Search in sources :

Example 26 with SummarizerConfiguration

use of org.apache.accumulo.core.client.summary.SummarizerConfiguration in project accumulo by apache.

the class EntryLengthSummarizersTest method testBasicVisibility.

@Test
public void testBasicVisibility() {
    SummarizerConfiguration sc = SummarizerConfiguration.builder(EntryLengthSummarizer.class).build();
    EntryLengthSummarizer entrySum = new EntryLengthSummarizer();
    Key k1 = new Key("r1", "f1", "q1", "v1");
    Key k2 = new Key("r2", "f2", "q2", "v2");
    Key k3 = new Key("r3", "f3", "q3", "v3");
    Collector collector = entrySum.collector(sc);
    collector.accept(k1, new Value(""));
    collector.accept(k2, new Value(""));
    collector.accept(k3, new Value(""));
    HashMap<String, Long> stats = new HashMap<>();
    collector.summarize(stats::put);
    HashMap<String, Long> expected = new HashMap<>();
    expected.put("key.min", 8L);
    expected.put("key.max", 8L);
    expected.put("key.sum", 24L);
    // Log2 Histogram
    expected.put("key.logHist.3", 3L);
    expected.put("row.min", 2L);
    expected.put("row.max", 2L);
    expected.put("row.sum", 6L);
    // Log2 Histogram
    expected.put("row.logHist.1", 3L);
    expected.put("family.min", 2L);
    expected.put("family.max", 2L);
    expected.put("family.sum", 6L);
    // Log2 Histogram
    expected.put("family.logHist.1", 3L);
    expected.put("qualifier.min", 2L);
    expected.put("qualifier.max", 2L);
    expected.put("qualifier.sum", 6L);
    // Log2 Histogram
    expected.put("qualifier.logHist.1", 3L);
    expected.put("visibility.min", 2L);
    expected.put("visibility.max", 2L);
    expected.put("visibility.sum", 6L);
    // Log2 Histogram
    expected.put("visibility.logHist.1", 3L);
    expected.put("value.min", 0L);
    expected.put("value.max", 0L);
    expected.put("value.sum", 0L);
    // Log2 Histogram
    expected.put("value.logHist.0", 3L);
    expected.put("total", 3L);
    Assert.assertEquals(expected, stats);
}
Also used : HashMap(java.util.HashMap) EntryLengthSummarizer(org.apache.accumulo.core.client.summary.summarizers.EntryLengthSummarizer) Collector(org.apache.accumulo.core.client.summary.Summarizer.Collector) Value(org.apache.accumulo.core.data.Value) SummarizerConfiguration(org.apache.accumulo.core.client.summary.SummarizerConfiguration) Key(org.apache.accumulo.core.data.Key) Test(org.junit.Test)

Example 27 with SummarizerConfiguration

use of org.apache.accumulo.core.client.summary.SummarizerConfiguration in project accumulo by apache.

the class EntryLengthSummarizersTest method testBasicQualifier.

