Search in sources :

Example 6 with ImplicitColumnManager

use of org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager in project drill by apache.

the class TestImplicitColumnProjection method testProjectList.

/**
 * Test the file projection planner with metadata.
 */
@Test
public void testProjectList() {
    Path filePath = new Path("hdfs:///w/x/y/z.csv");
    ImplicitColumnManager metadataManager = new ImplicitColumnManager(fixture.getOptionManager(), standardOptions(filePath));
    ScanLevelProjection scanProj = ScanLevelProjection.build(RowSetTestUtils.projectList(ScanTestUtils.FILE_NAME_COL, "a", ScanTestUtils.partitionColName(0)), ScanTestUtils.parsers(metadataManager.projectionParser()));
    assertEquals(3, scanProj.columns().size());
    // Scan-level projection: defines the columns
    {
        assertTrue(scanProj.columns().get(0) instanceof FileMetadataColumn);
        FileMetadataColumn col0 = (FileMetadataColumn) scanProj.columns().get(0);
        assertTrue(col0 instanceof FileMetadataColumn);
        assertEquals(ScanTestUtils.FILE_NAME_COL, col0.name());
        assertEquals(MinorType.VARCHAR, col0.schema().getType().getMinorType());
        assertEquals(DataMode.REQUIRED, col0.schema().getType().getMode());
        ColumnProjection col1 = scanProj.columns().get(1);
        assertTrue(col1 instanceof UnresolvedColumn);
        assertEquals("a", col1.name());
        assertTrue(scanProj.columns().get(2) instanceof PartitionColumn);
        PartitionColumn col2 = (PartitionColumn) scanProj.columns().get(2);
        assertTrue(col2 instanceof PartitionColumn);
        assertEquals(ScanTestUtils.partitionColName(0), col2.name());
        assertEquals(MinorType.VARCHAR, col2.schema().getType().getMinorType());
        assertEquals(DataMode.OPTIONAL, col2.schema().getType().getMode());
    }
    // Schema-level projection, fills in values.
    TupleMetadata tableSchema = new SchemaBuilder().add("a", MinorType.VARCHAR).buildSchema();
    metadataManager.startFile(filePath);
    NullColumnBuilder builder = new NullBuilderBuilder().build();
    ResolvedRow rootTuple = new ResolvedRow(builder);
    new ExplicitSchemaProjection(scanProj, tableSchema, rootTuple, ScanTestUtils.resolvers(metadataManager));
    List<ResolvedColumn> columns = rootTuple.columns();
    assertEquals(3, columns.size());
    {
        assertTrue(columns.get(0) instanceof FileMetadataColumn);
        FileMetadataColumn col0 = (FileMetadataColumn) columns.get(0);
        assertTrue(col0 instanceof FileMetadataColumn);
        assertEquals(ScanTestUtils.FILE_NAME_COL, col0.name());
        assertEquals("z.csv", col0.value());
        assertEquals(MinorType.VARCHAR, col0.schema().getType().getMinorType());
        assertEquals(DataMode.REQUIRED, col0.schema().getType().getMode());
        ResolvedColumn col1 = columns.get(1);
        assertEquals("a", col1.name());
        assertTrue(columns.get(2) instanceof PartitionColumn);
        PartitionColumn col2 = (PartitionColumn) columns.get(2);
        assertTrue(col2 instanceof PartitionColumn);
        assertEquals(ScanTestUtils.partitionColName(0), col2.name());
        assertEquals("x", col2.value());
        assertEquals(MinorType.VARCHAR, col2.schema().getType().getMinorType());
        assertEquals(DataMode.OPTIONAL, col2.schema().getType().getMode());
    }
    // Verify that the file metadata columns were picked out
    assertEquals(2, metadataManager.metadataColumns().size());
    assertSame(columns.get(0), metadataManager.metadataColumns().get(0));
    assertSame(columns.get(2), metadataManager.metadataColumns().get(1));
}
Also used : Path(org.apache.hadoop.fs.Path) ImplicitColumnManager(org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager) ScanLevelProjection(org.apache.drill.exec.physical.impl.scan.project.ScanLevelProjection) PartitionColumn(org.apache.drill.exec.physical.impl.scan.file.PartitionColumn) NullBuilderBuilder(org.apache.drill.exec.physical.impl.scan.project.NullColumnBuilder.NullBuilderBuilder) ExplicitSchemaProjection(org.apache.drill.exec.physical.impl.scan.project.ExplicitSchemaProjection) ResolvedColumn(org.apache.drill.exec.physical.impl.scan.project.ResolvedColumn) NullColumnBuilder(org.apache.drill.exec.physical.impl.scan.project.NullColumnBuilder) ResolvedRow(org.apache.drill.exec.physical.impl.scan.project.ResolvedTuple.ResolvedRow) UnresolvedColumn(org.apache.drill.exec.physical.impl.scan.project.AbstractUnresolvedColumn.UnresolvedColumn) ColumnProjection(org.apache.drill.exec.physical.impl.scan.project.ColumnProjection) TupleMetadata(org.apache.drill.exec.record.metadata.TupleMetadata) SchemaBuilder(org.apache.drill.exec.record.metadata.SchemaBuilder) FileMetadataColumn(org.apache.drill.exec.physical.impl.scan.file.FileMetadataColumn) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 7 with ImplicitColumnManager

