFormatTest.java │ │ └── test │ │ └── util │ │ └── ObjectSerializationUtilTest.java │ └── resources │ └── data.csv ├── index ├── bloom │ ├── pom.xml │ └── src │ │ └── main │ │ └── java │ │ └── org │ │ └── apache │ │ ├── carbondata │ │ └── index │ │ │ └── bloom │ │ │ ├── AbstractBloomIndexWriter.java │ │ │ ├── BloomCacheKeyValue.java │ │ │ ├── BloomCoarseGrainIndex.java │ │ │ ├── BloomCoarseGrainIndexFactory.java │ │ │ ├── BloomIndexBuilder.java │ │ │ ├── BloomIndexCache.java │ │ │ ├── BloomIndexFileStore.java │ │ │ ├── BloomIndexInputSplit.java │ │ │ ├── BloomIndexModel.java │ │ │ ├── BloomIndexWriter.java │ │ │ └── DataConvertUtil.java │ │ └── hadoop │ │ └── util │ │ └── bloom │ │ └── CarbonBloomFilter.java ├── examples │ ├── pom.xml │ └── src │ │ └── minmaxindex │ │ ├── main │ │ └── java │ │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ └── index │ │ │ └── examples │ │ │ ├── BlockletMinMax.java │ │ │ ├── MinMaxDataWriter.java │ │ │ ├── MinMaxIndex.java │ │ │ ├── MinMaxIndexBlockDetails.java │ │ │ └── MinMaxIndexIndexFactory.java │ │ └── test │ │ └── scala │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── index │ │ └── examples │ │ └── MinMaxIndexSuite.scala ├── lucene │ ├── pom.xml │ └── src │ │ └── main │ │ └── java │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── index │ │ └── lucene │ │ ├── LuceneFineGrainIndex.java │ │ ├── LuceneFineGrainIndexFactory.java │ │ ├── LuceneIndexBuilder.java │ │ ├── LuceneIndexFactoryBase.java │ │ ├── LuceneIndexInputSplit.java │ │ └── LuceneIndexWriter.java └── secondary-index │ ├── pom.xml │ └── src │ └── test │ └── scala │ └── org │ └── apache │ ├── carbondata │ └── spark │ │ └── testsuite │ │ ├── mergedata │ │ └── CarbonDataFileMergeTestCaseOnSI.scala │ │ ├── mergeindex │ │ └── CarbonIndexFileMergeTestCaseWithSI.scala │ │ └── secondaryindex │ │ ├── DropTableTest.scala │ │ ├── InsertIntoCarbonTableTestCase.scala │ │ ├── TestAlterTableColumnRenameWithIndex.scala │ │ ├── TestBroadCastSIFilterPushJoinWithUDF.scala │ │ ├── TestCTASWithIndex.scala │ │ ├── TestCacheOperationsForSI.scala │ │ ├── TestCarbonInternalMetastore.scala │ │ ├── TestCarbonJoin.scala │ │ ├── TestCreateIndexForCleanAndDeleteSegment.scala │ │ ├── TestCreateIndexTable.scala │ │ ├── TestCreateIndexWithLoadAndCompaction.scala │ │ ├── TestIndexModelForORFilterPushDown.scala │ │ ├── TestIndexModelWithAggQueries.scala │ │ ├── TestIndexModelWithIUD.scala │ │ ├── TestIndexModelWithLocalDictionary.scala │ │ ├── TestIndexModelWithUnsafeColumnPage.scala │ │ ├── TestIndexRepair.scala │ │ ├── TestIndexWithIndexModelOnFirstColumnAndSortColumns.scala │ │ ├── TestLikeQueryWithIndex.scala │ │ ├── TestNIQueryWithIndex.scala │ │ ├── TestQueryWithSkipSI.scala │ │ ├── TestRegisterIndexCarbonTable.scala │ │ ├── TestSIWithAddSegment.scala │ │ ├── TestSIWithComplexArrayType.scala │ │ ├── TestSIWithInsertOverwrite.scala │ │ ├── TestSIWithPartition.scala │ │ ├── TestSIWithRangeColumn.scala │ │ ├── TestSIWithSecondaryIndex.scala │ │ └── TestSecondaryIndexUtils.scala │ └── spark │ └── util │ └── TestCarbonSegmentUtil.scala ├── integration ├── flink-build │ └── pom.xml ├── flink-proxy │ ├── pom.xml │ └── src │ │ └── main │ │ ├── java │ │ └── org │ │ │ └── apache │ │ │ └── carbon │ │ │ └── flink │ │ │ ├── ProxyFileSystem.java │ │ │ ├── ProxyFileSystemFactory.java │ │ │ ├── ProxyFileWriter.java │ │ │ ├── ProxyFileWriterFactory.java │ │ │ ├── ProxyRecoverable.java │ │ │ ├── ProxyRecoverableOutputStream.java │ │ │ ├── ProxyRecoverableSerializer.java │ │ │ └── ProxyRecoverableWriter.java │ │ └── resources │ │ └── META-INF │ │ └── services │ │ └── org.apache.flink.core.fs.FileSystemFactory ├── flink │ ├── pom.xml │ └── src │ │ ├── main │ │ ├── java │ │ │ └── org │ │ │ │ └── apache │ │ │ │ └── carbon │ │ │ │ ├── core │ │ │ │ └── metadata │ │ │ │ │ └── StageManager.java │ │ │ │ └── flink │ │ │ │ ├── CarbonLocalProperty.java │ │ │ │ ├── CarbonLocalWriter.java │ │ │ │ ├── CarbonLocalWriterFactory.java │ │ │ │ ├── CarbonLocalWriterFactoryBuilder.java │ │ │ │ ├── CarbonS3Property.java │ │ │ │ ├── CarbonS3Writer.java │ │ │ │ ├── CarbonS3WriterFactory.java │ │ │ │ ├── CarbonS3WriterFactoryBuilder.java │ │ │ │ ├── CarbonWriter.java │ │ │ │ ├── CarbonWriterFactory.java │ │ │ │ └── CarbonWriterFactoryBuilder.java │ │ └── resources │ │ │ └── META-INF │ │ │ └── services │ │ │ └── org.apache.carbon.flink.CarbonWriterFactoryBuilder │ │ └── test │ │ └── scala │ │ └── org │ │ └── apache │ │ └── carbon │ │ └── flink │ │ ├── TestCarbonPartitionWriter.scala │ │ ├── TestCarbonWriter.scala │ │ ├── TestDeleteStageFiles.scala │ │ └── TestSource.scala ├── hive │ ├── pom.xml │ └── src │ │ ├── main │ │ ├── java │ │ │ └── org │ │ │ │ └── apache │ │ │ │ └── carbondata │ │ │ │ └── hive │ │ │ │ ├── CarbonFileHiveSerDe.java │ │ │ │ ├── CarbonHiveInputSplit.java │ │ │ │ ├── CarbonHiveRecordReader.java │ │ │ │ ├── CarbonHiveRow.java │ │ │ │ ├── CarbonHiveSerDe.java │ │ │ │ ├── CarbonStorageHandler.java │ │ │ │ ├── Hive2CarbonExpression.java │ │ │ │ ├── HiveDataTypeUtils.java │ │ │ │ ├── MapredCarbonInputFormat.java │ │ │ │ ├── MapredCarbonOutputCommitter.java │ │ │ │ ├── MapredCarbonOutputFormat.java │ │ │ │ ├── WritableReadSupport.java │ │ │ │ ├── test │ │ │ │ └── server │ │ │ │ │ └── HiveEmbeddedServer2.java │ │ │ │ └── util │ │ │ │ ├── DataTypeUtil.java │ │ │ │ └── HiveCarbonUtil.java │ │ ├── resources │ │ │ ├── array │ │ │ │ └── complexArray.csv │ │ │ ├── complex │ │ │ │ └── complex.csv │ │ │ ├── csv │ │ │ │ └── data.csv │ │ │ ├── log4j.properties │ │ │ ├── map │ │ │ │ └── complexMap.csv │ │ │ ├── struct │ │ │ │ └── struct.csv │ │ │ └── text │ │ │ │ └── string.txt │ │ └── scala │ │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ └── hive │ │ │ └── CarbonHiveMetastoreListener.scala │ │ └── test │ │ └── java │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── hive │ │ ├── Hive2CarbonExpressionTest.java │ │ ├── HiveCarbonTest.java │ │ ├── HiveTestUtils.java │ │ └── TestCarbonSerDe.java ├── presto │ ├── pom.xml │ └── src │ │ ├── main │ │ ├── java │ │ │ └── org │ │ │ │ └── apache │ │ │ │ └── carbondata │ │ │ │ └── presto │ │ │ │ ├── CarbonVectorBatch.java │ │ │ │ ├── ColumnarVectorWrapperDirect.java │ │ │ │ ├── PrestoCarbonVectorizedRecordReader.java │ │ │ │ ├── Types.java │ │ │ │ └── impl │ │ │ │ ├── CarbonLocalInputSplit.java │ │ │ │ ├── CarbonLocalMultiBlockSplit.java │ │ │ │ ├── CarbonTableCacheModel.java │ │ │ │ └── CarbonTableConfig.java │ │ ├── prestodb │ │ │ └── org │ │ │ │ └── apache │ │ │ │ └── carbondata │ │ │ │ └── presto │ │ │ │ ├── CarbondataConnectorFactory.java │ │ │ │ ├── CarbondataModule.java │ │ │ │ ├── CarbondataPageSource.java │ │ │ │ ├── CarbondataPageSourceProvider.java │ │ │ │ ├── CarbondataPlugin.java │ │ │ │ ├── CarbondataSplitManager.java │ │ │ │ ├── PrestoFilterUtil.java │ │ │ │ ├── impl │ │ │ │ └── CarbonTableReader.java │ │ │ │ └── readers │ │ │ │ ├── BooleanStreamReader.java │ │ │ │ ├── ByteStreamReader.java │ │ │ │ ├── ComplexTypeStreamReader.java │ │ │ │ ├── DecimalSliceStreamReader.java │ │ │ │ ├── DoubleStreamReader.java │ │ │ │ ├── FloatStreamReader.java │ │ │ │ ├── IntegerStreamReader.java │ │ │ │ ├── LongStreamReader.java │ │ │ │ ├── ObjectStreamReader.java │ │ │ │ ├── PrestoVectorBlockBuilder.java │ │ │ │ ├── ShortStreamReader.java │ │ │ │ ├── SliceStreamReader.java │ │ │ │ └── TimestampStreamReader.java │ │ ├── prestosql │ │ │ └── org │ │ │ │ └── apache │ │ │ │ └── carbondata │ │ │ │ └── presto │ │ │ │ ├── CarbonDataConnector.java │ │ │ │ ├── CarbonDataFileWriter.java │ │ │ │ ├── CarbonDataFileWriterFactory.java │ │ │ │ ├── CarbonDataHandleResolver.java │ │ │ │ ├── CarbonDataInsertTableHandle.java │ │ │ │ ├── CarbonDataLocationService.java │ │ │ │ ├── CarbonDataMetaData.java │ │ │ │ ├── CarbonDataPageSinkProvider.java │ │ │ │ ├── CarbonDataWriterFactory.java │ │ │ │ ├── CarbonMetadataFactory.java │ │ │ │ ├── CarbondataConnectorFactory.java │ │ │ │ ├── CarbondataModule.java │ │ │ │ ├── CarbondataPageSource.java │ │ │ │ ├── CarbondataPageSourceProvider.java │ │ │ │ ├── CarbondataPlugin.java │ │ │ │ ├── CarbondataSplitManager.java │ │ │ │ ├── InternalCarbonDataConnectorFactory.java │ │ │ │ ├── PrestoFilterUtil.java │ │ │ │ ├── impl │ │ │ │ └── CarbonTableReader.java │ │ │ │ └── readers │ │ │ │ ├── BooleanStreamReader.java │ │ │ │ ├── ByteStreamReader.java │ │ │ │ ├── ComplexTypeStreamReader.java │ │ │ │ ├── DecimalSliceStreamReader.java │ │ │ │ ├── DoubleStreamReader.java │ │ │ │ ├── FloatStreamReader.java │ │ │ │ ├── IntegerStreamReader.java │ │ │ │ ├── LongStreamReader.java │ │ │ │ ├── PrestoVectorBlockBuilder.java │ │ │ │ ├── ShortStreamReader.java │ │ │ │ ├── SliceStreamReader.java │ │ │ │ └── TimestampStreamReader.java │ │ ├── resources │ │ │ └── log4j.properties │ │ └── scala │ │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ └── presto │ │ │ └── CarbonPrestoDecodeReadSupport.scala │ │ └── test │ │ ├── prestodb │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ └── presto │ │ │ └── server │ │ │ ├── PrestoServer.scala │ │ │ └── PrestoTestUtil.scala │ │ ├── prestosql │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ └── presto │ │ │ └── server │ │ │ ├── PrestoServer.scala │ │ │ └── PrestoTestUtil.scala │ │ ├── resources │ │ ├── alldatatype.csv │ │ └── log4j.properties │ │ └── scala │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── presto │ │ ├── integrationtest │ │ ├── PrestoAllDataTypeLocalDictTest.scala │ │ ├── PrestoAllDataTypeTest.scala │ │ ├── PrestoInsertIntoTableTestCase.scala │ │ ├── PrestoTestNonTransactionalTableFiles.scala │ │ └── PrestoTestUsingSparkStore.scala │ │ └── util │ │ └── CarbonDataStoreCreator.scala ├── spark-common-cluster-test │ ├── pom.xml │ └── src │ │ └── test │ │ ├── resources │ │ ├── hdfs-site.xml │ │ └── testdatafileslist.txt │ │ └── scala │ │ └── org │ │ └── apache │ │ ├── carbondata │ │ └── cluster │ │ │ └── sdv │ │ │ ├── generated │ │ │ ├── AlterTableTestCase.scala │ │ │ ├── BadRecordTestCase.scala │ │ │ ├── BloomFilterIndexTestCase.scala │ │ │ ├── ComplexDataTypeTestCase.scala │ │ │ ├── CreateTableAsSelectTestCase.scala │ │ │ ├── CreateTableWithLocalDictionaryTestCase.scala │ │ │ ├── DataLoadingIUDTestCase.scala │ │ │ ├── DataLoadingTestCase.scala │ │ │ ├── DataLoadingV3TestCase.scala │ │ │ ├── GlobalSortTestCase.scala │ │ │ ├── InvertedindexTestCase.scala │ │ │ ├── LoadTableWithLocalDictionaryTestCase.scala │ │ │ ├── LuceneTestCase.scala │ │ │ ├── MergeIndexTestCase.scala │ │ │ ├── OffheapQuery1TestCase.scala │ │ │ ├── OffheapQuery2TestCase.scala │ │ │ ├── OffheapSort1TestCase.scala │ │ │ ├── OffheapSort2TestCase.scala │ │ │ ├── PrestoSampleTestCase.scala │ │ │ ├── QueriesBVATestCase.scala │ │ │ ├── QueriesBasicTestCase.scala │ │ │ ├── QueriesCompactionTestCase.scala │ │ │ ├── QueriesNormalTestCase.scala │ │ │ ├── QueriesRangeFilterTestCase.scala │ │ │ ├── QueriesSparkBlockDistTestCase.scala │ │ │ ├── SDKwriterTestCase.scala │ │ │ ├── SetParameterTestCase.scala │ │ │ ├── ShowLoadsTestCase.scala │ │ │ ├── SortColumnExcudeDictTestCase.scala │ │ │ ├── SortColumnTestCase.scala │ │ │ ├── StandardPartitionTestCase.scala │ │ │ ├── TableCommentAlterTableTestCase.scala │ │ │ ├── TestPartitionWithGlobalSort.scala │ │ │ ├── TimestamptypesTestCase.scala │ │ │ ├── V3offheapvectorTestCase.scala │ │ │ ├── Vector1TestCase.scala │ │ │ ├── Vector2TestCase.scala │ │ │ └── datasource │ │ │ │ ├── CreateTableUsingSparkCarbonFileFormatTestCase.scala │ │ │ │ └── SparkCarbonDataSourceTestCase.scala │ │ │ ├── register │ │ │ └── TestRegisterCarbonTable.scala │ │ │ └── suite │ │ │ └── SDVSuites.scala │ │ └── spark │ │ └── sql │ │ └── common │ │ └── util │ │ ├── CarbonFunSuite.scala │ │ ├── DataSourceTestUtil.scala │ │ ├── PlanTest.scala │ │ ├── QueryTest.scala │ │ └── Tags.scala └── spark │ ├── pom.xml │ └── src │ ├── main │ ├── antlr4 │ │ └── org │ │ │ └── apache │ │ │ └── spark │ │ │ └── sql │ │ │ └── parser │ │ │ └── CarbonSqlBase.g4 │ ├── common2.3and2.4 │ │ └── org │ │ │ └── apache │ │ │ └── spark │ │ │ └── sql │ │ │ ├── CarbonDataSourceScanHelper.scala │ │ │ ├── SparkVersionAdapter.scala │ │ │ ├── execution │ │ │ └── CarbonCodegenSupport.scala │ │ │ ├── hive │ │ │ ├── CarbonAnalyzer.scala │ │ │ ├── CarbonSqlAstBuilder.scala │ │ │ ├── SqlAstBuilderHelper.scala │ │ │ └── execution │ │ │ │ └── command │ │ │ │ └── CarbonResetCommand.scala │ │ │ └── parser │ │ │ ├── CarbonExtensionSqlParser.scala │ │ │ ├── CarbonSparkSqlParser.scala │ │ │ └── SparkSqlAstBuilderWrapper.scala │ ├── common2.4and3.1 │ │ └── org │ │ │ └── apache │ │ │ └── spark │ │ │ └── sql │ │ │ ├── CarbonBoundReference.scala │ │ │ ├── avro │ │ │ └── AvroFileFormatFactory.scala │ │ │ └── execution │ │ │ └── CreateDataSourceTableCommand.scala │ ├── java │ │ └── org │ │ │ └── apache │ │ │ ├── carbondata │ │ │ ├── converter │ │ │ │ └── SparkDataTypeConverterImpl.java │ │ │ ├── index │ │ │ │ └── IndexProvider.java │ │ │ ├── spark │ │ │ │ ├── InitInputMetrics.java │ │ │ │ ├── exception │ │ │ │ │ └── ProcessMetaDataException.java │ │ │ │ ├── load │ │ │ │ │ └── DecimalSerializableComparator.java │ │ │ │ ├── readsupport │ │ │ │ │ ├── SparkGenericRowReadSupportImpl.java │ │ │ │ │ └── SparkRowReadSupportImpl.java │ │ │ │ ├── util │ │ │ │ │ └── Util.java │ │ │ │ └── vectorreader │ │ │ │ │ ├── ColumnarVectorWrapper.java │ │ │ │ │ ├── ColumnarVectorWrapperDirect.java │ │ │ │ │ └── VectorizedCarbonRecordReader.java │ │ │ └── stream │ │ │ │ └── CarbonStreamRecordReader.java │ │ │ └── spark │ │ │ └── sql │ │ │ ├── CarbonAntlrSqlVisitor.java │ │ │ ├── CarbonDictionaryWrapper.java │ │ │ ├── CarbonMergeIntoSQLCommand.scala │ │ │ ├── CarbonVectorProxy.java │ │ │ ├── ColumnVectorFactory.java │ │ │ ├── index │ │ │ └── IndexTableUtil.java │ │ │ ├── merge │ │ │ └── model │ │ │ │ ├── CarbonJoinExpression.java │ │ │ │ ├── CarbonMergeIntoModel.java │ │ │ │ ├── ColumnModel.java │ │ │ │ └── TableModel.java │ │ │ └── secondaryindex │ │ │ ├── exception │ │ │ ├── IndexTableExistException.java │ │ │ └── SecondaryIndexException.java │ │ │ ├── jobs │ │ │ ├── BlockletIndexDetailsWithSchema.java │ │ │ ├── BlockletIndexInputFormat.java │ │ │ └── CarbonBlockLoaderHelper.java │ │ │ ├── load │ │ │ ├── CarbonInternalLoaderUtil.java │ │ │ └── RowComparator.java │ │ │ └── query │ │ │ ├── CarbonSecondaryIndexExecutor.java │ │ │ └── SecondaryIndexQueryResultProcessor.java │ ├── scala │ │ └── org │ │ │ └── apache │ │ │ ├── carbondata │ │ │ ├── api │ │ │ │ └── CarbonStore.scala │ │ │ ├── events │ │ │ │ ├── AlterTableEvents.scala │ │ │ │ ├── CacheEvents.scala │ │ │ │ ├── CarbonInitEvents.scala │ │ │ │ ├── CleanFilesEvents.scala │ │ │ │ ├── CreateCarbonRelationEvent.scala │ │ │ │ ├── CreateDatabaseEvents.scala │ │ │ │ ├── CreateTableEvents.scala │ │ │ │ ├── DeleteSegmentEvents.scala │ │ │ │ ├── DropIndexEvents.scala │ │ │ │ ├── DropTableEvents.scala │ │ │ │ ├── Events.scala │ │ │ │ ├── IUDEvents.scala │ │ │ │ ├── IndexEvents.scala │ │ │ │ ├── IndexServerEvents.scala │ │ │ │ ├── LookupRelationEvents.scala │ │ │ │ ├── RefreshTableEvents.scala │ │ │ │ ├── exception │ │ │ │ │ └── EventExceptions.scala │ │ │ │ └── package.scala │ │ │ ├── geo │ │ │ │ ├── GeoUdfRegister.scala │ │ │ │ ├── GeoUtilUDFs.scala │ │ │ │ ├── GeoUtils.scala │ │ │ │ └── InPolygonUDF.scala │ │ │ ├── index │ │ │ │ ├── CarbonMergeBloomIndexFilesRDD.scala │ │ │ │ ├── IndexRebuildRDD.scala │ │ │ │ ├── TextMatchUDF.scala │ │ │ │ └── secondary │ │ │ │ │ ├── SecondaryIndex.java │ │ │ │ │ ├── SecondaryIndexFactory.java │ │ │ │ │ └── SecondaryIndexModel.java │ │ │ ├── indexserver │ │ │ │ ├── DistributedCountRDD.scala │ │ │ │ ├── DistributedPruneRDD.scala │ │ │ │ ├── DistributedRDDUtils.scala │ │ │ │ ├── DistributedShowCacheRDD.scala │ │ │ │ ├── IndexJobs.scala │ │ │ │ ├── IndexServer.scala │ │ │ │ ├── InvalidateSegmentCacheRDD.scala │ │ │ │ └── SegmentPruneRDD.scala │ │ │ ├── recovery │ │ │ │ └── tablestatus │ │ │ │ │ └── TableStatusRecovery.scala │ │ │ ├── spark │ │ │ │ ├── CarbonColumnValidator.scala │ │ │ │ ├── CarbonOption.scala │ │ │ │ ├── CarbonSparkFactory.scala │ │ │ │ ├── KeyVal.scala │ │ │ │ ├── StreamingOption.scala │ │ │ │ ├── load │ │ │ │ │ ├── CsvRDDHelper.scala │ │ │ │ │ ├── DataLoadProcessBuilderOnSpark.scala │ │ │ │ │ ├── DataLoadProcessorStepOnSpark.scala │ │ │ │ │ └── GlobalSortHelper.scala │ │ │ │ ├── rdd │ │ │ │ │ ├── CarbonDataRDDFactory.scala │ │ │ │ │ ├── CarbonDeltaRowScanRDD.scala │ │ │ │ │ ├── CarbonDropPartitionRDD.scala │ │ │ │ │ ├── CarbonGlobalDictionaryRDD.scala │ │ │ │ │ ├── CarbonMergerRDD.scala │ │ │ │ │ ├── CarbonRDD.scala │ │ │ │ │ ├── CarbonScanRDD.scala │ │ │ │ │ ├── CarbonSparkPartition.scala │ │ │ │ │ ├── CarbonTableCompactor.scala │ │ │ │ │ ├── CompactionFactory.scala │ │ │ │ │ ├── CompactionTaskCompletionListener.scala │ │ │ │ │ ├── Compactor.scala │ │ │ │ │ ├── InsertTaskCompletionListener.scala │ │ │ │ │ ├── NewCarbonDataLoadRDD.scala │ │ │ │ │ ├── QueryTaskCompletionListener.scala │ │ │ │ │ ├── SparkReadSupport.scala │ │ │ │ │ ├── StreamHandoffRDD.scala │ │ │ │ │ └── UpdateDataLoad.scala │ │ │ │ ├── thriftserver │ │ │ │ │ └── CarbonThriftServer.scala │ │ │ │ └── util │ │ │ │ │ ├── CarbonScalaUtil.scala │ │ │ │ │ ├── CarbonSparkUtil.scala │ │ │ │ │ ├── CommonUtil.scala │ │ │ │ │ ├── DataGenerator.scala │ │ │ │ │ └── DataTypeConverterUtil.scala │ │ │ ├── store │ │ │ │ └── SparkCarbonStore.scala │ │ │ ├── stream │ │ │ │ └── StreamJobManager.scala │ │ │ ├── streamer │ │ │ │ ├── AvroDFSSource.scala │ │ │ │ ├── AvroKafkaSource.scala │ │ │ │ ├── CarbonDStream.scala │ │ │ │ ├── CarbonDataStreamer.scala │ │ │ │ ├── CarbonDataStreamerException.scala │ │ │ │ ├── CarbonStreamerConfig.scala │ │ │ │ ├── SchemaSource.scala │ │ │ │ ├── Source.scala │ │ │ │ └── SourceFactory.scala │ │ │ ├── streaming │ │ │ │ ├── CarbonSparkStreamingListener.scala │ │ │ │ ├── CarbonStreamSparkStreaming.scala │ │ │ │ ├── CarbonStreamingQueryListener.scala │ │ │ │ └── StreamSinkFactory.scala │ │ │ ├── trash │ │ │ │ └── DataTrashManager.scala │ │ │ └── view │ │ │ │ ├── MVCatalogInSpark.scala │ │ │ │ ├── MVEvents.scala │ │ │ │ ├── MVField.scala │ │ │ │ ├── MVFunctions.scala │ │ │ │ ├── MVHelper.scala │ │ │ │ ├── MVManagerInSpark.scala │ │ │ │ ├── MVPlanWrapper.scala │ │ │ │ ├── MVRefresher.scala │ │ │ │ ├── MVSchemaWrapper.scala │ │ │ │ └── MVTimeGranularity.scala │ │ │ └── spark │ │ │ ├── CarbonInputMetrics.scala │ │ │ ├── DataSkewRangePartitioner.scala │ │ │ ├── rdd │ │ │ ├── CarbonMergeFilesRDD.scala │ │ │ ├── DataLoadCoalescedRDD.scala │ │ │ └── DataLoadPartitionCoalescer.scala │ │ │ ├── sql │ │ │ ├── CarbonBoundReference.scala │ │ │ ├── CarbonCatalystOperators.scala │ │ │ ├── CarbonCountStar.scala │ │ │ ├── CarbonDataFrameWriter.scala │ │ │ ├── CarbonDatasourceHadoopRelation.scala │ │ │ ├── CarbonEnv.scala │ │ │ ├── CarbonExpressions.scala │ │ │ ├── CarbonExtensions.scala │ │ │ ├── CarbonSession.scala │ │ │ ├── CarbonSource.scala │ │ │ ├── CarbonSparkStreamingFactory.scala │ │ │ ├── CarbonThreadUtil.scala │ │ │ ├── CustomDeterministicExpression.scala │ │ │ ├── EnvHelper.scala │ │ │ ├── SQLConf.scala │ │ │ ├── SparkUnknownExpression.scala │ │ │ ├── carbondata │ │ │ │ └── execution │ │ │ │ │ └── datasources │ │ │ │ │ ├── CarbonFileIndex.scala │ │ │ │ │ ├── CarbonFileIndexReplaceRule.scala │ │ │ │ │ ├── CarbonSparkDataSourceUtil.scala │ │ │ │ │ ├── SparkCarbonFileFormat.scala │ │ │ │ │ ├── readsupport │ │ │ │ │ └── SparkUnsafeRowReadSupport.scala │ │ │ │ │ └── tasklisteners │ │ │ │ │ └── CarbonTaskCompletionListener.scala │ │ │ ├── catalyst │ │ │ │ ├── AbstractCarbonSparkSQLParser.scala │ │ │ │ ├── CarbonDDLSqlParser.scala │ │ │ │ ├── CarbonParserUtil.scala │ │ │ │ ├── CarbonTableIdentifierImplicit.scala │ │ │ │ └── analysis │ │ │ │ │ └── EmptyRule.scala │ │ │ ├── events │ │ │ │ ├── MergeBloomIndexEventListener.scala │ │ │ │ └── MergeIndexEventListener.scala │ │ │ ├── execution │ │ │ │ ├── CarbonTakeOrderedAndProjectExec.scala │ │ │ │ ├── CastExpressionOptimization.scala │ │ │ │ ├── command │ │ │ │ │ ├── cache │ │ │ │ │ │ ├── CacheUtil.scala │ │ │ │ │ │ ├── CarbonDropCacheCommand.scala │ │ │ │ │ │ └── CarbonShowCacheCommand.scala │ │ │ │ │ ├── carbonTableSchemaCommon.scala │ │ │ │ │ ├── index │ │ │ │ │ │ ├── CarbonCreateIndexCommand.scala │ │ │ │ │ │ ├── CarbonRefreshIndexCommand.scala │ │ │ │ │ │ ├── DropIndexCommand.scala │ │ │ │ │ │ ├── IndexRepairCommand.scala │ │ │ │ │ │ └── ShowIndexesCommand.scala │ │ │ │ │ ├── management │ │ │ │ │ │ ├── CarbonAddLoadCommand.scala │ │ │ │ │ │ ├── CarbonAlterTableCompactionCommand.scala │ │ │ │ │ │ ├── CarbonAlterTableFinishStreaming.scala │ │ │ │ │ │ ├── CarbonCleanFilesCommand.scala │ │ │ │ │ │ ├── CarbonCliCommand.scala │ │ │ │ │ │ ├── CarbonDeleteLoadByIdCommand.scala │ │ │ │ │ │ ├── CarbonDeleteLoadByLoadDateCommand.scala │ │ │ │ │ │ ├── CarbonDeleteStageFilesCommand.scala │ │ │ │ │ │ ├── CarbonInsertFromStageCommand.scala │ │ │ │ │ │ ├── CarbonInsertIntoCommand.scala │ │ │ │ │ │ ├── CarbonInsertIntoHadoopFsRelationCommand.scala │ │ │ │ │ │ ├── CarbonInsertIntoWithDf.scala │ │ │ │ │ │ ├── CarbonLoadDataCommand.scala │ │ │ │ │ │ ├── CarbonLoadParams.scala │ │ │ │ │ │ ├── CarbonShowSegmentsAsSelectCommand.scala │ │ │ │ │ │ ├── CarbonShowSegmentsCommand.scala │ │ │ │ │ │ ├── CommonLoadUtils.scala │ │ │ │ │ │ └── RefreshCarbonTableCommand.scala │ │ │ │ │ ├── mutation │ │ │ │ │ │ ├── CarbonProjectForDeleteCommand.scala │ │ │ │ │ │ ├── CarbonProjectForUpdateCommand.scala │ │ │ │ │ │ ├── CarbonTruncateCommand.scala │ │ │ │ │ │ ├── DeleteExecution.scala │ │ │ │ │ │ ├── HorizontalCompaction.scala │ │ │ │ │ │ ├── HorizontalCompactionException.scala │ │ │ │ │ │ ├── IUDCommonUtil.scala │ │ │ │ │ │ └── merge │ │ │ │ │ │ │ ├── CarbonMergeDataSetCommand.scala │ │ │ │ │ │ │ ├── CarbonMergeDataSetException.scala │ │ │ │ │ │ │ ├── CarbonMergeDataSetUtil.scala │ │ │ │ │ │ │ ├── HistoryTableLoadHelper.scala │ │ │ │ │ │ │ ├── MergeDataSetBuilder.scala │ │ │ │ │ │ │ ├── MergeHandler.scala │ │ │ │ │ │ │ ├── MergeOperationType.scala │ │ │ │ │ │ │ ├── MergeProjection.scala │ │ │ │ │ │ │ ├── MergeUtil.scala │ │ │ │ │ │ │ ├── MutationAction.scala │ │ │ │ │ │ │ ├── TranxManager.scala │ │ │ │ │ │ │ ├── UpsertBuilder.scala │ │ │ │ │ │ │ ├── interfaces.scala │ │ │ │ │ │ │ └── udf │ │ │ │ │ │ │ └── BlockPathsUDF.scala │ │ │ │ │ ├── package.scala │ │ │ │ │ ├── partition │ │ │ │ │ │ ├── CarbonAlterTableAddHivePartitionCommand.scala │ │ │ │ │ │ └── CarbonAlterTableDropHivePartitionCommand.scala │ │ │ │ │ ├── schema │ │ │ │ │ │ ├── CarbonAlterTableAddColumnCommand.scala │ │ │ │ │ │ ├── CarbonAlterTableColRenameDataTypeChangeCommand.scala │ │ │ │ │ │ ├── CarbonAlterTableDropColumnCommand.scala │ │ │ │ │ │ ├── CarbonAlterTableRenameCommand.scala │ │ │ │ │ │ ├── CarbonAlterTableSetCommand.scala │ │ │ │ │ │ └── CarbonAlterTableUnsetCommand.scala │ │ │ │ │ ├── stream │ │ │ │ │ │ ├── CarbonCreateStreamCommand.scala │ │ │ │ │ │ ├── CarbonDropStreamCommand.scala │ │ │ │ │ │ └── CarbonShowStreamsCommand.scala │ │ │ │ │ ├── table │ │ │ │ │ │ ├── CarbonCreateDataSourceTableCommand.scala │ │ │ │ │ │ ├── CarbonCreateTableAsSelectCommand.scala │ │ │ │ │ │ ├── CarbonCreateTableCommand.scala │ │ │ │ │ │ ├── CarbonCreateTableLikeCommand.scala │ │ │ │ │ │ ├── CarbonDescribeFormattedCommand.scala │ │ │ │ │ │ ├── CarbonDropTableCommand.scala │ │ │ │ │ │ ├── CarbonExplainCommand.scala │ │ │ │ │ │ ├── CarbonShowCreateTableCommand.scala │ │ │ │ │ │ └── CarbonShowTablesCommand.scala │ │ │ │ │ └── view │ │ │ │ │ │ ├── CarbonCreateMVCommand.scala │ │ │ │ │ │ ├── CarbonDropMVCommand.scala │ │ │ │ │ │ ├── CarbonRefreshMVCommand.scala │ │ │ │ │ │ └── CarbonShowMVCommand.scala │ │ │ │ ├── datasources │ │ │ │ │ └── SparkCarbonTableFormat.scala │ │ │ │ ├── joins │ │ │ │ │ └── BroadCastPolygonFilterPushJoin.scala │ │ │ │ ├── strategy │ │ │ │ │ ├── CarbonDataSourceScan.scala │ │ │ │ │ ├── CarbonPlanHelper.scala │ │ │ │ │ ├── CarbonSourceStrategy.scala │ │ │ │ │ ├── DDLHelper.scala │ │ │ │ │ ├── DDLStrategy.scala │ │ │ │ │ ├── DMLHelper.scala │ │ │ │ │ ├── DMLStrategy.scala │ │ │ │ │ ├── MixedFormatHandler.scala │ │ │ │ │ ├── PushDownHelper.scala │ │ │ │ │ └── StreamingTableStrategy.scala │ │ │ │ └── streaming │ │ │ │ │ └── CarbonAppendableStreamSink.scala │ │ │ ├── hive │ │ │ │ ├── CarbonAnalysisRules.scala │ │ │ │ ├── CarbonFileMetastore.scala │ │ │ │ ├── CarbonHiveIndexMetadataUtil.scala │ │ │ │ ├── CarbonHiveMetaStore.scala │ │ │ │ ├── CarbonMetaStore.scala │ │ │ │ ├── CarbonRelation.scala │ │ │ │ ├── CarbonSQLConf.scala │ │ │ │ ├── CarbonSessionCatalog.scala │ │ │ │ ├── CarbonSessionCatalogUtil.scala │ │ │ │ ├── CarbonSessionUtil.scala │ │ │ │ ├── CreateCarbonSourceTableAsSelectCommand.scala │ │ │ │ ├── DistributionUtil.scala │ │ │ │ ├── cli │ │ │ │ │ └── CarbonSQLCLIDriver.scala │ │ │ │ └── execution │ │ │ │ │ └── command │ │ │ │ │ └── CarbonHiveCommands.scala │ │ │ ├── index │ │ │ │ └── CarbonIndexUtil.scala │ │ │ ├── listeners │ │ │ │ ├── DropCacheEventListeners.scala │ │ │ │ ├── MVListeners.scala │ │ │ │ ├── PrePrimingListener.scala │ │ │ │ └── ShowCacheEventListener.scala │ │ │ ├── optimizer │ │ │ │ ├── CarbonFilters.scala │ │ │ │ ├── CarbonIUDRule.scala │ │ │ │ ├── CarbonUDFTransformRule.scala │ │ │ │ ├── MVMatcher.scala │ │ │ │ ├── MVRewrite.scala │ │ │ │ └── MVRewriteRule.scala │ │ │ ├── parser │ │ │ │ ├── CarbonAntlrParser.scala │ │ │ │ ├── CarbonExtensionSpark2SqlParser.scala │ │ │ │ ├── CarbonSpark2SqlParser.scala │ │ │ │ ├── CarbonSparkSqlParserUtil.scala │ │ │ │ └── MVQueryParser.scala │ │ │ ├── profiler │ │ │ │ ├── Profiler.scala │ │ │ │ ├── ProfilerListener.scala │ │ │ │ └── ProfilerLogger.scala │ │ │ ├── secondaryindex │ │ │ │ ├── command │ │ │ │ │ ├── RegisterIndexTableCommand.scala │ │ │ │ │ ├── SICreationCommand.scala │ │ │ │ │ ├── SILoadCommand.scala │ │ │ │ │ └── SIRebuildSegmentRunner.scala │ │ │ │ ├── events │ │ │ │ │ ├── AlterTableColumnRenameEventListener.scala │ │ │ │ │ ├── AlterTableCompactionPostEventListener.scala │ │ │ │ │ ├── AlterTableDropColumnEventListener.scala │ │ │ │ │ ├── AlterTableMergeIndexSIEventListener.scala │ │ │ │ │ ├── AlterTableRenameEventListener.scala │ │ │ │ │ ├── CleanFilesPostEventListener.scala │ │ │ │ │ ├── CreateCarbonRelationEventListener.scala │ │ │ │ │ ├── DeleteFromTableEventListener.scala │ │ │ │ │ ├── DeleteSegmentByDateListener.scala │ │ │ │ │ ├── DeleteSegmentByIdListener.scala │ │ │ │ │ ├── DropCacheSIEventListener.scala │ │ │ │ │ ├── LoadSIEvents.scala │ │ │ │ │ ├── SIDropEventListener.scala │ │ │ │ │ ├── SILoadEventListener.scala │ │ │ │ │ ├── SILoadEventListenerForFailedSegments.scala │ │ │ │ │ ├── SIRefreshEventListener.scala │ │ │ │ │ ├── ShowCacheSIEventListener.scala │ │ │ │ │ └── UpdateTablePreEventListener.scala │ │ │ │ ├── hive │ │ │ │ │ └── CarbonInternalMetastore.scala │ │ │ │ ├── jobs │ │ │ │ │ ├── SparkBlockletIndexLoaderJob.scala │ │ │ │ │ └── StringProjectionQueryJob.scala │ │ │ │ ├── joins │ │ │ │ │ └── BroadCastSIFilterPushJoin.scala │ │ │ │ ├── load │ │ │ │ │ └── Compactor.scala │ │ │ │ ├── optimizer │ │ │ │ │ ├── CarbonSITransformationRule.scala │ │ │ │ │ └── CarbonSecondaryIndexOptimizer.scala │ │ │ │ ├── rdd │ │ │ │ │ ├── CarbonSIRebuildRDD.scala │ │ │ │ │ ├── CarbonSecondaryIndexRDD.scala │ │ │ │ │ └── SecondaryIndexCreator.scala │ │ │ │ └── util │ │ │ │ │ ├── FileInternalUtil.scala │ │ │ │ │ ├── InternalKeyVal.scala │ │ │ │ │ └── SecondaryIndexUtil.scala │ │ │ ├── test │ │ │ │ ├── ResourceRegisterAndCopier.scala │ │ │ │ ├── SparkTestQueryExecutor.scala │ │ │ │ ├── TestQueryExecutor.scala │ │ │ │ └── util │ │ │ │ │ ├── CarbonFunSuite.scala │ │ │ │ │ ├── PlanTest.scala │ │ │ │ │ └── QueryTest.scala │ │ │ └── util │ │ │ │ ├── CarbonException.scala │ │ │ │ ├── CarbonMetastoreTypes.scala │ │ │ │ ├── CreateTableCommonUtil.scala │ │ │ │ ├── SparkSQLUtil.scala │ │ │ │ └── SparkTypeConverter.scala │ │ │ └── util │ │ │ ├── AlterTableUtil.scala │ │ │ ├── CarbonReflectionUtils.scala │ │ │ ├── CleanFiles.scala │ │ │ ├── Compaction.scala │ │ │ ├── DeleteSegmentByDate.scala │ │ │ ├── DeleteSegmentById.scala │ │ │ ├── FileUtils.scala │ │ │ ├── MergeIndexUtil.scala │ │ │ ├── PartitionCacheManager.scala │ │ │ ├── ScalaCompilerUtil.scala │ │ │ ├── SparkUtil.scala │ │ │ ├── TableAPIUtil.scala │ │ │ └── TableLoader.scala │ ├── spark2.3 │ │ ├── com │ │ │ └── databricks │ │ │ │ └── spark │ │ │ │ └── avro │ │ │ │ └── AvroWriter.scala │ │ └── org │ │ │ └── apache │ │ │ └── spark │ │ │ └── sql │ │ │ ├── CarbonBoundReference.scala │ │ │ ├── CarbonToSparkAdapter.scala │ │ │ ├── SparkSqlAdapter.scala │ │ │ ├── avro │ │ │ └── AvroFileFormatFactory.scala │ │ │ ├── execution │ │ │ └── CreateDataSourceTableCommand.scala │ │ │ └── hive │ │ │ └── CarbonSessionStateBuilder.scala │ ├── spark2.4 │ │ └── org │ │ │ └── apache │ │ │ └── spark │ │ │ └── sql │ │ │ ├── CarbonToSparkAdapter.scala │ │ │ ├── SparkSqlAdapter.scala │ │ │ └── hive │ │ │ └── CarbonSessionStateBuilder.scala │ └── spark3.1 │ │ └── org │ │ └── apache │ │ └── spark │ │ └── sql │ │ ├── CarbonDataSourceScanHelper.scala │ │ ├── CarbonToSparkAdapter.scala │ │ ├── SparkSqlAdapter.scala │ │ ├── SparkVersionAdapter.scala │ │ ├── execution │ │ └── CarbonCodegenSupport.scala │ │ ├── hive │ │ ├── CarbonAnalyzer.scala │ │ ├── CarbonSessionStateBuilder.scala │ │ ├── CarbonSqlAstBuilder.scala │ │ ├── SqlAstBuilderHelper.scala │ │ └── execution │ │ │ └── command │ │ │ └── CarbonResetCommand.scala │ │ └── parser │ │ ├── CarbonExtensionSqlParser.scala │ │ ├── CarbonSparkSqlParser.scala │ │ └── SparkSqlAstBuilderWrapper.scala │ ├── resources │ └── META-INF │ │ └── services │ │ ├── org.apache.spark.sql.sources.DataSourceRegister │ │ └── org.apache.spark.sql.test.TestQueryExecutorRegister │ └── test │ ├── java │ └── org │ │ └── apache │ │ └── carbondata │ │ ├── sdk │ │ └── util │ │ │ └── BinaryUtil.java │ │ └── stream │ │ └── CarbonStreamRecordReaderTest.java │ ├── resources │ ├── 100_olap.csv │ ├── 10dim_4msr.csv │ ├── 32000char.csv │ ├── Array.csv │ ├── IUD │ │ ├── T_Hive1.csv │ │ ├── bad_record.csv │ │ ├── badrecord.csv │ │ ├── comp1.csv │ │ ├── comp2.csv │ │ ├── comp3.csv │ │ ├── comp4.csv │ │ ├── dest.csv │ │ ├── negativevalue.csv │ │ ├── other.csv │ │ ├── sample.csv │ │ ├── sample_updated.csv │ │ ├── source2.csv │ │ ├── source3.csv │ │ ├── update01.csv │ │ └── updateinpartition.csv │ ├── MoreThan32KChar.csv │ ├── OLDFORMATTABLE.csv │ ├── OLDFORMATTABLEHIVE.csv │ ├── Struct.csv │ ├── StructofStruct.csv │ ├── Test_Data1_Logrithmic.csv │ ├── adap.csv │ ├── adap_double1.csv │ ├── adap_double2.csv │ ├── adap_double3.csv │ ├── adap_double4.csv │ ├── adap_int1.csv │ ├── adap_int2.csv │ ├── adap_int3.csv │ ├── alldatatypeforpartition.csv │ ├── alldictionary │ │ ├── complex │ │ │ └── 20160423 │ │ │ │ └── 1400_1405 │ │ │ │ └── complex.dictionary │ │ └── sample │ │ │ └── 20160423 │ │ │ └── 1400_1405 │ │ │ └── sample.dictionary │ ├── array1.csv │ ├── arrayColumnEmpty.csv │ ├── avgTest.csv │ ├── badrecords │ │ ├── bigtab.csv │ │ ├── bigtabbad.csv │ │ ├── complexdata.csv │ │ ├── datasample.csv │ │ ├── dummy.csv │ │ ├── dummy2.csv │ │ ├── emptyTimeStampValue.csv │ │ ├── emptyValues.csv │ │ ├── insufficientColumns.csv │ │ └── seriazableValue.csv │ ├── bigIntData.csv │ ├── bigIntDataWithHeader.csv │ ├── bigIntDataWithoutHeader.csv │ ├── big_decimal_without_header.csv │ ├── big_int_Decimal.csv │ ├── binaryDataBase64.csv │ ├── binaryDataHex.csv │ ├── binaryStringNullData.csv │ ├── binarystringdata.csv │ ├── binarystringdata2.csv │ ├── binarystringdatawithHead.csv │ ├── bool │ │ ├── supportBoolean.csv │ │ ├── supportBooleanBadRecords.csv │ │ ├── supportBooleanDifferentFormat.csv │ │ ├── supportBooleanOnlyBoolean.csv │ │ ├── supportBooleanTwoBooleanColumns.csv │ │ └── supportBooleanWithFileHeader.csv │ ├── carriage_return_in_string.csv │ ├── channelsId.csv │ ├── character_carbon.csv │ ├── character_hive.csv │ ├── columndictionary │ │ ├── country.csv │ │ └── name.csv │ ├── comment.csv │ ├── compaction │ │ ├── compaction1.csv │ │ ├── compaction1_forhive.csv │ │ ├── compaction2.csv │ │ ├── compaction3.csv │ │ ├── compactionIUD1.csv │ │ ├── compactionIUD2.csv │ │ ├── compactionIUD3.csv │ │ ├── compactionIUD4.csv │ │ ├── compactioncard2.csv │ │ ├── compactioncard2_forhive.csv │ │ └── nodictionary_compaction.csv │ ├── complexTypeDecimal.csv │ ├── complexTypeDecimalNested.csv │ ├── complexTypeDecimalNestedHive.csv │ ├── complexbinary.csv │ ├── complexdata.csv │ ├── complexdata1.csv │ ├── complexdata2.csv │ ├── complexdata3.csv │ ├── complexdatareordered.csv │ ├── complexdatastructextra.csv │ ├── complextypeWithEmptyRecords.csv │ ├── complextypediffentcolheaderorder.csv │ ├── complextypesample.csv │ ├── complextypespecialchardelimiter.csv │ ├── data.csv │ ├── data1.csv │ ├── data2.csv │ ├── data2_DiffTimeFormat.csv │ ├── dataIncrement.csv │ ├── dataWithEmptyRows.csv │ ├── dataWithNegativeValues.csv │ ├── dataWithNullFirstLine.csv │ ├── dataWithSingleQuote.csv │ ├── data_alltypes.csv │ ├── data_beyond68yrs.csv │ ├── data_big.csv │ ├── data_partition_badrecords.csv │ ├── data_sort.csv │ ├── data_timestamp.csv │ ├── data_withCAPSHeader.csv │ ├── data_withMixedHeader.csv │ ├── data_with_all_types.csv │ ├── data_with_special_char.csv │ ├── datadelimiter.csv │ ├── datanullmeasurecol.csv │ ├── dataretention1.csv │ ├── dataretention11.csv │ ├── dataretention2.csv │ ├── dataretention3.csv │ ├── datasample.csv │ ├── datasamplecomplex.csv │ ├── datasamplefordate.csv │ ├── datasamplenull.csv │ ├── datasingleCol.csv │ ├── datasingleComplexCol.csv │ ├── datawithNegeativewithoutHeader.csv │ ├── datawithNegtiveNumber.csv │ ├── datawithbackslash.csv │ ├── datawithblanklines.csv │ ├── datawithcomplexspecialchar.csv │ ├── datawithescapecharacter.csv │ ├── datawithmaxbigint.csv │ ├── datawithmaxinteger.csv │ ├── datawithmaxminbigint.csv │ ├── datawithmaxmininteger.csv │ ├── datawithminbigint.csv │ ├── datawithmininteger.csv │ ├── datawithnullmeasure.csv │ ├── datawithnullmsrs.csv │ ├── datawithoutheader.csv │ ├── datawithspecialcharacter.csv │ ├── datedatafile.csv │ ├── dblocation │ │ └── test.csv │ ├── decimalBoundaryDataCarbon.csv │ ├── decimalBoundaryDataHive.csv │ ├── decimalData.csv │ ├── decimalDataWithHeader.csv │ ├── decimalDataWithoutHeader.csv │ ├── decimal_int_range.csv │ ├── deviceInformationId.csv │ ├── deviceInformationId2.csv │ ├── dimSample.csv │ ├── dimTableSample.csv │ ├── double.csv │ ├── double │ │ ├── data_notitle_AdaptiveFloating_byte.csv │ │ ├── data_notitle_AdaptiveFloating_int.csv │ │ ├── data_notitle_AdaptiveFloating_short.csv │ │ ├── data_notitle_AdaptiveFloating_short_int.csv │ │ ├── data_notitle_byte.csv │ │ ├── data_notitle_int.csv │ │ ├── data_notitle_long.csv │ │ ├── data_notitle_short.csv │ │ └── data_notitle_short_int.csv │ ├── emp.csv │ ├── emptyDimensionData.csv │ ├── emptyDimensionDataHive.csv │ ├── emptylines.csv │ ├── emptyrow │ │ ├── csvwithonlyspacechar.csv │ │ └── emptyRows.csv │ ├── encoding_types.csv │ ├── filter │ │ ├── betweenFilter.csv │ │ ├── datagrtlrt.csv │ │ ├── datawithnull.csv │ │ ├── datawithoutnull.csv │ │ ├── emp2.csv │ │ ├── emp2allnull.csv │ │ ├── emp2nonull.csv │ │ ├── notEqualToFilter.csv │ │ └── notNullFilter.csv │ ├── floatSample.csv │ ├── geodata.csv │ ├── geodata2.csv │ ├── geodata3.csv │ ├── geodataWithCorrectSpatialIndex.csv │ ├── geodataWithErrorSpatialIndex.csv │ ├── globalsort │ │ ├── sample1.csv │ │ ├── sample2.csv │ │ └── sample3.csv │ ├── hiverangenodictionarycompare.csv │ ├── invalidMeasures.csv │ ├── j2.csv │ ├── join │ │ ├── data1.csv │ │ ├── data2.csv │ │ ├── emp.csv │ │ ├── employee.csv │ │ ├── mgr.csv │ │ └── mobile.csv │ ├── jsonFiles │ │ ├── data │ │ │ ├── PrimitiveTypeWithNull.json │ │ │ ├── StructOfAllTypes.json │ │ │ ├── allPrimitiveType.json │ │ │ ├── allPrimitiveTypeBadRecord.json │ │ │ ├── arrayOfStructOfStruct.json │ │ │ ├── arrayOfarrayOfarrayOfStruct.json │ │ │ └── similarSchemaFiles │ │ │ │ ├── JsonReaderTest │ │ │ │ ├── MultipleRowSingleLineJson.json │ │ │ │ ├── SingleRowSingleLineJson.json │ │ │ │ └── withRecordIdentifier │ │ │ │ │ ├── MultipleRowMultipleLineJsonWithRecordIdentifier.json │ │ │ │ │ ├── SingleRowMultipleLineJsonWithRecordIdentifier.json │ │ │ │ │ └── SingleRowSingleLineJsonWithRecordIdentifier.json │ │ │ │ ├── allPrimitiveTypeMultipleRows.json │ │ │ │ └── allPrimitiveTypeSingleArray.json │ │ └── schema │ │ │ ├── StructOfAllTypes.avsc │ │ │ ├── arrayOfStructOfStruct.avsc │ │ │ └── arrayOfarrayOfarrayOfStruct.avsc │ ├── lessthandatacolumndata.csv │ ├── loadMultiFiles │ │ ├── .invisibilityfile │ │ ├── _SUCCESS │ │ ├── data.csv │ │ ├── emptyfile.csv │ │ ├── nestedfolder1 │ │ │ ├── data.csv │ │ │ ├── data1.csv │ │ │ └── nestedfolder2 │ │ │ │ └── data.csv │ │ └── non-csv │ ├── localdictionary.csv │ ├── locationInfoActiveCountry.csv │ ├── mac.csv │ ├── measureinsertintotest.csv │ ├── mobileimei.csv │ ├── mv_sampledata.csv │ ├── newsample.csv │ ├── noneCsvFormat.cs │ ├── nontransactional.csv │ ├── nontransactional1.csv │ ├── nullSample.csv │ ├── nullandnonparsableValue.csv │ ├── nullmeasurevalue.csv │ ├── nullvalueserialization.csv │ ├── numeric_column_invalid_values.csv │ ├── oscon_10.csv │ ├── outofrange.csv │ ├── overwriteTable1_noRecord.csv │ ├── overwriteTable1_someRecord.csv │ ├── overwriteTable2_noRecord.csv │ ├── overwriteTable2_someRecord.csv │ ├── partData.csv │ ├── partition_data.csv │ ├── partition_data_example.csv │ ├── predefdic │ │ ├── allpredefdictionary.csv │ │ ├── data3.csv │ │ └── dicfilepath.csv │ ├── products.csv │ ├── range_column │ │ └── dataskew.csv │ ├── rangedata.csv │ ├── rangedatasample.csv │ ├── rangenodictionarycompare.csv │ ├── restructure │ │ ├── data1.csv │ │ ├── data2.csv │ │ ├── data3.csv │ │ ├── data4.csv │ │ ├── data5.csv │ │ ├── data6.csv │ │ ├── data7.csv │ │ └── data_2000.csv │ ├── sales_data.csv │ ├── sample │ ├── sample.csv │ ├── sample.csv.bz2 │ ├── sample.csv.gz │ ├── sampleComplex.csv │ ├── sample_withDelimiter017.csv │ ├── secindex │ │ ├── IUD │ │ │ ├── sample_1.csv │ │ │ └── sample_2.csv │ │ ├── array.csv │ │ ├── array2.csv │ │ ├── data_10000.csv │ │ ├── datafile_100.csv │ │ ├── dest.csv │ │ ├── dest1.csv │ │ ├── dest2.csv │ │ ├── dest3.csv │ │ ├── firstunique.csv │ │ ├── index.csv │ │ ├── secondaryIndexLikeTest.csv │ │ ├── secondunique.csv │ │ └── source3.csv │ ├── seq_20Records.csv │ ├── shortintboundary.csv │ ├── shortolap.csv │ ├── sort_columns │ │ ├── alldatatype1.csv │ │ └── alldatatype2.csv │ ├── source.csv │ ├── source_without_header.csv │ ├── streamSample.csv │ ├── streamSample_with_long_string.csv │ ├── struct_all.csv │ ├── structofarray.csv │ ├── structusingstruct.csv │ ├── temp │ │ └── data1.csv │ ├── test.json │ ├── testBigInt_boundary_value.csv │ ├── testShortAndIntDataType.csv │ ├── test_json.json │ ├── timeStampFormatData1.csv │ ├── timeStampFormatData2.csv │ ├── timeseriestest.csv │ ├── timestamp.csv │ ├── timestampdata.csv │ ├── timestampdatafile.csv │ ├── tpch │ │ ├── customers.csv │ │ ├── lineitem.csv │ │ ├── nation.csv │ │ ├── orders.csv │ │ ├── region.csv │ │ └── supplier.csv │ ├── unicodechar.csv │ ├── uniq.csv │ ├── uniqwithoutheader.csv │ ├── vardhandaterestruct.csv │ └── verticalDelimitedData.csv │ └── scala │ └── org │ └── apache │ ├── carbondata │ ├── geo │ │ ├── GeoQueryTest.scala │ │ └── GeoTest.scala │ ├── index │ │ ├── bloom │ │ │ ├── BloomCoarseGrainIndexFunctionSuite.scala │ │ │ ├── BloomCoarseGrainIndexSuite.scala │ │ │ └── BloomCoarseGrainIndexTestUtil.scala │ │ └── lucene │ │ │ ├── LuceneCoarseGrainIndexSuite.scala │ │ │ └── LuceneFineGrainIndexSuite.scala │ ├── integration │ │ └── spark │ │ │ └── testsuite │ │ │ ├── aggquery │ │ │ └── IntegerDataTypeTestCase.scala │ │ │ ├── bigdecimal │ │ │ ├── TestBigInt.scala │ │ │ └── TestDimensionWithDecimalDataType.scala │ │ │ ├── binary │ │ │ └── TestBinaryDataType.scala │ │ │ ├── complexType │ │ │ ├── TestAdaptiveComplexType.scala │ │ │ ├── TestAdaptiveEncodingForNullValues.scala │ │ │ ├── TestAdaptiveEncodingSafeColumnPageForComplexDataType.scala │ │ │ ├── TestAdaptiveEncodingUnsafeColumnPageForComplexDataType.scala │ │ │ ├── TestAdaptiveEncodingUnsafeHeapColumnPageForComplexDataType.scala │ │ │ ├── TestAllComplexDataType.scala │ │ │ ├── TestArrayContainsPushDown.scala │ │ │ ├── TestCompactionComplexType.scala │ │ │ ├── TestComplexDataType.scala │ │ │ ├── TestComplexTypeQuery.scala │ │ │ ├── TestComplexTypeWithBigArray.scala │ │ │ └── TestCreateTableWithDouble.scala │ │ │ ├── dataload │ │ │ ├── MultiFilesDataLoagdingTestCase.scala │ │ │ ├── SparkStoreCreatorForPresto.scala │ │ │ ├── TestLoadDataGeneral.scala │ │ │ ├── TestLoadDataWithAutoLoadMerge.scala │ │ │ ├── TestLoadDataWithBlankLine.scala │ │ │ ├── TestLoadDataWithCompression.scala │ │ │ ├── TestLoadDataWithEmptyArrayColumns.scala │ │ │ ├── TestLoadDataWithJunkChars.scala │ │ │ ├── TestLoadDataWithMaxMinBigInt.scala │ │ │ ├── TestLoadDataWithMaxMinInteger.scala │ │ │ ├── TestLoadDataWithNullMeasures.scala │ │ │ ├── TestLoadDataWithSortColumnBounds.scala │ │ │ ├── TestLoadDataWithStaleDataInSegmentFolder.scala │ │ │ ├── TestLoadDataWithUnsafeMemory.scala │ │ │ ├── TestLoadDataWithYarnLocalDirs.scala │ │ │ └── TestNoInvertedIndexLoadAndQuery.scala │ │ │ ├── emptyrow │ │ │ ├── TestCSVHavingOnlySpaceChar.scala │ │ │ ├── TestEmptyRows.scala │ │ │ └── TestSkipEmptyLines.scala │ │ │ ├── primitiveTypes │ │ │ ├── ArrayDataTypeTestCase.scala │ │ │ ├── DoubleDataTypeTestCase.scala │ │ │ ├── FloatDataTypeTestCase.scala │ │ │ ├── MapDataTypeTestCase.scala │ │ │ └── TestAdaptiveEncodingForPrimitiveTypes.scala │ │ │ └── recovery │ │ │ └── TableStatusRecoveryTest.scala │ ├── spark │ │ ├── testsuite │ │ │ ├── ShowTable │ │ │ │ └── TestShowTable.scala │ │ │ ├── TestCarbonCli.scala │ │ │ ├── addsegment │ │ │ │ └── AddSegmentTestCase.scala │ │ │ ├── aggquery │ │ │ │ ├── AllDataTypesTestCaseAggregate.scala │ │ │ │ └── AverageQueryTestCase.scala │ │ │ ├── allqueries │ │ │ │ ├── AllDataTypesTestCase.scala │ │ │ │ ├── DoubleDataTypeTest.scala │ │ │ │ ├── InsertIntoCarbonTableSpark2TestCase.scala │ │ │ │ ├── InsertIntoCarbonTableTestCase.scala │ │ │ │ ├── MeasureOnlyTableTestCases.scala │ │ │ │ ├── TestPruneUsingSegmentMinMax.scala │ │ │ │ ├── TestQueryWithColumnMetCacheAndCacheLevelProperty.scala │ │ │ │ ├── TestQueryWithoutDataLoad.scala │ │ │ │ └── TestTableNameHasDbName.scala │ │ │ ├── alterTable │ │ │ │ ├── TestAlterTableAddColumns.scala │ │ │ │ ├── TestAlterTableCompactionLevelThreshold.scala │ │ │ │ ├── TestAlterTableSortColumnsProperty.scala │ │ │ │ └── TestAlterTableWithColumnMetCacheAndCacheLevelProperty.scala │ │ │ ├── badrecordloger │ │ │ │ ├── BadRecordActionTest.scala │ │ │ │ ├── BadRecordEmptyDataTest.scala │ │ │ │ └── BadRecordLoggerTest.scala │ │ │ ├── bigdecimal │ │ │ │ ├── TestAvgForBigInt.scala │ │ │ │ ├── TestBigDecimal.scala │ │ │ │ ├── TestNullAndEmptyFields.scala │ │ │ │ └── TestNullAndEmptyFieldsUnsafe.scala │ │ │ ├── blockprune │ │ │ │ ├── BlockPruneQueryTestCase.scala │ │ │ │ └── CarbonCustomBlockDistributionTest.scala │ │ │ ├── booleantype │ │ │ │ ├── BooleanDataTypesBaseTest.scala │ │ │ │ ├── BooleanDataTypesBigFileTest.scala │ │ │ │ ├── BooleanDataTypesFilterTest.scala │ │ │ │ ├── BooleanDataTypesInsertTest.scala │ │ │ │ ├── BooleanDataTypesLoadTest.scala │ │ │ │ ├── BooleanDataTypesParameterTest.scala │ │ │ │ ├── BooleanDataTypesSortTest.scala │ │ │ │ └── compress │ │ │ │ │ └── TestBooleanCompressSuite.scala │ │ │ ├── cleanfiles │ │ │ │ ├── TestCleanFileCommand.scala │ │ │ │ └── TestCleanFilesCommandPartitionTable.scala │ │ │ ├── cloud │ │ │ │ ├── AllDataSourceTestCase.scala │ │ │ │ └── CacheRefreshTestCase.scala │ │ │ ├── compaction │ │ │ │ └── TestHybridCompaction.scala │ │ │ ├── createTable │ │ │ │ ├── TestAlterTableWithTableComment.scala │ │ │ │ ├── TestCarbonFileInputFormatWithExternalCarbonTable.scala │ │ │ │ ├── TestCreateDDLForComplexMapType.scala │ │ │ │ ├── TestCreateExternalTable.scala │ │ │ │ ├── TestCreateHiveTableWithCarbonDS.scala │ │ │ │ ├── TestCreateTableAsSelect.scala │ │ │ │ ├── TestCreateTableIfNotExists.scala │ │ │ │ ├── TestCreateTableLike.scala │ │ │ │ ├── TestCreateTablePath.scala │ │ │ │ ├── TestCreateTableWithBlockletSize.scala │ │ │ │ ├── TestCreateTableWithColumnComment.scala │ │ │ │ ├── TestCreateTableWithColumnMetCacheAndCacheLevelProperty.scala │ │ │ │ ├── TestCreateTableWithCompactionOptions.scala │ │ │ │ ├── TestCreateTableWithDatabaseNameCaseChange.scala │ │ │ │ ├── TestCreateTableWithPageSizeInMb.scala │ │ │ │ ├── TestCreateTableWithSortScope.scala │ │ │ │ ├── TestCreateTableWithSpaceInColumnName.scala │ │ │ │ ├── TestCreateTableWithTableComment.scala │ │ │ │ ├── TestNonTransactionalCarbonTable.scala │ │ │ │ ├── TestNonTransactionalCarbonTableForBinary.scala │ │ │ │ ├── TestNonTransactionalCarbonTableForMapType.scala │ │ │ │ ├── TestNonTransactionalCarbonTableJsonWriter.scala │ │ │ │ ├── TestNonTransactionalCarbonTableWithAvroDataType.scala │ │ │ │ ├── TestNonTransactionalCarbonTableWithComplexType.scala │ │ │ │ └── TestRenameTableWithIndex.scala │ │ │ ├── datacompaction │ │ │ │ ├── CarbonIndexFileMergeTestCase.scala │ │ │ │ ├── CompactionSupportGlobalSortBigFileTest.scala │ │ │ │ ├── CompactionSupportGlobalSortFunctionTest.scala │ │ │ │ ├── CompactionSupportGlobalSortParameterTest.scala │ │ │ │ ├── CompactionSupportSpecifiedSegmentsTest.scala │ │ │ │ ├── DataCompactionBlockletBoundryTest.scala │ │ │ │ ├── DataCompactionBoundaryConditionsTest.scala │ │ │ │ ├── DataCompactionCardinalityBoundryTest.scala │ │ │ │ ├── DataCompactionLockTest.scala │ │ │ │ ├── MajorCompactionIgnoreInMinorTest.scala │ │ │ │ ├── MajorCompactionStopsAfterCompaction.scala │ │ │ │ ├── MajorCompactionWithMeasureSortColumns.scala │ │ │ │ └── TableLevelCompactionOptionTest.scala │ │ │ ├── dataload │ │ │ │ ├── TestDataLoadPartitionCoalescer.scala │ │ │ │ ├── TestDataLoadWithColumnsMoreThanSchema.scala │ │ │ │ ├── TestDataLoadWithFileName.scala │ │ │ │ ├── TestDataLoadWithOverWrite.scala │ │ │ │ ├── TestGlobalSortDataLoad.scala │ │ │ │ ├── TestLoadDataFrame.scala │ │ │ │ ├── TestLoadDataUseAllDictionary.scala │ │ │ │ ├── TestLoadDataWithDictionaryExcludeAndInclude.scala │ │ │ │ ├── TestLoadDataWithDiffTimestampFormat.scala │ │ │ │ ├── TestLoadDataWithFileHeaderException.scala │ │ │ │ ├── TestLoadDataWithHiveSyntaxDefaultFormat.scala │ │ │ │ ├── TestLoadDataWithHiveSyntaxUnsafe.scala │ │ │ │ ├── TestLoadDataWithMalformedCarbonCommandException.scala │ │ │ │ ├── TestLoadDataWithNoMeasure.scala │ │ │ │ ├── TestLoadDataWithNotProperInputFile.scala │ │ │ │ ├── TestLoadOptions.scala │ │ │ │ ├── TestLoadTblNameIsKeyword.scala │ │ │ │ ├── TestLoadWithSortTempCompressed.scala │ │ │ │ ├── TestRangeColumnDataLoad.scala │ │ │ │ ├── TestTableLevelBlockSize.scala │ │ │ │ └── TestTableLoadMinSize.scala │ │ │ ├── dataretention │ │ │ │ └── DataRetentionTestCase.scala │ │ │ ├── dblocation │ │ │ │ └── DBLocationCarbonTableTestCase.scala │ │ │ ├── deleteTable │ │ │ │ └── TestDeleteTableNewDDL.scala │ │ │ ├── describeTable │ │ │ │ └── TestDescribeTable.scala │ │ │ ├── detailquery │ │ │ │ ├── AllQueriesSpark2TestCase.scala │ │ │ │ ├── CastColumnTestCase.scala │ │ │ │ ├── ColumnPropertyValidationTestCase.scala │ │ │ │ ├── ExpressionWithNullTestCase.scala │ │ │ │ ├── HighCardinalityDataTypesTestCase.scala │ │ │ │ ├── IntegerDataTypeTestCase.scala │ │ │ │ ├── NoDictionaryColumnTestCase.scala │ │ │ │ ├── RangeFilterAllDataTypesTestCases.scala │ │ │ │ ├── RangeFilterTestCase.scala │ │ │ │ ├── SubqueryWithFilterAndSortTestCase.scala │ │ │ │ └── ValueCompressionDataTypeTestCase.scala │ │ │ ├── directdictionary │ │ │ │ ├── DateDataTypeDirectDictionaryTest.scala │ │ │ │ ├── DateDataTypeDirectDictionaryWithNoDictTestCase.scala │ │ │ │ ├── DateDataTypeDirectDictionaryWithOffHeapSortDisabledTest.scala │ │ │ │ ├── DateDataTypeNullDataTest.scala │ │ │ │ ├── TimestampDataTypeDirectDictionaryTestCase.scala │ │ │ │ ├── TimestampDataTypeDirectDictionaryWithNoDictTestCase.scala │ │ │ │ ├── TimestampDataTypeNullDataTest.scala │ │ │ │ ├── TimestampNoDictionaryColumnCastTestCase.scala │ │ │ │ └── TimestampNoDictionaryColumnTestCase.scala │ │ │ ├── filterexpr │ │ │ │ ├── AllDataTypesTestCaseFilter.scala │ │ │ │ ├── CountStarTestCase.scala │ │ │ │ ├── FilterProcessorTestCase.scala │ │ │ │ ├── GrtLtFilterProcessorTestCase.scala │ │ │ │ ├── IntegerDataTypeTestCase.scala │ │ │ │ ├── NullMeasureValueTestCaseFilter.scala │ │ │ │ ├── TestAndEqualFilterEmptyOperandValue.scala │ │ │ │ ├── TestBetweenFilter.scala │ │ │ │ ├── TestGrtLessFilter.scala │ │ │ │ ├── TestImplicitFilterExpression.scala │ │ │ │ ├── TestInFilter.scala │ │ │ │ ├── TestIsNullFilter.scala │ │ │ │ └── TestNotNullFilter.scala │ │ │ ├── flatfolder │ │ │ │ └── FlatFolderTableLoadingTestCase.scala │ │ │ ├── index │ │ │ │ ├── CGIndexTestCase.scala │ │ │ │ ├── FGIndexTestCase.scala │ │ │ │ ├── IndexWriterSuite.scala │ │ │ │ ├── TestIndexCommand.scala │ │ │ │ └── TestIndexStatus.scala │ │ │ ├── insertQuery │ │ │ │ └── InsertIntoNonCarbonTableTestCase.scala │ │ │ ├── iud │ │ │ │ ├── DeleteCarbonTableTestCase.scala │ │ │ │ ├── HorizontalCompactionTestCase.scala │ │ │ │ ├── MergeIntoCarbonTableTestCase.scala │ │ │ │ ├── TestInsertAndOtherCommandConcurrent.scala │ │ │ │ ├── TestUpdateAndDeleteWithLargeData.scala │ │ │ │ ├── UpdateCarbonTableTestCase.scala │ │ │ │ └── UpdateCarbonTableTestCaseWithBadRecord.scala │ │ │ ├── joinquery │ │ │ │ ├── AllDataTypesTestCaseJoin.scala │ │ │ │ ├── IntegerDataTypeTestCase.scala │ │ │ │ ├── JoinWithoutDictionaryColumn.scala │ │ │ │ └── OrderByLimitTestCase.scala │ │ │ ├── localdictionary │ │ │ │ ├── LocalDictionarySupportAlterTableTest.scala │ │ │ │ ├── LocalDictionarySupportCreateTableTest.scala │ │ │ │ └── LocalDictionarySupportLoadTableTest.scala │ │ │ ├── longstring │ │ │ │ └── VarcharDataTypesBasicTestCase.scala │ │ │ ├── measurenullvalue │ │ │ │ └── NullMeasureValueTestCaseAggregate.scala │ │ │ ├── merge │ │ │ │ └── MergeTestCase.scala │ │ │ ├── nullvalueserialization │ │ │ │ └── TestNullValueSerialization.scala │ │ │ ├── partition │ │ │ │ ├── TestShowPartitions.scala │ │ │ │ └── TestUpdateForPartitionTable.scala │ │ │ ├── sdk │ │ │ │ └── TestSDKWithTransactionalTable.scala │ │ │ ├── segment │ │ │ │ └── ShowSegmentTestCase.scala │ │ │ ├── segmentreading │ │ │ │ ├── TestSegmentReading.scala │ │ │ │ └── TestSegmentReadingForMultiThreading.scala │ │ │ ├── sortcolumns │ │ │ │ ├── TestSortColumns.scala │ │ │ │ └── TestSortColumnsWithUnsafe.scala │ │ │ ├── sortexpr │ │ │ │ ├── AllDataTypesTestCaseSort.scala │ │ │ │ └── IntegerDataTypeTestCase.scala │ │ │ ├── standardpartition │ │ │ │ ├── StandardPartitionBadRecordLoggerTest.scala │ │ │ │ ├── StandardPartitionComplexDataTypeTestCase.scala │ │ │ │ ├── StandardPartitionGlobalSortTestCase.scala │ │ │ │ ├── StandardPartitionTableCleanTestCase.scala │ │ │ │ ├── StandardPartitionTableCompactionTestCase.scala │ │ │ │ ├── StandardPartitionTableDropTestCase.scala │ │ │ │ ├── StandardPartitionTableLoadingTestCase.scala │ │ │ │ ├── StandardPartitionTableOverwriteTestCase.scala │ │ │ │ └── StandardPartitionTableQueryTestCase.scala │ │ │ └── windowsexpr │ │ │ │ └── WindowsExprTestCase.scala │ │ └── util │ │ │ ├── BadRecordUtil.scala │ │ │ └── DataTypeConverterUtilSuite.scala │ ├── sql │ │ └── commands │ │ │ ├── StoredAsCarbondataSuite.scala │ │ │ ├── TestCarbonDropCacheCommand.scala │ │ │ ├── TestCarbonShowCacheCommand.scala │ │ │ └── UsingCarbondataSuite.scala │ ├── store │ │ └── SparkCarbonStoreTest.scala │ └── view │ │ ├── MVTest.scala │ │ ├── plans │ │ ├── ExtractJoinConditionsSuite.scala │ │ ├── IsSPJGHSuite.scala │ │ ├── LogicalToModularPlanSuite.scala │ │ ├── ModularToSQLSuite.scala │ │ └── SignatureSuite.scala │ │ ├── rewrite │ │ ├── MVCoalesceTestCase.scala │ │ ├── MVCountAndCaseTestCase.scala │ │ ├── MVCreateTestCase.scala │ │ ├── MVExceptionTestCase.scala │ │ ├── MVFilterAndJoinTest.scala │ │ ├── MVIncrementalLoadingTestcase.scala │ │ ├── MVInvalidTestCase.scala │ │ ├── MVMultiJoinTestCase.scala │ │ ├── MVRewriteTestCase.scala │ │ ├── MVSampleTestCase.scala │ │ ├── MVTPCDSTestCase.scala │ │ ├── MVTpchTestCase.scala │ │ ├── SelectAllColumnsSuite.scala │ │ ├── TestAllOperationsOnMV.scala │ │ ├── TestPartitionWithMV.scala │ │ ├── TestSQLSuite.scala │ │ ├── Tpcds_1_4_Suite.scala │ │ └── matching │ │ │ ├── TestSQLBatch.scala │ │ │ └── TestTPCDS_1_4_Batch.scala │ │ ├── testutil │ │ ├── ModularPlanTest.scala │ │ ├── TestSQLBatch.scala │ │ ├── TestSQLBatch2.scala │ │ ├── Tpcds_1_4_QueryBatch.scala │ │ └── Tpcds_1_4_Tables.scala │ │ └── timeseries │ │ ├── TestCreateMVWithTimeSeries.scala │ │ ├── TestMVTimeSeriesLoadAndQuery.scala │ │ └── TestMVTimeSeriesQueryRollUp.scala │ ├── indexserver │ ├── DistributedRDDUtilsTest.scala │ └── IndexServerTest.scala │ └── spark │ ├── SparkCommandSuite.scala │ ├── carbondata │ ├── BadRecordPathLoadOptionTest.scala │ ├── CarbonDataSourceSuite.scala │ ├── DataLoadFailAllTypeSortTest.scala │ ├── TableStatusBackupTest.scala │ ├── TestStreamingTableOpName.scala │ ├── TestStreamingTableQueryFilter.scala │ ├── TestStreamingTableWithLongString.scala │ ├── TestStreamingTableWithRowParser.scala │ ├── bucketing │ │ └── TableBucketingTestCase.scala │ ├── commands │ │ └── SetCommandTestCase.scala │ ├── datatype │ │ └── NumericDimensionBadRecordTest.scala │ ├── deletetable │ │ └── DeleteTableTestCase.scala │ ├── iud │ │ └── DeleteCarbonTableSubqueryTestCase.scala │ ├── query │ │ ├── ReusedExchangeTestSuite.scala │ │ ├── SubQueryJoinTestSuite.scala │ │ ├── SubQueryTestSuite.scala │ │ └── TestNotEqualToFilter.scala │ ├── register │ │ └── TestRegisterCarbonTable.scala │ ├── restructure │ │ ├── AlterTableRevertTestCase.scala │ │ ├── AlterTableUpgradeSegmentTest.scala │ │ ├── AlterTableValidationTestCase.scala │ │ └── vectorreader │ │ │ ├── AddColumnTestCases.scala │ │ │ ├── AlterTableColumnRenameTestCase.scala │ │ │ ├── ChangeDataTypeTestCases.scala │ │ │ └── DropColumnTestCases.scala │ └── vectorreader │ │ └── VectorReaderTestCase.scala │ ├── sql │ ├── CarbonExtensionSuite.scala │ ├── DynamicPartitionPruningTestCase.scala │ ├── GetDataSizeAndIndexSizeTest.scala │ ├── carbondata │ │ └── datasource │ │ │ ├── SparkCarbonDataSourceBinaryTest.scala │ │ │ ├── SparkCarbonDataSourceTest.scala │ │ │ └── TestCreateTableUsingSparkCarbonFileFormat.scala │ ├── common │ │ └── util │ │ │ └── Tags.scala │ ├── execution │ │ └── command │ │ │ ├── CarbonTableSchemaCommonSuite.scala │ │ │ └── mutation │ │ │ └── CarbonTruncateCommandTest.scala │ └── profiler │ │ └── ProfilerSuite.scala │ └── util │ ├── CarbonCommandSuite.scala │ ├── SparkUtil4Test.scala │ └── SparkUtilTest.scala ├── licenses-binary ├── LICENSE-paranamer.txt └── LICENSE-zstd-jni.txt ├── mv └── plan │ ├── pom.xml │ └── src │ └── main │ ├── common2.3and2.4 │ └── org │ │ └── apache │ │ └── carbondata │ │ └── mv │ │ └── plans │ │ └── modular │ │ └── SparkVersionHelper.scala │ ├── scala │ └── org │ │ └── apache │ │ └── carbondata │ │ └── mv │ │ ├── dsl │ │ └── package.scala │ │ ├── expressions │ │ └── modular │ │ │ └── subquery.scala │ │ └── plans │ │ ├── modular │ │ ├── AggregatePushDown.scala │ │ ├── Flags.scala │ │ ├── Harmonizer.scala │ │ ├── ModularPatterns.scala │ │ ├── ModularPlan.scala │ │ ├── ModularPlanSignatureGenerator.scala │ │ ├── ModularRelation.scala │ │ ├── Modularizer.scala │ │ ├── basicOperators.scala │ │ └── queryGraph.scala │ │ ├── package.scala │ │ └── util │ │ ├── BirdcageOptimizer.scala │ │ ├── Logical2ModularExtractions.scala │ │ ├── LogicalPlanSignatureGenerator.scala │ │ ├── Printers.scala │ │ ├── SQLBuild.scala │ │ ├── SQLBuildDSL.scala │ │ ├── SQLBuilder.scala │ │ ├── Signature.scala │ │ └── TableCluster.scala │ ├── spark2.3 │ └── org │ │ └── apache │ │ └── carbondata │ │ └── mv │ │ └── plans │ │ └── modular │ │ └── ExpressionHelper.scala │ ├── spark2.4 │ └── org │ │ └── apache │ │ └── carbondata │ │ └── mv │ │ └── plans │ │ └── modular │ │ └── ExpressionHelper.scala │ └── spark3.1 │ └── org │ └── apache │ └── carbondata │ └── mv │ └── plans │ └── modular │ ├── ExpressionHelper.scala │ └── SparkVersionHelper.scala ├── pom.xml ├── processing ├── CARBON_PROCESSINGLogResource.properties ├── pom.xml └── src │ ├── main │ └── java │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── processing │ │ ├── datatypes │ │ ├── ArrayDataType.java │ │ ├── GenericDataType.java │ │ ├── PrimitiveDataType.java │ │ └── StructDataType.java │ │ ├── exception │ │ ├── DataLoadingException.java │ │ ├── MultipleMatchingException.java │ │ └── SliceMergerException.java │ │ ├── index │ │ ├── IndexWriterException.java │ │ └── IndexWriterListener.java │ │ ├── loading │ │ ├── AbstractDataLoadProcessorStep.java │ │ ├── BadRecordsLogger.java │ │ ├── BadRecordsLoggerProvider.java │ │ ├── CarbonDataLoadConfiguration.java │ │ ├── ComplexDelimitersEnum.java │ │ ├── DataField.java │ │ ├── DataLoadExecutor.java │ │ ├── DataLoadProcessBuilder.java │ │ ├── FailureCauses.java │ │ ├── TableProcessingOperations.java │ │ ├── complexobjects │ │ │ ├── ArrayObject.java │ │ │ └── StructObject.java │ │ ├── constants │ │ │ └── DataLoadProcessorConstants.java │ │ ├── converter │ │ │ ├── BadRecordLogHolder.java │ │ │ ├── FieldConverter.java │ │ │ ├── RowConverter.java │ │ │ └── impl │ │ │ │ ├── BinaryFieldConverterImpl.java │ │ │ │ ├── ComplexFieldConverterImpl.java │ │ │ │ ├── DirectDictionaryFieldConverterImpl.java │ │ │ │ ├── FieldEncoderFactory.java │ │ │ │ ├── MeasureFieldConverterImpl.java │ │ │ │ ├── NonDictionaryFieldConverterImpl.java │ │ │ │ ├── RowConverterImpl.java │ │ │ │ ├── SpatialIndexFieldConverterImpl.java │ │ │ │ └── binary │ │ │ │ ├── Base64BinaryDecoder.java │ │ │ │ ├── BinaryDecoder.java │ │ │ │ ├── DefaultBinaryDecoder.java │ │ │ │ └── HexBinaryDecoder.java │ │ ├── csvinput │ │ │ ├── BlockDetails.java │ │ │ ├── BoundedInputStream.java │ │ │ ├── CSVInputFormat.java │ │ │ ├── CSVRecordReaderIterator.java │ │ │ └── StringArrayWritable.java │ │ ├── dictionary │ │ │ └── DirectDictionary.java │ │ ├── events │ │ │ └── LoadEvents.java │ │ ├── exception │ │ │ ├── BadRecordFoundException.java │ │ │ ├── CarbonDataLoadingException.java │ │ │ └── NoRetryException.java │ │ ├── iterator │ │ │ └── CarbonOutputIteratorWrapper.java │ │ ├── jsoninput │ │ │ ├── JsonInputFormat.java │ │ │ └── JsonStreamReader.java │ │ ├── model │ │ │ ├── CarbonDataLoadSchema.java │ │ │ ├── CarbonLoadModel.java │ │ │ ├── CarbonLoadModelBuilder.java │ │ │ └── LoadOption.java │ │ ├── parser │ │ │ ├── CarbonParserFactory.java │ │ │ ├── ComplexParser.java │ │ │ ├── GenericParser.java │ │ │ ├── RowParser.java │ │ │ └── impl │ │ │ │ ├── ArrayParserImpl.java │ │ │ │ ├── JsonRowParser.java │ │ │ │ ├── MapParserImpl.java │ │ │ │ ├── PrimitiveParserImpl.java │ │ │ │ ├── RangeColumnParserImpl.java │ │ │ │ ├── RowParserImpl.java │ │ │ │ └── StructParserImpl.java │ │ ├── partition │ │ │ ├── Partitioner.java │ │ │ └── impl │ │ │ │ ├── HashPartitionerImpl.java │ │ │ │ ├── RangePartitionerImpl.java │ │ │ │ ├── RawRowComparator.java │ │ │ │ └── SparkHashExpressionPartitionerImpl.java │ │ ├── row │ │ │ ├── CarbonRowBatch.java │ │ │ └── IntermediateSortTempRow.java │ │ ├── sort │ │ │ ├── AbstractMergeSorter.java │ │ │ ├── CarbonPriorityQueue.java │ │ │ ├── SortStepRowHandler.java │ │ │ ├── Sorter.java │ │ │ ├── SorterFactory.java │ │ │ ├── impl │ │ │ │ ├── ParallelReadMergeSorterImpl.java │ │ │ │ ├── ParallelReadMergeSorterWithColumnRangeImpl.java │ │ │ │ ├── ThreadStatusObserver.java │ │ │ │ ├── UnsafeParallelReadMergeSorterImpl.java │ │ │ │ └── UnsafeParallelReadMergeSorterWithColumnRangeImpl.java │ │ │ └── unsafe │ │ │ │ ├── UnsafeCarbonRowPage.java │ │ │ │ ├── UnsafeSortDataRows.java │ │ │ │ ├── comparator │ │ │ │ ├── UnsafeRowComparator.java │ │ │ │ └── UnsafeRowComparatorForNormalDims.java │ │ │ │ ├── holder │ │ │ │ ├── SortTempChunkHolder.java │ │ │ │ ├── UnsafeCarbonRow.java │ │ │ │ ├── UnsafeCarbonRowForMerge.java │ │ │ │ ├── UnsafeFinalMergePageHolder.java │ │ │ │ ├── UnsafeInmemoryHolder.java │ │ │ │ ├── UnsafeInmemoryMergeHolder.java │ │ │ │ └── UnsafeSortTempFileChunkHolder.java │ │ │ │ ├── merger │ │ │ │ ├── UnsafeInMemoryIntermediateDataMerger.java │ │ │ │ ├── UnsafeIntermediateFileMerger.java │ │ │ │ ├── UnsafeIntermediateMerger.java │ │ │ │ └── UnsafeSingleThreadFinalSortFilesMerger.java │ │ │ │ └── sort │ │ │ │ ├── SortDataFormat.java │ │ │ │ ├── TimSort.java │ │ │ │ └── UnsafeIntSortDataFormat.java │ │ └── steps │ │ │ ├── CarbonRowDataWriterProcessorStepImpl.java │ │ │ ├── DataConverterProcessorStepImpl.java │ │ │ ├── DataWriterProcessorStepImpl.java │ │ │ ├── InputProcessorStepImpl.java │ │ │ ├── InputProcessorStepWithNoConverterImpl.java │ │ │ ├── JsonInputProcessorStepImpl.java │ │ │ └── SortProcessorStepImpl.java │ │ ├── merger │ │ ├── AbstractResultProcessor.java │ │ ├── CarbonCompactionExecutor.java │ │ ├── CarbonCompactionUtil.java │ │ ├── CarbonDataMergerUtil.java │ │ ├── CarbonDataMergerUtilResult.java │ │ ├── CompactionResultSortProcessor.java │ │ ├── CompactionType.java │ │ ├── NodeBlockRelation.java │ │ ├── NodeMultiBlockRelation.java │ │ └── RowResultMergerProcessor.java │ │ ├── sort │ │ ├── DummyRowUpdater.java │ │ ├── SchemaBasedRowUpdater.java │ │ ├── SortTempRowUpdater.java │ │ ├── exception │ │ │ └── CarbonSortKeyAndGroupByException.java │ │ └── sortdata │ │ │ ├── FileMergeSortComparator.java │ │ │ ├── InMemorySortTempChunkHolder.java │ │ │ ├── IntermediateFileMerger.java │ │ │ ├── IntermediateSortTempRowComparator.java │ │ │ ├── NewRowComparator.java │ │ │ ├── NewRowComparatorForNormalDims.java │ │ │ ├── SingleThreadFinalSortFilesMerger.java │ │ │ ├── SortDataRows.java │ │ │ ├── SortIntermediateFileMerger.java │ │ │ ├── SortObserver.java │ │ │ ├── SortParameters.java │ │ │ ├── SortTempFileChunkHolder.java │ │ │ └── TableFieldStat.java │ │ ├── store │ │ ├── CarbonDataFileAttributes.java │ │ ├── CarbonDataWriterFactory.java │ │ ├── CarbonFactDataHandlerColumnar.java │ │ ├── CarbonFactDataHandlerModel.java │ │ ├── CarbonFactHandler.java │ │ ├── CarbonFactHandlerFactory.java │ │ ├── TablePage.java │ │ ├── messages │ │ │ └── messages_en_US.properties │ │ └── writer │ │ │ ├── AbstractFactDataWriter.java │ │ │ ├── CarbonFactDataWriter.java │ │ │ └── v3 │ │ │ ├── BlockletDataHolder.java │ │ │ └── CarbonFactDataWriterImplV3.java │ │ └── util │ │ ├── Auditor.java │ │ ├── CarbonBadRecordUtil.java │ │ ├── CarbonDataProcessorUtil.java │ │ ├── CarbonLoaderUtil.java │ │ ├── CarbonQueryUtil.java │ │ └── TableOptionConstant.java │ └── test │ ├── java │ └── org │ │ └── apache │ │ └── carbondata │ │ ├── core │ │ └── keygenerator │ │ │ └── directdictionary │ │ │ └── timestamp │ │ │ └── TimeStampDirectDictionaryGeneratorTest.java │ │ ├── lcm │ │ └── locks │ │ │ ├── LocalFileLockTest.java │ │ │ └── ZooKeeperLockingTest.java │ │ └── processing │ │ ├── loading │ │ ├── csvinput │ │ │ └── CSVInputFormatTest.java │ │ └── partition │ │ │ └── impl │ │ │ └── RawRowComparatorTest.java │ │ ├── sort │ │ └── sortdata │ │ │ ├── FileMergeSortComparatorTest.java │ │ │ ├── IntermediateSortTempRowComparatorTest.java │ │ │ └── NewRowComparatorTest.java │ │ └── util │ │ └── CarbonLoaderUtilTest.java │ └── resources │ ├── DATA_FACT_SMALL.csv │ ├── csv │ ├── csv_with_bom.csv │ ├── csv_with_bom.csv.bz2 │ ├── csv_with_bom.csv.gz │ ├── data.csv │ ├── data.csv.bz2 │ ├── data.csv.gz │ ├── data.csv.lz4 │ └── data.csv.snappy │ ├── input │ ├── 100.csv │ ├── 2col.csv │ ├── 3col.csv │ └── 5col.csv │ └── schemas │ ├── default │ └── carbon │ │ ├── loadmetadata.metadata │ │ ├── meta.lock │ │ └── metadata │ └── modifiedTime.mdt ├── python ├── README.md ├── __init__.py ├── pycarbon │ ├── __init__.py │ ├── core │ │ ├── Constants.py │ │ ├── __init__.py │ │ ├── carbon.py │ │ ├── carbon_arrow_reader_worker.py │ │ ├── carbon_dataset_metadata.py │ │ ├── carbon_fs_utils.py │ │ ├── carbon_local_memory_cache.py │ │ ├── carbon_py_dict_reader_worker.py │ │ ├── carbon_reader.py │ │ ├── carbon_tf_utils.py │ │ └── carbon_utils.py │ ├── integration │ │ ├── __init__.py │ │ ├── pytorch.py │ │ └── tensorflow.py │ ├── reader.py │ ├── sdk │ │ ├── ArrowCarbonReader.py │ │ ├── CarbonReader.py │ │ ├── CarbonSchemaReader.py │ │ ├── CarbonWriter.py │ │ ├── Configuration.py │ │ ├── Constants.py │ │ ├── PaginationCarbonReader.py │ │ ├── SDKUtil.py │ │ └── __init__.py │ └── tests │ │ ├── .coveragerc │ │ ├── __init__.py │ │ ├── conftest.py │ │ ├── core │ │ ├── __init__.py │ │ ├── test_carbon.py │ │ ├── test_carbon_common.py │ │ ├── test_carbon_end_to_end.py │ │ ├── test_carbon_fs_utils.py │ │ ├── test_carbon_memory_cache.py │ │ ├── test_carbon_predicates.py │ │ ├── test_carbon_reader.py │ │ ├── test_carbon_tf_dataset.py │ │ ├── test_carbon_tf_utils.py │ │ └── test_reader.py │ │ ├── hello_world │ │ ├── README.md │ │ ├── __init__.py │ │ ├── dataset_with_normal_schema │ │ │ ├── __init__.py │ │ │ ├── generate_dataset_carbon.py │ │ │ ├── python_hello_world_carbon.py │ │ │ ├── pytorch_hello_world_carbon.py │ │ │ ├── tensorflow_hello_world_carbon.py │ │ │ └── tests │ │ │ │ └── test_generate_dataset_carbon_with_normal_schema.py │ │ └── dataset_with_unischema │ │ │ ├── __init__.py │ │ │ ├── generate_pycarbon_dataset.py │ │ │ ├── pyspark_hello_world_carbon.py │ │ │ ├── python_hello_world_carbon.py │ │ │ ├── pytorch_hello_world_carbon.py │ │ │ ├── tensorflow_hello_world_carbon.py │ │ │ └── tests │ │ │ └── test_generate_dataset.py │ │ ├── im │ │ ├── __init__.py │ │ └── test.py │ │ ├── mnist │ │ ├── README.md │ │ ├── __init__.py │ │ ├── dataset_with_normal_schema │ │ │ ├── README.md │ │ │ ├── __init__.py │ │ │ ├── generate_mnist_carbon.py │ │ │ ├── tf_carbon.py │ │ │ └── tf_external_example_carbon_unified_api.py │ │ └── dataset_with_unischema │ │ │ ├── README.md │ │ │ ├── __init__.py │ │ │ ├── generate_pycarbon_mnist.py │ │ │ ├── pytorch_example_carbon.py │ │ │ ├── pytorch_example_carbon_unified_api.py │ │ │ ├── schema.py │ │ │ ├── tests │ │ │ ├── __init__.py │ │ │ ├── conftest.py │ │ │ └── test_tf_mnist_carbon.py │ │ │ ├── tf_example_carbon.py │ │ │ └── tf_example_carbon_unified_api.py │ │ ├── resources │ │ ├── carbondatalogo.jpg │ │ ├── carbondatalogo2.jpg │ │ ├── flowers │ │ │ ├── 10686568196_b1915544a8.jpg │ │ │ ├── 10686568196_b1915544a8.txt │ │ │ ├── 10712722853_5632165b04.jpg │ │ │ ├── 10712722853_5632165b04.txt │ │ │ └── subfolder │ │ │ │ ├── 10841136265_af473efc60.jpg │ │ │ │ └── 10841136265_af473efc60.txt │ │ ├── voc │ │ │ ├── 2007_000027.jpg │ │ │ ├── 2007_000027.xml │ │ │ ├── 2007_000032.jpg │ │ │ ├── 2007_000032.xml │ │ │ ├── 2007_000033.jpg │ │ │ ├── 2007_000033.xml │ │ │ ├── 2007_000039.jpg │ │ │ ├── 2007_000039.xml │ │ │ ├── 2009_001444.jpg │ │ │ └── 2009_001444.xml │ │ └── vocForSegmentationClass │ │ │ ├── 2007_000032.jpg │ │ │ ├── 2007_000032.png │ │ │ ├── 2007_000033.jpg │ │ │ ├── 2007_000033.png │ │ │ ├── 2007_000042.jpg │ │ │ └── 2007_000042.png │ │ ├── sdk │ │ ├── __init__.py │ │ └── test_read_write_carbon.py │ │ └── test.py ├── setup.cfg └── setup.py ├── scalastyle-config.xml ├── sdk ├── CSDK │ ├── CMakeLists.txt │ ├── src │ │ ├── CarbonProperties.cpp │ │ ├── CarbonProperties.h │ │ ├── CarbonReader.cpp │ │ ├── CarbonReader.h │ │ ├── CarbonRow.cpp │ │ ├── CarbonRow.h │ │ ├── CarbonSchemaReader.cpp │ │ ├── CarbonSchemaReader.h │ │ ├── CarbonWriter.cpp │ │ ├── CarbonWriter.h │ │ ├── Configuration.cpp │ │ ├── Configuration.h │ │ ├── Schema.cpp │ │ └── Schema.h │ └── test │ │ └── main.cpp └── sdk │ ├── pom.xml │ └── src │ ├── main │ ├── java │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ ├── sdk │ │ │ └── file │ │ │ │ ├── ArrowCarbonReader.java │ │ │ │ ├── AvroCarbonWriter.java │ │ │ │ ├── CSVCarbonWriter.java │ │ │ │ ├── CarbonIUD.java │ │ │ │ ├── CarbonReader.java │ │ │ │ ├── CarbonReaderBuilder.java │ │ │ │ ├── CarbonSchemaReader.java │ │ │ │ ├── CarbonWriter.java │ │ │ │ ├── CarbonWriterBuilder.java │ │ │ │ ├── JsonCarbonWriter.java │ │ │ │ ├── ORCCarbonWriter.java │ │ │ │ ├── PaginationCarbonReader.java │ │ │ │ ├── ParquetCarbonWriter.java │ │ │ │ ├── RowUtil.java │ │ │ │ ├── Schema.java │ │ │ │ ├── TestUtil.java │ │ │ │ ├── arrow │ │ │ │ ├── ArrowConverter.java │ │ │ │ ├── ArrowFieldWriter.java │ │ │ │ ├── ArrowUtils.java │ │ │ │ └── ArrowWriter.java │ │ │ │ ├── cache │ │ │ │ └── BlockletRows.java │ │ │ │ └── utils │ │ │ │ └── SDKUtil.java │ │ │ └── store │ │ │ ├── CarbonRowReadSupport.java │ │ │ ├── CarbonStore.java │ │ │ ├── LocalCarbonStore.java │ │ │ └── MetaCachedCarbonStore.java │ └── resources │ │ └── log4j.properties │ └── test │ ├── java │ └── org │ │ └── apache │ │ └── carbondata │ │ ├── sdk │ │ └── file │ │ │ ├── ArrowCarbonReaderTest.java │ │ │ ├── AvroCarbonWriterTest.java │ │ │ ├── CSVCarbonWriterTest.java │ │ │ ├── CarbonIUDTest.java │ │ │ ├── CarbonReaderTest.java │ │ │ ├── CarbonSchemaReaderTest.java │ │ │ ├── ConcurrentAvroSdkWriterTest.java │ │ │ ├── ConcurrentSdkReaderTest.java │ │ │ ├── ConcurrentSdkWriterTest.java │ │ │ ├── ImageTest.java │ │ │ ├── JSONCarbonWriterTest.java │ │ │ ├── MinMaxTest.java │ │ │ ├── MultithreadSDKBlockletReaderTest.java │ │ │ ├── ORCCarbonWriterTest.java │ │ │ ├── PaginationCarbonReaderTest.java │ │ │ └── ParquetCarbonWriterTest.java │ │ ├── store │ │ └── LocalCarbonStoreTest.java │ │ └── util │ │ └── BinaryUtil.java │ └── resources │ ├── file │ ├── NestedMap.parquet │ ├── avro_files │ │ ├── users.avro │ │ ├── users_2.avro │ │ └── users_3.avro │ ├── csv_files │ │ ├── primitive_data.csv │ │ ├── primitive_data_2.csv │ │ └── primitive_data_3.csv │ ├── json_files │ │ ├── allPrimitiveType.json │ │ ├── allPrimitiveTypeMultipleRows.json │ │ └── allPrimitiveTypeSingleArray.json │ ├── nested_schema.avro │ ├── orc_files │ │ ├── sample.orc │ │ ├── sample_2.orc │ │ └── sample_3.orc │ ├── parquet_files │ │ ├── file1.parquet │ │ ├── file2.parquet │ │ └── file3.parquet │ ├── repeated-schema.parquet │ ├── testTimestamp.orc │ ├── userdata1.avro │ ├── userdata1.parquet │ ├── userdata1_orc │ └── weather.avro │ └── image │ ├── carbondatalogo.jpg │ ├── flowers │ ├── 10686568196_b1915544a8.jpg │ ├── 10686568196_b1915544a8.txt │ ├── 10712722853_5632165b04.jpg │ ├── 10712722853_5632165b04.txt │ └── subfolder │ │ ├── 10841136265_af473efc60.jpg │ │ └── 10841136265_af473efc60.txt │ ├── voc │ ├── 2007_000027.jpg │ ├── 2007_000027.xml │ ├── 2007_000032.jpg │ ├── 2007_000032.xml │ ├── 2007_000033.jpg │ ├── 2007_000033.xml │ ├── 2007_000039.jpg │ ├── 2007_000039.xml │ ├── 2009_001444.jpg │ └── 2009_001444.xml │ └── vocForSegmentationClass │ ├── 2007_000032.jpg │ ├── 2007_000032.png │ ├── 2007_000033.jpg │ ├── 2007_000033.png │ ├── 2007_000042.jpg │ └── 2007_000042.png ├── streaming ├── pom.xml └── src │ ├── main │ ├── java │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ └── streaming │ │ │ ├── CarbonStreamException.java │ │ │ ├── CarbonStreamOutputFormat.java │ │ │ ├── CarbonStreamRecordWriter.java │ │ │ ├── StreamBlockletWriter.java │ │ │ ├── index │ │ │ └── StreamFileIndex.java │ │ │ ├── parser │ │ │ ├── CSVStreamParserImp.java │ │ │ └── CarbonStreamParser.java │ │ │ └── segment │ │ │ └── StreamSegment.java │ ├── scala │ │ └── org │ │ │ └── apache │ │ │ └── carbondata │ │ │ └── streaming │ │ │ └── parser │ │ │ ├── FieldConverter.scala │ │ │ └── RowStreamParserImp.scala │ ├── spark2.x │ │ └── org.apache.carbondata.util │ │ │ └── SparkStreamingUtil.scala │ └── spark3.1 │ │ └── org │ │ └── apache │ │ └── carbondata │ │ └── util │ │ └── SparkStreamingUtil.scala │ └── test │ └── java │ └── org │ └── apache │ └── carbondata │ └── streaming │ └── CarbonStreamOutputFormatTest.java └── tools └── cli ├── pom.xml └── src ├── main └── java │ └── org │ └── apache │ └── carbondata │ └── tool │ ├── CarbonCli.java │ ├── Command.java │ ├── DataFile.java │ ├── DataSummary.java │ ├── FileCollector.java │ ├── ScanBenchmark.java │ ├── ShardPrinter.java │ └── TableFormatter.java └── test └── java └── org └── apache └── carbondata └── tool └── CarbonCliTest.java /.DS_Store: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/.DS_Store -------------------------------------------------------------------------------- /.github/PULL_REQUEST_TEMPLATE.md: -------------------------------------------------------------------------------- 1 | ### Why is this PR needed? 2 | 3 | 4 | ### What changes were proposed in this PR? 5 | 6 | 7 | ### Does this PR introduce any user interface change? 8 | - No 9 | - Yes. (please explain the change and update document) 10 | 11 | ### Is any new testcase added? 12 | - No 13 | - Yes 14 | 15 | 16 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | *~ 2 | *.#* 3 | *#*# 4 | *.swp 5 | *.ipr 6 | *.iml 7 | *.iws 8 | *.pyc 9 | *.pyo 10 | .idea/ 11 | .idea_modules/ 12 | .settings 13 | .cache 14 | target/ 15 | store/CSDK/cmake-build-debug/* 16 | .project 17 | .classpath 18 | .DS_Store 19 | metastore_db/ 20 | derby.log 21 | python/.idea/ 22 | */.cache-main 23 | */.cache-tests 24 | */*/.cache-main 25 | */*/.cache-tests 26 | */*/*/.cache-main 27 | */*/*/.cache-tests 28 | *.flattened-pom.xml 29 | python/pycarbon/.pylintrc 30 | -------------------------------------------------------------------------------- /NOTICE: -------------------------------------------------------------------------------- 1 | Apache CarbonData 2 | Copyright 2016 and onwards The Apache Software Foundation. 3 | 4 | This product includes software developed at 5 | The Apache Software Foundation (http://www.apache.org/). 6 | 7 | Based on source code originally developed by 8 | Huawei (http://www.huawei.com/). 9 | -------------------------------------------------------------------------------- /build/docker/carbondata-notebook/Dockerfile: -------------------------------------------------------------------------------- 1 | FROM jupyter/all-spark-notebook:spark-3.1.1 2 | USER root 3 | WORKDIR . 4 | RUN wget https://dlcdn.apache.org/carbondata/2.3.0/apache-carbondata-2.3.0-bin-spark3.1.1-hadoop2.7.2.jar -P /usr/local/spark-3.1.1-bin-hadoop3.2/jars/ 5 | -------------------------------------------------------------------------------- /core/CARBON_CORELogResource.properties: -------------------------------------------------------------------------------- 1 | # 2 | # Licensed to the Apache Software Foundation (ASF) under one 3 | # or more contributor license agreements. See the NOTICE file 4 | # distributed with this work for additional information 5 | # regarding copyright ownership. The ASF licenses this file 6 | # to you under the Apache License, Version 2.0 (the 7 | # "License"); you may not use this file except in compliance 8 | # with the License. You may obtain a copy of the License at 9 | # 10 | # http://www.apache.org/licenses/LICENSE-2.0 11 | # 12 | # Unless required by applicable law or agreed to in writing, software 13 | # distributed under the License is distributed on an "AS IS" BASIS, 14 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 15 | # See the License for the specific language governing permissions and 16 | # limitations under the License. 17 | # 18 | carbon.core = {0} -------------------------------------------------------------------------------- /core/src/main/java/org/apache/carbondata/core/datastore/filesystem/CarbonFileFilter.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.core.datastore.filesystem; 19 | 20 | public interface CarbonFileFilter { 21 | boolean accept(CarbonFile file); 22 | } 23 | -------------------------------------------------------------------------------- /core/src/main/java/org/apache/carbondata/core/fileoperations/FileWriteOperation.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.core.fileoperations; 19 | 20 | public enum FileWriteOperation { 21 | 22 | APPEND, OVERWRITE 23 | } 24 | -------------------------------------------------------------------------------- /core/src/main/java/org/apache/carbondata/core/index/status/IndexStatus.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.core.index.status; 19 | 20 | /** 21 | * Index status 22 | */ 23 | public enum IndexStatus { 24 | ENABLED, DISABLED, DROPPED 25 | } 26 | -------------------------------------------------------------------------------- /core/src/main/java/org/apache/carbondata/core/memory/MemoryType.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.core.memory; 19 | 20 | public enum MemoryType { 21 | OFFHEAP, ONHEAP 22 | } 23 | -------------------------------------------------------------------------------- /core/src/main/java/org/apache/carbondata/core/scan/expression/LeafExpression.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.core.scan.expression; 19 | 20 | public abstract class LeafExpression extends Expression { 21 | 22 | } 23 | -------------------------------------------------------------------------------- /core/src/main/java/org/apache/carbondata/core/scan/filter/executer/FilterBitSetUpdater.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.core.scan.filter.executer; 19 | 20 | import java.util.BitSet; 21 | 22 | public interface FilterBitSetUpdater { 23 | void updateBitset(BitSet bitSet, int bitIndex); 24 | } 25 | 26 | -------------------------------------------------------------------------------- /core/src/main/java/org/apache/carbondata/core/scan/filter/intf/FilterOptimizer.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.core.scan.filter.intf; 19 | 20 | import org.apache.carbondata.core.scan.expression.Expression; 21 | 22 | public interface FilterOptimizer { 23 | Expression optimizeFilter(); 24 | } 25 | -------------------------------------------------------------------------------- /core/src/main/java/org/apache/carbondata/core/view/MVStatus.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.core.view; 19 | 20 | /** 21 | * MV status 22 | */ 23 | public enum MVStatus { 24 | ENABLED, DISABLED, DROPPED 25 | } 26 | -------------------------------------------------------------------------------- /core/src/test/java/org/apache/carbondata/core/cache/dictionary/AbstractDictionaryCacheTest.java: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/core/src/test/java/org/apache/carbondata/core/cache/dictionary/AbstractDictionaryCacheTest.java -------------------------------------------------------------------------------- /core/src/test/java/org/apache/carbondata/core/writer/CarbonDictionaryWriterImplTest.java: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/core/src/test/java/org/apache/carbondata/core/writer/CarbonDictionaryWriterImplTest.java -------------------------------------------------------------------------------- /core/src/test/resources/carbonTest.properties: -------------------------------------------------------------------------------- 1 | # 2 | # Licensed to the Apache Software Foundation (ASF) under one 3 | # or more contributor license agreements. See the NOTICE file 4 | # distributed with this work for additional information 5 | # regarding copyright ownership. The ASF licenses this file 6 | # to you under the Apache License, Version 2.0 (the 7 | # "License"); you may not use this file except in compliance 8 | # with the License. You may obtain a copy of the License at 9 | # 10 | # http://www.apache.org/licenses/LICENSE-2.0 11 | # 12 | # Unless required by applicable law or agreed to in writing, software 13 | # distributed under the License is distributed on an "AS IS" BASIS, 14 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 15 | # See the License for the specific language governing permissions and 16 | # limitations under the License. 17 | # 18 | database=testSchema 19 | tableName=carbon 20 | storePath=carbonStore 21 | -------------------------------------------------------------------------------- /dev/java.header: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | -------------------------------------------------------------------------------- /docs/images/2-1_1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-1_1.png -------------------------------------------------------------------------------- /docs/images/2-1_1_latest.PNG: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-1_1_latest.PNG -------------------------------------------------------------------------------- /docs/images/2-2_1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-2_1.png -------------------------------------------------------------------------------- /docs/images/2-3_1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-3_1.png -------------------------------------------------------------------------------- /docs/images/2-3_2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-3_2.png -------------------------------------------------------------------------------- /docs/images/2-3_3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-3_3.png -------------------------------------------------------------------------------- /docs/images/2-3_4.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-3_4.png -------------------------------------------------------------------------------- /docs/images/2-4_1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-4_1.png -------------------------------------------------------------------------------- /docs/images/2-5_1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-5_1.png -------------------------------------------------------------------------------- /docs/images/2-5_2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-5_2.png -------------------------------------------------------------------------------- /docs/images/2-5_3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-5_3.png -------------------------------------------------------------------------------- /docs/images/2-6_1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/2-6_1.png -------------------------------------------------------------------------------- /docs/images/CarbonData_logo.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/CarbonData_logo.png -------------------------------------------------------------------------------- /docs/images/QRCode_WechatGroup.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/QRCode_WechatGroup.png -------------------------------------------------------------------------------- /docs/images/carbon_data_file_structure_new.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/carbon_data_file_structure_new.png -------------------------------------------------------------------------------- /docs/images/carbon_data_format_new.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/carbon_data_format_new.png -------------------------------------------------------------------------------- /docs/images/carbondata-performance.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/carbondata-performance.png -------------------------------------------------------------------------------- /docs/images/carbondata-streamer-tool-pipeline.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/carbondata-streamer-tool-pipeline.png -------------------------------------------------------------------------------- /docs/images/codegen.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/codegen.png -------------------------------------------------------------------------------- /docs/images/spatial-index-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/spatial-index-1.png -------------------------------------------------------------------------------- /docs/images/spatial-index-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/spatial-index-2.png -------------------------------------------------------------------------------- /docs/images/spatial-index-polygonlist.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/spatial-index-polygonlist.png -------------------------------------------------------------------------------- /docs/images/spatial-index-polylinelist.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/spatial-index-polylinelist.png -------------------------------------------------------------------------------- /docs/images/spatial-index-rangelist.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/spatial-index-rangelist.png -------------------------------------------------------------------------------- /docs/images/using-carbondata-in-notebook-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/using-carbondata-in-notebook-1.png -------------------------------------------------------------------------------- /docs/images/using-carbondata-in-notebook-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/using-carbondata-in-notebook-2.png -------------------------------------------------------------------------------- /docs/images/using-carbondata-in-notebook-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/using-carbondata-in-notebook-3.png -------------------------------------------------------------------------------- /docs/images/using-carbondata-in-notebook-visualization-0.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/using-carbondata-in-notebook-visualization-0.png -------------------------------------------------------------------------------- /docs/images/using-carbondata-in-notebook-visualization-1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/using-carbondata-in-notebook-visualization-1.png -------------------------------------------------------------------------------- /docs/images/using-carbondata-in-notebook-visualization-2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/using-carbondata-in-notebook-visualization-2.png -------------------------------------------------------------------------------- /docs/images/using-carbondata-in-notebook-visualization-3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/images/using-carbondata-in-notebook-visualization-3.png -------------------------------------------------------------------------------- /docs/zh_cn/images/SortColumns.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/docs/zh_cn/images/SortColumns.png -------------------------------------------------------------------------------- /examples/spark/src/main/resources/data.csv: -------------------------------------------------------------------------------- 1 | shortField,intField,bigintField,doubleField,stringField,timestampField,decimalField,dateField,charField,floatField,complexData 2 | 1,10,1100,48.4,spark,2015/4/23 12:01:01,1.23,2015/4/23,aaa,2.5,'foo'#'bar'#'world' 3 | 5,17,1140,43.4,spark,2015/7/27 12:01:02,3.45,2015/7/27,bbb,2.5,'foo'#'bar'#'world' 4 | 1,11,1100,44.4,flink,2015/5/23 12:01:03,23.23,2015/5/23,ccc,2.5,'foo'#'bar'#'world' 5 | 1,10,1150,43.4,spark,2015/7/24 12:01:04,254.12,2015/7/24,ddd,2.5,'foo'#'bar'#'world' 6 | 1,10,1100,47.4,spark,2015/7/23 12:01:05,876.14,2015/7/23,eeee,3.5,'foo'#'bar'#'world' 7 | 3,14,1160,43.4,hive,2015/7/26 12:01:06,3454.32,2015/7/26,ff,2.5,'foo'#'bar'#'world' 8 | 2,10,1100,43.4,impala,2015/7/23 12:01:07,456.98,2015/7/23,ggg,2.5,'foo'#'bar'#'world' 9 | 1,10,1100,43.4,spark,2015/5/23 12:01:08,32.53,2015/5/23,hhh,2.5,'foo'#'bar'#'world' 10 | 4,16,1130,42.4,impala,2015/7/23 12:01:09,67.23,2015/7/23,iii,2.5,'foo'#'bar'#'world' 11 | 1,10,1100,43.4,spark,2015/7/23 12:01:10,832.23,2015/7/23,jjj,2.5,'foo'#'bar'#'world' 12 | -------------------------------------------------------------------------------- /examples/spark/src/main/resources/data1.csv: -------------------------------------------------------------------------------- 1 | shortField,intField,bigintField,doubleField,stringField,timestampField,decimalField,dateField,charField,floatField 2 | 1,10,1100,48.4,spark,2015-4-23 12:01:01,1.23,2015-4-23,aaa,2.5 3 | 5,17,1140,43.4,spark,2015-7-27 12:01:02,3.45,2015-7-27,bbb,2.5 4 | 1,11,1100,44.4,flink,2015-5-23 12:01:03,23.23,2015-5-23,ccc,2.5 5 | 1,10,1150,43.4,spark,2015-7-24 12:01:04,254.12,2015-7-24,ddd,2.5 6 | 1,10,1100,47.4,spark,2015-7-23 12:01:05,876.14,2015-7-23,eeee,3.5 7 | 3,14,1160,43.4,hive,2015-7-26 12:01:06,3454.32,2015-7-26,ff,2.5 8 | 2,10,1100,43.4,impala,2015-7-23 12:01:07,456.98,2015-7-23,ggg,2.5 9 | 1,10,1100,43.4,spark,2015-5-23 12:01:08,32.53,2015-5-23,hhh,2.5 10 | 4,16,1130,42.4,impala,2015-7-23 12:01:09,67.23,2015-7-23,iii,2.5 11 | 1,10,1100,43.4,spark,2015-7-23 12:01:10,832.23,2015-7-23,jjj,2.5 12 | -------------------------------------------------------------------------------- /examples/spark/src/main/resources/dataSample.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary,floatField 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,15000,2.34 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892,15001,2.34 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,15002,2.34 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,15003,2.34 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,15004,2.34 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961,15005,3.5 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875,15006,2.34 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,15007,2.34 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008,2.34 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009,2.34 -------------------------------------------------------------------------------- /examples/spark/src/main/resources/dimSample.csv: -------------------------------------------------------------------------------- 1 | id,name,city 2 | 1,David,Beijing 3 | 2,Mark,Paris 4 | 3,Bill,NewYork 5 | 4,Sara,Tokyo 6 | 5,John,Beijing 7 | 6,Michel,Chicago 8 | 7,Robert,Houston 9 | 8,Sunny,Boston 10 | 9,Mary,Tokyo 11 | 10,Edward,Paris 12 | 11,James,Washington 13 | 12,Maria,Berlin 14 | 13,Adam,Athens 15 | 14,Peter,Boston 16 | 15,George,Paris 17 | 16,Paul,Shanghai 18 | 17,Lisa,Hangzhou 19 | 18,Angel,Beijing 20 | 19,Emily,Bangalore 21 | 20,Kevin,Singapore -------------------------------------------------------------------------------- /examples/spark/src/main/resources/log4j.properties: -------------------------------------------------------------------------------- 1 | log4j.logger.org.apache.spark.sql.profiler.ProfilerLogger$=INFO,F1 2 | log4j.appender.F1=org.apache.log4j.RollingFileAppender 3 | log4j.appender.F1.File=${path.target}/profiler.log 4 | log4j.appender.F1.MaxFileSize=4024KB 5 | log4j.appender.F1.MaxBackupIndex=20 6 | log4j.appender.F1.layout=org.apache.log4j.PatternLayout 7 | log4j.appender.F1.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss,SSS} %5p %c - %m%n -------------------------------------------------------------------------------- /examples/spark/src/main/resources/sample.csv: -------------------------------------------------------------------------------- 1 | ID,NAME,SALARY 2 | 1,'liang',200000 3 | 2,'anubhav',20000 -------------------------------------------------------------------------------- /examples/spark/src/main/resources/streamSample.csv: -------------------------------------------------------------------------------- 1 | id,name,city,salary,file 2 | 100000001,batch_1,city_1,0.1,school_1:school_11$20 3 | 100000002,batch_2,city_2,0.2,school_2:school_22$30 4 | 100000003,batch_3,city_3,0.3,school_3:school_33$40 5 | 100000004,batch_4,city_4,0.4,school_4:school_44$50 6 | 100000005,batch_5,city_5,0.5,school_5:school_55$60 7 | -------------------------------------------------------------------------------- /integration/flink-proxy/src/main/resources/META-INF/services/org.apache.flink.core.fs.FileSystemFactory: -------------------------------------------------------------------------------- 1 | org.apache.carbon.flink.ProxyFileSystemFactory -------------------------------------------------------------------------------- /integration/flink/src/main/resources/META-INF/services/org.apache.carbon.flink.CarbonWriterFactoryBuilder: -------------------------------------------------------------------------------- 1 | org.apache.carbon.flink.CarbonLocalWriterFactoryBuilder 2 | org.apache.carbon.flink.CarbonS3WriterFactoryBuilder -------------------------------------------------------------------------------- /integration/hive/src/main/java/org/apache/carbondata/hive/CarbonFileHiveSerDe.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.hive; 19 | 20 | public class CarbonFileHiveSerDe extends CarbonHiveSerDe { 21 | } 22 | -------------------------------------------------------------------------------- /integration/hive/src/main/resources/array/complexArray.csv: -------------------------------------------------------------------------------- 1 | abc$def$ghijkl,1,1$2,123456$2555$9999999,1.1$2.2,1.2323$2.3$9.98,1.2323$2.3$9.89,hello$world$china,true$false$true,k$a,122$123$124,2014-01-11$2014-02-20 2 | abc$pqrst,1$2,1$2,123456$2555,1.1$2.2,2.2929$2.3,2.2929$2.3$6.789,hello$world,true$false,k$a,122$123$124,2014-01-11$2014-02-20 3 | abc$def,1$2$3,1$2,123456$26262,1.1$2.2,2.2929$2.39,2.2929$2.39,hello$world,true$false,k$a,122$123$124,2014-01-11$2014-02-20 4 | abc$def,1$2$3$4,1$2,56$1555,1.1$2.2,1.2$2.999,1.2$2.999,hello$world,true$false,k$a,122$123$124,2014-01-11$2014-02-20 5 | abc$def,1$2$3$4$5,1$2,123456$2555,1.1$2.2,1.2$2.3$0.09,1.2$2.3,hello$world,true$false,k$a,122$123$124,2014-01-11$2014-02-20 6 | -------------------------------------------------------------------------------- /integration/hive/src/main/resources/complex/complex.csv: -------------------------------------------------------------------------------- 1 | k$a,Key1@Val1$Key2@Val2,bangalore$560066 -------------------------------------------------------------------------------- /integration/hive/src/main/resources/csv/data.csv: -------------------------------------------------------------------------------- 1 | 1,10,1100,48.4,spark,2015-04-23 12:01:01,1.23,2015-04-23,aaa,2.5 2 | 5,17,1140,43.4,spark,2015-07-27 12:01:02,3.45,2015-07-27,bbb,2.5 3 | 1,11,1100,44.4,flink,2015-05-23 12:01:03,23.23,2015-05-23,ccc,2.5 4 | 1,10,1150,43.4,spark,2015-07-24 12:01:04,254.12,2015-07-24,ddd,2.5 5 | 1,10,1100,47.4,spark,2015-07-23 12:01:05,876.14,2015-07-23,eeee,3.5 6 | 3,14,1160,43.4,hive,2015-07-26 12:01:06,3454.32,2015-07-26,ff,2.5 7 | 2,10,1100,43.4,impala,2015-07-23 12:01:07,456.98,2015-07-23,ggg,2.5 8 | 1,10,1100,43.4,spark,2015-05-23 12:01:08,32.53,2015-05-23,hhh,2.5 9 | 4,16,1130,42.4,impala,2015-07-23 12:01:09,67.23,2015-07-23,iii,2.5 10 | 1,10,1100,43.4,spark,2015-07-23 12:01:10,832.23,2015-07-23,jjj,2.5 11 | -------------------------------------------------------------------------------- /integration/hive/src/main/resources/log4j.properties: -------------------------------------------------------------------------------- 1 | # Root logger option 2 | log4j.rootLogger=INFO,stdout 3 | 4 | 5 | # Redirect log messages to console 6 | log4j.appender.debug=org.apache.log4j.RollingFileAppender 7 | log4j.appender.stdout=org.apache.log4j.ConsoleAppender 8 | log4j.appender.stdout.Target=System.out 9 | log4j.appender.stdout.layout=org.apache.log4j.PatternLayout 10 | log4j.appender.stdout.