public void testDump()
{
// conf.set(HiveConf.ConfVars.HIVE_ORC_ENCODING_STRATEGY.varname, "COMPRESSION");
using (Stream file = File.OpenWrite(TestFilePath))
{
OrcFile.WriterOptions options = new OrcFile.WriterOptions(new Properties(), conf);
options.inspector(ObjectInspectorFactory.getReflectionObjectInspector(typeof(MyRecord)));
options.stripeSize(100000);
options.compress(CompressionKind.ZLIB);
options.bufferSize(10000);
options.rowIndexStride(1000);
using (Writer writer = OrcFile.createWriter(TestFilePath, file, options))
{
Random r1 = new Random(1);
for (int i = 0; i < 21000; ++i)
{
writer.addRow(new MyRecord(r1.Next(), r1.NextLong(),
TestHelpers.words[r1.Next(TestHelpers.words.Length)]));
}
}
}
string outputFilename = "orc-file-dump.out";
using (CaptureStdout capture = new CaptureStdout(Path.Combine(workDir, outputFilename)))
{
FileDump.Main(new string[] { TestFilePath.ToString(), "--rowindex=1,2,3" });
}
TestHelpers.CompareFilesByLine(outputFilename, Path.Combine(workDir, outputFilename));
}
public void SimpleTest()
{
OrcFile.WriterOptions options = new OrcFile.WriterOptions(new Properties(), new Configuration());
options.inspector(PrimitiveObjectInspectorFactory.writableStringObjectInspector);
using (Stream file = File.Create(filename))
using (Writer writer = OrcFile.createWriter(filename, file, options))
{
writer.addRow("hello");
}
Reader reader = OrcFile.createReader(() => File.OpenRead(filename), filename);
using (RecordReader recordReader = reader.rows())
{
object value = recordReader.next();
Assert.True(value is string);
Assert.Equal("hello", value);
}
}
public void testDictionaryThreshold()
{
// conf.set(HiveConf.ConfVars.HIVE_ORC_ENCODING_STRATEGY.varname, "COMPRESSION");
// conf.setFloat(HiveConf.ConfVars.HIVE_ORC_DICTIONARY_KEY_SIZE_THRESHOLD.varname, 0.49f);
using (Stream file = File.OpenWrite(TestFilePath))
{
OrcFile.WriterOptions options = new OrcFile.WriterOptions(new Properties(), conf);
options.inspector(ObjectInspectorFactory.getReflectionObjectInspector(typeof(MyRecord)));
options.stripeSize(100000);
options.compress(CompressionKind.ZLIB);
options.bufferSize(10000);
options.rowIndexStride(1000);
using (Writer writer = OrcFile.createWriter(TestFilePath, file, options))
{
Random r1 = new Random(1);
int nextInt = 0;
for (int i = 0; i < 21000; ++i)
{
// Write out the same string twice, this guarantees the fraction of rows with
// distinct strings is 0.5
if (i % 2 == 0)
{
nextInt = r1.Next(TestHelpers.words.Length);
// Append the value of i to the word, this guarantees when an index or word is repeated
// the actual string is unique.