@Test
public void testBasicQualifier() {
    SummarizerConfiguration sc = SummarizerConfiguration.builder(EntryLengthSummarizer.class).build();
    EntryLengthSummarizer entrySum = new EntryLengthSummarizer();
    Key k1 = new Key("r1", "f1", "q1");
    Key k2 = new Key("r2", "f2", "q2");
    Key k3 = new Key("r3", "f3", "q3");
    Collector collector = entrySum.collector(sc);
    collector.accept(k1, new Value(""));
    collector.accept(k2, new Value(""));
    collector.accept(k3, new Value(""));
    HashMap<String, Long> stats = new HashMap<>();
    collector.summarize(stats::put);
    HashMap<String, Long> expected = new HashMap<>();
    expected.put("key.min", 6L);
    expected.put("key.max", 6L);
    expected.put("key.sum", 18L);
    // Log2 Histogram
    expected.put("key.logHist.3", 3L);
    expected.put("row.min", 2L);
    expected.put("row.max", 2L);
    expected.put("row.sum", 6L);
    // Log2 Histogram
    expected.put("row.logHist.1", 3L);
    expected.put("family.min", 2L);
    expected.put("family.max", 2L);
    expected.put("family.sum", 6L);
    // Log2 Histogram
    expected.put("family.logHist.1", 3L);
    expected.put("qualifier.min", 2L);
    expected.put("qualifier.max", 2L);
    expected.put("qualifier.sum", 6L);
    // Log2 Histogram
    expected.put("qualifier.logHist.1", 3L);
    expected.put("visibility.min", 0L);
    expected.put("visibility.max", 0L);
    expected.put("visibility.sum", 0L);
    // Log2 Histogram
    expected.put("visibility.logHist.0", 3L);
    expected.put("value.min", 0L);
    expected.put("value.max", 0L);
    expected.put("value.sum", 0L);
    // Log2 Histogram
    expected.put("value.logHist.0", 3L);
    expected.put("total", 3L);
    Assert.assertEquals(expected, stats);
}
Also used : HashMap(java.util.HashMap) EntryLengthSummarizer(org.apache.accumulo.core.client.summary.summarizers.EntryLengthSummarizer) Collector(org.apache.accumulo.core.client.summary.Summarizer.Collector) Value(org.apache.accumulo.core.data.Value) SummarizerConfiguration(org.apache.accumulo.core.client.summary.SummarizerConfiguration) Key(org.apache.accumulo.core.data.Key) Test(org.junit.Test)

Example 28 with SummarizerConfiguration

use of org.apache.accumulo.core.client.summary.SummarizerConfiguration in project accumulo by apache.

the class EntryLengthSummarizersTest method testComplexFamily.

@Test
public void testComplexFamily() {
    SummarizerConfiguration sc = SummarizerConfiguration.builder(EntryLengthSummarizer.class).build();
    EntryLengthSummarizer entrySum = new EntryLengthSummarizer();
    Key k1 = new Key("r1", "family1");
    Key k2 = new Key("row2", "columnFamily2");
    Key k3 = new Key("columnRow3", "f3");
    Collector collector = entrySum.collector(sc);
    collector.accept(k1, new Value(""));
    collector.accept(k2, new Value(""));
    collector.accept(k3, new Value(""));
    HashMap<String, Long> stats = new HashMap<>();
    collector.summarize(stats::put);
    HashMap<String, Long> expected = new HashMap<>();
    expected.put("key.min", 9L);
    expected.put("key.max", 17L);
    expected.put("key.sum", 38L);
    // Log2 Histogram
    expected.put("key.logHist.3", 1L);
    expected.put("key.logHist.4", 2L);
    expected.put("row.min", 2L);
    expected.put("row.max", 10L);
    expected.put("row.sum", 16L);
    // Log2 Histogram
    expected.put("row.logHist.1", 1L);
    expected.put("row.logHist.2", 1L);
    expected.put("row.logHist.3", 1L);
    expected.put("family.min", 2L);
    expected.put("family.max", 13L);
    expected.put("family.sum", 22L);
    // Log2 Histogram
    expected.put("family.logHist.1", 1L);
    expected.put("family.logHist.3", 1L);
    expected.put("family.logHist.4", 1L);
    expected.put("qualifier.min", 0L);
    expected.put("qualifier.max", 0L);
    expected.put("qualifier.sum", 0L);
    // Log2 Histogram
    expected.put("qualifier.logHist.0", 3L);
    expected.put("visibility.min", 0L);
    expected.put("visibility.max", 0L);
    expected.put("visibility.sum", 0L);
    // Log2 Histogram
    expected.put("visibility.logHist.0", 3L);
    expected.put("value.min", 0L);
    expected.put("value.max", 0L);
    expected.put("value.sum", 0L);
    // Log2 Histogram
    expected.put("value.logHist.0", 3L);
    expected.put("total", 3L);
    Assert.assertEquals(expected, stats);
}
Also used : HashMap(java.util.HashMap) EntryLengthSummarizer(org.apache.accumulo.core.client.summary.summarizers.EntryLengthSummarizer) Collector(org.apache.accumulo.core.client.summary.Summarizer.Collector) Value(org.apache.accumulo.core.data.Value) SummarizerConfiguration(org.apache.accumulo.core.client.summary.SummarizerConfiguration) Key(org.apache.accumulo.core.data.Key) Test(org.junit.Test)