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss} %-5p %c{1}:%L - %m%n 11 | -------------------------------------------------------------------------------- /integration/hive/src/main/resources/map/complexMap.csv: -------------------------------------------------------------------------------- 1 | Key1@Val1$Key2@Val2,1@key1$2@key2,1.12@1.34$2.12@2.34,1@2$3@4,1.23@varchar1$2.34@varchar2,1234@2014-01-10$124@2014-01-10$1@2014-01-10,hello@1$world@2$china@3,1234567@true$123456@false$123456789@true 2 | Key21@Va2l1$Key22@Va2l2,1@key1$2@key2,2.12@2.99$2.12@2.89,1@2$3@4,1.99@varchar21$2.99@varchar22,1234@2014-01-10$124@2014-01-10$1@2014-01-10,hello@1$world@2$china@3,1234567@true$123456@false$123456789@true 3 | -------------------------------------------------------------------------------- /integration/hive/src/main/resources/struct/struct.csv: -------------------------------------------------------------------------------- 1 | Egypt$123$560066$123456789$1.23$2.34567$char$true$varchar1$123$2014-02-20$1.234 2 | China$123$560066$123996789$1.23$2.34567$char$true$varchar2$123$2014-02-20$1.234 3 | India$123$561066$123456789$1.23$2.34567$char$true$varchar3$123$2014-02-20$1.234 4 | Egypt$123$5066$123456989$1.23$2.34567$char$true$varchar4$123$2014-02-20$1.234 5 | Sri Lanka$123$56006$123456789$1.23$2.34567$char$true$varchar5$123$2014-02-20$1.234 6 | -------------------------------------------------------------------------------- /integration/hive/src/main/resources/text/string.txt: -------------------------------------------------------------------------------- 1 | abc 1 2 | test 2 3 | test 3 4 |  8 5 |  9 6 | testtest 4 7 | testtest 5 8 | testtest 6 9 | -------------------------------------------------------------------------------- /integration/presto/src/main/resources/log4j.properties: -------------------------------------------------------------------------------- 1 | # Root logger option 2 | log4j.rootLogger=INFO,stdout 3 | 4 | 5 | # Redirect log messages to console 6 | log4j.appender.debug=org.apache.log4j.RollingFileAppender 7 | log4j.appender.stdout=org.apache.log4j.ConsoleAppender 8 | log4j.appender.stdout.Target=System.out 9 | log4j.appender.stdout.layout=org.apache.log4j.PatternLayout 10 | log4j.appender.stdout.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss} %-5p %c{1}:%L - %m%n 11 | 12 | -------------------------------------------------------------------------------- /integration/presto/src/test/resources/alldatatype.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary,bonus,monthlyBonus,dob,shortfield,isCurrentEmployee 2 | 1,2015-07-23,china,anubhav,phone197,ASD69643,5000000.00,1234.444,12.1234,2016-04-14 15/00/09,10,true 3 | 2,2015-07-24,china,jatin,phone756,ASD42892,150010.999,1234.5555,15.13,2016-04-14 15:00:09,10,null 4 | 3,2015-07-25,china,liang,phone1904,ASD37014,15002.110,600.777,16.181,2016-01-14 15:07:09,8,true 5 | 4,2015-07-26,china,prince,phone2435,ASD66902,15003.00,9999.999,17.3654,1992-04-14 13:00:09,4,true 6 | 5,2015-07-27,china,bhavya,phone2441,ASD90633,15004.00,5000.999,12.11,2010-06-19 14:10:06,11,true 7 | 6,2015-07-28,china,akash,phone294,ASD59961,15005.00,500.59,18.65,2013-07-19 12:10:08,18,false 8 | 7,2015-07-29,china,sahil,phone610,ASD14875,15006.00,500.99,,19.65,2007-04-19 11:10:06,17,false 9 | 8,2015-07-30,china,geetika,phone1848,ASD57308,15007.500,500.88,200.97,2008-09-21 11:10:06,10,true 10 | 9,2015-07-18,china,ravindra,phone706,ASD86717,15008.00,700.999,45.25,2009-06-19 15:10:06,1,true 11 | 9,2015/07/18,china,jitesh,phone706,ASD86717,15008.00,500.414,11.655,2001-08-29 13:09:03,12,true 12 | 13 | -------------------------------------------------------------------------------- /integration/presto/src/test/resources/log4j.properties: -------------------------------------------------------------------------------- 1 | # Root logger option 2 | log4j.rootLogger=INFO,stdout 3 | 4 | 5 | # Redirect log messages to console 6 | log4j.appender.debug=org.apache.log4j.RollingFileAppender 7 | log4j.appender.stdout=org.apache.log4j.ConsoleAppender 8 | log4j.appender.stdout.Target=System.out 9 | log4j.appender.stdout.layout=org.apache.log4j.PatternLayout 10 | log4j.appender.stdout.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss} %-5p %c{1}:%L - %m%n 11 | 12 | -------------------------------------------------------------------------------- /integration/spark-common-cluster-test/src/test/resources/hdfs-site.xml: -------------------------------------------------------------------------------- 1 | 2 | 3 | 16 | 17 | 18 | 19 | 20 | 21 | dfs.client.block.write.replace-datanode-on-failure.policy 22 | NEVER 23 | 24 | 25 | -------------------------------------------------------------------------------- /integration/spark-common-cluster-test/src/test/scala/org/apache/spark/sql/common/util/Tags.scala: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | package org.apache.spark.sql.common.util 18 | 19 | import org.scalatest.Tag 20 | 21 | object Include extends Tag("Include") 22 | 23 | object Exclude extends Tag("Exclude") 24 | -------------------------------------------------------------------------------- /integration/spark/src/main/common2.3and2.4/org/apache/spark/sql/execution/CarbonCodegenSupport.scala: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | package org.apache.spark.sql.execution 18 | 19 | import org.apache.spark.sql.execution.joins.HashJoin 20 | 21 | trait CarbonCodegenSupport extends SparkPlan with HashJoin { 22 | 23 | } 24 | -------------------------------------------------------------------------------- /integration/spark/src/main/scala/org/apache/spark/sql/SQLConf.scala: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | package org.apache.spark.sql 18 | 19 | class SQLConf extends org.apache.spark.sql.internal.SQLConf { 20 | val CASE_SENSITIVE = true 21 | 22 | val STARSCHEMA_DETECTION = true 23 | } 24 | -------------------------------------------------------------------------------- /integration/spark/src/resources/META-INF/services/org.apache.spark.sql.test.TestQueryExecutorRegister: -------------------------------------------------------------------------------- 1 | ## ------------------------------------------------------------------------ 2 | ## Licensed to the Apache Software Foundation (ASF) under one or more 3 | ## contributor license agreements. See the NOTICE file distributed with 4 | ## this work for additional information regarding copyright ownership. 5 | ## The ASF licenses this file to You under the Apache License, Version 2.0 6 | ## (the "License"); you may not use this file except in compliance with 7 | ## the License. You may obtain a copy of the License at 8 | ## 9 | ## http://www.apache.org/licenses/LICENSE-2.0 10 | ## 11 | ## Unless required by applicable law or agreed to in writing, software 12 | ## distributed under the License is distributed on an "AS IS" BASIS, 13 | ## WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | ## See the License for the specific language governing permissions and 15 | ## limitations under the License. 16 | ## ------------------------------------------------------------------------ 17 | org.apache.spark.sql.test.SparkTestQueryExecutor -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/bad_record.csv: -------------------------------------------------------------------------------- 1 | item,name 2 | 2,Apple -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/badrecord.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | ravi,2,kiran,huawei 3 | manohar,4,vanam,huawei -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/comp1.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,1,aa,aaa 3 | b,2,bb,bbb 4 | c,3,cc,ccc 5 | d,4,dd,ddd 6 | e,5,ee,eee 7 | f,6,ff,fff 8 | g,7,gg,ggg 9 | h,8,hh,hhh 10 | i,9,ii,iii 11 | j,10,jj,jjj 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/comp2.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,11,aa,aaa 3 | b,12,bb,bbb 4 | c,13,cc,ccc 5 | d,14,dd,ddd 6 | e,15,ee,eee 7 | f,16,ff,fff 8 | g,17,gg,ggg 9 | h,18,hh,hhh 10 | i,19,ii,iii 11 | j,20,jj,jjj 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/comp3.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,21,aa,aaa 3 | b,22,bb,bbb 4 | c,23,cc,ccc 5 | d,24,dd,ddd 6 | e,25,ee,eee 7 | f,26,ff,fff 8 | g,27,gg,ggg 9 | h,28,hh,hhh 10 | i,29,ii,iii 11 | j,30,jj,jjj 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/comp4.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,31,aa,aaa 3 | b,32,bb,bbb 4 | c,33,cc,ccc 5 | d,34,dd,ddd 6 | e,35,ee,eee 7 | f,36,ff,fff 8 | g,37,gg,ggg 9 | h,38,hh,hhh 10 | i,39,ii,iii 11 | j,40,jj,jjj 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/dest.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,1,aa,aaa 3 | b,2,bb,bbb 4 | c,3,cc,ccc 5 | d,4,dd,ddd 6 | e,5,ee,eee -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/negativevalue.csv: -------------------------------------------------------------------------------- 1 | -30000,aaa,-300 2 | 0,ddd,0 3 | -20000,bbb,-200 4 | 70000,ggg,700 5 | 10000,eee,100, 6 | -10000,ccc,-100, 7 | null,null,null -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/other.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,1,MGM,Disco 3 | b,2,RGK,Music 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/sample.csv: -------------------------------------------------------------------------------- 1 | ID,country,name,phonetype,serialname,salary 2 | 4,china,ravz,jio,ASD66902,15003 3 | 1,india,ravi,airtel,ASD90633,15004 4 | 6,usa,manu,alkatel,ASD59961,15005 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/sample_updated.csv: -------------------------------------------------------------------------------- 1 | part0/segment_0/part-0-0-1475753917000.carbondata/0/0,4,china,ravz,Aircel,ASD66902,200000 2 | part0/segment_0/part-0-0-1475753917000.carbondata/0/2,6,usa,manu,Aircel,ASD59961,15005 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/source2.csv: -------------------------------------------------------------------------------- 1 | c11,c22,c33,c55,c66 2 | a,1,MGM,Disco,10 3 | b,2,RGK,Music,8 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/source3.csv: -------------------------------------------------------------------------------- 1 | c11,c22,c33,c55,c66 2 | a,1,MGM,Disco,10 3 | b,2,RGK,Music,8 4 | d,4,YDY,Weather,9 5 | e,5,ZAZ,Election,11 6 | g,7,YTY,Hello,12 7 | h,8,TBT,Yeh,13 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/update01.csv: -------------------------------------------------------------------------------- 1 | imei,age,task,num,level,name 2 | imei0,2147,9279,100.05,100.055,fegt 3 | imei1,-2148,-9807,10.05,100.05,lrhkr 4 | imei2,2147,9279,100.05,100.055,dfegt 5 | imei3,-217,-9206,100.005,100.05,lrhkr 6 | imei4,10,0,15.5,45,Lily -------------------------------------------------------------------------------- /integration/spark/src/test/resources/IUD/updateinpartition.csv: -------------------------------------------------------------------------------- 1 | id,sales,dtm 2 | 001,0,20200907 3 | 002,0,20200907 4 | 003,0,20200907 5 | 004,0,20200907 6 | 005,0,20200907 7 | 006,0,20200907 8 | 007,0,20200907 9 | 008,0,20200907 10 | 009,0,20200907 11 | 010,0,20200907 12 | 011,0,20200908 13 | 012,0,20200908 14 | 013,0,20200908 15 | 014,0,20200908 16 | 015,0,20200908 17 | 016,0,20200908 18 | 017,0,20200908 19 | 018,0,20200908 20 | 019,0,20200908 21 | 020,0,20200908 22 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/Struct.csv: -------------------------------------------------------------------------------- 1 | 1,11$abc$10.00 2 | 2,12$abcd$10.01 3 | 3,13$abce$10.02 4 | 4,14$abcr$10.03 5 | 5,15$abct$10.04 6 | 6,16$abcn$10.05 7 | 7,17$abcq$10.06 8 | 8,18$abcs$10.07 9 | 9,19$abcm$10.08 10 | 10,20$abck$10.09 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/StructofStruct.csv: -------------------------------------------------------------------------------- 1 | 1,11&abc&10.00 2 | 2,12&abcd&10.01 3 | 3,13&abce&10.02 4 | 4,14&abcr&10.03 5 | 5,15&abct&10.04 6 | 6,16&abcn&10.05 7 | 7,17&abcq&10.06 8 | 8,18&abcs&10.07 9 | 9,19&abcm&10.08 10 | 10,20&abck&10.09 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/adap.csv: -------------------------------------------------------------------------------- 1 | 1,500$abc$20:30:40 2 | 2,600$abc$20:30:40 3 | 3,600$abc$20:30:40 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/adap_double1.csv: -------------------------------------------------------------------------------- 1 | 1,1.323$abc$2.2:3.3:4.4 2 | 2,1.323$abc$2.2:3.3:4.4 3 | 3,1.323$abc$2.2:3.3:4.4 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/adap_double2.csv: -------------------------------------------------------------------------------- 1 | 1,1.323$abc$20.2:30.3:40.4 2 | 2,2.323$abc$20.2:30.3:40.4 3 | 3,4.323$abc$20.2:30.3:40.4 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/adap_double3.csv: -------------------------------------------------------------------------------- 1 | 1,1.323$abc$20.2:30.3:500.423 2 | 2,2.323$abc$20.2:30.3:500.423 3 | 3,50.323$abc$20.2:30.3:500.423 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/adap_double4.csv: -------------------------------------------------------------------------------- 1 | 1,1.323$abc$20.2:30.3:50000.423 2 | 2,2.323$abc$20.2:30.3:50000.423 3 | 3,50000.323$abc$20.2:30.3:50000.423 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/adap_int1.csv: -------------------------------------------------------------------------------- 1 | 1,500$abc$200:300:400 2 | 2,700$abc$200:300:400 3 | 3,800$abc$200:300:400 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/adap_int2.csv: -------------------------------------------------------------------------------- 1 | 1,50000$abc$2000000:3000000:4000000 2 | 2,70000$abc$2000000:3000000:4000000 3 | 3,100000$abc$2000000:3000000:4000000 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/adap_int3.csv: -------------------------------------------------------------------------------- 1 | 1,500000$abc$200:300:52000000 2 | 2,7000000$abc$200:300:52000000 3 | 3,10000000$abc$200:300:52000000 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/alldatatypeforpartition.csv: -------------------------------------------------------------------------------- 1 | smallIntField,intField,bigIntField,floatField,doubleField,decimalField,timestampField,dateField,stringField,varcharField,charField,arrayField,structField 2 | -32768,-2147483648,-9223372036854775808,-2147483648.1,-9223372036854775808.1,-9223372036854775808.1234,2017-06-11 00:00:01,2017-06-11,abc1,abcd1,abcde1,a$b$c$1,a$b$1 3 | 128,32768,2147483648,2147483647.1,9223372036854775807.1,9223372036854775807.1234,2017-06-12 23:59:59,2017-06-12,abc2,abcd2,abcde2,a$b$c$2,a$b$2 4 | 32767,2147483647,9223372036854775807,2147483648.1,9223372036854775808.1,9223372036854775808.1234,2017-06-13 23:59:59,2017-06-13,abc3,abcd3,abcde3,a$b$c$3,a$b$3 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/alldictionary/complex/20160423/1400_1405/complex.dictionary: -------------------------------------------------------------------------------- 1 | 0,100080 2 | 1,2355 3 | 2,1ROM size 4 | 3,29-11-2015 5 | 4,1AA100080$2BB100080 6 | 5,MAC283$MAC284$MAC285 7 | 6,1:Chinese:Guangdong Province:shenzhen:longgang:matishan$1:India:Guangdong Province:shenzhen:longgang:matishan 8 | 7,02-03-2016$02-03-2016:02-03-2016 9 | 8,2355 10 | 9,954 11 | 0,100081 12 | 1,1650 13 | 2,6ROM size 14 | 3,29-11-2015 15 | 4,1AA100081$2BB100081 16 | 5,MAC286$MAC287$MAC288 17 | 6,1:Chinese:Guangdong Province:shenzhen:longgang:matishan$1:India:Guangdong Province:shenzhen:longgang:matishan 18 | 7,03-03-2016$03-03-2016:03-03-2016 19 | 8,1650 20 | 9,613 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/alldictionary/sample/20160423/1400_1405/sample.dictionary: -------------------------------------------------------------------------------- 1 | 0,1 2 | 1,david 3 | 2,shenzhen 4 | 0,2 5 | 1,eason 6 | 0,3 7 | 1,jarry 8 | 2,wuhan 9 | 2,Bangalore -------------------------------------------------------------------------------- /integration/spark/src/test/resources/array1.csv: -------------------------------------------------------------------------------- 1 | 1,hello$bye,12345$5678,123456$3456,12.0$13.0,2017-07-09 12:00:00$2016-07-09 13:00:00,123456789$987654321 2 | 2,welcome$hi,9876$1234,13456$356,15.0$18.0,2019-07-09 12:00:00$2015-07-09 13:00:00,8888888888$99999999999 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/arrayColumnEmpty.csv: -------------------------------------------------------------------------------- 1 | imei,age,productdate,gamePointId,reserved6,mobile 2 | ime100004,14,2016-05-02 05:19:24,95,,April/100 3 | ime100003,10,2016-05-01 13:09:20,12,,Apr-98 4 | ime100003,13,2016-05-01 16:08:51,67,,Mar-91 5 | ime100004,13,2016-05-01 19:11:53,67,,Apr-72 6 | ime100005,10,2016-05-02 02:17:36,12,,May-91 7 | ime100004,13,2016-05-01 16:37:11,67,,Mar-47 8 | ime100005,11,2016-05-01 06:27:23,34.5,,Apr-40 9 | ime100002,10,2016-05-02 05:19:24,12,,Apr-37 10 | ime100005,10,2016-05-01 14:56:12,12,,May-32 11 | ime100003,10,2016-05-01 16:39:25,12,,20-Mar 12 | ime100002,10,2016-05-01 23:00:29,12,,May-36 13 | ime100002,13,2016-05-01 12:42:14,67,,16-Jul 14 | ime100005,11,2016-05-01 20:12:10,34.5,,Jun-61 15 | ime100005,13,2016-05-02 06:06:05,67,,May-99 16 | ime100005,10,2016-05-01 23:58:34,12,,Mar-55 17 | ime100005,12,2016-05-01 09:46:31,34.6,,Apr-86 18 | ime100005,12,2016-05-01 13:27:11,34.6,,Jun-34 19 | ime100003,14,2016-05-01 05:36:34,95,,Jul-89 20 | ime100001,14,2016-05-01 08:13:27,95,,Jun-47 21 | ime100001,14,2016-05-01 13:40:14,95,,Mar-75 22 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/avgTest.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,count,phonetype,serialname,salary 2 | 1,2015-7-23 00:00:00,china,1234,phone197,ASD69643,15000 3 | 2,2015-7-24 00:00:00,india,1233,phone756,ASD42892,15001 4 | 3,2015-7-25 00:00:00,usa,1000,phone1904,ASD37014,15002 5 | 4,2015-7-26 00:00:00,china,2222,phone2435,ASD66902,15003 6 | 5,2015-7-27 00:00:00,china,1300,phone2441,ASD90633,15004 7 | 6,2015-7-28 00:00:00,usa,2100,phone294,ASD59961,15005 8 | 7,2015-7-29 00:00:00,china,1500,phone610,ASD14875,15006 9 | 8,2015-7-30 00:00:00,china,1600,phone1848,ASD57308,15007 10 | 9,2015-7-18 00:00:00,china,1000,phone706,ASD86717,15008 11 | 10,2015-7-19 00:00:00,usa,2100,phone685,ASD30505,15009 12 | 11,2015-7-18 00:00:00,china,2200,phone1554,ASD26101,15010 13 | 12,2015-7-19 00:00:00,india,1300,phone1781,ASD85711,15011 14 | 13,2015-7-20 00:00:00,china,2000,phone943,ASD39200,15012 15 | 14,2015-7-21 00:00:00,india,1500,phone1954,ASD80468,15013 16 | 15,2015-7-22 00:00:00,china,6200,phone451,ASD1954,15014 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/badrecords/complexdata.csv: -------------------------------------------------------------------------------- 1 | arrayColumn,structColumn,arrayStruct 2 | 1997-03-20 14:00:09,1$1997-03-20 14:00:09,1#1997-03-20 14:00:09 3 | 1997-03-32 14:00:09,2$1997-03-20 14:00:10,2#1997-03-20 14:00:10 4 | 1997-03-33 14:00:09,3$1997-03-20 14:00:11,3#1997-03-20 14:00:11 5 | 1997-03-31 14:00:09,4$1997-03-20 14:00:12,4#1997-03-20 14:00:12 6 | 1997-03-20 14:00:09,a$1997-03-20 14:00:13,5#1997-03-20 14:00:13 7 | 1997-03-21 14:00:09,b$1997-03-20 14:00:14,6#1997-03-20 14:00:14 8 | 1997-03-22 14:00:09,5$1997-03-20 14:00:15,5#1997-03-20 14:00:15 9 | 1997-03-23 14:00:09,6$1997-03-20 14:00:16,6#1997-03-20 14:00:16 10 | 1997-03-24 14:00:09,7$1997-03-20 14:00:17,7#1997-03-20 14:00:17 11 | 1997-03-25 14:00:09,8$1997-03-20 14:00:18,8#1997-03-50 14:00:18 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/badrecords/datasample.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,actual_price,Quantity,sold_price 2 | 10000,2015/7/23,china,120000.45,3,140000.377 3 | 10001,null,china,120000.45,3,140000.377 4 | 10003,2015/7/23,null,120000.45,3,140000.377 5 | 10004,2015/7/23,china,120000.45ghf,3,140000.377 6 | 10005,2015/7/23,china,120000.45,3ghf,140000.377 7 | 10006,2015/7/23,china,120000.45,3,140000.377ghf 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/badrecords/dummy.csv: -------------------------------------------------------------------------------- 1 | name,dob,weight 2 | \N,\N,1 3 | ,,xfds 4 | "","","" -------------------------------------------------------------------------------- /integration/spark/src/test/resources/badrecords/dummy2.csv: -------------------------------------------------------------------------------- 1 | name,dob,weight 2 | "","","" -------------------------------------------------------------------------------- /integration/spark/src/test/resources/badrecords/emptyTimeStampValue.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,actual_price,Quantity,sold_price 2 | \N,,\N,\N,\N,\N 3 | 10003,2015/7/23,xyz,120003.45,3,140000.377 4 | 10003gh,2015/7/23,xyz,120003.45,3,140000.377 5 | 10003,20/7/2016,xyz,120003.45,3,140000.377 6 | 10003,2015/7/23,xyz,120003.45gf,3,140000.377 7 | 8 | 10003,2015/7/23,xyz,120003.45,3,140000.377gf -------------------------------------------------------------------------------- /integration/spark/src/test/resources/badrecords/emptyValues.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,actual_price,Quantity,sold_price 2 | 10001,2015/7/23,,120003.45,3,140000.377 3 | ,2015/7/23,,120003.45,3,140000.377 4 | ,,,120003.45,3,140000.377 5 | ,,,,3,140000.377 6 | ,,,,,140000.377 7 | ,,,,, 8 | 10003,2015/7/23,india,120003.45,3,140000.377 9 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/badrecords/insufficientColumns.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,actual_price,Quantity,sold_price 2 | 100, 3 | 10001,2015/7/23, 4 | 10003,2015/7/23,null,120003.45,3,140000.377 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/badrecords/seriazableValue.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,actual_price,Quantity,sold_price 2 | \N,\N,\N,\N,\N,\N 3 | 10003,2015/7/23,null,120003.45,3,140000.377 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bigIntData.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,10000001 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892,10000000 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,10000000 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,10000000 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,10000000 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961,10000000 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875,10000000 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,10000000 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,10000000 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,10000000 12 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,10000000 13 | 12,2015/7/19,china,aaa12,phone1781,ASD85711,10000000 14 | 13,2015/7/19,china,aaa13,phone17851,ASD85721,10000000 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bigIntDataWithHeader.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,150003452628 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892,452345254658824 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,6521652222 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,15003 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,1500043525 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961,2745654656565662345 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875,150064555555 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,150074422222 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008023 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,1500043525 12 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,1501000 13 | 12,2015/7/19,china,aaa12,phone1781,ASD85711,1501120 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bigIntDataWithoutHeader.csv: -------------------------------------------------------------------------------- 1 | 1,2015/7/23,china,aaa1,phone197,ASD69643,150003452628 2 | 2,2015/7/24,china,aaa2,phone756,ASD42892,452345254658824 3 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,6521652222 4 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,15003 5 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,1500043525 6 | 6,2015/7/28,china,aaa6,phone294,ASD59961,2745654656565662345 7 | 7,2015/7/29,china,aaa7,phone610,ASD14875,150064555555 8 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,150074422222 9 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008023 10 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,1500043525 11 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,1501000 12 | 12,2015/7/19,china,aaa12,phone1781,ASD85711,1501120 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/big_decimal_without_header.csv: -------------------------------------------------------------------------------- 1 | 1,32473289848372638424.8218378712 2 | 2,99487323423232324232.2434323233 3 | 3,12773443434389239382.4309238238 4 | 4,38488747823423323726.3589238237 5 | 5,93838663748166353423.4273832762 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/big_int_Decimal.csv: -------------------------------------------------------------------------------- 1 | imei,age,task,name,country,city,sale,num,level,quest,productdate,enddate,pointid,score 2 | imei0,2147,9279,fegt,china,hangzhou,10000,100.05,100.055,10,2016-05-01 12:25:36,2016-05-01 21:14:48,1,1.005 3 | imei1,-2148,-9807,lrhkr,America,NewYork,1000,10.05,100.05,100,2016-05-02 19:25:15,2016-05-02 22:25:46,2,1.05 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/binaryStringNullData.csv: -------------------------------------------------------------------------------- 1 | 2|false|2.png|history|true 2 | 3|false|3.png|biology|false 3 | 3|false|3.png||false 4 | 1|true|1.png|education|true -------------------------------------------------------------------------------- /integration/spark/src/test/resources/binarystringdata.csv: -------------------------------------------------------------------------------- 1 | 2|false|2.png|history|true 2 | 3|false|3.png|biology|false 3 | 1|true|1.png|education|true -------------------------------------------------------------------------------- /integration/spark/src/test/resources/binarystringdata2.csv: -------------------------------------------------------------------------------- 1 | 2|false|2.png|abc|true 2 | 3|false|3.png|binary|false 3 | 1|true|1.png|^Ayard duty^B|true -------------------------------------------------------------------------------- /integration/spark/src/test/resources/binarystringdatawithHead.csv: -------------------------------------------------------------------------------- 1 | id|label|name|autolabel|binaryfield 2 | 2|false|2.png|true|binary 3 | 3|false|3.png|false|1 4 | 1|true|1.png|true|Hello world -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bool/supportBoolean.csv: -------------------------------------------------------------------------------- 1 | 1,true,10,1100,48.4,spark,2015/4/23 12:01:01,1.23,2015/4/23,aaa,2.5,'foo'#'bar'#'world' 2 | 5,false,17,1140,43.4,spark,2015/7/27 12:01:02,3.45,2015/7/27,bbb,2.5,'foo'#'bar'#'world' 3 | 1,false,11,1100,44.4,flink,2015/5/23 12:01:03,23.23,2015/5/23,ccc,2.5,'foo'#'bar'#'world' 4 | 1,true,10,1150,43.4,spark,2015/7/24 12:01:04,254.12,2015/7/24,ddd,2.5,'foo'#'bar'#'world' 5 | 1,true,10,1100,47.4,spark,2015/7/23 12:01:05,876.14,2015/7/23,eeee,3.5,'foo'#'bar'#'world' 6 | 3,true,14,1160,43.4,hive,2015/7/26 12:01:06,3454.32,2015/7/26,ff,2.5,'foo'#'bar'#'world' 7 | 2,false,10,1100,43.4,impala,2015/7/23 12:01:07,456.98,2015/7/23,ggg,2.5,'foo'#'bar'#'world' 8 | 1,false,10,1100,43.4,spark,2015/5/23 12:01:08,32.53,2015/5/23,hhh,2.5,'foo'#'bar'#'world' 9 | 4,false,16,1130,42.4,impala,2015/7/23 12:01:09,67.23,2015/7/23,iii,2.5,'foo'#'bar'#'world' 10 | 1,false,10,1100,43.4,spark,2015/7/23 12:01:10,832.23,2015/7/23,jjj,2.5,'foo'#'bar'#'world' 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bool/supportBooleanBadRecords.csv: -------------------------------------------------------------------------------- 1 | 1,true,10,1100,48.4,spark,2015/4/23 12:01:01,1.23,2015/4/23,aaa,2.5,'foo'#'bar'#'world',true 2 | 5,falsee,17,1140,43.4,spark,2015/7/27 12:01:02,3.45,2015/7/27,bbb,2.5,'foo'#'bar'#'world',true 3 | 1,f,11,1100,44.4,flink,2015/5/23 12:01:03,23.23,2015/5/23,ccc,2.5,'foo'#'bar'#'world',true 4 | 1,truee,10,1150,43.4,spark,2015/7/24 12:01:04,254.12,2015/7/24,ddd,2.5,'foo'#'bar'#'world',true 5 | 1,truea,10,1100,47.4,spark,2015/7/23 12:01:05,876.14,2015/7/23,eeee,3.5,'foo'#'bar'#'world',true 6 | 3,true,14,1160,43.4,hive,2015/7/26 12:01:06,3454.32,2015/7/26,ff,2.5,'foo'#'bar'#'world',falsee 7 | 2,false,10,1100,43.4,impala,2015/7/23 12:01:07,456.98,2015/7/23,ggg,2.5,'foo'#'bar'#'world',falsea 8 | 1,false,10,1100,43.4,spark,2015/5/23 12:01:08,32.53,2015/5/23,hhh,2.5,'foo'#'bar'#'world',falsef 9 | 4,false,16,1130,42.4,impala,2015/7/23 12:01:09,67.23,2015/7/23,iii,2.5,'foo'#'bar'#'world',falsea 10 | 1,false,10,1100,43.4,spark,2015/7/23 12:01:10,832.23,2015/7/23,jjj,2.5,'foo'#'bar'#'world',false -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bool/supportBooleanDifferentFormat.csv: -------------------------------------------------------------------------------- 1 | True 2 | TRUE 3 | true 4 | "true" 5 | false 6 | False 7 | FALSE 8 | "FALSE" 9 | 10 | null 11 | NULL 12 | TRUEA 13 | true" 14 | 'true' 15 | truee 16 | 'false' 17 | falsee 18 | FFALSE 19 | f 20 | t -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bool/supportBooleanOnlyBoolean.csv: -------------------------------------------------------------------------------- 1 | True 2 | TRUE 3 | true 4 | "true" 5 | false 6 | False 7 | FALSE 8 | "FALSE" 9 | 10 | null 11 | NULL -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bool/supportBooleanTwoBooleanColumns.csv: -------------------------------------------------------------------------------- 1 | 1,true,10,1100,48.4,spark,2015/4/23 12:01:01,1.23,2015/4/23,aaa,2.5,'foo'#'bar'#'world',true 2 | 5,false,17,1140,43.4,spark,2015/7/27 12:01:02,3.45,2015/7/27,bbb,2.5,'foo'#'bar'#'world',true 3 | 1,false,11,1100,44.4,flink,2015/5/23 12:01:03,23.23,2015/5/23,ccc,2.5,'foo'#'bar'#'world',true 4 | 1,true,10,1150,43.4,spark,2015/7/24 12:01:04,254.12,2015/7/24,ddd,2.5,'foo'#'bar'#'world',true 5 | 1,true,10,1100,47.4,spark,2015/7/23 12:01:05,876.14,2015/7/23,eeee,3.5,'foo'#'bar'#'world',true 6 | 3,true,14,1160,43.4,hive,2015/7/26 12:01:06,3454.32,2015/7/26,ff,2.5,'foo'#'bar'#'world',false 7 | 2,false,10,1100,43.4,impala,2015/7/23 12:01:07,456.98,2015/7/23,ggg,2.5,'foo'#'bar'#'world',false 8 | 1,false,10,1100,43.4,spark,2015/5/23 12:01:08,32.53,2015/5/23,hhh,2.5,'foo'#'bar'#'world',false 9 | 4,false,16,1130,42.4,impala,2015/7/23 12:01:09,67.23,2015/7/23,iii,2.5,'foo'#'bar'#'world',false 10 | 1,false,10,1100,43.4,spark,2015/7/23 12:01:10,832.23,2015/7/23,jjj,2.5,'foo'#'bar'#'world',false -------------------------------------------------------------------------------- /integration/spark/src/test/resources/bool/supportBooleanWithFileHeader.csv: -------------------------------------------------------------------------------- 1 | shortField,booleanField,intField,bigintField,doubleField,stringField,timestampField,decimalField,dateField,charField,floatField,complexData 2 | 1,true,10,1100,48.4,spark,2015/4/23 12:01:01,1.23,2015/4/23,aaa,2.5,'foo'#'bar'#'world' 3 | 5,false,17,1140,43.4,spark,2015/7/27 12:01:02,3.45,2015/7/27,bbb,2.5,'foo'#'bar'#'world' 4 | 1,false,11,1100,44.4,flink,2015/5/23 12:01:03,23.23,2015/5/23,ccc,2.5,'foo'#'bar'#'world' 5 | 1,true,10,1150,43.4,spark,2015/7/24 12:01:04,254.12,2015/7/24,ddd,2.5,'foo'#'bar'#'world' 6 | 1,true,10,1100,47.4,spark,2015/7/23 12:01:05,876.14,2015/7/23,eeee,3.5,'foo'#'bar'#'world' 7 | 3,true,14,1160,43.4,hive,2015/7/26 12:01:06,3454.32,2015/7/26,ff,2.5,'foo'#'bar'#'world' 8 | 2,false,10,1100,43.4,impala,2015/7/23 12:01:07,456.98,2015/7/23,ggg,2.5,'foo'#'bar'#'world' 9 | 1,false,10,1100,43.4,spark,2015/5/23 12:01:08,32.53,2015/5/23,hhh,2.5,'foo'#'bar'#'world' 10 | 4,false,16,1130,42.4,impala,2015/7/23 12:01:09,67.23,2015/7/23,iii,2.5,'foo'#'bar'#'world' 11 | 1,false,10,1100,43.4,spark,2015/7/23 12:01:10,832.23,2015/7/23,jjj,2.5,'foo'#'bar'#'world' 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/carriage_return_in_string.csv: -------------------------------------------------------------------------------- 1 | 1,2 ,3 2 | 4,5,6 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/channelsId.csv: -------------------------------------------------------------------------------- 1 | 1123 2 | 1234 3 | 1111, 4 | 1431, 5 | 1421 6 | 1421| 7 | 1234, 8 | 1245, 9 | 1231| 10 | 2234| -------------------------------------------------------------------------------- /integration/spark/src/test/resources/columndictionary/country.csv: -------------------------------------------------------------------------------- 1 | usa 2 | china 3 | uk 4 | france 5 | brazil -------------------------------------------------------------------------------- /integration/spark/src/test/resources/columndictionary/name.csv: -------------------------------------------------------------------------------- 1 | aaa1 2 | aaa2 3 | aaa3 4 | aaa4 5 | aaa5 6 | aaa6 7 | aaa7 8 | aaa8 9 | aaa9 10 | aaa10 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/comment.csv: -------------------------------------------------------------------------------- 1 | .