Examples with GetColumnStatisticsForTableRequest com.amazonaws.services.glue.model.GetColumnStatisticsForTableRequest used on opensource projects

Example 1 with GetColumnStatisticsForTableRequest

use of com.amazonaws.services.glue.model.GetColumnStatisticsForTableRequest in project alluxio by Alluxio.

the class GlueDatabase method getTable.

@Override
public UdbTable getTable(String tableName, UdbBypassSpec bypassSpec) throws IOException {
    Table table;
    List<Partition> partitions;
    try {
        GetTableRequest tableRequest = new GetTableRequest().withCatalogId(mGlueConfiguration.get(Property.CATALOG_ID)).withDatabaseName(mGlueDbName).withName(tableName);
        table = getClient().getTable(tableRequest).getTable();
        partitions = batchGetPartitions(getClient(), tableName);
        PathTranslator pathTranslator = mountAlluxioPaths(table, partitions, bypassSpec);
        List<Column> partitionColumns;
        if (table.getPartitionKeys() == null) {
            partitionColumns = Collections.emptyList();
        } else {
            partitionColumns = table.getPartitionKeys();
        }
        // Get table parameters
        Map<String, String> tableParameters = table.getParameters() == null ? Collections.emptyMap() : table.getParameters();
        // Get column statistics info for table
        List<String> columnNames = table.getStorageDescriptor().getColumns().stream().map(Column::getName).collect(Collectors.toList());
        GetColumnStatisticsForTableRequest getColumnStatisticsForTableRequest = new GetColumnStatisticsForTableRequest().withCatalogId(mGlueConfiguration.get(Property.CATALOG_ID)).withDatabaseName(mGlueDbName).withTableName(tableName).withColumnNames(columnNames);
        List<ColumnStatisticsInfo> columnStatisticsTableData = new ArrayList<>();
        if (mGlueConfiguration.getBoolean(Property.TABLE_COLUMN_STATISTICS_ENABLE)) {
            columnStatisticsTableData = getTableColumnStatistics(mGlueDbName, tableName, getColumnStatisticsForTableRequest);
        }
        // Get column statistics info for partitions
        // potential expensive call
        Map<String, List<ColumnStatisticsInfo>> statsMap = new HashMap<>();
        if (mGlueConfiguration.getBoolean(Property.PARTITION_COLUMN_STATISTICS_ENABLE)) {
            for (Partition partition : partitions) {
                List<String> partitionValue = partition.getValues();
                if (partitionValue != null) {
                    GetColumnStatisticsForPartitionRequest getColumnStatisticsForPartitionRequest = new GetColumnStatisticsForPartitionRequest().withCatalogId(mGlueConfiguration.get(Property.CATALOG_ID)).withDatabaseName(mGlueDbName).withTableName(tableName).withColumnNames(columnNames).withPartitionValues(partitionValue);
                    String partName = GlueUtils.makePartitionName(partitionColumns, partition.getValues());
                    statsMap.put(partName, getPartitionColumnStatistics(mGlueDbName, tableName, getColumnStatisticsForPartitionRequest));
                }
            }
        }
        PartitionInfo partitionInfo = PartitionInfo.newBuilder().setDbName(mGlueDbName).setTableName(tableName).addAllDataCols(GlueUtils.toProto(table.getStorageDescriptor().getColumns())).setStorage(GlueUtils.toProto(table.getStorageDescriptor(), pathTranslator)).putAllParameters(tableParameters).build();
        Layout layout = Layout.newBuilder().setLayoutType(HiveLayout.TYPE).setLayoutData(partitionInfo.toByteString()).build();
        List<UdbPartition> udbPartitions = new ArrayList<>();
        if (partitionColumns.isEmpty()) {
            PartitionInfo.Builder partitionInfoBuilder = PartitionInfo.newBuilder().setDbName(mGlueDbName).setTableName(tableName).addAllDataCols(GlueUtils.toProto(table.getStorageDescriptor().getColumns())).setStorage(GlueUtils.toProto(table.getStorageDescriptor(), pathTranslator)).setPartitionName(tableName).putAllParameters(tableParameters);
            udbPartitions.add(new GluePartition(new HiveLayout(partitionInfoBuilder.build(), Collections.emptyList())));
        } else {
            for (Partition partition : partitions) {
                String partName = GlueUtils.makePartitionName(partitionColumns, partition.getValues());
                PartitionInfo.Builder partitionInfoBuilder = PartitionInfo.newBuilder().setDbName(mGlueDbName).setTableName(tableName).addAllDataCols(GlueUtils.toProto(partition.getStorageDescriptor().getColumns())).setStorage(GlueUtils.toProto(partition.getStorageDescriptor(), pathTranslator)).setPartitionName(partName).putAllParameters(partition.getParameters() == null ? Collections.emptyMap() : partition.getParameters());
                if (partition.getValues() != null) {
                    partitionInfoBuilder.addAllValues(partition.getValues());
                }
                udbPartitions.add(new GluePartition(new HiveLayout(partitionInfoBuilder.build(), statsMap.getOrDefault(partName, Collections.emptyList()))));
            }
        }
        return new GlueTable(this, pathTranslator, tableName, GlueUtils.toProtoSchema(table.getStorageDescriptor().getColumns()), columnStatisticsTableData, // Get FieldSchema from partition keys
        GlueUtils.toProto(table.getPartitionKeys()), udbPartitions, layout, table);
    } catch (EntityNotFoundException e) {
        throw new NotFoundException("Table " + tableName + " does not exist in Database: " + mGlueDbName + "; Catalog ID: " + mGlueConfiguration.get(Property.CATALOG_ID) + ".", e);
    } catch (ValidationException e) {
        throw new IOException("Failed to get table: " + tableName + " in Database: " + mGlueDbName + "; Catalog ID: " + mGlueConfiguration.get(Property.CATALOG_ID) + " with validation error: " + e.getMessage(), e);
    } catch (GlueEncryptionException e) {
        throw new IOException("Failed to get table: " + tableName + " in Database: " + mGlueDbName + "; Catalog ID: " + mGlueConfiguration.get(Property.CATALOG_ID) + " error: " + e.getMessage(), e);
    }
}