Search in sources :

Example 96 with Options

use of com.google.cloud.teleport.v2.transforms.DeleteBigQueryDataFn.Options in project DataflowTemplates by GoogleCloudPlatform.

the class DataplexBigQueryToGcsFilterTest method test_whenTablesSet_filterExcludesTablesByName.

@Test
public void test_whenTablesSet_filterExcludesTablesByName() {
    BigQueryTable.Builder includedTable1 = table().setTableName("includedTable1");
    BigQueryTable.Builder includedTable2 = table().setTableName("includedTable2");
    BigQueryTable.Builder excludedTable = table().setTableName("excludedTable");
    BigQueryTablePartition p = partition().build();
    options.setTables("includedTable1,includedTable2");
    options.setExportDataModifiedBeforeDateTime(null);
    Filter f = new DataplexBigQueryToGcsFilter(options, new ArrayList<String>());
    assertThat(f.shouldSkipUnpartitionedTable(includedTable1)).isFalse();
    assertThat(f.shouldSkipUnpartitionedTable(includedTable2)).isFalse();
    assertThat(f.shouldSkipUnpartitionedTable(excludedTable)).isTrue();
    assertThat(f.shouldSkipPartitionedTable(includedTable1, Collections.singletonList(p))).isFalse();
    assertThat(f.shouldSkipPartitionedTable(includedTable2, Collections.singletonList(p))).isFalse();
    assertThat(f.shouldSkipPartitionedTable(excludedTable, Collections.singletonList(p))).isTrue();
    assertThat(f.shouldSkipPartition(includedTable1, p)).isFalse();
    assertThat(f.shouldSkipPartition(includedTable2, p)).isFalse();
    // Should NOT skip PARTITIONS, only tables as a whole because of their name:
    assertThat(f.shouldSkipPartition(excludedTable, p)).isFalse();
}
Also used : BigQueryTablePartition(com.google.cloud.teleport.v2.values.BigQueryTablePartition) Filter(com.google.cloud.teleport.v2.utils.BigQueryMetadataLoader.Filter) BigQueryTable(com.google.cloud.teleport.v2.values.BigQueryTable) Test(org.junit.Test)

Example 97 with Options

use of com.google.cloud.teleport.v2.transforms.DeleteBigQueryDataFn.Options in project DataflowTemplates by GoogleCloudPlatform.

the class DataplexBigQueryToGcsFilterTest method test_whenBeforeDateIs1Day3HoursDuration_dateParsedCorrectly.

@Test
public void test_whenBeforeDateIs1Day3HoursDuration_dateParsedCorrectly() {
    // current time in the DEFAULT time zone minus 1 day 3 hours:
    long micros = Instant.now().minus(Duration.millis(27 * 60 * 60 * 1000)).getMillis() * 1000L;
    BigQueryTable.Builder olderTable = table().setLastModificationTime(micros - 100000L);
    BigQueryTable.Builder newerTable = table().setLastModificationTime(micros + 100000L);
    options.setTables(null);
    options.setExportDataModifiedBeforeDateTime("-p1dt3h");
    Filter f = new DataplexBigQueryToGcsFilter(options, new ArrayList<String>());
    assertThat(f.shouldSkipUnpartitionedTable(newerTable)).isTrue();
    assertThat(f.shouldSkipUnpartitionedTable(olderTable)).isFalse();
}
Also used : Filter(com.google.cloud.teleport.v2.utils.BigQueryMetadataLoader.Filter) BigQueryTable(com.google.cloud.teleport.v2.values.BigQueryTable) Test(org.junit.Test)

Example 98 with Options

use of com.google.cloud.teleport.v2.transforms.DeleteBigQueryDataFn.Options in project DataflowTemplates by GoogleCloudPlatform.

the class DataplexBigQueryToGcsFilterTest method test_whenBeforeDateSet_filterExcludesTablesAndPartitions.

@Test
public void test_whenBeforeDateSet_filterExcludesTablesAndPartitions() {
    BigQueryTable.Builder olderTable = table().setLastModificationTime(TS_MICROS_2021_01_01_15_00_00_UTC - 1000L);
    BigQueryTable.Builder newerTable = table().setLastModificationTime(TS_MICROS_2021_01_01_15_00_00_UTC + 1000L);
    BigQueryTablePartition olderPartition = partition().setPartitionName("p1").setLastModificationTime(TS_MICROS_2021_01_01_15_00_00_UTC - 1000L).build();
    BigQueryTablePartition newerPartition = partition().setPartitionName("p2").setLastModificationTime(TS_MICROS_2021_01_01_15_00_00_UTC + 1000L).build();
    List<BigQueryTablePartition> partitions = Arrays.asList(olderPartition, newerPartition);
    options.setTables(null);
    options.setExportDataModifiedBeforeDateTime("2021-01-01T15:00:00Z");
    Filter f = new DataplexBigQueryToGcsFilter(options, new ArrayList<String>());
    assertThat(f.shouldSkipUnpartitionedTable(newerTable)).isTrue();
    assertThat(f.shouldSkipUnpartitionedTable(olderTable)).isFalse();
    // If a table is partitioned, we should filter individual partitions by modification time,
    // so the table itself should NOT be skipped no matter what the table modification time is.
    // Expecting shouldSkip = false for both newer and older tables:
    assertThat(f.shouldSkipPartitionedTable(newerTable, partitions)).isFalse();
    assertThat(f.shouldSkipPartitionedTable(olderTable, partitions)).isFalse();
    assertThat(f.shouldSkipPartition(olderTable, newerPartition)).isTrue();
    assertThat(f.shouldSkipPartition(olderTable, olderPartition)).isFalse();
}
Also used : BigQueryTablePartition(com.google.cloud.teleport.v2.values.BigQueryTablePartition) Filter(com.google.cloud.teleport.v2.utils.BigQueryMetadataLoader.Filter) BigQueryTable(com.google.cloud.teleport.v2.values.BigQueryTable) Test(org.junit.Test)