use of org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager in project drill by apache.

the class TestColumnsArrayParser method testMetadataColumnsWithColumnsArray.

/**
 * The `columns` column is special: can't be used with other column names.
 * Make sure that the rule <i>does not</i> apply to implicit columns.
 */
@Test
public void testMetadataColumnsWithColumnsArray() {
    Path filePath = new Path("hdfs:///w/x/y/z.csv");
    ImplicitColumnManager metadataManager = new ImplicitColumnManager(fixture.getOptionManager(), standardOptions(filePath));
    ScanLevelProjection scanProj = ScanLevelProjection.build(RowSetTestUtils.projectList(ScanTestUtils.FILE_NAME_COL, ColumnsScanFramework.COLUMNS_COL, ScanTestUtils.SUFFIX_COL), ScanTestUtils.parsers(new ColumnsArrayParser(true), metadataManager.projectionParser()));
    assertFalse(scanProj.projectAll());
    assertEquals(3, scanProj.columns().size());
    assertEquals(ScanTestUtils.FILE_NAME_COL, scanProj.columns().get(0).name());
    assertEquals(ColumnsScanFramework.COLUMNS_COL, scanProj.columns().get(1).name());
    assertEquals(ScanTestUtils.SUFFIX_COL, scanProj.columns().get(2).name());
    // Verify column type
    assertTrue(scanProj.columns().get(0) instanceof FileMetadataColumn);
    assertTrue(scanProj.columns().get(1) instanceof UnresolvedColumnsArrayColumn);
    assertTrue(scanProj.columns().get(2) instanceof FileMetadataColumn);
}
Also used : SchemaPath(org.apache.drill.common.expression.SchemaPath) Path(org.apache.hadoop.fs.Path) ImplicitColumnManager(org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager) UnresolvedColumnsArrayColumn(org.apache.drill.exec.physical.impl.scan.columns.UnresolvedColumnsArrayColumn) ScanLevelProjection(org.apache.drill.exec.physical.impl.scan.project.ScanLevelProjection) ColumnsArrayParser(org.apache.drill.exec.physical.impl.scan.columns.ColumnsArrayParser) FileMetadataColumn(org.apache.drill.exec.physical.impl.scan.file.FileMetadataColumn) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 8 with ImplicitColumnManager

use of org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager in project drill by apache.

the class TestImplicitColumnParser method testRevisedWildcardAndPartition.

