Examples with TrinoException io.trino.spi.TrinoException used on opensource projects

Example 71 with TrinoException

use of io.trino.spi.TrinoException in project trino by trinodb.

the class HiveMetadata method finishInsert.

@Override
public Optional<ConnectorOutputMetadata> finishInsert(ConnectorSession session, ConnectorInsertTableHandle insertHandle, Collection<Slice> fragments, Collection<ComputedStatistics> computedStatistics) {
    HiveInsertTableHandle handle = (HiveInsertTableHandle) insertHandle;
    List<PartitionUpdate> partitionUpdates = fragments.stream().map(Slice::getBytes).map(partitionUpdateCodec::fromJson).collect(toImmutableList());
    HiveStorageFormat tableStorageFormat = handle.getTableStorageFormat();
    partitionUpdates = PartitionUpdate.mergePartitionUpdates(partitionUpdates);
    Table table = metastore.getTable(handle.getSchemaName(), handle.getTableName()).orElseThrow(() -> new TableNotFoundException(handle.getSchemaTableName()));
    if (!table.getStorage().getStorageFormat().getInputFormat().equals(tableStorageFormat.getInputFormat()) && isRespectTableFormat(session)) {
        throw new TrinoException(HIVE_CONCURRENT_MODIFICATION_DETECTED, "Table format changed during insert");
    }
    if (handle.getBucketProperty().isPresent() && isCreateEmptyBucketFiles(session)) {
        List<PartitionUpdate> partitionUpdatesForMissingBuckets = computePartitionUpdatesForMissingBuckets(session, handle, table, false, partitionUpdates);
        // replace partitionUpdates before creating the empty files so that those files will be cleaned up if we end up rollback
        partitionUpdates = PartitionUpdate.mergePartitionUpdates(concat(partitionUpdates, partitionUpdatesForMissingBuckets));
        for (PartitionUpdate partitionUpdate : partitionUpdatesForMissingBuckets) {
            Optional<Partition> partition = table.getPartitionColumns().isEmpty() ? Optional.empty() : Optional.of(buildPartitionObject(session, table, partitionUpdate));
            if (handle.isTransactional() && partition.isPresent()) {
                PartitionStatistics statistics = PartitionStatistics.builder().setBasicStatistics(partitionUpdate.getStatistics()).build();
                metastore.addPartition(session, handle.getSchemaName(), handle.getTableName(), partition.get(), partitionUpdate.getWritePath(), Optional.of(partitionUpdate.getFileNames()), statistics, handle.isRetriesEnabled());
            }
            createEmptyFiles(session, partitionUpdate.getWritePath(), table, partition, partitionUpdate.getFileNames());
        }
    }
    List<String> partitionedBy = table.getPartitionColumns().stream().map(Column::getName).collect(toImmutableList());
    Map<String, Type> columnTypes = handle.getInputColumns().stream().collect(toImmutableMap(HiveColumnHandle::getName, column -> column.getHiveType().getType(typeManager)));
    Map<List<String>, ComputedStatistics> partitionComputedStatistics = createComputedStatisticsToPartitionMap(computedStatistics, partitionedBy, columnTypes);
    for (PartitionUpdate partitionUpdate : partitionUpdates) {
        if (partitionUpdate.getName().isEmpty()) {
            // insert into unpartitioned table
            if (!table.getStorage().getStorageFormat().getInputFormat().equals(handle.getPartitionStorageFormat().getInputFormat()) && isRespectTableFormat(session)) {
                throw new TrinoException(HIVE_CONCURRENT_MODIFICATION_DETECTED, "Table format changed during insert");
            }
            PartitionStatistics partitionStatistics = createPartitionStatistics(partitionUpdate.getStatistics(), columnTypes, getColumnStatistics(partitionComputedStatistics, ImmutableList.of()));
            if (partitionUpdate.getUpdateMode() == OVERWRITE) {
                // get privileges from existing table
                PrincipalPrivileges principalPrivileges = fromHivePrivilegeInfos(metastore.listTablePrivileges(handle.getSchemaName(), handle.getTableName(), Optional.empty()));
                // first drop it
                metastore.dropTable(session, handle.getSchemaName(), handle.getTableName());
                // create the table with the new location