~carbon,.,~carbon,~carbon,~carbon,~carbon,~carbon,~carbon,~carbon 2 | ,carbon,,carbon,,carbon,,carbon,,carbon,,carbon,,carbon,,carbon 3 | #?carbon,#carbon,#carbon,#carbon,#carbon,#carbon,#carbon,#carbon 4 | ?carbon,#carbon,#carbon,#carbon,#carbon,#carbon,#carbon,#carbon 5 | ".carbon,"carbon,"carbon,"carbon,"carbon,"carbon,"carbon,"carbon -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/compaction1.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,07/23/2015,china,aaa1,phone197,ASD69643,15000 3 | 2,07/24/2015,chile,aaa2,phone756,ASD42892,15001 4 | 3,07/25/2015,america,aaa3,phone1904,ASD37014,15002 5 | 4,07/26/2015,canada,aaa4,phone2435,ASD66902,15003 6 | 5,07/27/2015,england,aaa5,phone2441,ASD90633,15004 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/compaction1_forhive.csv: -------------------------------------------------------------------------------- 1 | 1,07/23/2015,china,aaa1,phone197,ASD69643,15000 2 | 2,07/24/2015,chile,aaa2,phone756,ASD42892,15001 3 | 3,07/25/2015,america,aaa3,phone1904,ASD37014,15002 4 | 4,07/26/2015,canada,aaa4,phone2435,ASD66902,15003 5 | 5,07/27/2015,england,aaa5,phone2441,ASD90633,15004 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/compaction2.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 6,07/28/2015,newzealand,aaa6,phone294,ASD59961,15005 3 | 7,07/29/2015,mexico,aaa7,phone610,ASD14875,15006 4 | 8,07/30/2015,westindies,aaa8,phone1848,ASD57308,15007 5 | 9,07/18/2015,burma,aaa9,phone706,ASD86717,15008 6 | 10,07/19/2015,butan,aaa10,phone685,ASD30505,1500 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/compaction3.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 11,07/23/2015,china,aaa1,phone197,ASD69643,15005 3 | 12,07/24/2015,india,aaa2,phone756,ASD42892,15006 4 | 13,07/26/2015,iran,aaa4,phone2435,ASD66902,15008 5 | 14,07/27/2015,iraq,aaa5,phone2441,ASD90633,15009 6 | 15,07/28/2015,ireland,aaa6,phone2441,ASD90633,15010 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/compactionIUD1.csv: -------------------------------------------------------------------------------- 1 | FirstName,LastName,date,phonetype,serialname,ID,salary 2 | FirstOne,LastOne,07/24/2015, phone197,ASD69643,1,15000 3 | FirstSecond,LastSecond,07/24/2015,phone756,ASD42892,2,15001 4 | FirstThird,LastThird,07/25/2015,phone1904,ASD37014,3,15002 5 | FirstFour,LastFour,07/26/2015,phone2435,ASD66902,4,15003 6 | FirstFive,LastFive,07/27/2015,phone2441,ASD90633,5,15004 7 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/compactionIUD2.csv: -------------------------------------------------------------------------------- 1 | FirstName,LastName,date, phonetype,serialname,ID,salary 2 | FirstSix,LastSix,07/24/2015, phone197,ASD69643,6, 15000 3 | FirstSeven, LastSeven, 07/24/2015,phone756,ASD42892,7,15001 4 | FirstEight, LastEight, 07/25/2015,phone1904,ASD37014,8,15002 5 | FirstNine, LastNine, 07/26/2015,phone2435,ASD66902,9,15003 6 | FirstTen, LastTen, 07/27/2015,phone2441,ASD90633,10,15004 7 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/compactionIUD3.csv: -------------------------------------------------------------------------------- 1 | FirstName,LastName,date, phonetype,serialname,ID,salary 2 | FirstEleven,LastEleven,07/24/2015, phone197,ASD69643,11, 15000 3 | FirstTwelve, LastTwelve, 07/24/2015,phone756,ASD42892,12,15001 4 | FirstThirteen, LastThirteen, 07/25/2015,phone1904,ASD37014,13,15002 5 | FirstFourteen, LastFourteen, 07/26/2015,phone2435,ASD66902,14,15003 6 | FirstFifteen, LastFifteen, 07/27/2015,phone2441,ASD90633,15,15004 7 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/compactionIUD4.csv: -------------------------------------------------------------------------------- 1 | FirstName,LastName,date, phonetype,serialname,ID,salary 2 | FirstSixteen,LastSixteen,07/24/2015, phone197,ASD69643,16, 15000 3 | FirstSeventeen, LastSeventeen, 07/24/2015,phone756,ASD42892,17,15001 4 | FirstEighteen, LastEighteen, 07/25/2015,phone1904,ASD37014,18,15002 5 | FirstNineteen, LastNineteen, 07/26/2015,phone2435,ASD66902,19,15003 6 | FirstTwenty, LastTwenty, 07/27/2015,phone2441,ASD90633,20,15004 7 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/compaction/nodictionary_compaction.csv: -------------------------------------------------------------------------------- 1 | code1,code2,country_code,category_id,product_id,date,count1,count2,count3 2 | "51job, Inc.",21695-534,FR,610,60,2017-11-27,4483,0,510 3 | Intercontinental Exchange Inc.,22100-020,TH,87,4,2017-10-16,2,647,69630 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/complexTypeDecimal.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary,complex 2 | 1.2,2015/07/23,china,aaa1,phone197,ASD69643,15000,3.113$3.33 3 | 2,2015/07/24,china,aaa2,phone756,ASD42892,15001,3.123$7.33 4 | 4.3,2015/07/26,china,aaa4,phone2435,ASD66902,15003,3.123$56.33 5 | 5,2015/07/27,china,aaa5,phone2441,ASD90633,15004,3.133$5.33 6 | 6.5,2015/07/28,china,aaa6,phone294,ASD59961,15005,3.133$54.33 7 | 8,2015/07/30,china,aaa8,phone1848,ASD57308,15007,32.13$56.33 8 | 9.1,2015/07/18,china,aaa9,phone706,ASD86717,15008,3.213$44.33 9 | 10,2015/07/19,usa,aaa10,phone685,ASD30505,15009,32.13$33.33 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/complexTypeDecimalNested.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary,complex 2 | 1.2,2015/7/23,china,aaa1,phone197,ASD69643,15000,3.113:imei$3.33:imsi 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892,15001,3.123:imei$7.33:imsi 4 | 4.3,2015/7/26,china,aaa4,phone2435,ASD66902,15003,3.123:imei$56.33:imsi 5 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,15004,3.133:imei$5.33:imsi 6 | 6.5,2015/7/28,china,aaa6,phone294,ASD59961,15005,3.133:imei$54.33:imsi 7 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,15007,32.13:imei$56.33:imsi 8 | 9.1,2015/7/18,china,aaa9,phone706,ASD86717,15008,3.213:imei$44.33:imsi 9 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009,32.13:imei$33.33:imsi -------------------------------------------------------------------------------- /integration/spark/src/test/resources/complexTypeDecimalNestedHive.csv: -------------------------------------------------------------------------------- 1 | 1.2,2015-7-23 00:00:00,china,aaa1,phone197,ASD69643,15000,3.113:imei$3.33:imsi 2 | 2,2015/7/24,china,aaa2,phone756,ASD42892,15001,3.123:imei$7.33:imsi 3 | 4.3,2015/7/26,china,aaa4,phone2435,ASD66902,15003,3.123:imei$56.33:imsi 4 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,15004,3.133:imei$5.33:imsi 5 | 6.5,2015/7/28,china,aaa6,phone294,ASD59961,15005,3.133:imei$54.33:imsi 6 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,15007,32.13:imei$56.33:imsi 7 | 9.1,2015/7/18,china,aaa9,phone706,ASD86717,15008,3.213:imei$44.33:imsi 8 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009,32.13:imei$33.33:imsi -------------------------------------------------------------------------------- /integration/spark/src/test/resources/complexbinary.csv: -------------------------------------------------------------------------------- 1 | 1,true,abc,binary1$binary2,binary1,1&binary1 2 | 2,false,abcd,binary11$binary12,binary11,1&binary2 3 | 3,true,abcde,binary13$binary13,binary13,1&binary3 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/complexdatareordered.csv: -------------------------------------------------------------------------------- 1 | 1,MAC1$MAC2$MAC3,109,4ROM size,29-11-2015,109,2738.562 2 | 10,MAC4$MAC5$MAC6,93,1ROM size,29-11-2015,93,1714.635 3 | 100,MAC7$MAC8$MAC9,2591,2ROM size,29-11-2015,2591,1271 4 | 1000,MAC10$$MAC12,2531,2ROM size,29-11-2015,2531,692 5 | 10000,MAC13$$MAC15,2408,0ROM size,29-11-2015,2408,2175 6 | 100000,MAC16$$MAC18,1815,0ROM size,29-11-2015,1815,136 7 | 1000000,MAC19$$MAC21,2479,4ROM size,29-11-2015,2479,1600 8 | 100001,MAC22$$MAC24,1845,7ROM size,29-11-2015,1845,505 9 | 100002,MAC25$$MAC27,2008,1ROM size,29-11-2015,2008,1341 10 | 100003,MAC28$$MAC30,1121,5ROM size,29-11-2015,1121,2239 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/complextypeWithEmptyRecords.csv: -------------------------------------------------------------------------------- 1 | 1,109,4ROM size,Intel,29-11-2015,,MAC1:1,7:Chinese:Hubei Province:yichang:yichang:yichang$7:India:New Delhi:delhi:delhi:delhi,29-11-2015$29-11-2015:29-11-2015,109,2738.562,, 2 | 1,109,4ROM size,Intel,29-11-2015,1AA1$2,,7:Chinese:Hubei Province:yichang:yichang:yichang$7:India:New Delhi:delhi:delhi:delhi,29-11-2015$29-11-2015:29-11-2015,109,2738.562,, 3 | 1,109,4ROM size,Intel,29-11-2015,1AA1$2,MAC1:1,,29-11-2015$29-11-2015:29-11-2015,109,2738.562,, -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 11,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96,5040 3 | 12,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95,7124 4 | 13,madhan,TPL,07-07-2009,2,tester,10,network,928478,07-08-2009,30-12-2016,88,99,9054 5 | 14,anandh,SA,29-12-2010,3,manager,11,protocol,928278,29-01-2011,29-06-2016,77,92,11248 6 | 15,ayushi,SSA,09-11-2011,1,developer,12,security,928375,09-12-2011,29-05-2016,99,91,13245 7 | 16,pramod,SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2012,29-12-2016,86,93,5040 8 | 17,gawrav,PL,22-09-2013,2,tester,12,security,928778,22-10-2013,15-11-2016,78,97,9574 9 | 18,sibi,TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,84,98,7245 10 | 19,shivani,PL,12-05-2015,1,developer,10,network,928977,12-06-2015,12-11-2016,88,91,11254 11 | 20,bill,PM,01-12-2015,3,manager,14,Learning,928479,01-01-2016,30-11-2016,75,94,13547 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data1.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 101,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96,5040 3 | 120,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95,7124 4 | 103,madhan,TPL,07-07-2009,2,tester,10,network,928478,07-08-2009,30-12-2016,88,99,9054 5 | 140,anandh,SA,29-12-2010,3,manager,11,protocol,928278,29-01-2011,29-06-2016,77,92,11248 6 | 15,anu,SSA,09-11-2011,1,developer,12,security,928375,09-12-2011,29-05-2016,99,91,13245 7 | 160,pramod,SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2012,29-12-2016,86,93,5040 8 | 107,gawrav,PL,22-09-2013,2,tester,12,security,928778,22-10-2013,15-11-2016,78,97,9574 9 | 181,sibi,TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,84,98,7245 10 | 119,shivani,PL,12-05-2015,1,developer,10,network,928977,12-06-2015,12-11-2016,88,91,11254 11 | 210,bill,PM,01-12-2015,3,manager,14,Learning,928479,01-01-2016,30-11-2016,75,94,13547 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data2.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 4,2014-01-21 00:00:00,china,aaa4,phone2435,ASD66902,15003 3 | abc,2014-01-22 00:00:00,china,aaa5,phone2441,ASD90633,15004 4 | 6,2014-03-07 00:00:00,china,aaa6,phone294,ASD59961,15005 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data2_DiffTimeFormat.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 4,07-10-2014 00:00:00,china,aaa4,phone2435,ASD66902,4 3 | 8,07-20-2014 00:00:00,china,aaa5,phone2441,ASD90633,10 4 | 6,07-25-2014 00:00:00,china,aaa6,phone294,ASD59961,15005 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dataWithEmptyRows.csv: -------------------------------------------------------------------------------- 1 | 29000,cust_name_2000,active_emui_version_2000,2010-10-04 01:00:01,12345678 2 | ,,,,0 3 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dataWithNegativeValues.csv: -------------------------------------------------------------------------------- 1 | -30000,aaa,-300 2 | 0,ddd,0 3 | -20000,bbb,-200 4 | 70000,ggg,700 5 | 10000,eee,100 6 | -10000,ccc,-100 7 | null,null,null -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dataWithNullFirstLine.csv: -------------------------------------------------------------------------------- 1 | 2 | 11,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96.2,5040.56 3 | 12,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95.1,7124.21 4 | 13,madhan,TPL,07-07-2009,2,tester,10,network,928478,07-08-2009,30-12-2016,88,99,9054.235 5 | 14,anandh,SA,29-12-2010,3,manager,11,protocol,928278,29-01-2011,29-06-2016,77,92.2,11248.25 6 | 15,ayushi,SSA,09-11-2011,1,developer,12,security,928375,09-12-2011,29-05-2016,99,91.5,13245.48 7 | 16,pramod,SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2012,29-12-2016,86,93,5040.56 8 | 17,gawrav,PL,22-09-2013,2,tester,12,security,928778,22-10-2013,15-11-2016,78,97.45,9574.24 9 | 18,sibi,TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,84,98.23,7245.25 10 | 19,shivani,PL,12-05-2015,1,developer,10,network,928977,12-06-2015,12-11-2016,88,91.678,11254.24 11 | 20,bill,PM,01-12-2015,3,manager,14,Learning,928479,01-01-2016,30-11-2016,75,94.22,13547.25 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dataWithSingleQuote.csv: -------------------------------------------------------------------------------- 1 | id,name 2 | 1,Tom 3 | 2,"Tony 4 | 3,Lily" 5 | 4,Games" 6 | 5,"prival\" 7 | 6,"hello\" -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data_alltypes.csv: -------------------------------------------------------------------------------- 1 | 1,10,100,48.4,spark,2015/4/23 12:01:01,1.23,2015/4/23 11:01:01,aaa 2 | 5,17,140,43.4,spark,2015/7/27 12:01:02,3.45,2015/7/27 11:01:02,bbb 3 | 1,11,100,44.4,flink,2015/5/23 12:01:03,23.23,2015/5/23 11:01:03,ccc 4 | 1,10,150,43.4,spark,2015/7/24 12:01:04,254.12,2015/7/24 11:01:04,ddd 5 | 1,10,100,47.4,spark,2015/7/23 12:01:05,876.14,2015/7/23 11:01:05,eeee 6 | 3,14,160,43.4,hive,2015/7/26 12:01:06,3454.32,2015/7/26 11:01:06,ff 7 | 2,10,100,43.4,impala,2015/7/23 12:01:07,456.98,2015/7/23 11:01:07,ggg 8 | 1,10,100,43.4,spark,2015/5/23 12:01:08,32.53,2015/5/23 11:01:08,hhh 9 | 4,16,130,42.4,impala,2015/7/23 12:01:09,67.23,2015/7/23 11:01:09,iii 10 | 1,10,100,43.4,spark,2015/7/23 12:01:10,832.23,2015/7/23 11:01:10,jjj 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data_beyond68yrs.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 11,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-1800,29-11-1900,96,96,5040 3 | 12,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-1802,30-12-1902,85,95,7124 4 | 13,madhan,TPL,7/7/2009,2,tester,10,network,928478,7/8/2009,30-12-2016,88,99,9054 5 | 14,anandh,SA,29-12-2010,3,manager,11,protocol,928278,29-01-2000,29-06-2016,77,92,11248 6 | 15,ayushi,SSA,9/7/2011,1,developer,12,security,928375,9/12/2011,29-05-2016,99,91,13245 7 | 16,pramod,SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2038,29-12-2041,86,93,5040 8 | 17,gawrav,PL,22-09-2013,2,tester,12,security,928778,22-10-3000,15-11-3002,78,97,9574 9 | 18,sibi,TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,84,98,7245 10 | 19,shivani,PL,12/5/2015,1,developer,10,network,928977,12/6/2015,12/11/2016,88,91,11254 11 | 20,bill,PM,1/12/2015,3,manager,14,Learning,928479,1/1/2016,30-11-2016,75,94,13547 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data_partition_badrecords.csv: -------------------------------------------------------------------------------- 1 | intField1, stringField1, intField2 2 | 3 | , 4 | ,, 5 | 1, 6 | 2,b 7 | 3,c,13 8 | 4,d,14,d 9 | 5,e, 10 | 6,f, ,16 11 | 7,g,g 12 | 8,h,h,18 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data_timestamp.csv: -------------------------------------------------------------------------------- 1 | col 2 | 2014-01-01 18:00:00 3 | 2014-01-02 18:00:00 4 | 2014-01-03 18:00:00 5 | 6 | 2014-01-03 18:00:00 7 | 0 8 | 2014-01-03 18:00:00 9 | 10 | 2014-01-03 18:00:00 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data_withCAPSHeader.csv: -------------------------------------------------------------------------------- 1 | EMPNO,EMPNAME,DESIGNATION,DOJ,WORKGROUPCATEGORY,WORKGROUPCATEGORYNAME,DEPTNO,DEPTNAME,PROJECTCODE,PROJECTJOINDATE,PROJECTENDDATE,ATTENDANCE,UTILIZATION,SALARY 2 | 11,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96.2,5040.56 3 | 12,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95.1,7124.21 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data_withMixedHeader.csv: -------------------------------------------------------------------------------- 1 | EMPNO,EMPNAME,Designation,Doj,WORKGROUPCATEGORY,WorkGroupCategoryName,DEPTNO,DeptName,PROJECTCODE,PROJECTJOINDATE,PROJECTENDDATE,Attendance,Utilization,Salary 2 | 11,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96.2,5040.56 3 | 12,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95.1,7124.21 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data_with_all_types.csv: -------------------------------------------------------------------------------- 1 | 1,10,100,48.4,spark,2015/4/23 12:01:01,1.23,2015/4/23 11:01:01,aaa,2.5 2 | 5,17,140,43.4,spark,2015/7/27 12:01:02,3.45,2015/7/27 11:01:02,bbb,2.5 3 | 1,11,100,44.4,flink,2015/5/23 12:01:03,23.23,2015/5/23 11:01:03,ccc,2.5 4 | 1,10,150,43.4,spark,2015/7/24 12:01:04,254.12,2015/7/24 11:01:04,ddd,2.5 5 | 1,10,100,47.4,spark,2015/7/23 12:01:05,876.14,2015/7/23 11:01:05,eeee,3.5 6 | 3,14,160,43.4,hive,2015/7/26 12:01:06,3454.32,2015/7/26 11:01:06,ff,2.5 7 | 2,10,100,43.4,impala,2015/7/23 12:01:07,456.98,2015/7/23 11:01:07,ggg,2.5 8 | 1,10,100,43.4,spark,2015/5/23 12:01:08,32.53,2015/5/23 11:01:08,hhh,2.5 9 | 4,16,130,42.4,impala,2015/7/23 12:01:09,67.23,2015/7/23 11:01:09,iii,2.5 10 | 1,10,100,43.4,spark,2015/7/23 12:01:10,832.23,2015/7/23 11:01:10,jjj,2.5 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/data_with_special_char.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 11,"arvind,ss",SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96,5040 3 | 12,"krithin$ks",SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95,7124 4 | 13,"madhan%rr",TPL,07-07-2009,2,tester,10,network,928478,07-08-2009,30-12-2016,88,99,9054 5 | 14,"anandh(y)",SA,29-12-2010,3,manager,11,protocol,928278,29-01-2011,29-06-2016,77,92,11248 6 | 15,"ayushi*ty",SSA,09-11-2011,1,developer,12,security,928375,09-12-2011,29-05-2016,99,91,13245 7 | 16,"pramod&56",SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2012,29-12-2016,86,93,5040 8 | 17,"gawrav@66",PL,22-09-2013,2,tester,12,security,928778,22-10-2013,15-11-2016,78,97,9574 9 | 18,"sibi=56",TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,84,98,7245 10 | 19,shivani,PL,12-05-2015,1,developer,10,network,928977,12-06-2015,12-11-2016,88,91,11254 11 | 20,bill,PM,01-12-2015,3,manager,14,Learning,928479,01-01-2016,30-11-2016,75,94,13547 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datadelimiter.csv: -------------------------------------------------------------------------------- 1 | empno|empname|designation|doj|workgroupcategory|workgroupcategoryname|deptno|deptname|projectcode|projectjoindate|projectenddate|attendance|utilization|salary 2 | 11|arvind|SE|17-01-2007|1|developer|10|network|928478|17-02-2007|29-11-2016|96|96|5040 3 | 12|krithin|SSE|29-05-2008|1|developer|11|protocol|928378|29-06-2008|30-12-2016|85|95|7124 4 | 13|madhan|TPL|07-07-2009|2|tester|10|network|928478|07-08-2009|30-12-2016|88|99|9054 5 | 14|anandh|SA|29-12-2010|3|manager|11|protocol|928278|29-01-2011|29-06-2016|77|92|11248 6 | 15|ayushi|SSA|09-11-2011|1|developer|12|security|928375|09-12-2011|29-05-2016|99|91|13245 7 | 16|pramod|SE|14-10-2012|1|developer|13|configManagement|928478|14-11-2012|29-12-2016|86|93|5040 8 | 17|gawrav|PL|22-09-2013|2|tester|12|security|928778|22-10-2013|15-11-2016|78|97|9574 9 | 18|sibi|TL|15-08-2014|2|tester|14|Learning|928176|15-09-2014|29-05-2016|84|98|7245 10 | 19|shivani|PL|12-05-2015|1|developer|10|network|928977|12-06-2015|12-11-2016|88|91|11254 11 | 20|bill|PM|01-12-2015|3|manager|14|Learning|928479|01-01-2016|30-11-2016|75|94|13547 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datanullmeasurecol.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,A234 3 | 2,2015/7/24,china,aaa2,phone756,A453 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dataretention1.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,ind,aaa1,phone197,ASD69643,15000 3 | 2,2015/7/24,ind,aaa2,phone756,ASD42892,15001 4 | 3,2015/7/25,ind,aaa3,phone1904,ASD37014,15002 5 | 4,2015/7/26,ind,aaa4,phone2435,ASD66902,15003 6 | 5,2015/7/27,ind,aaa5,phone2441,ASD90633,15004 7 | 6,2015/7/28,ind,aaa6,phone294,ASD59961,15005 8 | 7,2015/7/29,ind,aaa7,phone610,ASD14875,15006 9 | 8,2015/7/30,ind,aaa8,phone1848,ASD57308,15007 10 | 9,2015/7/18,ind,aaa9,phone706,ASD86717,15008 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dataretention11.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,ind,aaa1,phone197,"ASD69643 3 | a",15000 4 | 2,2015/7/24,ind,aaa2,phone756,"ASD42892 5 | b",15001 6 | 3,2015/7/25,ind,aaa3,phone1904,ASD37014,15002 7 | 4,2015/7/26,ind,aaa4,phone2435,ASD66902,15003 8 | 5,2015/7/27,ind,aaa5,phone2441,ASD90633,15004 9 | 6,2015/7/28,ind,aaa6,phone294,ASD59961,15005 10 | 7,2015/7/29,ind,aaa7,phone610,ASD14875,15006 11 | 8,2015/7/30,ind,aaa8,phone1848,ASD57308,15007 12 | 9,2015/7/18,ind,aaa9,phone706,ASD86717,15008 13 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dataretention2.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,aus,aaa1,phone197,ASD69643,15000 3 | 2,2015/7/24,aus,aaa2,phone756,ASD42892,15001 4 | 3,2015/7/25,aus,aaa3,phone1904,ASD37014,15002 5 | 4,2015/7/26,aus,aaa4,phone2435,ASD66902,15003 6 | 5,2015/7/27,aus,aaa5,phone2441,ASD90633,15004 7 | 6,2015/7/28,aus,aaa6,phone294,ASD59961,15005 8 | 7,2015/7/29,aus,aaa7,phone610,ASD14875,15006 9 | 8,2015/7/30,aus,aaa8,phone1848,ASD57308,15007 10 | 9,2015/7/18,aus,aaa9,phone706,ASD86717,15008 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dataretention3.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,eng,aaa1,phone197,ASD69643,15000 3 | 2,2015/7/24,eng,aaa2,phone756,ASD42892,15001 4 | 3,2015/7/25,eng,aaa3,phone1904,ASD37014,15002 5 | 4,2015/7/26,eng,aaa4,phone2435,ASD66902,15003 6 | 5,2015/7/27,eng,aaa5,phone2441,ASD90633,15004 7 | 6,2015/7/28,eng,aaa6,phone294,ASD59961,15005 8 | 7,2015/7/29,eng,aaa7,phone610,ASD14875,15006 9 | 8,2015/7/30,eng,aaa8,phone1848,ASD57308,15007 10 | 9,2015/7/18,eng,aaa9,phone706,ASD86717,15008 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datasample.csv: -------------------------------------------------------------------------------- 1 | empno,doj,salary 2 | 11,2016-04-14 15:00:09,5040.56 3 | 12,2016-03-14 15:00:09,1040.56 4 | 13,,1040.56 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datasamplecomplex.csv: -------------------------------------------------------------------------------- 1 | 11,2016-03-14 08:30:00.000,2016-03-14 08:30:09.000$2016-03-14 15:00:09.000$2016-03-14 17:30:35.000,2016-03-14 08:30:09.000$2016-03-14 17:30:35.000,5040.56 2 | 12,2016-04-14 08:30:00.000,2016-04-14 09:30:09.000$2016-04-14 15:30:09.000$2016-04-14 17:30:35.000,2016-04-14 08:30:09.000$2016-04-14 18:30:35.000,1040.56 3 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datasamplefordate.csv: -------------------------------------------------------------------------------- 1 | empno,doj,salary 2 | 11,2016-04-14,5040.56 3 | 12,2016-03-14,1040.56 4 | 13,,1040.56 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datasamplenull.csv: -------------------------------------------------------------------------------- 1 | ID,dateField,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,15000 3 | 2,,china,aaa2,phone756,ASD42892,15001 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datasingleCol.csv: -------------------------------------------------------------------------------- 1 | CA 2 | LA 3 | AD 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datasingleComplexCol.csv: -------------------------------------------------------------------------------- 1 | CA:272126 2 | LA:272125 3 | AD:272128 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithNegeativewithoutHeader.csv: -------------------------------------------------------------------------------- 1 | imei0,2147483647,9223372036854775807 2 | imei1,-2147483648,-9223372036854775807 3 | imei2,2147483647,9223372036854775807 4 | imei3,-2147483648,-9223372036854775808 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithNegtiveNumber.csv: -------------------------------------------------------------------------------- 1 | imei,age,num 2 | imei0,2147483647,9223372036854775807 3 | imei1,-2147483648,-9223372036854775807 4 | imei2,2147483647,9223372036854775807 5 | imei3,-2147483648,-9223372036854775808 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithbackslash.csv: -------------------------------------------------------------------------------- 1 | "ID","date","country","name","phonetype","serialname","salary" 2 | "1","2015/7/23","china","aaa1",""phone197","ASD69643","15000" 3 | "2","2015/7/24","china","aaa2","phon\e756","ASD42892","15001" 4 | "3","2015/7/25","china","aaa3","phone1904","ASD37014","15002" 5 | "4","2015/7/26","china","aaa4","phone2435","ASD66902","15003" 6 | "5","2015/7/27","china","aaa5","phone\2441","ASD90633","15004" 7 | "6","2015/7/28","china","aaa6","phone294","ASD59961","15005" 8 | "7","2015/7/29","china","aaa7","phone610","ASD14875","15006" 9 | "8","2015/7/30","china","aaa8","phone1848","ASD57308","15007" 10 | "9","2015/7/18","china","aaa9","phone\706","ASD86717","15008" 11 | "10","2015/7/19","usa","aaa10","phone685","ASD30505","15009" -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithescapecharacter.csv: -------------------------------------------------------------------------------- 1 | imei,specialchar 2 | 1AA1,hash#124 3 | 1AA2,space 125 4 | 1AA3,ampersand&&hi 5 | 1AA4,escape\\esc 6 | 1AA44,"escape\esc" 7 | 1AA5,not!hi 8 | 1AA6,braces(hi) 9 | 1AA7,percentage%hi 10 | 1AA8,Tilde~~ 11 | 1AA9,dollar$hi 12 | 1AA10,star***hi 13 | 1AA11,colon:hi 14 | 1AA12,semi;colon 15 | 1AA13,quote'1'22 16 | 1AA14,underscore_hi 17 | 1AA15,equals=hi 18 | 1AA16,plus+hi 19 | 1232,"ayush@b.com" 20 | 12323,"ayush@@b.com" 21 | 12345,"西安\咸阳" 22 | 12346,"西安\\咸阳" 23 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithmaxbigint.csv: -------------------------------------------------------------------------------- 1 | imei,age 2 | 1AA1,10 3 | 1AA2,26 4 | 1AA3,10 5 | 1AA4,10 6 | 1AA5,20 7 | 1AA6,10 8 | 1AA7,10 9 | 1AA8,10 10 | 1AA9,10 11 | 1AA10,10 12 | 1AA11,9223372036854775807 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithmaxinteger.csv: -------------------------------------------------------------------------------- 1 | imei,age 2 | 1AA1,10 3 | 1AA2,26 4 | 1AA3,10 5 | 1AA4,10 6 | 1AA5,20 7 | 1AA6,10 8 | 1AA7,10 9 | 1AA8,10 10 | 1AA9,10 11 | 1AA10,10 12 | 1AA12,2147483647 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithmaxminbigint.csv: -------------------------------------------------------------------------------- 1 | imei,age 2 | 1AA1,10 3 | 1AA2,26 4 | 1AA3,10 5 | 1AA4,10 6 | 1AA5,20 7 | 1AA6,10 8 | 1AA7,10 9 | 1AA8,10 10 | 1AA9,10 11 | 1AA10,10 12 | 1AA11,-9223372036854775808 13 | 1AA11,9223372036854775807 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithmaxmininteger.csv: -------------------------------------------------------------------------------- 1 | imei,age 2 | 1AA1,10 3 | 1AA2,26 4 | 1AA3,10 5 | 1AA4,10 6 | 1AA5,20 7 | 1AA6,10 8 | 1AA7,10 9 | 1AA8,10 10 | 1AA9,10 11 | 1AA10,10 12 | 1AA11,-2147483648 13 | 1AA12,2147483647 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithminbigint.csv: -------------------------------------------------------------------------------- 1 | imei,age 2 | 1AA1,10 3 | 1AA2,26 4 | 1AA3,10 5 | 1AA4,10 6 | 1AA5,20 7 | 1AA6,10 8 | 1AA7,10 9 | 1AA8,10 10 | 1AA9,10 11 | 1AA10,10 12 | 1AA11,-9223372036854775808 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithmininteger.csv: -------------------------------------------------------------------------------- 1 | imei,age 2 | 1AA1,10 3 | 1AA2,26 4 | 1AA3,10 5 | 1AA4,10 6 | 1AA5,20 7 | 1AA6,10 8 | 1AA7,10 9 | 1AA8,10 10 | 1AA9,10 11 | 1AA10,10 12 | 1AA11,-2147483648 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithnullmeasure.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,0 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892, 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,0 6 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithnullmsrs.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 11,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,,, 3 | 12,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95.1, 4 | 13,madhan,TPL,07-07-2009,2,tester,10,network,928478,07-08-2009,30-12-2016,88,,9054.235 5 | 14,anandh,SA,29-12-2010,3,manager,11,protocol,928278,29-01-2011,29-06-2016,77,92.2, 6 | 15,ayushi,SSA,09-11-2011,1,developer,12,security,928375,09-12-2011,29-05-2016,,91.5,13245.48 7 | 16,pramod,SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2012,29-12-2016,86,, 8 | 17,gawrav,PL,22-09-2013,2,tester,12,security,928778,22-10-2013,15-11-2016,78,97.45,9574.24 9 | 18,sibi,TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,,98.23,7245.25 10 | 19,shivani,PL,12-05-2015,1,developer,10,network,928977,12-06-2015,12-11-2016,88,91.678,11254.24 11 | 20,bill,PM,01-12-2015,3,manager,14,Learning,928479,01-01-2016,30-11-2016,75,, -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datawithoutheader.csv: -------------------------------------------------------------------------------- 1 | 11,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96,5040 2 | 12,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95,7124 3 | 13,madhan,TPL,07-07-2009,2,tester,10,network,928478,07-08-2009,30-12-2016,88,99,9054 4 | 14,anandh,SA,29-12-2010,3,manager,11,protocol,928278,29-01-2011,29-06-2016,77,92,11248 5 | 15,ayushi,SSA,09-11-2011,1,developer,12,security,928375,09-12-2011,29-05-2016,99,91,13245 6 | 16,pramod,SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2012,29-12-2016,86,93,5040 7 | 17,gawrav,PL,22-09-2013,2,tester,12,security,928778,22-10-2013,15-11-2016,78,97,9574 8 | 18,sibi,TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,84,98,7245 9 | 19,shivani,PL,12-05-2015,1,developer,10,network,928977,12-06-2015,12-11-2016,88,91,11254 10 | 20,bill,PM,01-12-2015,3,manager,14,Learning,928479,01-01-2016,30-11-2016,75,94,13547 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/datedatafile.csv: -------------------------------------------------------------------------------- 1 | datetype1 2 | 2018-09-11 3 | 2018-09-12 4 | 2018-09-13 5 | 2018-09-14 6 | 2018-09-15 7 | 2018-09-16 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dblocation/test.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,1,aa,aaa 3 | b,2,bb,bbb 4 | c,3,cc,ccc 5 | d,4,dd,ddd 6 | e,5,ee,eee 7 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/decimalBoundaryDataCarbon.