/**
 * Include both a wildcard and a partition column. The wildcard, in
 * legacy mode, will create partition columns for any partitions not
 * mentioned in the project list.
 * <p>
 * Tests proposed functionality: included only requested partition
 * columns.
 */
@Test
public void testRevisedWildcardAndPartition() {
    Path filePath = new Path("hdfs:///w/x/y/z.csv");
    ImplicitColumnManager implictColManager = new ImplicitColumnManager(fixture.getOptionManager(), standardOptions(filePath));
    ScanLevelProjection scanProj = ScanLevelProjection.build(RowSetTestUtils.projectList(SchemaPath.DYNAMIC_STAR, ScanTestUtils.partitionColName(8)), Lists.newArrayList(implictColManager.projectionParser()));
    List<ColumnProjection> cols = scanProj.columns();
    assertEquals(2, cols.size());
    assertTrue(scanProj.columns().get(0) instanceof UnresolvedWildcardColumn);
    assertTrue(scanProj.columns().get(1) instanceof PartitionColumn);
}
Also used : SchemaPath(org.apache.drill.common.expression.SchemaPath) Path(org.apache.hadoop.fs.Path) ImplicitColumnManager(org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager) ScanLevelProjection(org.apache.drill.exec.physical.impl.scan.project.ScanLevelProjection) ColumnProjection(org.apache.drill.exec.physical.impl.scan.project.ColumnProjection) UnresolvedWildcardColumn(org.apache.drill.exec.physical.impl.scan.project.AbstractUnresolvedColumn.UnresolvedWildcardColumn) PartitionColumn(org.apache.drill.exec.physical.impl.scan.file.PartitionColumn) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 9 with ImplicitColumnManager

use of org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager in project drill by apache.

the class TestImplicitColumnParser method testFileMetadataColumnSelection.

/**
 * Test including file metadata (AKA "implicit columns") in the project
 * list.
 */
@Test
public void testFileMetadataColumnSelection() {
    Path filePath = new Path("hdfs:///w/x/y/z.csv");
    ImplicitColumnManager implictColManager = new ImplicitColumnManager(fixture.getOptionManager(), standardOptions(filePath));
    // Simulate SELECT a, fqn, filEPath, filename, suffix ...
    ScanLevelProjection scanProj = ScanLevelProjection.build(RowSetTestUtils.projectList("a", ScanTestUtils.FULLY_QUALIFIED_NAME_COL, // Sic, to test case sensitivity
    "filEPath", ScanTestUtils.FILE_NAME_COL, ScanTestUtils.SUFFIX_COL), Lists.newArrayList(implictColManager.projectionParser()));
    assertFalse(scanProj.projectAll());
    assertEquals(5, scanProj.requestedCols().size());
    assertEquals(5, scanProj.columns().size());
    assertEquals("a", scanProj.columns().get(0).name());
    assertEquals(ScanTestUtils.FULLY_QUALIFIED_NAME_COL, scanProj.columns().get(1).name());
    assertEquals("filEPath", scanProj.columns().get(2).name());
    assertEquals(ScanTestUtils.FILE_NAME_COL, scanProj.columns().get(3).name());
    assertEquals(ScanTestUtils.SUFFIX_COL, scanProj.columns().get(4).name());
    // Verify column type
    assertTrue(scanProj.columns().get(0) instanceof UnresolvedColumn);
    assertTrue(scanProj.columns().get(1) instanceof FileMetadataColumn);
    assertTrue(scanProj.columns().get(2) instanceof FileMetadataColumn);
    assertTrue(scanProj.columns().get(3) instanceof FileMetadataColumn);
    assertTrue(scanProj.columns().get(4) instanceof FileMetadataColumn);
    assertTrue(implictColManager.hasImplicitCols());
}
Also used : SchemaPath(org.apache.drill.common.expression.SchemaPath) Path(org.apache.hadoop.fs.Path) ImplicitColumnManager(org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager) ScanLevelProjection(org.apache.drill.exec.physical.impl.scan.project.ScanLevelProjection) UnresolvedColumn(org.apache.drill.exec.physical.impl.scan.project.AbstractUnresolvedColumn.UnresolvedColumn) FileMetadataColumn(org.apache.drill.exec.physical.impl.scan.file.FileMetadataColumn) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Example 10 with ImplicitColumnManager

use of org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager in project drill by apache.

the class TestImplicitColumnParser method testPartitionColumnSelection.

