Examples with AnalyzeInfoProvider org.apache.drill.exec.metastore.analyze.AnalyzeInfoProvider used on opensource projects

Example 1 with AnalyzeInfoProvider

use of org.apache.drill.exec.metastore.analyze.AnalyzeInfoProvider in project drill by apache.

the class MetastoreAnalyzeTableHandler method convertToDrel.

/**
 * Converts to Drill logical plan
 */
private DrillRel convertToDrel(RelNode relNode, SqlMetastoreAnalyzeTable sqlAnalyzeTable, DrillTableInfo drillTableInfo) throws ForemanSetupException, IOException {
    RelBuilder relBuilder = LOGICAL_BUILDER.create(relNode.getCluster(), null);
    DrillTable table = drillTableInfo.drillTable();
    AnalyzeInfoProvider analyzeInfoProvider = table.getGroupScan().getAnalyzeInfoProvider();
    List<String> schemaPath = drillTableInfo.schemaPath();
    String pluginName = schemaPath.get(0);
    String workspaceName = Strings.join(schemaPath.subList(1, schemaPath.size()), AbstractSchema.SCHEMA_SEPARATOR);
    String tableName = drillTableInfo.tableName();
    TableInfo tableInfo = TableInfo.builder().name(tableName).owner(table.getUserName()).type(analyzeInfoProvider.getTableTypeName()).storagePlugin(pluginName).workspace(workspaceName).build();
    ColumnNamesOptions columnNamesOptions = new ColumnNamesOptions(context.getOptions());
    List<String> segmentColumns = analyzeInfoProvider.getSegmentColumns(table, columnNamesOptions).stream().map(SchemaPath::getRootSegmentPath).collect(Collectors.toList());
    List<NamedExpression> segmentExpressions = segmentColumns.stream().map(partitionName -> new NamedExpression(SchemaPath.getSimplePath(partitionName), FieldReference.getWithQuotedRef(partitionName))).collect(Collectors.toList());
    List<MetadataInfo> rowGroupsInfo = Collections.emptyList();
    List<MetadataInfo> filesInfo = Collections.emptyList();
    Multimap<Integer, MetadataInfo> segments = ArrayListMultimap.create();
    BasicTablesRequests basicRequests;
    try {
        basicRequests = context.getMetastoreRegistry().get().tables().basicRequests();
    } catch (MetastoreException e) {
        logger.error("Error when obtaining Metastore instance for table {}", tableName, e);
        DrillRel convertedRelNode = convertToRawDrel(relBuilder.values(new String[] { MetastoreAnalyzeConstants.OK_FIELD_NAME, MetastoreAnalyzeConstants.SUMMARY_FIELD_NAME }, false, e.getMessage()).build());
        return new DrillScreenRel(convertedRelNode.getCluster(), convertedRelNode.getTraitSet(), convertedRelNode);
    }
    MetadataType metadataLevel = getMetadataType(sqlAnalyzeTable);
    List<SchemaPath> interestingColumns = sqlAnalyzeTable.getFieldNames();
    MetastoreTableInfo metastoreTableInfo = basicRequests.metastoreTableInfo(tableInfo);
    List<MetadataInfo> allMetaToHandle = null;
    List<MetadataInfo> metadataToRemove = new ArrayList<>();
    // whether incremental analyze may be produced
    if (metastoreTableInfo.isExists()) {
        RelNode finalRelNode = relNode;
        CheckedSupplier<TableScan, SqlUnsupportedException> tableScanSupplier = () -> DrillRelOptUtil.findScan(convertToDrel(finalRelNode.getInput(0)));
        MetadataInfoCollector metadataInfoCollector = analyzeInfoProvider.getMetadataInfoCollector(basicRequests, tableInfo, (FormatSelection) table.getSelection(), context.getPlannerSettings(), tableScanSupplier, interestingColumns, metadataLevel, segmentColumns.size());
        if (!metadataInfoCollector.isOutdated()) {
            DrillRel convertedRelNode = convertToRawDrel(relBuilder.values(new String[] { MetastoreAnalyzeConstants.OK_FIELD_NAME, MetastoreAnalyzeConstants.SUMMARY_FIELD_NAME }, false, "Table metadata is up to date, analyze wasn't performed.").build());
            return new DrillScreenRel(convertedRelNode.getCluster(), convertedRelNode.getTraitSet(), convertedRelNode);
        }
        // updates scan to read updated / new files, pass removed files into metadata handler
        relNode = relNode.copy(relNode.getTraitSet(), Collections.singletonList(metadataInfoCollector.getPrunedScan()));
        filesInfo = metadataInfoCollector.getFilesInfo();
        segments = metadataInfoCollector.getSegmentsInfo();
        rowGroupsInfo = metadataInfoCollector.getRowGroupsInfo();
        allMetaToHandle = metadataInfoCollector.getAllMetaToHandle();
        metadataToRemove = metadataInfoCollector.getMetadataToRemove();
    }
    // Step 2: constructs plan for producing analyze
    DrillRel convertedRelNode = convertToRawDrel(relNode);
