use of com.linkedin.pinot.common.data.DimensionFieldSpec in project pinot by linkedin.
the class DictionariesTest method testPaddedConflict.
/**
* Tests SegmentDictionaryCreator for case when there is one empty string
* and a string with a single padding character
*
* This test asserts that the padded length of the empty string is 1
* in actual padded dictionary), and not 0.
*
* @throws Exception
*/
@Test
public void testPaddedConflict() throws Exception {
File indexDir = new File("/tmp/dict.test");
indexDir.deleteOnExit();
FieldSpec fieldSpec = new DimensionFieldSpec("test", DataType.STRING, true);
String[] inputStrings = new String[2];
String[] paddedStrings = new String[2];
char paddingChar = '%';
try {
inputStrings[0] = "";
inputStrings[1] = "%";
// Sorted order: {"", "%"}
Arrays.sort(inputStrings);
SegmentDictionaryCreator dictionaryCreator = new SegmentDictionaryCreator(false, inputStrings, fieldSpec, indexDir, paddingChar);
boolean[] isSorted = new boolean[1];
isSorted[0] = true;
dictionaryCreator.build(isSorted);
} catch (Exception e) {
Assert.assertEquals(e.getMessage(), "Number of entries in dictionary != number of unique values in the data in column test");
} finally {
FileUtils.deleteQuietly(indexDir);
}
}
use of com.linkedin.pinot.common.data.DimensionFieldSpec in project pinot by linkedin.
the class DictionariesTest method testPaddedNoConflict.
/**
* Tests SegmentDictionaryCreator for case when there is one empty string
* and a string with a single '%' character
*
* This test asserts that the padded length of the empty string is 1
* in actual padded dictionary), and not 0.
*
* @throws Exception
*/
@Test
public void testPaddedNoConflict() throws Exception {
File indexDir = new File("/tmp/dict.test");
FieldSpec fieldSpec = new DimensionFieldSpec("test", DataType.STRING, true);
String[] inputStrings = new String[2];
String[] paddedStrings = new String[2];
char paddingChar = '\0';
inputStrings[0] = "";
inputStrings[1] = "%";
// Sorted order: {"", "%"}
Arrays.sort(inputStrings);
SegmentDictionaryCreator dictionaryCreator = new SegmentDictionaryCreator(false, inputStrings, fieldSpec, indexDir, paddingChar);
boolean[] isSorted = new boolean[1];
isSorted[0] = true;
dictionaryCreator.build(isSorted);
// Get the padded string as stored in the dictionary.
int targetPaddedLength = dictionaryCreator.getStringColumnMaxLength();
for (int i = 0; i < inputStrings.length; i++) {
paddedStrings[i] = SegmentDictionaryCreator.getPaddedString(inputStrings[i], targetPaddedLength, paddingChar);
}
// Sorted Order: {"abc def%%%%", "abc%%%%%%%"}
Arrays.sort(paddedStrings);
// Assert that indexOfSV for un-padded string returns the index of the corresponding padded string.
for (int i = 0; i < inputStrings.length; i++) {
int paddedIndex = dictionaryCreator.indexOfSV(inputStrings[i]);
Assert.assertTrue(paddedStrings[paddedIndex].equals(SegmentDictionaryCreator.getPaddedString(inputStrings[i], targetPaddedLength, paddingChar)));
}
dictionaryCreator.close();
FileUtils.deleteQuietly(indexDir);
FileUtils.deleteQuietly(indexDir);
}
use of com.linkedin.pinot.common.data.DimensionFieldSpec in project pinot by linkedin.
the class DictionariesTest method testDoubleColumnPreIndexStatsCollector.