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,12345678901234510.1234567890123 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892,12345678901234520.1234567890123 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,12345678901234530.1234567890123 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,12345678901234560.1234567890123 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,22345678901234560.1234567890123 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961,32345678901234560.1234567890123 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875,42345678901234560.1234567890123 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,52345678901234560.1234567890123 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,62345678901234560.1234567890123 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,72345678901234560.1234567890123 12 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,82345678901234560.1234567890123 13 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/decimalBoundaryDataHive.csv: -------------------------------------------------------------------------------- 1 | 1,2015/7/23,china,aaa1,phone197,ASD69643,12345678901234510.1234567890123 2 | 2,2015/7/24,china,aaa2,phone756,ASD42892,12345678901234520.1234567890123 3 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,12345678901234530.1234567890123 4 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,12345678901234560.1234567890123 5 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,22345678901234560.1234567890123 6 | 6,2015/7/28,china,aaa6,phone294,ASD59961,32345678901234560.1234567890123 7 | 7,2015/7/29,china,aaa7,phone610,ASD14875,42345678901234560.1234567890123 8 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,52345678901234560.1234567890123 9 | 9,2015/7/18,china,aaa9,phone706,ASD86717,62345678901234560.1234567890123 10 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,72345678901234560.1234567890123 11 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,82345678901234560.1234567890123 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/decimalData.csv: -------------------------------------------------------------------------------- 1 | smallIntField,intField,bigIntField,floatField,doubleField,decimalField,timestampField,dateField,stringField,varcharField,charField,arrayField,structField 2 | -1,-1,-1,-1.1,-1.1,-1.1234,2017-06-11 00:00:01,2017-06-11,abc1,abcd1,abcde1,a$b$c$1,a$b$1 3 | 2,2,2,2.1,2.1,2.1234,2017-06-12 23:59:02,2017-06-12,abc2,abcd2,abcde2,a$b$c$2,a$b$2 4 | 3,3,3,3.1,3.1,3.1234,2017-06-13 23:59:03,2017-06-13,abc3,abcd3,abcde3,a$b$c$3,a$b$3 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/decimalDataWithHeader.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,15000.43 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892,45234525465882.24 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,652165.22 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,15003.21 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,15000.43 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961,274565465656566.23 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875,15006.45 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,1500744.22 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008.02 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15000.43 12 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,15010.00 13 | 12,2015/7/19,china,aaa12,phone1781,ASD85711,15011.20 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/decimalDataWithoutHeader.csv: -------------------------------------------------------------------------------- 1 | 1,2015/7/23,china,aaa1,phone197,ASD69643,15000.43 2 | 2,2015/7/24,china,aaa2,phone756,ASD42892,45234525465882.24 3 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,652165.22 4 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,15003.21 5 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,15000.43 6 | 6,2015/7/28,china,aaa6,phone294,ASD59961,274565465656566.23 7 | 7,2015/7/29,china,aaa7,phone610,ASD14875,15006.45 8 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,1500744.22 9 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008.02 10 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15000.43 11 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,15010.00 12 | 12,2015/7/19,china,aaa12,phone1781,ASD85711,15011.20 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/decimal_int_range.csv: -------------------------------------------------------------------------------- 1 | d1 2 | 111111 3 | 222222.120 4 | 333333.12345 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/deviceInformationId.csv: -------------------------------------------------------------------------------- 1 | 100044 2 | 100045, 3 | 100046, 4 | 100047 5 | 100048 6 | 100049, 7 | 10005 8 | 100050 9 | 100051 10 | 100052, 11 | 100053 12 | 100054, 13 | 100055 14 | 100056 15 | 100057 16 | 100058 17 | 100059 18 | 10006 19 | 100060 20 | 100061 21 | 100062 22 | 100063 23 | 100064 24 | 100065 25 | 100066 26 | 100067 27 | 100068, 28 | 100069 29 | 10007 30 | 100070 31 | 100071 32 | 100072 33 | 100073 34 | 100074 35 | 100075 36 | 100076 37 | 100077 38 | 100078 39 | 100079 40 | 10008 41 | 100080 42 | 100081 43 | 100082 44 | 100083 45 | 100084, 46 | 100085 47 | 10086, 48 | 100087 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/deviceInformationId2.csv: -------------------------------------------------------------------------------- 1 | 100082 2 | 100083 3 | 100084, 4 | 100085 5 | 10086, 6 | 10011 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dimSample.csv: -------------------------------------------------------------------------------- 1 | id,name,city 2 | 1,David,Beijing 3 | 2,Mark,Paris 4 | 3,Bill,NewYork 5 | 4,Sara,Tokyo 6 | 5,John,Beijing 7 | 6,Michel,Chicago 8 | 7,Robert,Houston 9 | 8,Sunny,Boston 10 | 9,Mary,Tokyo 11 | 10,Edward,Paris 12 | 11,James,Washington 13 | 12,Maria,Berlin 14 | 13,Adam,Athens 15 | 14,Peter,Boston 16 | 15,George,Paris 17 | 16,Paul,Shanghai 18 | 17,Lisa,Hangzhou 19 | 18,Angel,Beijing 20 | 19,Emily,Bangalore 21 | 20,Kevin,Singapore -------------------------------------------------------------------------------- /integration/spark/src/test/resources/dimTableSample.csv: -------------------------------------------------------------------------------- 1 | id,name,city 2 | 1,david,shenzhen 3 | 2,eason,shenzhen 4 | 3,jarry,wuhan -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double.csv: -------------------------------------------------------------------------------- 1 | empno, salary 2 | 'abc', 775678765456789098765432.789 3 | 'def', 876567898743456785232.44431 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_AdaptiveFloating_byte.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|0.0012|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|0.0013|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|0.0014|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|0.0015|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_AdaptiveFloating_int.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|100000.2812|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|100000.2813|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|100000.2814|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|200000.1415|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_AdaptiveFloating_short.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|0.1412|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|0.1413|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|0.1414|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|0.1415|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_AdaptiveFloating_short_int.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|300.1412|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|300.1413|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|300.1414|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|800.1415|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_byte.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|3.1412|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|3.1413|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|3.1414|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|3.1415|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_int.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|199161.2812|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|199161.2813|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|199161.2814|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|200000.1415|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_long.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|100000.1412|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|100000.1413|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|100000.1414|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|200000.1415|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_short.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|3.1412|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|3.1413|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|3.1414|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|5.1415|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/double/data_notitle_short_int.csv: -------------------------------------------------------------------------------- 1 | 12345|"test1"|922337203687|11.11|300.1412|2016-8-1 11:45:15 2 | 12346|"test2"|922337203688|11.12|300.1413|2016-8-2 11:45:15 3 | 12347|"test2"|922337203689|11.13|300.1414|2016-8-3 11:45:15 4 | 12348|"test3"|922337203680|11.14|900.1415|2016-8-4 11:45:15 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/emp.csv: -------------------------------------------------------------------------------- 1 | emp_no,ename,job,mgr_id,date_of_joining,salary,bonus,dept_no 2 | 7369,SMITH,CLERK,7902,1980-12-17,800,,20 3 | 7499,ALLEN,SALESMAN,7698,1981-02-21,1600,300,30 4 | 7521,WARD,SALESMAN,7698,1981-03-21,1250,500,30 5 | 7566,JONES,MANAGER,7839,1983-04-01,2975,,20 6 | 7654,MARTIN,SALESMAN,7698,1989-05-11,1250,1400,30 7 | 7698,BLAKE,MANAGER,7839,1981-02-21,2850,,30 8 | 7782,CLARK,MANAGER,7839,1985-11-201,2450,,10 9 | 7788,SCOTT,ANALYST,7566,2001-02-21,3000,,20 10 | 7839,KING,PRESIDENT,,1981-04-12,5000,,10 11 | 7844,TURNER,SALESMAN,7698,1981-02-21,1500,0,30 12 | 7876,ADAMS,CLERK,7788,1999-02-01,1100,,20 13 | 7900,JAMES,CLERK,7698,1997-12-31,950,,30 14 | 7902,FORD,ANALYST,7566,1984-05-25,3000,,20 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/emptyDimensionData.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,,aaa1,,,15000 3 | 2,2015/7/24,,,,,15001 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,15002 5 | 4,2015/7/26,china,aaa4,,ASD66902,15003 6 | 5,2015/7/27,china,aaa5,phone2441,,15004 7 | 6,2015/7/28,,aaa6,phone294,ASD59961,15005 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875,15006 9 | 8,2015/7/30,china,aaa8,,ASD57308,15007 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009 12 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,15010 13 | 12,2015/7/19,,aaa12,phone1781,ASD85711,15011 14 | 13,2015/7/20,china,,phone943,,15012 15 | 14,2015/7/21,china,aaa14,phone1954,ASD80468,15013 16 | 15,2015/7/22,china,aaa15,phone451,ASD1954,15014 17 | 16,2015/7/23,china,aaa16,phone390,ASD38513,15015 18 | 17,2015/7/24,,aaa17,phone1929,ASD86213,15016 19 | 18,2015/7/25,usa,aaa18,,ASD88812,15017 20 | 19,2015/7/26,china,aaa19,phone2151,ASD9316,15018 21 | 20,2015/7/27,china,aaa20,phone2625,ASD62597,15019 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/emptyDimensionDataHive.csv: -------------------------------------------------------------------------------- 1 | 1,2015/7/23,,aaa1,,,15000 2 | 2,2015/7/24,,,,,15001 3 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,15002 4 | 4,2015/7/26,china,aaa4,,ASD66902,15003 5 | 5,2015/7/27,china,aaa5,phone2441,,15004 6 | 6,2015/7/28,,aaa6,phone294,ASD59961,15005 7 | 7,2015/7/29,china,aaa7,phone610,ASD14875,15006 8 | 8,2015/7/30,china,aaa8,,ASD57308,15007 9 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008 10 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009 11 | 11,2015/7/18,china,aaa11,phone1554,ASD26101,15010 12 | 12,2015/7/19,,aaa12,phone1781,ASD85711,15011 13 | 13,2015/7/20,china,,phone943,,15012 14 | 14,2015/7/21,china,aaa14,phone1954,ASD80468,15013 15 | 15,2015/7/22,china,aaa15,phone451,ASD1954,15014 16 | 16,2015/7/23,china,aaa16,phone390,ASD38513,15015 17 | 17,2015/7/24,,aaa17,phone1929,ASD86213,15016 18 | 18,2015/7/25,usa,aaa18,,ASD88812,15017 19 | 19,2015/7/26,china,aaa19,phone2151,ASD9316,15018 20 | 20,2015/7/27,china,aaa20,phone2625,ASD62597,15019 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/emptylines.csv: -------------------------------------------------------------------------------- 1 | name,age 2 | a,25 3 | 4 | b,22 5 | 6 | c,23 7 | 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/emptyrow/csvwithonlyspacechar.csv: -------------------------------------------------------------------------------- 1 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/emptyrow/emptyRows.csv: -------------------------------------------------------------------------------- 1 | eid,ename,sal,presal,comm,deptno,Desc 2 | 17 3 | 4 | 18,,,,,,NullValue_check 5 | 19,Ravan,12345678900123456789098765432112345678.12345678900123456789098765432112345678,12345678900123456789098765432112345678.12345678900123456789098765432112345678,12345678900123456789098765432112345678.12345678900123456789098765432112345678,12345678900123456789098765432112345678.12345678900123456789098765432112345678,38_38_range 6 | 25,a,1 7 | 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/encoding_types.csv: -------------------------------------------------------------------------------- 1 | begin_time,name,begin_time1,begin_time2,begin_time3,begin_time4,begin_time5,begin_time6,begin_time7,begin_time8,begin_time9,begin_time10,begin_time11,begin_time12,begin_time13,begin_time14,begin_time15,begin_time16,begin_time17,begin_time18,begin_time19,begin_time20 2 | 1497376581,name1,10000,8388600,125,1497376581,8386600,10000,100,125,1497376581,1497423738,2139095000,1497376581,1497423738,32000,123.4,11.1,3200.1,214744460.2,1497376581,1497376581 3 | 1497408581,name2,32000,45000,25,10000,55000,32000,75,35,1497423838,1497423838,2147484000,1497423838,1497423838,31900,838860.7,12.3,127.1,214748360.2,1497408581,1497408581 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/betweenFilter.csv: -------------------------------------------------------------------------------- 1 | id,name,orders 2 | 1,Bhavya,10 3 | 2,Sandeep,20 4 | 3,Vijay,5 5 | 4,Manish,6 6 | 5,Kunal,10 7 | 6,Divya,11 8 | 7,India,12 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/datagrtlrt.csv: -------------------------------------------------------------------------------- 1 | 2015-7-23 12:07:28,china,15000 2 | 2018-7-24 12:07:28,china,15001 3 | 2017-7-25 12:07:28,china,15002 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/datawithnull.csv: -------------------------------------------------------------------------------- 1 | 1,emp1,1234 2 | 2,emp2,4321 3 | 3,emp3,xyz -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/datawithoutnull.csv: -------------------------------------------------------------------------------- 1 | 1,emp1,1234 2 | 2,emp2,4321 3 | 3,emp3,22 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/emp2.csv: -------------------------------------------------------------------------------- 1 | empid,ename,sal,deptno,mgr,gender,dob,comm,desc 2 | 1,abc,1233,10,2,,2014-07-01 12:07:28,1234.191,string_null 3 | 2,bcd,1322,,3,f,2014-07-01 12:07:28,19.99,int_null 4 | 3,cde,4322,,4,m,,16.996,date_null 5 | 4, ,43243,,5,m,,999.117,string_space 6 | 5,,43242,20,6,m,2017-07-01 12:07:28,99.999,string_null 7 | 6,ijk,,20,6,m,2017-07-01 12:07:28,50089,double_null 8 | 7,pqr,2422,20,6,m,2017-07-01 12:07:28,32.339,decimal_null 9 | 8 10 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/emp2allnull.csv: -------------------------------------------------------------------------------- 1 | empid,ename,sal,deptno,mgr,gender,dob,comm,desc 2 | 1,abc,1233,10,2,,,1234.191,string_null 3 | 2,bcd,1322,,3,f,,19.99,int_null 4 | 3,cde,4322,,4,m,,16.996,date_null 5 | 4, ,43243,,5,m,,999.117,string_space 6 | 5,,43242,20,6,m,,99.999,string_null 7 | 6,ijk,,20,6,m,,50089,double_null 8 | 7,pqr,2422,20,6,m,,32.339,decimal_null 9 | 8 10 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/emp2nonull.csv: -------------------------------------------------------------------------------- 1 | empid,ename,sal,deptno,mgr,gender,dob,comm,desc 2 | 1,abc,1233,10,2,,2014-07-01 12:07:28,1234.191,string_null 3 | 2,bcd,1322,,3,f,2014-07-01 12:07:28,19.99,int_null 4 | 3,cde,4322,,4,m,2014-07-01 12:07:28,16.996,date_null 5 | 4, ,43243,,5,m,2014-07-01 12:07:28,999.117,string_space 6 | 5,,43242,20,6,m,2017-07-01 12:07:28,99.999,string_null 7 | 6,ijk,,20,6,m,2017-07-01 12:07:28,50089,double_null 8 | 7,pqr,2422,20,6,m,2017-07-01 12:07:28,32.339,decimal_null 9 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/notEqualToFilter.csv: -------------------------------------------------------------------------------- 1 | 1,2015-07-23 00:00:00,china,aaa1,phone197,ASD69643,15000 2 | 7,2015-07-24 00:00:00,china,aaa2,phone756,ASD42892,15001 3 | 7,2015-07-25 00:00:00,china,aaa3,phone1904,ASD37014,15002 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/filter/notNullFilter.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary,floatField 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,15000,2.34 3 | vishal,2015/7/24,china,aaa2,phone756,ASD42892,15001,2.34 4 | raghu,2015/7/25,china,aaa3,phone1904,ASD37014,15002,2.34 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,15003,2.34 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,15004,2.34 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961,15005,3.5 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875,15006,2.34 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,15007,2.34 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008,2.34 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009,2.34 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/floatSample.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary,rating 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,15000,2.34 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892,15001,2.34 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,15002,2.34 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,15003,2.34 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,15004,2.34 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961,15005,3.5 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875,15006,2.34 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,15007,2.34 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008,2.34 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009,2.34 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/geodata.csv: -------------------------------------------------------------------------------- 1 | timevalue,longitude,latitude 2 | 1575428400000,116285807,40084087 3 | 1575428400000,116372142,40129503 4 | 1575428400000,116187332,39979316 5 | 1575428400000,116337069,39951887 6 | 1575428400000,116359102,40154684 7 | 1575428400000,116736367,39970323 8 | 1575428400000,116720179,40009893 9 | 1575428400000,116346961,40133550 10 | 1575428400000,116302895,39930753 11 | 1575428400000,116288955,39999101 12 | 1575428400000,116176090,40129953 13 | 1575428400000,116725575,39981115 14 | 1575428400000,116266922,40179415 15 | 1575428400000,116353706,40156483 16 | 1575428400000,116362699,39942444 17 | 1575428400000,116325378,39963129 18 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/geodata3.csv: -------------------------------------------------------------------------------- 1 | col1,col2,longitude,latitude 2 | 1,12,120177080,30326882 3 | 2,19,120180685,30326327 4 | 3,1,120184976,30327105 5 | 4,12,120189311,30327549 6 | 5,1,120194460,30329698 7 | 6,6,120186965,30329133 8 | 7,8,120177481,30328911 9 | 8,9,120169713,30325614 10 | 9,10,120164563,30322243 11 | 10,11,120171558,30319613 12 | 11,1,120176365,30320687 13 | 12,2,120179669,30323688 14 | 13,34,120181001,30320761 15 | 2,3,120187094,30323540 16 | 5,4,120193574,30323651 17 | 8,6,120186192,30320132 18 | 13,7,120190055,30317464 19 | 6,8,120195376,30318094 20 | 12,10,120160786,30317094 21 | 15,11,120168211,30318057 22 | 1,12,120173618,30316612 23 | 12,1,120181001,30317316 24 | 6,14,120185162,30315908 25 | 8,15,120192415,30315871 26 | 9,16,120161902,30325614 27 | 10,1,120164306,30328096 28 | 12,2,120197093,30325985 29 | 4,4,120196020,30321651 30 | 6,5,120198638,30323540 31 | 7,7,120165421,30314834 32 | 2,5,116285807,40084087 33 | 1,3,116337069,39951887 34 | 1,2,116288955,39999101 35 | 5,6,116325378,39963129 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/geodataWithCorrectSpatialIndex.csv: -------------------------------------------------------------------------------- 1 | mygeohash,timevalue,longitude,latitude 2 | 855280799612,1575428400000,116285807,40084087 3 | 855283635086,1575428400000,116372142,40129503 4 | 855279346102,1575428400000,116187332,39979316 5 | 855282156308,1575428400000,116337069,39951887 6 | 855283640154,1575428400000,116359102,40154684 7 | 855282440834,1575428400000,116736367,39970323 8 | 855282468370,1575428400000,116720179,40009893 9 | 855283633205,1575428400000,116346961,40133550 10 | 855279270226,1575428400000,116302895,39930753 11 | 855279368850,1575428400000,116288955,39999101 12 | 855280812709,1575428400000,116176090,40129953 13 | 855282443862,1575428400000,116725575,39981115 14 | 855280927196,1575428400000,116266922,40179415 15 | 855283640110,1575428400000,116353706,40156483 16 | 855282072206,1575428400000,116362699,39942444 17 | 855282157702,1575428400000,116325378,39963129 18 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/geodataWithErrorSpatialIndex.csv: -------------------------------------------------------------------------------- 1 | mygeohash,timevalue,longitude,latitude 2 | 0,1575428400000,116285807,40084087 3 | 0,1575428400000,116372142,40129503 4 | 0,1575428400000,116187332,39979316 5 | 0,1575428400000,116337069,39951887 6 | 0,1575428400000,116359102,40154684 7 | 0,1575428400000,116736367,39970323 8 | 0,1575428400000,116720179,40009893 9 | 0,1575428400000,116346961,40133550 10 | 0,1575428400000,116302895,39930753 11 | 0,1575428400000,116288955,39999101 12 | 0,1575428400000,116176090,40129953 13 | 0,1575428400000,116725575,39981115 14 | 0,1575428400000,116266922,40179415 15 | 0,1575428400000,116353706,40156483 16 | 0,1575428400000,116362699,39942444 17 | 0,1575428400000,116325378,39963129 18 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/globalsort/sample1.csv: -------------------------------------------------------------------------------- 1 | id,name,city,age 2 | 10,a,wuhan,10 3 | 4,y,hangzhou,20 4 | 7,z,beijing,30 5 | 1,d,shenzhen,40 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/globalsort/sample2.csv: -------------------------------------------------------------------------------- 1 | id,name,city,age 2 | 11,c,wuhan,50 3 | 2,f,hangzhou,60 4 | 5,m,beijing,70 5 | eight,b,shenzhen,80 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/globalsort/sample3.csv: -------------------------------------------------------------------------------- 1 | id,name,city,age 2 | 9,e,wuhan,90 3 | 6,x,hangzhou,100 4 | 3,k,beijing,110 5 | 12,l,shenzhen,120 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/invalidMeasures.csv: -------------------------------------------------------------------------------- 1 | India,15000854676378676765378647856378567846578365786347865783456783456783465783465783465783465763478563478567834567834653750834758093478534857348578345789345789347395873483784857348573485734895789347589347589375984759389358347589737583758937589789798437893475893758934758945783475893758947589347587348957389573489758347589734589347589347589347534897589347589347583475893475893475893457893478934575489758973847583947538947583947534897349575375347398733895453444787893758345943458783497874587783597358973589785934789357895378593789357893578935789357893578935785783789357897897893789578935789357893578935789357893578937895783953789578935789357893578935789357893578935789357893789578935789357835378578357835978935357897893535789378953789578935789357893578935789,22.435 2 | USA,234.43,2224444444444444444444444465558999.23 3 | Russia,, -------------------------------------------------------------------------------- /integration/spark/src/test/resources/j2.csv: -------------------------------------------------------------------------------- 1 | 1,gb3e5135-5533-4ee7-51b3-F61F1355b471,2,2,,,2,563FXN1S,2016-06-28,OORM1L,,,46315_4,,,,,,,,66116E013000000000000000,66116E013000000000000000,13.143.170.55,0.0.0.1,,1,1,ZOCERS,1,1,,,seach out for star wars starwars starwars@foxmovies.comAA,,,,64,6416557544541,26557544541,560111140564316,64075303555565,504,55,,,,63613316334514,,,,,,211111111111111111,,,1,1,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,11163575,20160628 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/join/data1.csv: -------------------------------------------------------------------------------- 1 | 1,11,100,44.4,flink,2015/5/23 12:01:03,23.23,2015/5/23,ccc,2.5,'foo'#'bar'#'world' 2 | 1,10,150,43.4,spark,2015/7/24 12:01:04,254.12,2015/7/24,ddd,2.5,'foo'#'bar'#'world' -------------------------------------------------------------------------------- /integration/spark/src/test/resources/join/data2.csv: -------------------------------------------------------------------------------- 1 | 1,10,100,48.4,spark,2015/4/23 12:01:01,1.23,2015/4/23,aaa,2.5,'foo'#'bar'#'world' 2 | 5,17,140,43.4,spark,2015/7/27 12:01:02,3.45,2015/7/27,bbb,2.5,'foo'#'bar'#'world' 3 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/join/emp.csv: -------------------------------------------------------------------------------- 1 | tom,t23717,h2399,99780207526 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/join/mgr.csv: -------------------------------------------------------------------------------- 1 | harry,h2399,v788232,99823230205 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/PrimitiveTypeWithNull.json: -------------------------------------------------------------------------------- 1 | { 2 | "stringField": null, 3 | "intField": 26 4 | } 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/StructOfAllTypes.json: -------------------------------------------------------------------------------- 1 | { 2 | "StructColumn":{ 3 | "stringField": "bob", 4 | "intField": 10, 5 | "longField": 12345678, 6 | "doubleField": 123400.78, 7 | "boolField": true, 8 | "FloorNum": [1,2,3,4,5,6], 9 | "FloorString": [ "abc", "def"], 10 | "FloorLong": [ 1234567, 2345678], 11 | "FloorDouble": [ 1.0, 2.0, 33.33], 12 | "FloorBool": [ true, false, false, true] 13 | } 14 | } 15 | 16 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/allPrimitiveType.json: -------------------------------------------------------------------------------- 1 | { 2 | "stringField": "ajantha\"bhat\"", 3 | "intField": 26, 4 | "shortField": 26, 5 | "longField": 1234567, 6 | "doubleField": 23.3333, 7 | "boolField": false, 8 | "dateField": "2019-03-02", 9 | "timeField": "2019-02-12 03:03:34", 10 | "decimalField" : 55.35, 11 | "binaryField" : "abc" 12 | } 13 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/allPrimitiveTypeBadRecord.json: -------------------------------------------------------------------------------- 1 | { 2 | "stringField": 123, 3 | "intField": "string", 4 | "shortField": 1234567, 5 | "longField": 23.5, 6 | "doubleField": "string", 7 | "boolField": 10, 8 | "dateField": 12345, 9 | "timeField": 12345, 10 | "decimalField" : "String" 11 | } 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/arrayOfStructOfStruct.json: -------------------------------------------------------------------------------- 1 | { 2 | "name": "bob", 3 | "age": 10, 4 | "doorNum": [ 5 | { 6 | "street": "abc", 7 | "city": "city1", 8 | "FloorNum": {"wing" : "a", "number" : 1} 9 | }, 10 | { 11 | "street": "def", 12 | "city": "city2", 13 | "FloorNum": {"wing" : "b", "number" : 0} 14 | }, 15 | { 16 | "street": "ghi", 17 | "city": "city3", 18 | "FloorNum": {"wing" : "a", "number" : 2} 19 | } 20 | ] 21 | } -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/arrayOfarrayOfarrayOfStruct.json: -------------------------------------------------------------------------------- 1 | { 2 | "name": "ajantha", 3 | "age": 26, 4 | "BuildNum": [ 5 | [ 6 | [ 7 | {"street":"abc", "city":"city1"}, 8 | {"street":"def", "city":"city2"}, 9 | {"street":"cfg", "city":"city3"} 10 | ], 11 | [ 12 | {"street":"abc1", "city":"city3"}, 13 | {"street":"def1", "city":"city4"}, 14 | {"street":"cfg1", "city":"city5"} 15 | ] 16 | ], 17 | [ 18 | [ 19 | {"street":"abc2", "city":"cityx"}, 20 | {"street":"abc3", "city":"cityy"}, 21 | {"street":"abc4", "city":"cityz"} 22 | ], 23 | [ 24 | {"street":"a1bc", "city":"cityA"}, 25 | {"street":"a1bc", "city":"cityB"}, 26 | {"street":"a1bc", "city":"cityc"} 27 | ] 28 | ] 29 | ] 30 | } 31 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/similarSchemaFiles/JsonReaderTest/MultipleRowSingleLineJson.json: -------------------------------------------------------------------------------- 1 | {"stringField": "kkkk","intField": 26,"shortField": 26,"longField": 1234567,"doubleField": 23.3333,"boolField": false,"dateField": "2019-03-02","timeField": "2019-02-12 03:03:34","decimalField" : 55.35} 2 | {"stringField": "bbbb","intField": 26,"shortField": 26,"longField": 1234567,"doubleField": 23.3333,"boolField": false,"dateField": "2019-03-02","timeField": "2019-02-12 03:03:34","decimalField" : 55.35} 3 | {"stringField": "cccc","intField": 26,"shortField": 26,"longField": 1234567,"doubleField": 23.3333,"boolField": false,"dateField": "2019-03-02","timeField": "2019-02-12 03:03:34","decimalField" : 55.35} 4 | {"stringField": "dddd","intField": 26,"shortField": 26,"longField": 1234567,"doubleField": 23.3333,"boolField": false,"dateField": "2019-03-02","timeField": "2019-02-12 03:03:34","decimalField" : 55.35} 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/similarSchemaFiles/JsonReaderTest/SingleRowSingleLineJson.json: -------------------------------------------------------------------------------- 1 | {"stringField": "kkkk","intField": 26,"shortField": 26,"longField": 1234567,"doubleField": 23.3333,"boolField": false,"dateField": "2019-03-02","timeField": "2019-02-12 03:03:34","decimalField" : 55.35} 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/similarSchemaFiles/JsonReaderTest/withRecordIdentifier/SingleRowMultipleLineJsonWithRecordIdentifier.json: -------------------------------------------------------------------------------- 1 | {"jsonData":{ 2 | "stringField": "ajantha", 3 | "intField": 26, 4 | "shortField": 26, 5 | "longField": 1234567, 6 | "doubleField": 23.3333, 7 | "boolField": false, 8 | "dateField": "2019-03-02", 9 | "timeField": "2019-02-12 03:03:34", 10 | "decimalField": 55.35 11 | } 12 | } 13 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/similarSchemaFiles/JsonReaderTest/withRecordIdentifier/SingleRowSingleLineJsonWithRecordIdentifier.json: -------------------------------------------------------------------------------- 1 | {"jsonField":{"stringField": "kkkk","intField": 26,"shortField": 26,"longField": 1234567,"doubleField": 23.3333,"boolField": false,"dateField": "2019-03-02","timeField": "2019-02-12 03:03:34","decimalField" : 55.35}} 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/data/similarSchemaFiles/allPrimitiveTypeSingleArray.json: -------------------------------------------------------------------------------- 1 | [ 2 | { 3 | "stringField": "ZZ", 4 | "intField": 100, 5 | "shortField": 100, 6 | "longField": 1234567, 7 | "doubleField": 23.3333, 8 | "boolField": false, 9 | "dateField": "2020-03-02", 10 | "timeField": "2020-02-12 03:03:34", 11 | "decimalField": 55.35 12 | } 13 | ] -------------------------------------------------------------------------------- /integration/spark/src/test/resources/jsonFiles/schema/arrayOfarrayOfarrayOfStruct.avsc: -------------------------------------------------------------------------------- 1 | { 2 | "name": "address", 3 | "type": "record", 4 | "fields": [ 5 | { 6 | "name": "name", 7 | "type": "string" 8 | }, 9 | { 10 | "name": "age", 11 | "type": "int" 12 | }, 13 | { 14 | "name": "BuildNum", 15 | "type": { 16 | "type": "array", 17 | "items": { 18 | "name": "FloorNum", 19 | "type": "array", 20 | "items": { 21 | "name": "doorNum", 22 | "type": "array", 23 | "items": { 24 | "name": "my_address", 25 | "type": "record", 26 | "fields": [ 27 | { 28 | "name": "street", 29 | "type": "string" 30 | }, 31 | { 32 | "name": "city", 33 | "type": "string" 34 | } 35 | ] 36 | } 37 | } 38 | } 39 | } 40 | } 41 | ] 42 | } 43 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/lessthandatacolumndata.