                metastore.createTable(session, table, principalPrivileges, Optional.of(partitionUpdate.getWritePath()), Optional.of(partitionUpdate.getFileNames()), false, partitionStatistics, handle.isRetriesEnabled());
            } else if (partitionUpdate.getUpdateMode() == NEW || partitionUpdate.getUpdateMode() == APPEND) {
                // insert into unpartitioned table
                metastore.finishInsertIntoExistingTable(session, handle.getSchemaName(), handle.getTableName(), partitionUpdate.getWritePath(), partitionUpdate.getFileNames(), partitionStatistics, handle.isRetriesEnabled());
            } else {
                throw new IllegalArgumentException("Unsupported update mode: " + partitionUpdate.getUpdateMode());
            }
        } else if (partitionUpdate.getUpdateMode() == APPEND) {
            // insert into existing partition
            List<String> partitionValues = toPartitionValues(partitionUpdate.getName());
            PartitionStatistics partitionStatistics = createPartitionStatistics(partitionUpdate.getStatistics(), columnTypes, getColumnStatistics(partitionComputedStatistics, partitionValues));
            metastore.finishInsertIntoExistingPartition(session, handle.getSchemaName(), handle.getTableName(), partitionValues, partitionUpdate.getWritePath(), partitionUpdate.getFileNames(), partitionStatistics, handle.isRetriesEnabled());
        } else if (partitionUpdate.getUpdateMode() == NEW || partitionUpdate.getUpdateMode() == OVERWRITE) {
            // insert into new partition or overwrite existing partition
            Partition partition = buildPartitionObject(session, table, partitionUpdate);
            if (!partition.getStorage().getStorageFormat().getInputFormat().equals(handle.getPartitionStorageFormat().getInputFormat()) && isRespectTableFormat(session)) {
                throw new TrinoException(HIVE_CONCURRENT_MODIFICATION_DETECTED, "Partition format changed during insert");
            }
            PartitionStatistics partitionStatistics = createPartitionStatistics(partitionUpdate.getStatistics(), columnTypes, getColumnStatistics(partitionComputedStatistics, partition.getValues()));
            if (partitionUpdate.getUpdateMode() == OVERWRITE) {
                if (handle.getLocationHandle().getWriteMode() == DIRECT_TO_TARGET_EXISTING_DIRECTORY) {
                    removeNonCurrentQueryFiles(session, partitionUpdate.getTargetPath());
                    if (handle.isRetriesEnabled()) {
                        HdfsContext hdfsContext = new HdfsContext(session);
                        cleanExtraOutputFiles(hdfsEnvironment, hdfsContext, session.getQueryId(), partitionUpdate.getTargetPath(), ImmutableSet.copyOf(partitionUpdate.getFileNames()));
                    }
                } else {
                    metastore.dropPartition(session, handle.getSchemaName(), handle.getTableName(), partition.getValues(), true);
                    metastore.addPartition(session, handle.getSchemaName(), handle.getTableName(), partition, partitionUpdate.getWritePath(), Optional.of(partitionUpdate.getFileNames()), partitionStatistics, handle.isRetriesEnabled());
                }
            } else {
                metastore.addPartition(session, handle.getSchemaName(), handle.getTableName(), partition, partitionUpdate.getWritePath(), Optional.of(partitionUpdate.getFileNames()), partitionStatistics, handle.isRetriesEnabled());
            }
        } else {
            throw new IllegalArgumentException(format("Unsupported update mode: %s", partitionUpdate.getUpdateMode()));
        }
    }
    if (isFullAcidTable(table.getParameters())) {
        HdfsContext context = new HdfsContext(session);
        for (PartitionUpdate update : partitionUpdates) {
            long writeId = handle.getTransaction().getWriteId();
            Path deltaDirectory = new Path(format("%s/%s/%s", table.getStorage().getLocation(), update.getName(), deltaSubdir(writeId, writeId, 0)));
            createOrcAcidVersionFile(context, deltaDirectory);
        }
    }
    return Optional.of(new HiveWrittenPartitions(partitionUpdates.stream().map(PartitionUpdate::getName).collect(toImmutableList())));
}

