public virtual void TestUncompressedInputDefaultDelimiterPosValue() { Configuration conf = new Configuration(); string inputData = "1234567890\r\n12\r\n345"; Path inputFile = CreateInputFile(conf, inputData); conf.SetInt("io.file.buffer.size", 10); conf.SetInt(LineRecordReader.MaxLineLength, int.MaxValue); FileSplit split = new FileSplit(inputFile, 0, 15, (string[])null); LineRecordReader reader = new LineRecordReader(conf, split, null); LongWritable key = new LongWritable(); Text value = new Text(); reader.Next(key, value); // Get first record:"1234567890" NUnit.Framework.Assert.AreEqual(10, value.GetLength()); // Position should be 12 right after "1234567890\r\n" NUnit.Framework.Assert.AreEqual(12, reader.GetPos()); reader.Next(key, value); // Get second record:"12" NUnit.Framework.Assert.AreEqual(2, value.GetLength()); // Position should be 16 right after "1234567890\r\n12\r\n" NUnit.Framework.Assert.AreEqual(16, reader.GetPos()); NUnit.Framework.Assert.IsFalse(reader.Next(key, value)); split = new FileSplit(inputFile, 15, 4, (string[])null); reader = new LineRecordReader(conf, split, null); // The second split dropped the first record "\n" // The position should be 16 right after "1234567890\r\n12\r\n" NUnit.Framework.Assert.AreEqual(16, reader.GetPos()); reader.Next(key, value); // Get third record:"345" NUnit.Framework.Assert.AreEqual(3, value.GetLength()); // Position should be 19 right after "1234567890\r\n12\r\n345" NUnit.Framework.Assert.AreEqual(19, reader.GetPos()); NUnit.Framework.Assert.IsFalse(reader.Next(key, value)); NUnit.Framework.Assert.AreEqual(19, reader.GetPos()); inputData = "123456789\r\r\n"; inputFile = CreateInputFile(conf, inputData); split = new FileSplit(inputFile, 0, 12, (string[])null); reader = new LineRecordReader(conf, split, null); reader.Next(key, value); // Get first record:"123456789" NUnit.Framework.Assert.AreEqual(9, value.GetLength()); // Position should be 10 right after "123456789\r" NUnit.Framework.Assert.AreEqual(10, reader.GetPos()); reader.Next(key, value); // Get second record:"" NUnit.Framework.Assert.AreEqual(0, value.GetLength()); // Position should be 12 right after "123456789\r\r\n" NUnit.Framework.Assert.AreEqual(12, reader.GetPos()); NUnit.Framework.Assert.IsFalse(reader.Next(key, value)); NUnit.Framework.Assert.AreEqual(12, reader.GetPos()); }
/// <exception cref="System.IO.IOException"/> public virtual long GetPos() { lock (this) { return(lineRecordReader.GetPos()); } }
public virtual void TestUncompressedInputCustomDelimiterPosValue() { Configuration conf = new Configuration(); conf.SetInt("io.file.buffer.size", 10); conf.SetInt(LineRecordReader.MaxLineLength, int.MaxValue); string inputData = "abcdefghij++kl++mno"; Path inputFile = CreateInputFile(conf, inputData); string delimiter = "++"; byte[] recordDelimiterBytes = Sharpen.Runtime.GetBytesForString(delimiter, Charsets .Utf8); // the first split must contain two records to make sure that it also pulls // in the record from the 2nd split int splitLength = 15; FileSplit split = new FileSplit(inputFile, 0, splitLength, (string[])null); LineRecordReader reader = new LineRecordReader(conf, split, recordDelimiterBytes); LongWritable key = new LongWritable(); Text value = new Text(); // Get first record: "abcdefghij" NUnit.Framework.Assert.IsTrue("Expected record got nothing", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong length for record value", 10, value.GetLength ()); // Position should be 12 right after "abcdefghij++" NUnit.Framework.Assert.AreEqual("Wrong position after record read", 12, reader.GetPos ()); // Get second record: "kl" NUnit.Framework.Assert.IsTrue("Expected record got nothing", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong length for record value", 2, value.GetLength ()); // Position should be 16 right after "abcdefghij++kl++" NUnit.Framework.Assert.AreEqual("Wrong position after record read", 16, reader.GetPos ()); // Get third record: "mno" NUnit.Framework.Assert.IsTrue("Expected record got nothing", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong length for record value", 3, value.GetLength ()); // Position should be 19 right after "abcdefghij++kl++mno" NUnit.Framework.Assert.AreEqual("Wrong position after record read", 19, reader.GetPos ()); NUnit.Framework.Assert.IsFalse(reader.Next(key, value)); NUnit.Framework.Assert.AreEqual("Wrong position after record read", 19, reader.GetPos ()); reader.Close(); // No record is in the second split because the second split will drop // the first