Example 29 with SummarizerConfiguration

use of org.apache.accumulo.core.client.summary.SummarizerConfiguration in project accumulo by apache.

the class EntryLengthSummarizersTest method testLog2Histogram.

@Test
public void testLog2Histogram() {
    SummarizerConfiguration sc = SummarizerConfiguration.builder(EntryLengthSummarizer.class).build();
    EntryLengthSummarizer entrySum = new EntryLengthSummarizer();
    Key k1 = new Key("row1");
    Key k2 = new Key("row2");
    Key k3 = new Key("row3");
    Collector collector = entrySum.collector(sc);
    collector.accept(k1, new Value("01"));
    collector.accept(k2, new Value("012345678"));
    collector.accept(k3, new Value("012345679"));
    HashMap<String, Long> stats = new HashMap<>();
    collector.summarize(stats::put);
    HashMap<String, Long> expected = new HashMap<>();
    expected.put("key.min", 4L);
    expected.put("key.max", 4L);
    expected.put("key.sum", 12L);
    // Log2 Histogram for Key
    expected.put("key.logHist.2", 3L);
    expected.put("row.min", 4L);
    expected.put("row.max", 4L);
    expected.put("row.sum", 12L);
    // Log2 Histogram for Row
    expected.put("row.logHist.2", 3L);
    expected.put("family.min", 0L);
    expected.put("family.max", 0L);
    expected.put("family.sum", 0L);
    // Log2 Histogram for Family
    expected.put("family.logHist.0", 3L);
    expected.put("qualifier.min", 0L);
    expected.put("qualifier.max", 0L);
    expected.put("qualifier.sum", 0L);
    // Log2 Histogram for Qualifier
    expected.put("qualifier.logHist.0", 3L);
    expected.put("visibility.min", 0L);
    expected.put("visibility.max", 0L);
    expected.put("visibility.sum", 0L);
    // Log2 Histogram for Visibility
    expected.put("visibility.logHist.0", 3L);
    expected.put("value.min", 2L);
    expected.put("value.max", 9L);
    expected.put("value.sum", 20L);
    // Log2 Histogram for Value
    expected.put("value.logHist.1", 1L);
    expected.put("value.logHist.3", 2L);
    expected.put("total", 3L);
    Assert.assertEquals(expected, stats);
}
Also used : HashMap(java.util.HashMap) EntryLengthSummarizer(org.apache.accumulo.core.client.summary.summarizers.EntryLengthSummarizer) Collector(org.apache.accumulo.core.client.summary.Summarizer.Collector) Value(org.apache.accumulo.core.data.Value) SummarizerConfiguration(org.apache.accumulo.core.client.summary.SummarizerConfiguration) Key(org.apache.accumulo.core.data.Key) Test(org.junit.Test)

Example 30 with SummarizerConfiguration

use of org.apache.accumulo.core.client.summary.SummarizerConfiguration in project accumulo by apache.

the class EntryLengthSummarizersTest method testCombine3.