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643,15000 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892,15001 4 | 3,2015/7/25,china 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902,15003 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633,15004 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961,15005 8 | 7,2015/7/29,china,aaa7,phone610, 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308,15007 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717,15008 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505,15009 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/loadMultiFiles/.invisibilityfile: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 19,shivani,PL,12-05-2015,1,developer,10,network,928977,12-06-2015,12-11-2016,88,91.678,11254.24 3 | 20,bill,PM,01-12-2015,3,manager,14,Learning,928479,01-01-2016,30-11-2016,75,94.22,13547.25 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/loadMultiFiles/_SUCCESS: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/integration/spark/src/test/resources/loadMultiFiles/_SUCCESS -------------------------------------------------------------------------------- /integration/spark/src/test/resources/loadMultiFiles/data.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 11,arvind,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96.2,5040.56 3 | 12,krithin,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95.1,7124.21 4 | 13,madhan,TPL,07-07-2009,2,tester,10,network,928478,07-08-2009,30-12-2016,88,99,9054.235 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/loadMultiFiles/emptyfile.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/integration/spark/src/test/resources/loadMultiFiles/emptyfile.csv -------------------------------------------------------------------------------- /integration/spark/src/test/resources/loadMultiFiles/nestedfolder1/data.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 16,pramod,SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2012,29-12-2016,86,93,5040.56 3 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/loadMultiFiles/nestedfolder1/data1.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 14,anandh,SA,29-12-2010,3,manager,11,protocol,928278,29-01-2011,29-06-2016,77,92.2,11248.25 3 | 15,ayushi,SSA,09-11-2011,1,developer,12,security,928375,09-12-2011,29-05-2016,99,91.5,13245.48 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/loadMultiFiles/nestedfolder1/nestedfolder2/data.csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 17,gawrav,PL,22-09-2013,2,tester,12,security,928778,22-10-2013,15-11-2016,78,97.45,9574.24 3 | 18,sibi,TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,84,98.23,7245.25 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/loadMultiFiles/non-csv: -------------------------------------------------------------------------------- 1 | empno,empname,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | 19,shivani,PL,12-05-2015,1,developer,10,network,928977,12-06-2015,12-11-2016,88,91.678,11254.24 3 | 20,bill,PM,01-12-2015,3,manager,14,Learning,928479,01-01-2016,30-11-2016,75,94.22,13547.25 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/localdictionary.csv: -------------------------------------------------------------------------------- 1 | name,age 2 | vishal,30 3 | akash,24 4 | praveen,22 5 | kumar,30 6 | brijoo,35 7 | ravindra,34 8 | 9 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/locationInfoActiveCountry.csv: -------------------------------------------------------------------------------- 1 | Chinese 2 | Chinese 3 | Chinese 4 | Chinese 5 | Chinese 6 | USA 7 | UK 8 | France 9 | Italy -------------------------------------------------------------------------------- /integration/spark/src/test/resources/mac.csv: -------------------------------------------------------------------------------- 1 | MAC165 2 | MAC168 3 | MAC171 4 | MAC174 5 | MAC177 6 | MAC180 7 | MAC183 8 | MAC186 9 | MAC189 10 | MAC262 11 | MAC263 12 | MAC265 13 | MAC266 14 | MAC268 15 | MAC269 16 | MAC271 17 | MAC272 18 | MAC274 19 | MAC275 20 | MAC277 21 | MAC278 22 | MAC280 23 | MAC281 24 | MAC283 25 | MAC284 26 | MAC285 27 | MAC286 28 | MAC287 29 | MAC288 30 | MAC289 31 | MAC290 32 | MAC291 33 | MAC292 34 | MAC292 35 | MAC293 36 | MAC294 37 | MAC295 38 | MAC296 39 | MAC400 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/measureinsertintotest.csv: -------------------------------------------------------------------------------- 1 | id,name,city,age 2 | 1,david,shenzhen,31 3 | 2,eason,shenzhen,27 4 | 3,jarry,wuhan,35 5 | 3,jarry,Bangalore,35 6 | 4,kunal,Delhi,26 7 | 4,vishal,Bangalore,29 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/mobileimei.csv: -------------------------------------------------------------------------------- 1 | 1AA100064 2 | 1AA100065 3 | 1AA100066 4 | 1AA100067 5 | 1AA100068 6 | 1AA100069 7 | 1AA10007 8 | 1AA100070 9 | 1AA100071 10 | 1AA100072 11 | 1AA100073 12 | 1AA100074 13 | 1AA100075 14 | 1AA100076 15 | 1AA100077 16 | 1AA100078 17 | 1AA100079 18 | 1AA10008 19 | 1AA100080 20 | 1AA100081 21 | 1AA100082 22 | 1AA100083 23 | 1AA100084 24 | 1AA100000 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/mv_sampledata.csv: -------------------------------------------------------------------------------- 1 | empno,empname,projectcode,projectjoindate,salary 2 | 11,joey,2,2016-02-23 09:01:30,300 3 | 12,chandler,5,2016-02-23 09:01:50,400 4 | 13,pheobe,1,2016-02-23 09:03:30,450 5 | 14,monica,5,2016-02-23 09:03:50,650 6 | 15,ross,5,2016-02-23 09:07:50,250.3 7 | 16,rachel,1,2016-02-23 09:08:30,300 8 | 17,gunther,8,2016-02-23 09:08:40,800.2 9 | 18,tag,5,2016-02-23 09:16:50,200 10 | 19,will,1,2016-03-23 09:17:30,200 11 | 20,akash,8,2016-03-23 10:18:40,200.2 12 | 21,smith,5,2016-03-29 10:02:50,150.6 13 | 22,cathy,1,2016-02-25 10:03:30,450.5 14 | 23,pablo,5,2016-04-23 11:06:50,350 15 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/newsample.csv: -------------------------------------------------------------------------------- 1 | id,name,time 2 | 1,'one',2014-01-01 08:00:00 3 | 2,'two',2014-01-02 08:02:00 4 | 3,'three',2014-01-03 08:03:00 5 | 4,'four',2014-01-04 08:04:00 6 | null,null,null 7 | 5,'five',2020-10-20 08:01:20 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/noneCsvFormat.cs: -------------------------------------------------------------------------------- 1 | Id,number,name,gamePoint,mac 2 | 1,1.5,Mark,1.2$2,3 3 | 2,2,Twin,2.0$3,1.5 4 | 3,3.0,Betty,5$2.0,2 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/nontransactional.csv: -------------------------------------------------------------------------------- 1 | name, age, height 2 | arvind, 33, 6.2 3 | bill, 35, 7.3 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/nontransactional1.csv: -------------------------------------------------------------------------------- 1 | arvind, 33, 6.2 2 | bill, 35, 7.3 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/nullSample.csv: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/integration/spark/src/test/resources/nullSample.csv -------------------------------------------------------------------------------- /integration/spark/src/test/resources/nullandnonparsableValue.csv: -------------------------------------------------------------------------------- 1 | 1,2015-17-23 00:00:00,china,aaa1,phone197,ASD69643,15000.43525 2 | 2,2015-7-23 00:00:00,china,aaa1,phone197,ASD69643,15000.43525fd 3 | 3, 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/nullmeasurevalue.csv: -------------------------------------------------------------------------------- 1 | ID,date,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,china,aaa1,phone197,ASD69643, 3 | 2,2015/7/24,china,aaa2,phone756,ASD42892, 4 | 3,2015/7/25,china,aaa3,phone1904,ASD37014, 5 | 4,2015/7/26,china,aaa4,phone2435,ASD66902, 6 | 5,2015/7/27,china,aaa5,phone2441,ASD90633, 7 | 6,2015/7/28,china,aaa6,phone294,ASD59961, 8 | 7,2015/7/29,china,aaa7,phone610,ASD14875, 9 | 8,2015/7/30,china,aaa8,phone1848,ASD57308, 10 | 9,2015/7/18,china,aaa9,phone706,ASD86717, 11 | 10,2015/7/19,usa,aaa10,phone685,ASD30505, -------------------------------------------------------------------------------- /integration/spark/src/test/resources/nullvalueserialization.csv: -------------------------------------------------------------------------------- 1 | 1,2015-17-23 00:00:00,china,aaa1,phone197,ASD69643,15000.43525 2 | 2,\N,\N,\N,\N,\N,\N 3 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/numeric_column_invalid_values.csv: -------------------------------------------------------------------------------- 1 | 1,Pallavi,25 2 | 2,Rahul,24 3 | 3,Prabhat,twenty six 4 | 7,Neha,25 5 | 2,Geetika,22 6 | 3,Sangeeta,26 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/outofrange.csv: -------------------------------------------------------------------------------- 1 | column1,column2,column3,column4,column5,column6 2 | aa,bb,2147483648,2147483648,2147483648,2147483648 3 | x,y,-2147483648,-2147483648,-2147483648,-2147483648 4 | xx,yy,-2147483648,9223372036854775808,-2147483648,9223372036854775808 5 | c,d,-2147483648,-9223372036854775808,-2147483648,-9223372036854775808 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/overwriteTable1_noRecord.csv: -------------------------------------------------------------------------------- 1 | id,name,salary 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/overwriteTable1_someRecord.csv: -------------------------------------------------------------------------------- 1 | id,name,salary 2 | 1,hello,2300 3 | 2,hi,2500 4 | 3,xyz,4000 5 | 4,xyz1,5000 6 | 5,xyz2,6000 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/overwriteTable2_noRecord.csv: -------------------------------------------------------------------------------- 1 | id,name,salary,age -------------------------------------------------------------------------------- /integration/spark/src/test/resources/overwriteTable2_someRecord.csv: -------------------------------------------------------------------------------- 1 | id,name,salary,age 2 | 9,abc,48,20 3 | 10,abc1,90,21 4 | 11,abc2,99,22 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/partData.csv: -------------------------------------------------------------------------------- 1 | 9000,CUST_NAME_00000,ACTIVE_EMUI_VERSION_00000,1970-01-01 01:00:03,1970-01-01 02:00:03,123372036854,-223372036854,12345678901.1234000000,22345678901.1234000000,11234567489.7976000000,-11234567489.7976000000,1 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/predefdic/allpredefdictionary.csv: -------------------------------------------------------------------------------- 1 | 1,phone756 2 | 1,phonetype 3 | 1,phone757 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/predefdic/data3.csv: -------------------------------------------------------------------------------- 1 | ID,phonetype 2 | 1,phone197 3 | 2,phone756 4 | 3,phone757 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/predefdic/dicfilepath.csv: -------------------------------------------------------------------------------- 1 | phone756 2 | phone757 3 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/products.csv: -------------------------------------------------------------------------------- 1 | product,amount 2 | Mobile,2000 3 | Laptop,3000 4 | Kettle,70 5 | Washing Machine,1000 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/range_column/dataskew.csv: -------------------------------------------------------------------------------- 1 | 1,abc_1001 2 | 2,null 3 | 3,abc_1001 4 | 4,null 5 | 5,abc_1005 6 | 6,null 7 | 7,abc_1007 8 | 8,null 9 | 9,null 10 | 10,abc_1001 11 | 11,abc_1001 12 | 12,null 13 | 13,abc_1003 14 | 14,null 15 | 15,abc_1005 16 | 16,null 17 | 17,abc_1001 18 | 18,null 19 | 19,null 20 | 20,null -------------------------------------------------------------------------------- /integration/spark/src/test/resources/rangedata.csv: -------------------------------------------------------------------------------- 1 | empname,empno,designation,doj,workgroupcategory,workgroupcategoryname,deptno,deptname,projectcode,projectjoindate,projectenddate,attendance,utilization,salary 2 | arvind,11,SE,17-01-2007,1,developer,10,network,928478,17-02-2007,29-11-2016,96,96,5040 3 | krithin,12,SSE,29-05-2008,1,developer,11,protocol,928378,29-06-2008,30-12-2016,85,95,7124 4 | madhan,13,TPL,7/7/2009,2,tester,10,network,928478,7/8/2009,30-12-2016,88,99,9054 5 | anandh,14,SA,29-12-2010,3,manager,11,protocol,928278,29-01-2011,29-06-2016,77,92,11248 6 | ayushi,15,SSA,9/11/2011,1,developer,12,security,928375,9/12/2011,29-05-2016,99,91,13245 7 | pramod,16,SE,14-10-2012,1,developer,13,configManagement,928478,14-11-2012,29-12-2016,86,93,5040 8 | gawrav,17,PL,22-09-2013,2,tester,12,security,928778,22-10-2013,15-11-2016,78,97,9574 9 | sibi,18,TL,15-08-2014,2,tester,14,Learning,928176,15-09-2014,29-05-2016,84,98,7245 10 | shivani,19,PL,12/5/2015,1,developer,10,network,928977,12/6/2015,12/11/2016,88,91,11254 11 | bill,20,PM,1/12/2015,3,manager,14,Learning,928479,1/1/2016,30-11-2016,75,94,13547 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/rangedatasample.csv: -------------------------------------------------------------------------------- 1 | empno,doj,salary 2 | 11,2016-03-14 15:00:09,5040 3 | 12,2016-03-14 15:00:10,1040 4 | 13,2016-03-14 15:00:11,1041 5 | 14,2016-03-14 15:00:12,1042 6 | 15,2016-03-14 15:00:13,1043 7 | 16,2016-03-14 15:00:14,1044 8 | 17,2016-03-14 15:00:15,1045 9 | 18,2016-03-14 15:00:16,1046 10 | 19,2016-03-14 15:00:17,1047 11 | 20,2016-03-14 15:00:18,1048 12 | 21,2016-03-14 15:00:20,5040 13 | 22,2016-03-14 15:00:25,1040 14 | 23,2016-03-14 15:00:31,1041 15 | 24,2016-03-14 15:00:38,1042 16 | 25,2016-03-14 15:00:39,1043 17 | 26,2016-03-14 15:00:19,1044 18 | 27,2016-03-14 15:00:49,1045 19 | 28,2016-03-14 15:00:50,1046 20 | 29,2016-03-14 15:00:24,1047 21 | 30,2016-03-14 15:00:35,1048 22 | 31,,1040 23 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/restructure/data1.csv: -------------------------------------------------------------------------------- 1 | 100,spark,abc,23-04-2015,21.23 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/restructure/data2.csv: -------------------------------------------------------------------------------- 1 | 101,spark1,2016-04-23 12:01:01,312.23,def 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/restructure/data3.csv: -------------------------------------------------------------------------------- 1 | 102,spark2,2017-04-23 12:01:01,22.2722,mkg 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/restructure/data4.csv: -------------------------------------------------------------------------------- 1 | 104,spark4,2018-04-23 12:01:01,411.23 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/restructure/data5.csv: -------------------------------------------------------------------------------- 1 | 106,spark6,pqr,27.13,2004-04-23 12:01:01 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/restructure/data6.csv: -------------------------------------------------------------------------------- 1 | 7,hello1 2 | 8,welcome1 3 | bye,11 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/restructure/data7.csv: -------------------------------------------------------------------------------- 1 | spark1,abc 2 | spark2,pqr -------------------------------------------------------------------------------- /integration/spark/src/test/resources/sales_data.csv: -------------------------------------------------------------------------------- 1 | product,quantity 2 | Mobile,1 3 | Laptop,10 4 | Chocolates,200 5 | Biscuits,800 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/sample: -------------------------------------------------------------------------------- 1 | id,name,city,age 2 | 1,david,shenzhen,31 3 | 2,eason,shenzhen,27 4 | 3,jarry,wuhan,35 5 | 3,jarry,Bangalore,35 6 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/sample.csv: -------------------------------------------------------------------------------- 1 | id,name,city,age 2 | 1,david,shenzhen,31 3 | 2,eason,shenzhen,27 4 | 3,jarry,wuhan,35 5 | 3,jarry,Bangalore,35 6 | 4,kunal,Delhi,26 7 | 4,vishal,Bangalore,29 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/sample.csv.bz2: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/integration/spark/src/test/resources/sample.csv.bz2 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/sample.csv.gz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/integration/spark/src/test/resources/sample.csv.gz -------------------------------------------------------------------------------- /integration/spark/src/test/resources/sampleComplex.csv: -------------------------------------------------------------------------------- 1 | Id,number,name,gamePoint,mac 2 | 1,1.5,Mark,1.2$2,3 3 | 2,2,Twin,2.0$3,1.5 4 | 3,3.0,Betty,5$2.0,2 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/sample_withDelimiter017.csv: -------------------------------------------------------------------------------- 1 | idnamecityage 2 | 1davidshenzhen31 3 | 2easonshenzhen27 4 | 3jarrywuhan35 5 | 3jarryBangalore35 6 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/IUD/sample_2.csv: -------------------------------------------------------------------------------- 1 | 5,china 2 | 5,china 3 | 5,china 4 | 5,china 5 | 5,china 6 | 5,china 7 | 5,china 8 | 5,china 9 | 5,china 10 | 5,china 11 | 5,china 12 | 5,china 13 | 5,china 14 | 5,china 15 | 5,china 16 | 5,china 17 | 5,china 18 | 5,china 19 | 5,china 20 | 5,china 21 | 5,china 22 | 5,china 23 | 5,china 24 | 5,china 25 | 5,china 26 | 5,china 27 | 5,china 28 | 5,china 29 | 5,china 30 | 5,china 31 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/array.csv: -------------------------------------------------------------------------------- 1 | 1,abc,china$india$us 2 | 2,xyz,sri$can 3 | 3,mno,rus$china 4 | 4,lok,hk$bang 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/array2.csv: -------------------------------------------------------------------------------- 1 | 1,abc,china$india$us,hello$world 2 | 2,xyz,sri$can,iron$man$jarvis 3 | 3,mno,rus$china,ex$ex2 4 | 4,lok,hk$bang,ex$ex3 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/dest.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,1,aa,aaa 3 | b,2,bb,bbb 4 | c,3,cc,ccc 5 | d,4,dd,ddd 6 | e,5,ee,eee 7 | f,6,ff,fff 8 | g,7,gg,ggg 9 | h,8,hh,hhh 10 | i,9,ii,iii 11 | j,10,jj,jjj 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/dest1.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,11,aa,aaa 3 | b,12,bb,bbb 4 | c,13,cc,ccc 5 | d,14,dd,ddd 6 | e,15,ee,eee 7 | f,16,ff,fff 8 | g,17,gg,ggg 9 | h,18,hh,hhh 10 | i,19,ii,iii 11 | j,20,jj,jjj 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/dest2.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,21,aa,aaa 3 | b,22,bb,bbb 4 | c,23,cc,ccc 5 | d,24,dd,ddd 6 | e,25,ee,eee 7 | f,26,ff,fff 8 | g,27,gg,ggg 9 | h,28,hh,hhh 10 | i,29,ii,iii 11 | j,30,jj,jjj 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/dest3.csv: -------------------------------------------------------------------------------- 1 | c1,c2,c3,c5 2 | a,31,aa,aaa 3 | b,32,bb,bbb 4 | c,33,cc,ccc 5 | d,34,dd,ddd 6 | e,35,ee,eee 7 | f,36,ff,fff 8 | g,37,gg,ggg 9 | h,38,hh,hhh 10 | i,39,ii,iii 11 | j,40,jj,jjj 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/firstunique.csv: -------------------------------------------------------------------------------- 1 | 1|1|Customer#000000001|IVhzIApeRb ot,c,E|15|25-989-741-2989|711.56|BUILDING|to the even, regular platelets. regular, ironic epitaphs nag e| 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/index.csv: -------------------------------------------------------------------------------- 1 | 1,2,2016-09-10 00:00:00,2015-07-09 00:00:00,hello,world -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/secondaryIndexLikeTest.csv: -------------------------------------------------------------------------------- 1 | ID,country,name,phonetype,serialname 2 | 1,china,aaa1,phone197,A234 3 | 2,china,aaa2,phone756,A453 4 | 3,china,aaa3,phone197,A234 5 | 4,china,aaa4,phone756,A453 6 | 5,china,aaa5,phone197,A234 7 | 6,china,aaa6,phone756,A453 8 | 7,china,aaa7,phone197,A234 9 | 8,china,aaa8,phone756,A453 10 | 9,china,aaa1,phone756,A455 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/secondunique.csv: -------------------------------------------------------------------------------- 1 | 1|1|Customer#000000001|IVhzIApeRb ot,c,E|15|25-989-741-2988|711.56|AUTOMOBILE|to the even, regular platelets. regular, ironic epitaphs nag e| 2 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/secindex/source3.csv: -------------------------------------------------------------------------------- 1 | c11,c22,c33,c55,c66 2 | a,1,MGM,Disco,10 3 | b,2,RGK,Music,8 4 | d,4,YDY,Weather,9 5 | e,5,ZAZ,Election,11 6 | g,7,YTY,Hello,12 7 | h,8,TBT,Yeh,13 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/shortintboundary.csv: -------------------------------------------------------------------------------- 1 | value, value2, name 2 | 0,0,foo 3 | 127,0,bar 4 | 128,0,foo 5 | -127,0,foo 6 | -128,0,bar 7 | 32767,0,bar 8 | -32767,0,bar 9 | 32768,0,bar 10 | -32768,0,bar 11 | 65535,0,bar 12 | -65535,0,bar 13 | 8388606,0,bar 14 | -8388606,0,foo 15 | 8388607,0,bar 16 | -8388607,0,foo 17 | 0,8388608,bar 18 | 0,-8388608,foo 19 | 0,8388609,bar 20 | 0,-8388609,foo -------------------------------------------------------------------------------- /integration/spark/src/test/resources/shortolap.csv: -------------------------------------------------------------------------------- 1 | imei0,2147,9279,100.05,100.055,2016-05-01 12:25:36,aa,11 2 | imei1,-2148,-9807,10.05,100.05,2016-05-02 19:25:15,bb,22 3 | imei2,2147,9279,100.05,100.055,2016-05-01 12:25:36,cc,33 4 | imei3,-217,-9206,100.005,100.05,2016-05-02 19:25:15,dd,44 5 | imei4,10,0,15.5,45,2016-05-02 19:25:15,ee,55 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/streamSample.csv: -------------------------------------------------------------------------------- 1 | id,name,city,salary,tax,percent,birthday,register,updated,file 2 | 100000001,batch_1,city_1,0.1,0.01,80.01,1990-01-01,2010-01-01 10:01:01,2010-01-01 10:01:01,school_1:school_11$20 3 | 100000002,batch_2,city_2,0.2,0.02,80.02,1990-01-02,2010-01-02 10:01:01,2010-01-02 10:01:01,school_2:school_22$30 4 | 100000003,batch_3,city_3,0.3,0.03,80.03,1990-01-03,2010-01-03 10:01:01,2010-01-03 10:01:01,school_3:school_33$40 5 | 100000004,batch_4,city_4,0.4,0.04,80.04,1990-01-04,2010-01-04 10:01:01,2010-01-04 10:01:01,school_4:school_44$50 6 | 100000005,batch_5,city_5,0.5,0.05,80.05,1990-01-05,2010-01-05 10:01:01,2010-01-05 10:01:01,school_5:school_55$60 7 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/struct_all.csv: -------------------------------------------------------------------------------- 1 | 10&10$10&10 2 | 20&20$20&20 3 | 30&30$30&30 4 | 40&40$40&40 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/structusingstruct.csv: -------------------------------------------------------------------------------- 1 | def$klm&abc$12 2 | pri$sac&pra$18 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/temp/data1.csv: -------------------------------------------------------------------------------- 1 | id,name,rating 2 | 1,xyz,5 3 | 2,ghj,2 4 | 3,ghj,3 5 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/test.json: -------------------------------------------------------------------------------- 1 | {"name":"Michael"} 2 | {"name":"Andy", "age":30} 3 | {"name":"Justin", "age":19} 4 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/testBigInt_boundary_value.csv: -------------------------------------------------------------------------------- 1 | Invalid values,92233720368547758071234 2 | Invalid values,def 3 | All_null_values,null 4 | All_zeros_values,0 5 | Max_range_values,9223372036854775807 6 | Max_range_values,9223372036854775807 7 | Max_range_values,9223372036854775807 8 | Max_range_values,9223372036854775807 9 | Min_range_values,-9223372036854775808 10 | Min_range_values,-9223372036854775807 11 | Min_range_values,-9223372036854775806 12 | Min_range_values,-9223372036854775805 13 | Normal_values,2345 14 | Normal_values,1234 15 | Normal_values,3456 16 | Normal_values,4567 17 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/testShortAndIntDataType.csv: -------------------------------------------------------------------------------- 1 | 9223372,abc,1 2 | 900719925,def,2 3 | 9007199254,ghi,32768 4 | -2147483648,jkl,9.22337203685478E+018 5 | 9.22337203685478E+018,lmn,32767 6 | 2147483647,opq,05 7 | 9.22337203685478E+018,rst,9.22337203685478E+018 8 | -2147483647,uvw,7 9 | 9.22337203685478E+018,xyz,-32767 10 | 2147483648,fgh,-32768 11 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/test_json.json: -------------------------------------------------------------------------------- 1 | {"username":"cust","age":20,"phone":"5367288","housenum":"A780","address": {"Address_Detail": {"Building_Detail": {"Society_name":"xxxyy","building_no":"A780","house_no":4,"Building_Type": {"Buildingname":"abcd","buildingarea":100.3,"Building_Criteria":{"f1": {"username1":"cust0","age1":20,"phone1":682973,"housenum1":"A899","address1":{"Address_Detail1": {"Building_Detail1": {"Society_name1":"xxxyy","building_no1":"A780","house_no1":4,"Building_Type1": {"Buildingname1":"abcd","buildingarea1":100.3,"Building_Criteria1":{"f11": 33,"f21":{"inner1":{"children":[]}}},"AR11":["abc","gdf","ehf"],"AR21":[3,4,5],"AR31":[27735,7981366,9873262],"AR41":[9.2436482,686263.09,3.48249824],"AR51":[9.463462333333E7,47.398759828E5,9.2846892E4],"AR61":[true,true]}}}}}},"f2":"er","AR1":["abc","gdf","ehf"],"AR2":[3,4,5],"AR3":[27735,7981366,9873262],"AR4":[9.2436482,686263.09,3.48249824],"AR5":[9.463462333333E7,47.398759828E5,9.2846892E4],"AR6":[true,true]}}}}} -------------------------------------------------------------------------------- /integration/spark/src/test/resources/timeStampFormatData1.csv: -------------------------------------------------------------------------------- 1 | ID,date,starttime,country,name,phonetype,serialname,salary 2 | 1,2015/7/23,2016-7-23 01:01:30,china,aaa1,phone197,ASD69643,15000 3 | 2,2015/7/24,2016-7-24 01:02:30,china,aaa2,phone756,ASD42892,15001 4 | 3,2015/7/25,2016-7-25 01:03:30,china,aaa3,phone1904,ASD37014,15002 5 | 4,2015/7/26,2016-7-26 01:04:30,china,aaa4,phone2435,ASD66902,15003 6 | 5,2015/7/27,2016-7-27 01:05:30,china,aaa5,phone2441,ASD90633,15004 7 | 6,2015/7/28,2016-7-28 01:06:30,china,aaa6,phone294,ASD59961,15005 8 | 7,2015/7/29,2016-7-29 01:07:30,china,aaa7,phone610,ASD14875,15006 9 | 8,2015/7/30,2016-7-30 01:08:30,china,aaa8,phone1848,ASD57308,15007 10 | 9,2015/7/18,2016-7-18 01:09:30,china,aaa9,phone706,ASD86717,15008 11 | 10,2015/7/19,2016-7-19 01:10:30,usa,aaa10,phone685,ASD30505,15009 12 | 13 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/timeStampFormatData2.csv: -------------------------------------------------------------------------------- 1 | ID,date,starttime,country,name,phonetype,serialname,salary 2 | 11,2015-7-18,2016/07/18 02:02:02,china,aaa11,phone1554,ASD26101,15010 3 | 12,2015-7-19,2016/07/19 02:12:02,china,aaa12,phone1781,ASD85711,15011 4 | 13,2015-7-20,2016/07/20 02:22:02,china,aaa13,phone943,ASD39200,15012 5 | 14,2015-7-21,2016/07/21 02:32:02,china,aaa14,phone1954,ASD80468,15013 6 | 15,2015-7-22,2016/07/22 02:42:02,china,aaa15,phone451,ASD1954,15014 7 | 16,2015-7-23,2016/07/23 02:52:02,china,aaa16,phone390,ASD38513,15015 8 | 17,2015-7-24,2016/07/24 02:42:02,china,aaa17,phone1929,ASD86213,15016 9 | 18,2015-7-25,2016/07/25 02:32:02,usa,aaa18,phone910,ASD88812,15017 10 | 19,2015-7-26,2016/07/26 02:22:02,china,aaa19,phone2151,ASD9316,15018 11 | 20,2015-7-27,2016/07/27 02:12:02,china,aaa20,phone2625,ASD62597,15019 12 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/timeseriestest.csv: -------------------------------------------------------------------------------- 1 | mytime,name,age 2 | 2016-2-23 09:01:30,vishal,10 3 | 2016-2-23 09:01:40,kunal,20 4 | 2016-2-23 09:01:50,shahid,30 5 | 2016-2-23 09:02:30,kk,40 6 | 2016-2-23 09:02:40,rahul,50 7 | 2016-2-23 09:02:50,ravi,50 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/timestampdata.csv: -------------------------------------------------------------------------------- 1 | imei001,rat$aaa1,111$111,2015-01-01 13:00:00.000$2015-01-01 13:00:00.000,16$64,babu$001 2 | imei002,rat$aaa2,111$112,2015-01-01 13:00:00.000$2015-01-02 13:00:00.000,16$65,babu$002 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/timestampdatafile.csv: -------------------------------------------------------------------------------- 1 | timestamptype 2 | 2018-09-11 00:00:00 3 | 2018-09-12 00:00:00 4 | 2018-09-13 00:00:00 5 | 2018-09-14 00:00:00 6 | 2018-09-15 00:00:00 7 | 2018-09-16 00:00:00 8 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/tpch/region.csv: -------------------------------------------------------------------------------- 1 | 0|AFRICA|lar deposits. blithely final packages cajole. regular waters are final requests. regular accounts are according to | 2 | 1|AMERICA|hs use ironic, even requests. s| 3 | 2|ASIA|ges. thinly even pinto beans ca| 4 | 3|EUROPE|ly final courts cajole furiously final excuse| 5 | 4|MIDDLE EAST|uickly special accounts cajole carefully blithely close requests. carefully final asymptotes haggle furiousl| 6 | -------------------------------------------------------------------------------- /integration/spark/src/test/resources/uniq.csv: -------------------------------------------------------------------------------- 1 | name,double_column 2 | a,11234567489.7976 3 | b,11234567489.7976000000 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/uniqwithoutheader.csv: -------------------------------------------------------------------------------- 1 | a,11234567489.7976 2 | b,11234567489.7976000000 -------------------------------------------------------------------------------- /integration/spark/src/test/resources/verticalDelimitedData.csv: -------------------------------------------------------------------------------- 1 | 100044|1232|1697 2 | 100045|1602|1234 3 | 100047|8392|1823 4 | 100048|1184|1234 5 | 100051|2320|1407 6 | 100052|2300|845 7 | 100053|1210|1655 8 | 100054|1689|1368 9 | 100055|2823|1728 10 | 100056|68|750 11 | 100057|716|2288 12 | 100058|864|1234 -------------------------------------------------------------------------------- /integration/spark/src/test/scala/org/apache/spark/sql/common/util/Tags.scala: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.spark.sql.common.util 19 | 20 | import org.scalatest.Tag 21 | 22 | object Include extends Tag("Include") 23 | 24 | object Exclude extends Tag("Exclude") 25 | -------------------------------------------------------------------------------- /processing/CARBON_PROCESSINGLogResource.properties: -------------------------------------------------------------------------------- 1 | # 2 | # Licensed to the Apache Software Foundation (ASF) under one 3 | # or more contributor license agreements. See the NOTICE file 4 | # distributed with this work for additional information 5 | # regarding copyright ownership. The ASF licenses this file 6 | # to you under the Apache License, Version 2.0 (the 7 | # "License"); you may not use this file except in compliance 8 | # with the License. You may obtain a copy of the License at 9 | # 10 | # http://www.apache.org/licenses/LICENSE-2.0 11 | # 12 | # Unless required by applicable law or agreed to in writing, software 13 | # distributed under the License is distributed on an "AS IS" BASIS, 14 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 15 | # See the License for the specific language governing permissions and 16 | # limitations under the License. 