Example 72 with TrinoException

use of io.trino.spi.TrinoException in project trino by trinodb.

the class HiveMetadata method buildColumnDomain.

private static Domain buildColumnDomain(ColumnHandle column, List<HivePartition> partitions) {
    checkArgument(!partitions.isEmpty(), "partitions cannot be empty");
    boolean hasNull = false;
    boolean hasNaN = false;
    List<Object> nonNullValues = new ArrayList<>();
    Type type = ((HiveColumnHandle) column).getType();
    for (HivePartition partition : partitions) {
        NullableValue value = partition.getKeys().get(column);
        if (value == null) {
            throw new TrinoException(HIVE_UNKNOWN_ERROR, format("Partition %s does not have a value for partition column %s", partition, column));
        }
        if (value.isNull()) {
            hasNull = true;
        } else {
            if (isFloatingPointNaN(type, value.getValue())) {
                hasNaN = true;
            }
            nonNullValues.add(value.getValue());
        }
    }
    Domain domain;
    if (nonNullValues.isEmpty()) {
        domain = Domain.none(type);
    } else if (hasNaN) {
        domain = Domain.notNull(type);
    } else {
        domain = Domain.multipleValues(type, nonNullValues);
    }
    if (hasNull) {
        domain = domain.union(Domain.onlyNull(type));
    }
    return domain;
}

Example 73 with TrinoException

use of io.trino.spi.TrinoException in project trino by trinodb.

the class HiveMetadata method getInsertLayout.

@Override
public Optional<ConnectorTableLayout> getInsertLayout(ConnectorSession session, ConnectorTableHandle tableHandle) {
    HiveTableHandle hiveTableHandle = (HiveTableHandle) tableHandle;
    SchemaTableName tableName = hiveTableHandle.getSchemaTableName();
    Table table = metastore.getTable(tableName.getSchemaName(), tableName.getTableName()).orElseThrow(() -> new TableNotFoundException(tableName));
    if (table.getStorage().getBucketProperty().isPresent()) {
        if (!isSupportedBucketing(table)) {
            throw new TrinoException(NOT_SUPPORTED, "Cannot write to a table bucketed on an unsupported type");
        }
    } else // Note: we cannot use hiveTableHandle.isInAcidTransaction() here as transaction is not yet set in HiveTableHandle when getInsertLayout is called
    if (isFullAcidTable(table.getParameters())) {
        table = Table.builder(table).withStorage(storage -> storage.setBucketProperty(Optional.of(new HiveBucketProperty(ImmutableList.of(), HiveBucketing.BucketingVersion.BUCKETING_V2, 1, ImmutableList.of())))).build();
    }
    Optional<HiveBucketHandle> hiveBucketHandle = getHiveBucketHandle(session, table, typeManager);
    List<Column> partitionColumns = table.getPartitionColumns();
    if (hiveBucketHandle.isEmpty()) {
        // return preferred layout which is partitioned by partition columns
        if (partitionColumns.isEmpty()) {
            return Optional.empty();
        }
        return Optional.of(new ConnectorTableLayout(partitionColumns.stream().map(Column::getName).collect(toImmutableList())));
    }
    HiveBucketProperty bucketProperty = table.getStorage().getBucketProperty().orElseThrow(() -> new NoSuchElementException("Bucket property should be set"));
    if (!bucketProperty.getSortedBy().isEmpty() && !isSortedWritingEnabled(session)) {
        throw new TrinoException(NOT_SUPPORTED, "Writing to bucketed sorted Hive tables is disabled");
    }
    ImmutableList.Builder<String> partitioningColumns = ImmutableList.builder();
    hiveBucketHandle.get().getColumns().stream().map(HiveColumnHandle::getName).forEach(partitioningColumns::add);
    partitionColumns.stream().map(Column::getName).forEach(partitioningColumns::add);
    HivePartitioningHandle partitioningHandle = new HivePartitioningHandle(hiveBucketHandle.get().getBucketingVersion(), hiveBucketHandle.get().getTableBucketCount(), hiveBucketHandle.get().getColumns().stream().map(HiveColumnHandle::getHiveType).collect(toImmutableList()), OptionalInt.of(hiveBucketHandle.get().getTableBucketCount()), !partitionColumns.isEmpty() && isParallelPartitionedBucketedWrites(session));
    return Optional.of(new ConnectorTableLayout(partitioningHandle, partitioningColumns.build()));
}

Example 74 with TrinoException

use of io.trino.spi.TrinoException in project trino by trinodb.

the class RcFileFileWriterFactory method createFileWriter.