/**
 * Verify that partition columns, in any case, work.
 */
@Test
public void testPartitionColumnSelection() {
    Path filePath = new Path("hdfs:///w/x/y/z.csv");
    ImplicitColumnManager implictColManager = new ImplicitColumnManager(fixture.getOptionManager(), standardOptions(filePath));
    String dir0 = ScanTestUtils.partitionColName(0);
    // Sic: case insensitivity, but name in project list
    // is preferred over "natural" name.
    String dir1 = "DIR1";
    String dir2 = ScanTestUtils.partitionColName(2);
    ScanLevelProjection scanProj = ScanLevelProjection.build(RowSetTestUtils.projectList(dir2, dir1, dir0, "a"), Lists.newArrayList(implictColManager.projectionParser()));
    assertEquals(4, scanProj.columns().size());
    assertEquals(dir2, scanProj.columns().get(0).name());
    assertEquals(dir1, scanProj.columns().get(1).name());
    assertEquals(dir0, scanProj.columns().get(2).name());
    assertEquals("a", scanProj.columns().get(3).name());
    // Verify column type
    assertTrue(scanProj.columns().get(0) instanceof PartitionColumn);
}
Also used : SchemaPath(org.apache.drill.common.expression.SchemaPath) Path(org.apache.hadoop.fs.Path) ImplicitColumnManager(org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager) ScanLevelProjection(org.apache.drill.exec.physical.impl.scan.project.ScanLevelProjection) PartitionColumn(org.apache.drill.exec.physical.impl.scan.file.PartitionColumn) SubOperatorTest(org.apache.drill.test.SubOperatorTest) Test(org.junit.Test)

Aggregations

ImplicitColumnManager (org.apache.drill.exec.physical.impl.scan.file.ImplicitColumnManager)24 Path (org.apache.hadoop.fs.Path)24 SubOperatorTest (org.apache.drill.test.SubOperatorTest)23 Test (org.junit.Test)23 SchemaPath (org.apache.drill.common.expression.SchemaPath)17 ScanLevelProjection (org.apache.drill.exec.physical.impl.scan.project.ScanLevelProjection)14 SchemaBuilder (org.apache.drill.exec.record.metadata.SchemaBuilder)13 TupleMetadata (org.apache.drill.exec.record.metadata.TupleMetadata)13 ColumnProjection (org.apache.drill.exec.physical.impl.scan.project.ColumnProjection)8 FileMetadataColumn (org.apache.drill.exec.physical.impl.scan.file.FileMetadataColumn)7 PartitionColumn (org.apache.drill.exec.physical.impl.scan.file.PartitionColumn)7 ResolvedRow (org.apache.drill.exec.physical.impl.scan.project.ResolvedTuple.ResolvedRow)7 MockScanBuilder (org.apache.drill.exec.physical.impl.scan.ScanTestUtils.MockScanBuilder)6 UnresolvedWildcardColumn (org.apache.drill.exec.physical.impl.scan.project.AbstractUnresolvedColumn.UnresolvedWildcardColumn)6 ScanSchemaOrchestrator (org.apache.drill.exec.physical.impl.scan.project.ScanSchemaOrchestrator)6 ScanOrchestratorBuilder (org.apache.drill.exec.physical.impl.scan.project.ScanSchemaOrchestrator.ScanOrchestratorBuilder)6 File (java.io.File)5 ReaderSchemaOrchestrator (org.apache.drill.exec.physical.impl.scan.project.ReaderSchemaOrchestrator)5 ResultSetLoader (org.apache.drill.exec.physical.resultSet.ResultSetLoader)5 SingleRowSet (org.apache.drill.exec.physical.rowSet.RowSet.SingleRowSet)5