Example 99 with Options

use of com.google.cloud.teleport.v2.transforms.DeleteBigQueryDataFn.Options in project DataflowTemplates by GoogleCloudPlatform.

the class DataplexBigQueryToGcsFilterTest method test_whenTargetFileExistsWithWriteDispositionSKIP_filterExcludesTables.

@Test
public void test_whenTargetFileExistsWithWriteDispositionSKIP_filterExcludesTables() {
    BigQueryTable.Builder t = table().setTableName("table1").setPartitioningColumn("p2");
    BigQueryTablePartition p = partition().setPartitionName("partition1").build();
    options.setTables(null);
    options.setExportDataModifiedBeforeDateTime(null);
    options.setFileFormat(FileFormatOptions.AVRO);
    options.setWriteDisposition(WriteDispositionOptions.SKIP);
    Filter f = new DataplexBigQueryToGcsFilter(options, Arrays.asList("table1/output-table1.avro", "table1/p2=partition1/output-table1-partition1.avro"));
    assertThat(f.shouldSkipUnpartitionedTable(t)).isTrue();
    assertThat(f.shouldSkipPartition(t, p)).isTrue();
}
Also used : BigQueryTablePartition(com.google.cloud.teleport.v2.values.BigQueryTablePartition) Filter(com.google.cloud.teleport.v2.utils.BigQueryMetadataLoader.Filter) BigQueryTable(com.google.cloud.teleport.v2.values.BigQueryTable) Test(org.junit.Test)

Example 100 with Options

use of com.google.cloud.teleport.v2.transforms.DeleteBigQueryDataFn.Options in project DataflowTemplates by GoogleCloudPlatform.

the class DataplexBigQueryToGcsFilterTest method test_whenBeforeDateIs1DayDuration_dateParsedCorrectly.

@Test
public void test_whenBeforeDateIs1DayDuration_dateParsedCorrectly() {
    // current time in the DEFAULT time zone minus one day:
    long micros = Instant.now().minus(Duration.standardDays(1)).getMillis() * 1000L;
    BigQueryTable.Builder olderTable = table().setLastModificationTime(micros - 100000L);
    BigQueryTable.Builder newerTable = table().setLastModificationTime(micros + 100000L);
    options.setTables(null);
    options.setExportDataModifiedBeforeDateTime("-P1D");
    Filter f = new DataplexBigQueryToGcsFilter(options, new ArrayList<String>());
    assertThat(f.shouldSkipUnpartitionedTable(newerTable)).isTrue();
    assertThat(f.shouldSkipUnpartitionedTable(olderTable)).isFalse();
}
Also used : Filter(com.google.cloud.teleport.v2.utils.BigQueryMetadataLoader.Filter) BigQueryTable(com.google.cloud.teleport.v2.values.BigQueryTable) Test(org.junit.Test)

Aggregations

Test (org.junit.Test)63 PCollectionTuple (org.apache.beam.sdk.values.PCollectionTuple)25 FailsafeElement (com.google.cloud.teleport.v2.values.FailsafeElement)20 Pipeline (org.apache.beam.sdk.Pipeline)19 CoderRegistry (org.apache.beam.sdk.coders.CoderRegistry)19 BigQueryTable (com.google.cloud.teleport.v2.values.BigQueryTable)15 GenericRecord (org.apache.avro.generic.GenericRecord)12 Category (org.junit.experimental.categories.Category)12 Filter (com.google.cloud.teleport.v2.utils.BigQueryMetadataLoader.Filter)10 BigQueryTablePartition (com.google.cloud.teleport.v2.values.BigQueryTablePartition)10 PubSubToElasticsearchOptions (com.google.cloud.teleport.v2.elasticsearch.options.PubSubToElasticsearchOptions)9 TableRow (com.google.api.services.bigquery.model.TableRow)8 DataplexClient (com.google.cloud.teleport.v2.clients.DataplexClient)8 FileFormatConversionOptions (com.google.cloud.teleport.v2.templates.FileFormatConversion.FileFormatConversionOptions)8 KV (org.apache.beam.sdk.values.KV)8 ArrayList (java.util.ArrayList)7 ElasticsearchWriteOptions (com.google.cloud.teleport.v2.elasticsearch.options.ElasticsearchWriteOptions)6 GCSToElasticsearchOptions (com.google.cloud.teleport.v2.elasticsearch.options.GCSToElasticsearchOptions)6 FileFormatConversionOptions (com.google.cloud.teleport.v2.templates.DataplexFileFormatConversion.FileFormatConversionOptions)6 PubSubProtoToBigQueryOptions (com.google.cloud.teleport.v2.templates.PubsubProtoToBigQuery.PubSubProtoToBigQueryOptions)6