use of org.apache.drill.exec.vector.accessor.ColumnAccessors.VarCharColumnWriter in project drill by apache.
the class TestVariableWidthWriter method makeWriter.
private VarCharColumnWriter makeWriter(VarCharVector vector, TestIndex index) {
VarCharColumnWriter writer = new VarCharColumnWriter(vector);
writer.bindIndex(index);
assertEquals(ValueType.STRING, writer.valueType());
return writer;
}
use of org.apache.drill.exec.vector.accessor.ColumnAccessors.VarCharColumnWriter in project drill by apache.
the class TestVariableWidthWriter method testRollover.
/**
* The rollover method is used during vector overflow.
*/
@Test
public void testRollover() {
try (VarCharVector vector = allocVector(1000)) {
TestIndex index = new TestIndex();
VarCharColumnWriter writer = makeWriter(vector, index);
writer.startWrite();
// Simulate doing an overflow of ten values.
String base = "sample-value";
for (int i = 0; i < 10; i++) {
index.index = i;
writer.startRow();
writer.setString(base + i);
writer.saveRow();
}
// Overflow occurs after writing the 11th row
index.index = 10;
writer.startRow();
String overflowValue = base + 10;
writer.setString(overflowValue);
// Overflow occurs
writer.preRollover();
// Simulate rollover
byte[] dummy = new byte[] { (byte) 0x55 };
for (int i = 0; i < 500; i++) {
vector.getMutator().setSafe(i, dummy);
}
for (int i = 1; i < 15; i++) {
vector.getOffsetVector().getMutator().set(i, 0xdeadbeef);
}
vector.getMutator().setSafe(0, overflowValue.getBytes(Charsets.UTF_8));
writer.postRollover();
index.index = 0;
writer.saveRow();
for (int i = 1; i < 5; i++) {
index.index = i;
writer.startRow();
writer.setString(base + (i + 10));
writer.saveRow();
}
writer.endWrite();
for (int i = 0; i < 5; i++) {
assertEquals(base + (10 + i), stringAt(vector, i));
}
}
}
use of org.apache.drill.exec.vector.accessor.ColumnAccessors.VarCharColumnWriter in project drill by apache.
the class TestVariableWidthWriter method testSizeLimit.
/**
* Test resize monitoring. Add a listener to an Varchar writer,
* capture each resize, and refuse a resize when the s
* of the vector exceeds 1 MB. This will trigger an overflow,
* which will throw an exception which we then check for.
*/
@Test
public void testSizeLimit() {
try (VarCharVector vector = allocVector(1000)) {
TestIndex index = new TestIndex();
VarCharColumnWriter writer = makeWriter(vector, index);
writer.bindListener(new ColumnWriterListener() {
// Because assumed array size is 10, so 10 * 1000 = 10,000
// rounded to 16K
int totalAlloc = 16384;
@Override
public void overflowed(ScalarWriter writer) {
throw new IllegalStateException("overflow called");
}
@Override
public boolean canExpand(ScalarWriter writer, int delta) {
totalAlloc += delta;
return totalAlloc < 1024 * 1024;
}
});
writer.startWrite();
byte[] value = new byte[423];
Arrays.fill(value, (byte) 'X');
try {
for (int i = 0; ; i++) {
index.index = i;
writer.startRow();
writer.setBytes(value, value.length);
writer.saveRow();
}
} catch (IllegalStateException e) {
assertTrue(e.getMessage().contains("overflow called"));
}
}
}
use of org.apache.drill.exec.vector.accessor.ColumnAccessors.VarCharColumnWriter in project drill by apache.
the class TestVariableWidthWriter method testRolloverWithEmpties.
/**
* Simulate the case in which the tail end of an overflow
* batch has empties. <tt>preRollover()</tt> should back-fill
* them with the next offset prior to rollover.
*/
@Test
public void testRolloverWithEmpties() {
try (VarCharVector vector = allocVector(1000)) {
TestIndex index = new TestIndex();
VarCharColumnWriter writer = makeWriter(vector, index);
writer.startWrite();
// Simulate doing an overflow of 15 values,
// of which 5 are empty.
String base = "sample-value";
for (int i = 0; i < 10; i++) {
index.index = i;
writer.startRow();
writer.setString(base + i);
writer.saveRow();
}
for (int i = 10; i < 15; i++) {
index.index = i;
writer.startRow();
writer.saveRow();
}
// Overflow occurs before writing the 16th row
index.index = 15;
writer.startRow();
// Overflow occurs. This should fill empty offsets.
writer.preRollover();
for (int i = 0; i < 10; i++) {
assertEquals(base + i, stringAt(vector, i));
}
for (int i = 10; i < 15; i++) {
assertEquals("", stringAt(vector, i));
}
// Simulate rollover
byte[] dummy = new byte[] { (byte) 0x55 };
for (int i = 0; i < 500; i++) {
vector.getMutator().setSafe(i, dummy);
}
for (int i = 1; i < 15; i++) {
vector.getOffsetVector().getMutator().set(i, 0xdeadbeef);
}
vector.getMutator().setSafe(0, new byte[] {});
writer.postRollover();
index.index = 0;
writer.saveRow();
for (int i = 1; i < 5; i++) {
index.index = i;
writer.startRow();
writer.saveRow();
}
for (int i = 5; i < 10; i++) {
index.index = i;
writer.startRow();
writer.setString(base + (i + 20));
writer.saveRow();
}
writer.endWrite();
for (int i = 0; i < 5; i++) {
assertEquals("", stringAt(vector, i));
}
for (int i = 5; i < 10; i++) {
assertEquals(base + (i + 20), stringAt(vector, i));
}
}
}
use of org.apache.drill.exec.vector.accessor.ColumnAccessors.VarCharColumnWriter in project drill by apache.
the class TestVariableWidthWriter method testFillEmpties.
/**
* Filling empties in a variable-width row means carrying forward
* offsets (as tested elsewhere), leaving zero-length values.
*/
@Test
public void testFillEmpties() {
try (VarCharVector vector = allocVector(1000)) {
TestIndex index = new TestIndex();
VarCharColumnWriter writer = makeWriter(vector, index);
writer.startWrite();
// Write values, skipping four out of five positions,
// forcing backfill.
// The number of values is odd, forcing the writer to
// back-fill at the end as well as between values.
String base = "sample-value";
for (int i = 0; i < 501; i += 5) {
index.index = i;
writer.startRow();
writer.setString(base + i);
writer.saveRow();
}
// At end, vector index defined to point one past the
// last row. That is, the vector index gives the row count.
index.index = 504;
writer.endWrite();
for (int i = 0; i < 504; i++) {
assertEquals("Mismatch on " + i, (i % 5) == 0 ? base + i : "", stringAt(vector, i));
}
}
}
Aggregations