/**
* Is this stream part of a dictionary?
* @return is this part of a dictionary?
*/
public static bool isDictionary(OrcProto.Stream.Types.Kind kind,
OrcProto.ColumnEncoding encoding)
{
Debug.Assert(kind != OrcProto.Stream.Types.Kind.DICTIONARY_COUNT);
OrcProto.ColumnEncoding.Types.Kind encodingKind = encoding.Kind;
return(kind == OrcProto.Stream.Types.Kind.DICTIONARY_DATA ||
(kind == OrcProto.Stream.Types.Kind.LENGTH &&
(encodingKind == OrcProto.ColumnEncoding.Types.Kind.DICTIONARY ||
encodingKind == OrcProto.ColumnEncoding.Types.Kind.DICTIONARY_V2)));
}
public void testHalfDistinctCheckDisabled()
{
ObjectInspector inspector = ObjectInspectorFactory.getReflectionObjectInspector(typeof(string));
int[] input = new int[20000];
// conf.setBoolean(ConfVars.HIVE_ORC_ROW_INDEX_STRIDE_DICTIONARY_CHECK.varname, false);
using (Stream file = File.OpenWrite(TestFilePath))
using (Writer writer = OrcFile.createWriter(TestFilePath, file, OrcFile.writerOptions(conf)
.inspector(inspector)
.compress(CompressionKind.NONE)
.bufferSize(10000)))
{
Random rand = new Random(123);
for (int i = 0; i < 20000; i++)
{
input[i] = rand.Next(10000);
}
for (int i = 0; i < 20000; i++)
{
writer.addRow(input[i].ToString());
}
}
Reader reader = OrcFile.createReader(TestFilePath, OrcFile.readerOptions(conf));
using (RecordReader rows = reader.rows())
{
int idx = 0;
while (rows.hasNext())
{
object row = rows.next();
Assert.Equal(input[idx++].ToString(), row);
}
// make sure the encoding type is correct
foreach (StripeInformation stripe in reader.getStripes())
{
// hacky but does the job, this casting will work as long this test resides
// within the same package as ORC reader
OrcProto.StripeFooter footer = ((RecordReaderImpl)rows).readStripeFooter(stripe);
for (int i = 0; i < footer.ColumnsCount; ++i)
{
OrcProto.ColumnEncoding encoding = footer.GetColumns(i);
Assert.Equal(OrcProto.ColumnEncoding.Types.Kind.DICTIONARY_V2, encoding.Kind);
}
}
}
}
public void testTooManyDistinctV11AlwaysDictionary()
{
ObjectInspector inspector = ObjectInspectorFactory.getReflectionObjectInspector(typeof(string));
using (Stream file = File.OpenWrite(TestFilePath))
using (Writer writer = OrcFile.createWriter(TestFilePath, file, OrcFile.writerOptions(conf)
.inspector(inspector)
.compress(CompressionKind.NONE)
.version(OrcFile.Version.V_0_11)
.bufferSize(10000)))
{
for (int i = 0; i < 20000; i++)
{
writer.addRow(i.ToString());
}
}
Reader reader = OrcFile.createReader(TestFilePath, OrcFile.readerOptions(conf));
using (RecordReader rows = reader.rows())
{
int idx = 0;
while (rows.hasNext())
{
object row = rows.next();
Assert.Equal((idx++).ToString(), row);
}
// make sure the encoding type is correct
foreach (StripeInformation stripe in reader.getStripes())
{
// hacky but does the job, this casting will work as long this test resides
// within the same package as ORC reader
OrcProto.StripeFooter footer = ((RecordReaderImpl)rows).readStripeFooter(stripe);
for (int i = 0; i < footer.ColumnsCount; ++i)
{
OrcProto.ColumnEncoding encoding = footer.GetColumns(i);
Assert.Equal(OrcProto.ColumnEncoding.Types.Kind.DICTIONARY, encoding.Kind);
}
}
}
}
public static void addRgFilteredStreamToRanges(OrcProto.Stream stream,
bool[] includedRowGroups, bool isCompressed, OrcProto.RowIndex index,
OrcProto.ColumnEncoding encoding, OrcProto.Type type, int compressionSize, bool hasNull,
long offset, long length, DiskRangeList.CreateHelper list, bool doMergeBuffers)
{
for (int group = 0; group < includedRowGroups.Length; ++group)
{
if (!includedRowGroups[group])
{
continue;
}
int posn = getIndexPosition(
encoding.Kind, type.Kind, stream.Kind, isCompressed, hasNull);
long start = (long)index.EntryList[group].PositionsList[posn];
long nextGroupOffset;
bool isLast = group == (includedRowGroups.Length - 1);
nextGroupOffset = isLast ? length : (int)index.EntryList[group + 1].PositionsList[posn];
start += offset;
long end = offset + estimateRgEndOffset(
isCompressed, isLast, nextGroupOffset, length, compressionSize);
list.addOrMerge(start, end, doMergeBuffers, true);
}
}
public static void printJsonMetaData(List <string> files, Configuration conf,
List <int> rowIndexCols, bool prettyPrint, bool printTimeZone)
{
JsonWriter writer = new JsonWriter();
bool multiFile = files.Count > 1;
if (multiFile)
{
writer.array();
}
else
{
writer.newObject();
}
foreach (string filename in files)
{
if (multiFile)
{
writer.newObject();
}
writer.key("fileName").value(Path.GetFileName(filename));
Reader reader = OrcFile.createReader(filename, OrcFile.readerOptions(conf));
writer.key("fileVersion").value(OrcFile.VersionHelper.getName(reader.getFileVersion()));