    boolean createNewAggregations = true;
    // List of columns for which statistics should be collected: interesting columns + segment columns
    List<SchemaPath> statisticsColumns = interestingColumns == null ? null : new ArrayList<>(interestingColumns);
    if (statisticsColumns != null) {
        segmentColumns.stream().map(SchemaPath::getSimplePath).forEach(statisticsColumns::add);
    }
    SchemaPath locationField = analyzeInfoProvider.getLocationField(columnNamesOptions);
    if (analyzeInfoProvider.supportsMetadataType(MetadataType.ROW_GROUP) && metadataLevel.includes(MetadataType.ROW_GROUP)) {
        MetadataHandlerContext handlerContext = MetadataHandlerContext.builder().tableInfo(tableInfo).metadataToHandle(rowGroupsInfo).metadataType(MetadataType.ROW_GROUP).depthLevel(segmentExpressions.size()).segmentColumns(segmentColumns).build();
        convertedRelNode = getRowGroupAggRelNode(segmentExpressions, convertedRelNode, createNewAggregations, statisticsColumns, handlerContext);
        createNewAggregations = false;
        locationField = SchemaPath.getSimplePath(MetastoreAnalyzeConstants.LOCATION_FIELD);
    }
    if (analyzeInfoProvider.supportsMetadataType(MetadataType.FILE) && metadataLevel.includes(MetadataType.FILE)) {
        MetadataHandlerContext handlerContext = MetadataHandlerContext.builder().tableInfo(tableInfo).metadataToHandle(filesInfo).metadataType(MetadataType.FILE).depthLevel(segmentExpressions.size()).segmentColumns(segmentColumns).build();
        convertedRelNode = getFileAggRelNode(segmentExpressions, convertedRelNode, createNewAggregations, statisticsColumns, locationField, handlerContext);
        locationField = SchemaPath.getSimplePath(MetastoreAnalyzeConstants.LOCATION_FIELD);
        createNewAggregations = false;
    }
    if (analyzeInfoProvider.supportsMetadataType(MetadataType.SEGMENT) && metadataLevel.includes(MetadataType.SEGMENT)) {
        for (int i = segmentExpressions.size(); i > 0; i--) {
            MetadataHandlerContext handlerContext = MetadataHandlerContext.builder().tableInfo(tableInfo).metadataToHandle(new ArrayList<>(segments.get(i - 1))).metadataType(MetadataType.SEGMENT).depthLevel(i).segmentColumns(segmentColumns.subList(0, i)).build();
            convertedRelNode = getSegmentAggRelNode(segmentExpressions, convertedRelNode, createNewAggregations, statisticsColumns, locationField, i, handlerContext);
            locationField = SchemaPath.getSimplePath(MetastoreAnalyzeConstants.LOCATION_FIELD);
            createNewAggregations = false;
        }
    }
    if (analyzeInfoProvider.supportsMetadataType(MetadataType.TABLE) && metadataLevel.includes(MetadataType.TABLE)) {
        MetadataHandlerContext handlerContext = MetadataHandlerContext.builder().tableInfo(tableInfo).metadataToHandle(Collections.emptyList()).metadataType(MetadataType.TABLE).depthLevel(segmentExpressions.size()).segmentColumns(segmentColumns).build();
        convertedRelNode = getTableAggRelNode(convertedRelNode, createNewAggregations, statisticsColumns, locationField, handlerContext);
    } else {
        throw new IllegalStateException("Analyze table with NONE level");
    }
    boolean useStatistics = context.getOptions().getOption(PlannerSettings.STATISTICS_USE);
    SqlNumericLiteral samplePercentLiteral = sqlAnalyzeTable.getSamplePercent();
    double samplePercent = samplePercentLiteral == null ? 100.0 : samplePercentLiteral.intValue(true);
    // Step 3: adds rel nodes for producing statistics analyze if required
    RelNode analyzeRel = useStatistics ? new DrillAnalyzeRel(convertedRelNode.getCluster(), convertedRelNode.getTraitSet(), convertToRawDrel(relNode), samplePercent) : convertToRawDrel(relBuilder.values(new String[] { "" }, "").build());
    MetadataControllerContext metadataControllerContext = MetadataControllerContext.builder().tableInfo(tableInfo).metastoreTableInfo(metastoreTableInfo).location(((FormatSelection) table.getSelection()).getSelection().getSelectionRoot()).interestingColumns(interestingColumns).segmentColumns(segmentColumns).metadataToHandle(allMetaToHandle).metadataToRemove(metadataToRemove).analyzeMetadataLevel(metadataLevel).build();
    convertedRelNode = new MetadataControllerRel(convertedRelNode.getCluster(), convertedRelNode.getTraitSet(), convertedRelNode, analyzeRel, metadataControllerContext);
    return new DrillScreenRel(convertedRelNode.getCluster(), convertedRelNode.getTraitSet(), convertedRelNode);
}