@Override
public Optional<FileWriter> createFileWriter(Path path, List<String> inputColumnNames, StorageFormat storageFormat, Properties schema, JobConf configuration, ConnectorSession session, OptionalInt bucketNumber, AcidTransaction transaction, boolean useAcidSchema, WriterKind writerKind) {
    if (!RCFileOutputFormat.class.getName().equals(storageFormat.getOutputFormat())) {
        return Optional.empty();
    }
    RcFileEncoding rcFileEncoding;
    if (LazyBinaryColumnarSerDe.class.getName().equals(storageFormat.getSerde())) {
        rcFileEncoding = new BinaryRcFileEncoding(timeZone);
    } else if (ColumnarSerDe.class.getName().equals(storageFormat.getSerde())) {
        rcFileEncoding = createTextVectorEncoding(schema);
    } else {
        return Optional.empty();
    }
    Optional<String> codecName = Optional.ofNullable(configuration.get(FileOutputFormat.COMPRESS_CODEC));
    // existing tables and partitions may have columns in a different order than the writer is providing, so build
    // an index to rearrange columns in the proper order
    List<String> fileColumnNames = getColumnNames(schema);
    List<Type> fileColumnTypes = getColumnTypes(schema).stream().map(hiveType -> hiveType.getType(typeManager, getTimestampPrecision(session))).collect(toList());
    int[] fileInputColumnIndexes = fileColumnNames.stream().mapToInt(inputColumnNames::indexOf).toArray();
    try {
        FileSystem fileSystem = hdfsEnvironment.getFileSystem(session.getIdentity(), path, configuration);
        OutputStream outputStream = fileSystem.create(path, false);
        Optional<Supplier<RcFileDataSource>> validationInputFactory = Optional.empty();
        if (isRcfileOptimizedWriterValidate(session)) {
            validationInputFactory = Optional.of(() -> {
                try {
                    return new HdfsRcFileDataSource(path.toString(), fileSystem.open(path), fileSystem.getFileStatus(path).getLen(), stats);
                } catch (IOException e) {
                    throw new TrinoException(HIVE_WRITE_VALIDATION_FAILED, e);
                }
            });
        }
        Callable<Void> rollbackAction = () -> {
            fileSystem.delete(path, false);
            return null;
        };
        return Optional.of(new RcFileFileWriter(outputStream, rollbackAction, rcFileEncoding, fileColumnTypes, codecName, fileInputColumnIndexes, ImmutableMap.<String, String>builder().put(PRESTO_VERSION_NAME, nodeVersion.toString()).put(PRESTO_QUERY_ID_NAME, session.getQueryId()).buildOrThrow(), validationInputFactory));
    } catch (Exception e) {
        throw new TrinoException(HIVE_WRITER_OPEN_ERROR, "Error creating RCFile file", e);
    }
}

Example 75 with TrinoException

use of io.trino.spi.TrinoException in project trino by trinodb.

the class HivePageSink method getWriterIndexes.

private int[] getWriterIndexes(Page page) {
    Page partitionColumns = extractColumns(page, partitionColumnsInputIndex);
    Block bucketBlock = buildBucketBlock(page);
    int[] writerIndexes = pagePartitioner.partitionPage(partitionColumns, bucketBlock);
    if (pagePartitioner.getMaxIndex() >= maxOpenWriters) {
        throw new TrinoException(HIVE_TOO_MANY_OPEN_PARTITIONS, format("Exceeded limit of %s open writers for partitions/buckets", maxOpenWriters));
    }
    // expand writers list to new size
    while (writers.size() <= pagePartitioner.getMaxIndex()) {
        writers.add(null);
    }
    // create missing writers
    for (int position = 0; position < page.getPositionCount(); position++) {
        int writerIndex = writerIndexes[position];
        HiveWriter writer = writers.get(writerIndex);
        if (writer != null) {
            // and file names have no random component (e.g. bucket_00000)
            if (bucketFunction != null || isTransactional || writer.getWrittenBytes() <= targetMaxFileSize) {
                continue;
            }
            // close current writer
            closeWriter(writer);
        }
        OptionalInt bucketNumber = OptionalInt.empty();
        if (bucketBlock != null) {
            bucketNumber = OptionalInt.of(bucketBlock.getInt(position, 0));
        }
        writer = writerFactory.createWriter(partitionColumns, position, bucketNumber);
        writers.set(writerIndex, writer);
    }
    verify(writers.size() == pagePartitioner.getMaxIndex() + 1);
    verify(!writers.contains(null));
    return writerIndexes;
}