record, which was already reported by the first split. split = new FileSplit(inputFile, splitLength, inputData.Length - splitLength, (string [])null); reader = new LineRecordReader(conf, split, recordDelimiterBytes); // The position should be 19 right after "abcdefghij++kl++mno" and should // not change NUnit.Framework.Assert.AreEqual("Wrong position after record read", 19, reader.GetPos ()); NUnit.Framework.Assert.IsFalse("Unexpected record returned", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong position after record read", 19, reader.GetPos ()); reader.Close(); // multi char delimiter with starting part of the delimiter in the data inputData = "abcd+efgh++ijk++mno"; inputFile = CreateInputFile(conf, inputData); splitLength = 5; split = new FileSplit(inputFile, 0, splitLength, (string[])null); reader = new LineRecordReader(conf, split, recordDelimiterBytes); // Get first record: "abcd+efgh" NUnit.Framework.Assert.IsTrue("Expected record got nothing", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong position after record read", 11, reader.GetPos ()); NUnit.Framework.Assert.AreEqual("Wrong length for record value", 9, value.GetLength ()); // should have jumped over the delimiter, no record NUnit.Framework.Assert.IsFalse("Unexpected record returned", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong position after record read", 11, reader.GetPos ()); reader.Close(); // next split: check for duplicate or dropped records split = new FileSplit(inputFile, splitLength, inputData.Length - splitLength, (string [])null); reader = new LineRecordReader(conf, split, recordDelimiterBytes); // Get second record: "ijk" first in this split NUnit.Framework.Assert.IsTrue("Expected record got nothing", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong position after record read", 16, reader.GetPos ()); NUnit.Framework.Assert.AreEqual("Wrong length for record value", 3, value.GetLength ()); // Get third record: "mno" second in this split NUnit.Framework.Assert.IsTrue("Expected record got nothing", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong position after record read", 19, reader.GetPos ()); NUnit.Framework.Assert.AreEqual("Wrong length for record value", 3, value.GetLength ()); // should be at the end of the input NUnit.Framework.Assert.IsFalse(reader.Next(key, value)); NUnit.Framework.Assert.AreEqual("Wrong position after record read", 19, reader.GetPos ()); reader.Close(); inputData = "abcd|efgh|+|ij|kl|+|mno|pqr"; inputFile = CreateInputFile(conf, inputData); delimiter = "|+|"; recordDelimiterBytes = Sharpen.Runtime.GetBytesForString(delimiter, Charsets.Utf8 ); // walking over the buffer and split sizes checks for proper processing // of the ambiguous bytes of the delimiter for (int bufferSize = 1; bufferSize <= inputData.Length; bufferSize++) { for (int splitSize = 1; splitSize < inputData.Length; splitSize++) { conf.SetInt("io.file.buffer.size", bufferSize); split = new FileSplit(inputFile, 0, bufferSize, (string[])null); reader = new LineRecordReader(conf, split, recordDelimiterBytes); // Get first record: "abcd|efgh" always possible NUnit.Framework.Assert.IsTrue("Expected record got nothing", reader.Next(key, value )); NUnit.Framework.Assert.IsTrue("abcd|efgh".Equals(value.ToString())); NUnit.Framework.Assert.AreEqual("Wrong position after record read", 9, value.GetLength ()); // Position should be 12 right after "|+|" int recordPos = 12; NUnit.Framework.Assert.AreEqual("Wrong position after record read", recordPos, reader .GetPos()); // get the next record: "ij|kl" if the split/buffer allows it if (reader.Next(key, value)) { // check the record info: "ij|kl" NUnit.Framework.Assert.IsTrue("ij|kl".Equals(value.ToString())); // Position should be 20 right after "|+|" recordPos = 20; NUnit.Framework.Assert.AreEqual("Wrong position after record read", recordPos, reader .GetPos()); } // get the third record: "mno|pqr" if the split/buffer allows it if (reader.Next(key, value)) { // check the record info: "mno|pqr" NUnit.Framework.Assert.IsTrue("mno|pqr".Equals(value.ToString())); // Position should be 27 at the end of the string now recordPos = inputData.Length; NUnit.Framework.Assert.AreEqual("Wrong position after record read", recordPos, reader .GetPos()); } // no more records can be read we should still be at the last position NUnit.Framework.Assert.IsFalse("Unexpected record returned", reader.Next(key, value )); NUnit.Framework.Assert.AreEqual("Wrong position after record read", recordPos, reader .GetPos()); reader.Close(); } } }