Example 2 with AnalyzeInfoProvider

use of org.apache.drill.exec.metastore.analyze.AnalyzeInfoProvider in project drill by apache.

the class MetastoreAnalyzeTableHandler method getPlan.

@Override
public PhysicalPlan getPlan(SqlNode sqlNode) throws ValidationException, RelConversionException, IOException, ForemanSetupException {
    if (!context.getOptions().getOption(ExecConstants.METASTORE_ENABLED_VALIDATOR)) {
        throw UserException.validationError().message("Running ANALYZE TABLE REFRESH METADATA command when Metastore is disabled (`metastore.enabled` is set to false)").build(logger);
    }
    // disables during analyze to prevent using data about locations from the Metastore
    context.getOptions().setLocalOption(ExecConstants.METASTORE_ENABLED, false);
    SqlMetastoreAnalyzeTable sqlAnalyzeTable = unwrap(sqlNode, SqlMetastoreAnalyzeTable.class);
    SqlNode tableRef = sqlAnalyzeTable.getTableRef();
    DrillTableInfo drillTableInfo = DrillTableInfo.getTableInfoHolder(tableRef, config);
    AnalyzeInfoProvider analyzeInfoProvider = drillTableInfo.drillTable().getGroupScan().getAnalyzeInfoProvider();
    if (analyzeInfoProvider == null) {
        throw UserException.validationError().message("ANALYZE is not supported for group scan [%s]", drillTableInfo.drillTable().getGroupScan()).build(logger);
    }
    ColumnNamesOptions columnNamesOptions = new ColumnNamesOptions(context.getOptions());
    // creates select with DYNAMIC_STAR column and analyze specific columns to obtain corresponding table scan
    SqlSelect scanSql = new SqlSelect(SqlParserPos.ZERO, SqlNodeList.EMPTY, getColumnList(analyzeInfoProvider.getProjectionFields(drillTableInfo.drillTable(), getMetadataType(sqlAnalyzeTable), columnNamesOptions)), tableRef, null, null, null, null, null, null, null);
    ConvertedRelNode convertedRelNode = validateAndConvert(rewrite(scanSql));
    RelDataType validatedRowType = convertedRelNode.getValidatedRowType();
    RelNode relScan = convertedRelNode.getConvertedNode();
    DrillRel drel = convertToDrel(relScan, sqlAnalyzeTable, drillTableInfo);
    Prel prel = convertToPrel(drel, validatedRowType);
    logAndSetTextPlan("Drill Physical", prel, logger);
    PhysicalOperator pop = convertToPop(prel);
    PhysicalPlan plan = convertToPlan(pop);
    log("Drill Plan", plan, logger);
    return plan;
}