TestHelpers.words[nextInt] += "-" + i;
}
writer.addRow(new MyRecord(r1.Next(), r1.NextLong(), TestHelpers.words[nextInt]));
}
}
}
string outputFilename = "orc-file-dump-dictionary-threshold.out";
using (CaptureStdout capture = new CaptureStdout(Path.Combine(workDir, outputFilename)))
{
FileDump.Main(new string[] { TestFilePath.ToString(), "--rowindex=1,2,3" });
}
TestHelpers.CompareFilesByLine(outputFilename, Path.Combine(workDir, outputFilename));
}
public void testJsonDump()
{
ObjectInspector inspector;
inspector = ObjectInspectorFactory.getReflectionObjectInspector(typeof(MyRecord));
// conf.set(HiveConf.ConfVars.HIVE_ORC_ENCODING_STRATEGY.varname, "COMPRESSION");
OrcFile.WriterOptions options = OrcFile.writerOptions(conf)
.inspector(inspector)
.stripeSize(100000)
.compress(CompressionKind.ZLIB)
.bufferSize(10000)
.rowIndexStride(1000)
.bloomFilterColumns("s");
using (Stream file = File.OpenWrite(TestFilePath))
using (Writer writer = OrcFile.createWriter(TestFilePath, file, options))
{
Random r1 = new Random(1);
for (int i = 0; i < 21000; ++i)
{
if (i % 100 == 0)
{
writer.addRow(new MyRecord(r1.Next(), r1.NextLong(), null));
}
else
{
writer.addRow(new MyRecord(r1.Next(), r1.NextLong(),
TestHelpers.words[r1.Next(TestHelpers.words.Length)]));
}
}
}
const string outputFilename = "orc-file-dump.json";
using (CaptureStdout capture = new CaptureStdout(Path.Combine(workDir, outputFilename)))
{
FileDump.Main(new string[] { TestFilePath.ToString(), "-j", "-p", "--rowindex=3" });
}
TestHelpers.CompareFilesByLine(outputFilename, Path.Combine(workDir, outputFilename));
}
public void testDataDump()
{
using (Stream file = File.OpenWrite(TestFilePath))
{
OrcFile.WriterOptions options = new OrcFile.WriterOptions(new Properties(), conf);
options.inspector(ObjectInspectorFactory.getReflectionObjectInspector(typeof(AllTypesRecord)));
options.stripeSize(100000);
options.compress(CompressionKind.NONE);
options.bufferSize(10000);
options.rowIndexStride(1000);
using (Writer writer = OrcFile.createWriter(TestFilePath, file, options))
{
Dictionary <string, string> m = new Dictionary <string, string>(2);
m.Add("k1", "v1");
writer.addRow(new AllTypesRecord(
true,
(sbyte)10,
(short)100,
1000,
10000L,
4.0f,
20.0,
HiveDecimal.Parse("4.2222"),
new Timestamp(1416967764000L),
new Date(1416967764000L),
"string",
m,
new List <int> {
100, 200
},
new AllTypesRecord.Struct(10, "foo")));
m.Clear();
m.Add("k3", "v3");
writer.addRow(new AllTypesRecord(
false,
(sbyte)20,
(short)200,
2000,
20000L,
8.0f,
40.0,
HiveDecimal.Parse("2.2222"),
new Timestamp(1416967364000L),
new Date(1411967764000L),
"abcd",
m,
new List <int> {
200, 300
},
new AllTypesRecord.Struct(20, "bar")));
}
}
string[] lines;
using (CaptureStdoutToMemory capture = new CaptureStdoutToMemory())
{
FileDump.Main(TestFilePath, "-d");
lines = capture.Text.Split(new[] { '\n' }, StringSplitOptions.RemoveEmptyEntries);
}
Assert.Equal(2, lines.Length);
// Don't be fooled by the big space in the middle, this line is quite long
Assert.Equal("{\"b\":true,\"bt\":10,\"s\":100,\"i\":1000,\"l\":10000,\"f\":4,\"d\":20,\"de\":\"4.2222\",\"t\":\"2014-11-25 18:09:24\",\"dt\":\"2014-11-25\",\"str\":\"string\",\"c\":\"hello \",\"vc\":\"hello\",\"m\":[{\"_key\":\"k1\",\"_value\":\"v1\"}],\"a\":[100,200],\"st\":{\"i\":10,\"s\":\"foo\"}}", lines[0]);
Assert.Equal("{\"b\":false,\"bt\":20,\"s\":200,\"i\":2000,\"l\":20000,\"f\":8,\"d\":40,\"de\":\"2.2222\",\"t\":\"2014-11-25 18:02:44\",\"dt\":\"2014-09-28\",\"str\":\"abcd\",\"c\":\"world \",\"vc\":\"world\",\"m\":[{\"_key\":\"k3\",\"_value\":\"v3\"}],\"a\":[200,300],\"st\":{\"i\":20,\"s\":\"bar\"}}", lines[1]);
}