@Test
public void testDoubleColumnPreIndexStatsCollector() throws Exception {
FieldSpec spec = new DimensionFieldSpec("column1", DataType.DOUBLE, true);
AbstractColumnStatisticsCollector statsCollector = new DoubleColumnPreIndexStatsCollector(spec);
statsCollector.collect(new Integer(1));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Float(2));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Long(3));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Double(4));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Integer(4));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Float(2));
Assert.assertFalse(statsCollector.isSorted());
statsCollector.collect(new Double(40));
Assert.assertFalse(statsCollector.isSorted());
statsCollector.collect(new Double(20));
Assert.assertFalse(statsCollector.isSorted());
statsCollector.seal();
Assert.assertEquals(statsCollector.getCardinality(), 6);
Assert.assertEquals(((Number) statsCollector.getMinValue()).intValue(), 1);
Assert.assertEquals(((Number) statsCollector.getMaxValue()).intValue(), 40);
Assert.assertFalse(statsCollector.isSorted());
}
use of com.linkedin.pinot.common.data.DimensionFieldSpec in project pinot by linkedin.
the class DictionariesTest method testSingleEmptyString.
/**
* Tests SegmentDictionaryCreator for case when there is only one string
* and it is empty
*
* This test asserts that the padded length of the empty string is 1
* in actual padded dictionary), and not 0.
*
* @throws Exception
*/
@Test
public void testSingleEmptyString() throws Exception {
File indexDir = new File("/tmp/dict.test");
indexDir.deleteOnExit();
FieldSpec fieldSpec = new DimensionFieldSpec("test", DataType.STRING, true);
String[] inputStrings = new String[1];
String[] paddedStrings = new String[1];
try {
inputStrings[0] = "";
// Sorted order: {""}
Arrays.sort(inputStrings);
boolean[] isSorted = new boolean[1];
isSorted[0] = true;
SegmentDictionaryCreator dictionaryCreator = new SegmentDictionaryCreator(false, inputStrings, fieldSpec, indexDir, V1Constants.Str.DEFAULT_STRING_PAD_CHAR);
dictionaryCreator.build(isSorted);
// Get the padded string as stored in the dictionary.
int targetPaddedLength = dictionaryCreator.getStringColumnMaxLength();
Assert.assertTrue(targetPaddedLength == 1);
for (int i = 0; i < inputStrings.length; i++) {
paddedStrings[i] = SegmentDictionaryCreator.getPaddedString(inputStrings[i], targetPaddedLength, V1Constants.Str.DEFAULT_STRING_PAD_CHAR);
}
// Sorted Order: {"%"}
Arrays.sort(paddedStrings);
// Assert that indexOfSV for un-padded string returns the index of the corresponding padded string.
for (int i = 0; i < inputStrings.length; i++) {
int paddedIndex = dictionaryCreator.indexOfSV(inputStrings[i]);
Assert.assertTrue(paddedStrings[paddedIndex].equals(SegmentDictionaryCreator.getPaddedString(inputStrings[i], targetPaddedLength, V1Constants.Str.DEFAULT_STRING_PAD_CHAR)));
}
// Verify that empty string got padded
Assert.assertTrue(paddedStrings[0].equals(SegmentDictionaryCreator.getPaddedString(inputStrings[0], targetPaddedLength, V1Constants.Str.DEFAULT_STRING_PAD_CHAR)));
dictionaryCreator.close();
} catch (Exception e) {
throw e;
} finally {
FileUtils.deleteQuietly(indexDir);
}
}
use of com.linkedin.pinot.common.data.DimensionFieldSpec in project pinot by linkedin.
the class DictionariesTest method testFloatColumnPreIndexStatsCollector.
@Test
public void testFloatColumnPreIndexStatsCollector() throws Exception {
FieldSpec spec = new DimensionFieldSpec("column1", DataType.FLOAT, true);
AbstractColumnStatisticsCollector statsCollector = new FloatColumnPreIndexStatsCollector(spec);
statsCollector.collect(new Integer(1));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Float(2));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Long(3));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Double(4));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Integer(4));
Assert.assertTrue(statsCollector.isSorted());
statsCollector.collect(new Float(2));
Assert.assertFalse(statsCollector.isSorted());
statsCollector.collect(new Double(40));
Assert.assertFalse(statsCollector.isSorted());
statsCollector.collect(new Double(20));
Assert.assertFalse(statsCollector.isSorted());
statsCollector.seal();
Assert.assertEquals(statsCollector.getCardinality(), 6);
Assert.assertEquals(((Number) statsCollector.getMinValue()).intValue(), 1);
Assert.assertEquals(((Number) statsCollector.getMaxValue()).intValue(), 40);
Assert.assertFalse(statsCollector.isSorted());
}
Aggregations