@Test
public void testCombine3() {
    SummarizerConfiguration sc = SummarizerConfiguration.builder(EntryLengthSummarizer.class).build();
    EntryLengthSummarizer entrySum = new EntryLengthSummarizer();
    Collector collector1 = entrySum.collector(sc);
    collector1.accept(new Key("r1", "f1"), new Value("v1"));
    HashMap<String, Long> stats1 = new HashMap<>();
    collector1.summarize(stats1::put);
    Collector collector2 = entrySum.collector(sc);
    collector2.accept(new Key("row1", "family1", "q1"), new Value(""));
    HashMap<String, Long> stats2 = new HashMap<>();
    collector2.summarize(stats2::put);
    Combiner combiner = entrySum.combiner(sc);
    combiner.merge(stats1, stats2);
    HashMap<String, Long> expected = new HashMap<>();
    expected.put("key.min", 4L);
    expected.put("key.max", 13L);
    expected.put("key.sum", 17L);
    // Log2 Histogram for Key
    expected.put("key.logHist.2", 1L);
    expected.put("key.logHist.4", 1L);
    expected.put("row.min", 2L);
    expected.put("row.max", 4L);
    expected.put("row.sum", 6L);
    // Log2 Histogram for Row
    expected.put("row.logHist.1", 1L);
    expected.put("row.logHist.2", 1L);
    expected.put("family.min", 2L);
    expected.put("family.max", 7L);
    expected.put("family.sum", 9L);
    // Log2 Histogram for Family
    expected.put("family.logHist.1", 1L);
    expected.put("family.logHist.3", 1L);
    expected.put("qualifier.min", 0L);
    expected.put("qualifier.max", 2L);
    expected.put("qualifier.sum", 2L);
    // Log2 Histogram for Qualifier
    expected.put("qualifier.logHist.0", 1L);
    expected.put("qualifier.logHist.1", 1L);
    expected.put("visibility.min", 0L);
    expected.put("visibility.max", 0L);
    expected.put("visibility.sum", 0L);
    // Log2 Histogram for Visibility
    expected.put("visibility.logHist.0", 2L);
    expected.put("value.min", 0L);
    expected.put("value.max", 2L);
    expected.put("value.sum", 2L);
    // Log2 Histogram for Value
    expected.put("value.logHist.0", 1L);
    expected.put("value.logHist.1", 1L);
    expected.put("total", 2L);
    Assert.assertEquals(expected, stats1);
}
Also used : HashMap(java.util.HashMap) EntryLengthSummarizer(org.apache.accumulo.core.client.summary.summarizers.EntryLengthSummarizer) Collector(org.apache.accumulo.core.client.summary.Summarizer.Collector) Value(org.apache.accumulo.core.data.Value) Combiner(org.apache.accumulo.core.client.summary.Summarizer.Combiner) SummarizerConfiguration(org.apache.accumulo.core.client.summary.SummarizerConfiguration) Key(org.apache.accumulo.core.data.Key) Test(org.junit.Test)

Aggregations

SummarizerConfiguration (org.apache.accumulo.core.client.summary.SummarizerConfiguration)41 Test (org.junit.Test)33 HashMap (java.util.HashMap)28 Key (org.apache.accumulo.core.data.Key)22 Value (org.apache.accumulo.core.data.Value)22 Collector (org.apache.accumulo.core.client.summary.Summarizer.Collector)19 EntryLengthSummarizer (org.apache.accumulo.core.client.summary.summarizers.EntryLengthSummarizer)16 Summary (org.apache.accumulo.core.client.summary.Summary)13 NewTableConfiguration (org.apache.accumulo.core.client.admin.NewTableConfiguration)10 CounterSummary (org.apache.accumulo.core.client.summary.CounterSummary)10 Connector (org.apache.accumulo.core.client.Connector)9 BatchWriter (org.apache.accumulo.core.client.BatchWriter)8 Text (org.apache.hadoop.io.Text)8 FamilySummarizer (org.apache.accumulo.core.client.summary.summarizers.FamilySummarizer)7 ArrayList (java.util.ArrayList)6 BatchWriterConfig (org.apache.accumulo.core.client.BatchWriterConfig)6 IOException (java.io.IOException)5 Collection (java.util.Collection)5 Map (java.util.Map)5 Entry (java.util.Map.Entry)5