writer.key("writerVersion").value(reader.getWriterVersion().ToString());
using (RecordReaderImpl rows = (RecordReaderImpl)reader.rows())
{
writer.key("numberOfRows").value(reader.getNumberOfRows());
writer.key("compression").value(reader.getCompression().ToString());
if (reader.getCompression() != CompressionKind.NONE)
{
writer.key("compressionBufferSize").value(reader.getCompressionSize());
}
writer.key("schemaString").value(reader.getObjectInspector().getTypeName());
writer.key("schema").array();
writeSchema(writer, reader.getTypes());
writer.endArray();
writer.key("stripeStatistics").array();
List <StripeStatistics> stripeStatistics = reader.getStripeStatistics();
for (int n = 0; n < stripeStatistics.Count; n++)
{
writer.newObject();
writer.key("stripeNumber").value(n + 1);
StripeStatistics ss = stripeStatistics[n];
writer.key("columnStatistics").array();
for (int i = 0; i < ss.getColumnStatistics().Length; i++)
{
writer.newObject();
writer.key("columnId").value(i);
writeColumnStatistics(writer, ss.getColumnStatistics()[i]);
writer.endObject();
}
writer.endArray();
writer.endObject();
}
writer.endArray();
ColumnStatistics[] stats = reader.getStatistics();
int colCount = stats.Length;
writer.key("fileStatistics").array();
for (int i = 0; i < stats.Length; ++i)
{
writer.newObject();
writer.key("columnId").value(i);
writeColumnStatistics(writer, stats[i]);
writer.endObject();
}
writer.endArray();
writer.key("stripes").array();
int stripeIx = -1;
foreach (StripeInformation stripe in reader.getStripes())
{
++stripeIx;
long stripeStart = stripe.getOffset();
OrcProto.StripeFooter footer = rows.readStripeFooter(stripe);
writer.newObject(); // start of stripe information
writer.key("stripeNumber").value(stripeIx + 1);
writer.key("stripeInformation");
writeStripeInformation(writer, stripe);
if (printTimeZone)
{
writer.key("writerTimezone").value(
footer.HasWriterTimezone ? footer.WriterTimezone : FileDump.UNKNOWN);
}
long sectionStart = stripeStart;
writer.key("streams").array();
foreach (OrcProto.Stream section in footer.StreamsList)
{
writer.newObject();
string kind = section.HasKind ? section.Kind.ToString() : FileDump.UNKNOWN;
writer.key("columnId").value(section.Column);
writer.key("section").value(kind);
writer.key("startOffset").value(sectionStart);
writer.key("length").value(section.Length);
sectionStart += (long)section.Length;
writer.endObject();
}
writer.endArray();
writer.key("encodings").array();
for (int i = 0; i < footer.ColumnsCount; ++i)
{
writer.newObject();
OrcProto.ColumnEncoding encoding = footer.ColumnsList[i];
writer.key("columnId").value(i);
writer.key("kind").value(encoding.Kind.ToString());
if (encoding.Kind == OrcProto.ColumnEncoding.Types.Kind.DICTIONARY ||
encoding.Kind == OrcProto.ColumnEncoding.Types.Kind.DICTIONARY_V2)
{
writer.key("dictionarySize").value(encoding.DictionarySize);
}
writer.endObject();
}
writer.endArray();
if (rowIndexCols != null && rowIndexCols.Count != 0)
{
// include the columns that are specified, only if the columns are included, bloom filter
// will be read
bool[] sargColumns = new bool[colCount];
foreach (int colIdx in rowIndexCols)
{
sargColumns[colIdx] = true;
}
RecordReaderImpl.Index indices = rows.readRowIndex(stripeIx, null, sargColumns);
writer.key("indexes").array();
foreach (int col in rowIndexCols)
{
writer.newObject();
writer.key("columnId").value(col);
writeRowGroupIndexes(writer, col, indices.getRowGroupIndex());
writeBloomFilterIndexes(writer, col, indices.getBloomFilterIndex());
writer.endObject();
}
writer.endArray();
}
writer.endObject(); // end of stripe information
}
writer.endArray();
long fileLen = new FileInfo(filename).Length;
long paddedBytes = FileDump.getTotalPaddingSize(reader);
// empty ORC file is ~45 bytes. Assumption here is file length always >0
double percentPadding = ((double)paddedBytes / (double)fileLen) * 100;
writer.key("fileLength").value(fileLen);
writer.key("paddingLength").value(paddedBytes);
writer.key("paddingRatio").value(percentPadding);
rows.close();
}
writer.endObject();
}
if (multiFile)
{
writer.endArray();
}
if (prettyPrint)
{
#if false
string prettyJson;
if (multiFile)
{
JSONArray jsonArray = new JSONArray(writer.toString());
prettyJson = jsonArray.toString(2);
}
else
{
JSONObject jsonObject = new JSONObject(writer.toString());
prettyJson = jsonObject.toString(2);
}
#else
string prettyJson = writer.ToString();
#endif
System.Console.WriteLine(prettyJson);
}
else
{
System.Console.WriteLine(writer.ToString());
}
}
public StreamReaderBuilder setColumnEncoding(OrcProto.ColumnEncoding encoding)
{
this.columnEncoding = encoding;
return this;
}