17 | # 18 | carbon.processing = {0} -------------------------------------------------------------------------------- /processing/src/main/java/org/apache/carbondata/processing/index/IndexWriterException.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.processing.index; 19 | 20 | public class IndexWriterException extends RuntimeException { 21 | public IndexWriterException(Throwable cause) { 22 | super(cause); 23 | } 24 | } 25 | -------------------------------------------------------------------------------- /processing/src/main/java/org/apache/carbondata/processing/loading/sort/unsafe/holder/UnsafeCarbonRow.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.processing.loading.sort.unsafe.holder; 19 | 20 | public class UnsafeCarbonRow { 21 | 22 | public long address; 23 | 24 | } 25 | -------------------------------------------------------------------------------- /processing/src/main/java/org/apache/carbondata/processing/loading/sort/unsafe/holder/UnsafeCarbonRowForMerge.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.processing.loading.sort.unsafe.holder; 19 | 20 | public class UnsafeCarbonRowForMerge extends UnsafeCarbonRow { 21 | 22 | public byte index; 23 | 24 | } 25 | -------------------------------------------------------------------------------- /processing/src/main/java/org/apache/carbondata/processing/store/messages/messages_en_US.properties: -------------------------------------------------------------------------------- 1 | # 2 | # Licensed to the Apache Software Foundation (ASF) under one 3 | # or more contributor license agreements. See the NOTICE file 4 | # distributed with this work for additional information 5 | # regarding copyright ownership. The ASF licenses this file 6 | # to you under the Apache License, Version 2.0 (the 7 | # "License"); you may not use this file except in compliance 8 | # with the License. You may obtain a copy of the License at 9 | # 10 | # http://www.apache.org/licenses/LICENSE-2.0 11 | # 12 | # Unless required by applicable law or agreed to in writing, software 13 | # distributed under the License is distributed on an "AS IS" BASIS, 14 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 15 | # See the License for the specific language governing permissions and 16 | # limitations under the License. 17 | # 18 | CarbonDataWriterDialog.Shell.Title=Carbon Data Writer 19 | Store.Location=Store Location 20 | Blocklet.Size=Blocklet Size 21 | Table.Name=Table Name 22 | Max.Node.In.File=Max Node In File 23 | -------------------------------------------------------------------------------- /processing/src/test/resources/csv/csv_with_bom.csv: -------------------------------------------------------------------------------- 1 | 1,2015/7/23,china,aaa1,phone197,ASD69643,15000 2 | 2,2015/7/24,china,aaa2,phone756,ASD42892,15001 3 | 3,2015/7/25,china,aaa3,phone1904,ASD37014,15002 -------------------------------------------------------------------------------- /processing/src/test/resources/csv/csv_with_bom.csv.bz2: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/csv/csv_with_bom.csv.bz2 -------------------------------------------------------------------------------- /processing/src/test/resources/csv/csv_with_bom.csv.gz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/csv/csv_with_bom.csv.gz -------------------------------------------------------------------------------- /processing/src/test/resources/csv/data.csv.bz2: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/csv/data.csv.bz2 -------------------------------------------------------------------------------- /processing/src/test/resources/csv/data.csv.gz: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/csv/data.csv.gz -------------------------------------------------------------------------------- /processing/src/test/resources/csv/data.csv.lz4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/csv/data.csv.lz4 -------------------------------------------------------------------------------- /processing/src/test/resources/csv/data.csv.snappy: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/csv/data.csv.snappy -------------------------------------------------------------------------------- /processing/src/test/resources/input/2col.csv: -------------------------------------------------------------------------------- 1 | "imei","deviceInformationId" 2 | "imei160456","column298686" 3 | "imei12295","column261458" 4 | "imei183681","column261649" 5 | "imei107676","column263862" 6 | "imei54726","column256022" 7 | "imei90131","column230268" 8 | "imei75181","column254159" 9 | "imei93410","column262900" 10 | "imei158496","column250670" 11 | "imei174111","column27019" 12 | -------------------------------------------------------------------------------- /processing/src/test/resources/input/3col.csv: -------------------------------------------------------------------------------- 1 | "imei","deviceInformationId","MAC","deviceInformationId","MAC" 2 | -------------------------------------------------------------------------------- /processing/src/test/resources/input/5col.csv: -------------------------------------------------------------------------------- 1 | "imei","deviceInformationId","MAC","deviceInformationId","MAC","deviceColor","device_backColor" 2 | -------------------------------------------------------------------------------- /processing/src/test/resources/schemas/default/carbon/loadmetadata.metadata: -------------------------------------------------------------------------------- 1 | [{"versionNumber":"1","timestamp":"05-04-2016 19:09:19","loadStatus":"Success","loadName":"0","partitionCount":"0","loadStartTime":"05-04-2016 19:06:19","visibility":"true"}] -------------------------------------------------------------------------------- /processing/src/test/resources/schemas/default/carbon/meta.lock: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/schemas/default/carbon/meta.lock -------------------------------------------------------------------------------- /processing/src/test/resources/schemas/default/carbon/metadata: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/schemas/default/carbon/metadata -------------------------------------------------------------------------------- /processing/src/test/resources/schemas/modifiedTime.mdt: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/processing/src/test/resources/schemas/modifiedTime.mdt -------------------------------------------------------------------------------- /python/__init__.py: -------------------------------------------------------------------------------- 1 | # Licensed to the Apache Software Foundation (ASF) under one or more 2 | # contributor license agreements. See the NOTICE file distributed with 3 | # this work for additional information regarding copyright ownership. 4 | # The ASF licenses this file to You under the Apache License, Version 2.0 5 | # (the "License"); you may not use this file except in compliance with 6 | # the License. You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | 16 | __version__ = '0.1.0' 17 | -------------------------------------------------------------------------------- /python/pycarbon/__init__.py: -------------------------------------------------------------------------------- 1 | # Licensed to the Apache Software Foundation (ASF) under one or more 2 | # contributor license agreements. See the NOTICE file distributed with 3 | # this work for additional information regarding copyright ownership. 4 | # The ASF licenses this file to You under the Apache License, Version 2.0 5 | # (the "License"); you may not use this file except in compliance with 6 | # the License. You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | 16 | __version__ = '0.1.0' 17 | -------------------------------------------------------------------------------- /python/pycarbon/core/Constants.py: -------------------------------------------------------------------------------- 1 | # Licensed to the Apache Software Foundation (ASF) under one or more 2 | # contributor license agreements. See the NOTICE file distributed with 3 | # this work for additional information regarding copyright ownership. 4 | # The ASF licenses this file to You under the Apache License, Version 2.0 5 | # (the "License"); you may not use this file except in compliance with 6 | # the License. You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | 16 | 17 | LOCAL_FILE_PREFIX = "file://" 18 | -------------------------------------------------------------------------------- /python/pycarbon/core/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/core/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/integration/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/integration/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/sdk/Constants.py: -------------------------------------------------------------------------------- 1 | # Licensed to the Apache Software Foundation (ASF) under one or more 2 | # contributor license agreements. See the NOTICE file distributed with 3 | # this work for additional information regarding copyright ownership. 4 | # The ASF licenses this file to You under the Apache License, Version 2.0 5 | # (the "License"); you may not use this file except in compliance with 6 | # the License. You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | 16 | LOCAL_FILE_PREFIX = "file://" 17 | -------------------------------------------------------------------------------- /python/pycarbon/sdk/SDKUtil.py: -------------------------------------------------------------------------------- 1 | # Licensed to the Apache Software Foundation (ASF) under one or more 2 | # contributor license agreements. See the NOTICE file distributed with 3 | # this work for additional information regarding copyright ownership. 4 | # The ASF licenses this file to You under the Apache License, Version 2.0 5 | # (the "License"); you may not use this file except in compliance with 6 | # the License. You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | 16 | 17 | class SDKUtil(object): 18 | def __init__(self): 19 | from jnius import autoclass 20 | self.SDKUtilClass = autoclass('org.apache.carbondata.sdk.file.utils.SDKUtil') 21 | 22 | def readBinary(self, path): 23 | return self.SDKUtilClass.readBinary(path) 24 | -------------------------------------------------------------------------------- /python/pycarbon/sdk/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/sdk/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/.coveragerc: -------------------------------------------------------------------------------- 1 | [run] 2 | omit = 3 | pycarbon/tests/* 4 | pycarbon/*/__init__.py 5 | pycarbon/__init__.py 6 | 7 | -------------------------------------------------------------------------------- /python/pycarbon/tests/core/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/core/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/hello_world/README.md: -------------------------------------------------------------------------------- 1 | external_dataset: generating carbon dataset which has non-unischema (standard carbon schema) 2 | 3 | pycarbon_dataset: generating carbon dataset which has unischema 4 | -------------------------------------------------------------------------------- /python/pycarbon/tests/hello_world/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/hello_world/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/hello_world/dataset_with_normal_schema/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/hello_world/dataset_with_normal_schema/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/hello_world/dataset_with_unischema/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/hello_world/dataset_with_unischema/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/im/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/im/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/im/test.py: -------------------------------------------------------------------------------- 1 | 2 | def print_string(str): 3 | print(str) -------------------------------------------------------------------------------- /python/pycarbon/tests/mnist/README.md: -------------------------------------------------------------------------------- 1 | external_dataset: generating carbon dataset which has non-unischema (standard carbon schema) 2 | 3 | pycarbon_dataset: generating carbon dataset which has unischema 4 | -------------------------------------------------------------------------------- /python/pycarbon/tests/mnist/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/mnist/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/mnist/dataset_with_normal_schema/__init__.py: -------------------------------------------------------------------------------- 1 | # Licensed to the Apache Software Foundation (ASF) under one or more 2 | # contributor license agreements. See the NOTICE file distributed with 3 | # this work for additional information regarding copyright ownership. 4 | # The ASF licenses this file to You under the Apache License, Version 2.0 5 | # (the "License"); you may not use this file except in compliance with 6 | # the License. You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | 16 | DEFAULT_MNIST_DATA_PATH = '/tmp/mnist_external' 17 | -------------------------------------------------------------------------------- /python/pycarbon/tests/mnist/dataset_with_unischema/__init__.py: -------------------------------------------------------------------------------- 1 | # Licensed to the Apache Software Foundation (ASF) under one or more 2 | # contributor license agreements. See the NOTICE file distributed with 3 | # this work for additional information regarding copyright ownership. 4 | # The ASF licenses this file to You under the Apache License, Version 2.0 5 | # (the "License"); you may not use this file except in compliance with 6 | # the License. You may obtain a copy of the License at 7 | # 8 | # http://www.apache.org/licenses/LICENSE-2.0 9 | # 10 | # Unless required by applicable law or agreed to in writing, software 11 | # distributed under the License is distributed on an "AS IS" BASIS, 12 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 | # See the License for the specific language governing permissions and 14 | # limitations under the License. 15 | 16 | DEFAULT_MNIST_DATA_PATH = '/tmp/mnist' 17 | -------------------------------------------------------------------------------- /python/pycarbon/tests/mnist/dataset_with_unischema/tests/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/mnist/dataset_with_unischema/tests/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/carbondatalogo.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/carbondatalogo.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/carbondatalogo2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/carbondatalogo2.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/flowers/10686568196_b1915544a8.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/flowers/10686568196_b1915544a8.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/flowers/10686568196_b1915544a8.txt: -------------------------------------------------------------------------------- 1 | tulips -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/flowers/10712722853_5632165b04.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/flowers/10712722853_5632165b04.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/flowers/10712722853_5632165b04.txt: -------------------------------------------------------------------------------- 1 | daisy -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/flowers/subfolder/10841136265_af473efc60.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/flowers/subfolder/10841136265_af473efc60.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/flowers/subfolder/10841136265_af473efc60.txt: -------------------------------------------------------------------------------- 1 | daisy -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/voc/2007_000027.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/voc/2007_000027.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/voc/2007_000032.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/voc/2007_000032.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/voc/2007_000033.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/voc/2007_000033.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/voc/2007_000039.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/voc/2007_000039.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/voc/2007_000039.xml: -------------------------------------------------------------------------------- 1 | 2 | VOC2012 3 | 2007_000039.jpg 4 | 5 | The VOC2007 Database 6 | PASCAL VOC2007 7 | flickr 8 | 9 | 10 | 500 11 | 375 12 | 3 13 | 14 | 1 15 | 16 | tvmonitor 17 | Frontal 18 | 0 19 | 0 20 | 21 | 156 22 | 89 23 | 344 24 | 279 25 | 26 | 27 | 28 | -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/voc/2009_001444.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/voc/2009_001444.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/voc/2009_001444.xml: -------------------------------------------------------------------------------- 1 | 2 | 2009_001444.jpg 3 | VOC2012 4 | 5 | cat 6 | 7 | 344 8 | 1 9 | 388 10 | 1 11 | 12 | 0 13 | 0 14 | Unspecified 15 | 1 16 | 17 | 1 18 | 19 | 3 20 | 388 21 | 500 22 | 23 | 24 | PASCAL VOC2009 25 | The VOC2009 Database 26 | flickr 27 | 28 | 29 | -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/vocForSegmentationClass/2007_000032.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/vocForSegmentationClass/2007_000032.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/vocForSegmentationClass/2007_000032.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/vocForSegmentationClass/2007_000032.png -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/vocForSegmentationClass/2007_000033.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/vocForSegmentationClass/2007_000033.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/vocForSegmentationClass/2007_000033.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/vocForSegmentationClass/2007_000033.png -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/vocForSegmentationClass/2007_000042.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/vocForSegmentationClass/2007_000042.jpg -------------------------------------------------------------------------------- /python/pycarbon/tests/resources/vocForSegmentationClass/2007_000042.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/resources/vocForSegmentationClass/2007_000042.png -------------------------------------------------------------------------------- /python/pycarbon/tests/sdk/__init__.py: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/python/pycarbon/tests/sdk/__init__.py -------------------------------------------------------------------------------- /python/pycarbon/tests/test.py: -------------------------------------------------------------------------------- 1 | import sys 2 | sys.path.append('../../') 3 | 4 | from pycarbon.tests.im.test import print_string 5 | 6 | if __name__ == '__main__': 7 | str = "hello" 8 | print_string(str) -------------------------------------------------------------------------------- /python/setup.cfg: -------------------------------------------------------------------------------- 1 | [bdist_wheel] 2 | universal = 1 3 | -------------------------------------------------------------------------------- /sdk/sdk/src/main/resources/log4j.properties: -------------------------------------------------------------------------------- 1 | # Root logger option 2 | log4j.rootLogger=INFO,stdout 3 | 4 | 5 | # Redirect log messages to console 6 | log4j.appender.debug=org.apache.log4j.RollingFileAppender 7 | log4j.appender.stdout=org.apache.log4j.ConsoleAppender 8 | log4j.appender.stdout.Target=System.out 9 | log4j.appender.stdout.layout=org.apache.log4j.PatternLayout 10 | log4j.appender.stdout.layout.ConversionPattern=%d{yyyy-MM-dd HH:mm:ss} %-5p %c{1}:%L - %m%n 11 | 12 | -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/NestedMap.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/NestedMap.parquet -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/avro_files/users.avro: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/avro_files/users.avro -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/avro_files/users_2.avro: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/avro_files/users_2.avro -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/avro_files/users_3.avro: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/avro_files/users_3.avro -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/csv_files/primitive_data.csv: -------------------------------------------------------------------------------- 1 | ID,country,name,salary 2 | 1,china,aaa1,15000 3 | 2,china,aaa2,15001 4 | 3,china,aaa3,15002 5 | 4,china,aaa4,15003 6 | 5,china,aaa5,15004 7 | 6,china,aaa6,15005 8 | 7,china,aaa7,15006 9 | 8,china,aaa8,15007 10 | 9,china,aaa9,15008 11 | 10,china,aaa10,15009 12 | -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/csv_files/primitive_data_2.csv: -------------------------------------------------------------------------------- 1 | ID,country,name,salary 2 | 11,china,aaa11,15010 3 | 12,china,aaa12,15011 4 | 13,china,aaa13,15012 5 | 14,china,aaa14,15013 6 | 15,china,aaa15,15014 7 | 16,china,aaa16,15015 8 | 17,china,aaa17,15016 9 | 18,china,aaa18,15017 10 | 19,china,aaa19,15018 11 | 20,china,aaa20,15019 12 | -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/csv_files/primitive_data_3.csv: -------------------------------------------------------------------------------- 1 | ID,country,name,salary 2 | 21,china,aaa21,15020 3 | 22,china,aaa22,15021 4 | 23,china,aaa23,15022 5 | 24,china,aaa24,15023 6 | 25,china,aaa25,15024 7 | 26,china,aaa26,15025 8 | 27,china,aaa27,15026 9 | 28,china,aaa28,15027 10 | 29,china,aaa29,15028 11 | 30,china,aaa30,15029 12 | -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/json_files/allPrimitiveType.json: -------------------------------------------------------------------------------- 1 | { 2 | "stringField": "nihal\"ojha\"", 3 | "intField": 26, 4 | "shortField": 26, 5 | "longField": 1234567, 6 | "doubleField": 23.3333, 7 | "boolField": false, 8 | "dateField": "2019-03-02", 9 | "timeField": "2019-02-12 03:03:34", 10 | "decimalField" : 55.35 11 | } 12 | -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/json_files/allPrimitiveTypeSingleArray.json: -------------------------------------------------------------------------------- 1 | [ 2 | { 3 | "stringField": "ZZ", 4 | "intField": 100, 5 | "shortField": 100, 6 | "longField": 1234567, 7 | "doubleField": 23.3333, 8 | "boolField": false, 9 | "dateField": "2020-03-02", 10 | "timeField": "2020-02-12 03:03:34", 11 | "decimalField": 55.35 12 | } 13 | ] 14 | -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/nested_schema.avro: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/nested_schema.avro -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/orc_files/sample.orc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/orc_files/sample.orc -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/orc_files/sample_2.orc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/orc_files/sample_2.orc -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/orc_files/sample_3.orc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/orc_files/sample_3.orc -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/parquet_files/file1.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/parquet_files/file1.parquet -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/parquet_files/file2.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/parquet_files/file2.parquet -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/parquet_files/file3.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/parquet_files/file3.parquet -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/repeated-schema.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/repeated-schema.parquet -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/testTimestamp.orc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/testTimestamp.orc -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/userdata1.avro: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/userdata1.avro -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/userdata1.parquet: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/userdata1.parquet -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/userdata1_orc: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/userdata1_orc -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/file/weather.avro: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/file/weather.avro -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/carbondatalogo.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/carbondatalogo.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/flowers/10686568196_b1915544a8.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/flowers/10686568196_b1915544a8.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/flowers/10686568196_b1915544a8.txt: -------------------------------------------------------------------------------- 1 | tulips -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/flowers/10712722853_5632165b04.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/flowers/10712722853_5632165b04.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/flowers/10712722853_5632165b04.txt: -------------------------------------------------------------------------------- 1 | daisy -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/flowers/subfolder/10841136265_af473efc60.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/flowers/subfolder/10841136265_af473efc60.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/flowers/subfolder/10841136265_af473efc60.txt: -------------------------------------------------------------------------------- 1 | daisy -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/voc/2007_000027.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/voc/2007_000027.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/voc/2007_000032.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/voc/2007_000032.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/voc/2007_000033.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/voc/2007_000033.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/voc/2007_000039.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/voc/2007_000039.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/voc/2007_000039.xml: -------------------------------------------------------------------------------- 1 | 2 | VOC2012 3 | 2007_000039.jpg 4 | 5 | The VOC2007 Database 6 | PASCAL VOC2007 7 | flickr 8 | 9 | 10 | 500 11 | 375 12 | 3 13 | 14 | 1 15 | 16 | tvmonitor 17 | Frontal 18 | 0 19 | 0 20 | 21 | 156 22 | 89 23 | 344 24 | 279 25 | 26 | 27 | 28 | -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/voc/2009_001444.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/voc/2009_001444.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/voc/2009_001444.xml: -------------------------------------------------------------------------------- 1 | 2 | 2009_001444.jpg 3 | VOC2012 4 | 5 | cat 6 | 7 | 344 8 | 1 9 | 388 10 | 1 11 | 12 | 0 13 | 0 14 | Unspecified 15 | 1 16 | 17 | 1 18 | 19 | 3 20 | 388 21 | 500 22 | 23 | 24 | PASCAL VOC2009 25 | The VOC2009 Database 26 | flickr 27 | 28 | 29 | -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000032.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000032.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000032.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000032.png -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000033.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000033.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000033.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000033.png -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000042.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000042.jpg -------------------------------------------------------------------------------- /sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000042.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/apache/carbondata/1ed4e2634cb26b3a1694588a756d2aa2ca546758/sdk/sdk/src/test/resources/image/vocForSegmentationClass/2007_000042.png -------------------------------------------------------------------------------- /tools/cli/src/main/java/org/apache/carbondata/tool/Command.java: -------------------------------------------------------------------------------- 1 | /* 2 | * Licensed to the Apache Software Foundation (ASF) under one or more 3 | * contributor license agreements. See the NOTICE file distributed with 4 | * this work for additional information regarding copyright ownership. 5 | * The ASF licenses this file to You under the Apache License, Version 2.0 6 | * (the "License"); you may not use this file except in compliance with 7 | * the License. You may obtain a copy of the License at 8 | * 9 | * http://www.apache.org/licenses/LICENSE-2.0 10 | * 11 | * Unless required by applicable law or agreed to in writing, software 12 | * distributed under the License is distributed on an "AS IS" BASIS, 13 | * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 14 | * See the License for the specific language governing permissions and 15 | * limitations under the License. 16 | */ 17 | 18 | package org.apache.carbondata.tool; 19 | 20 | import java.io.IOException; 21 | 22 | import org.apache.commons.cli.CommandLine; 23 | 24 | interface Command { 25 | void run(CommandLine line) throws IOException; 26 | } 27 | --------------------------------------------------------------------------------