CsvFieldIndexer: Fix calculation of offsets with unicode characters.

Fixes #4
This commit is contained in:
2023-08-18 15:47:09 +02:00
parent fb6d8d76a1
commit 6ed9718abb
4 changed files with 79 additions and 17 deletions

View File

@@ -137,6 +137,39 @@ public class CsvFieldIndexerTests
Assert.Equal(29, indexer.FieldIndex[1][3]);
}
[Fact]
public void GenerateIndex__TwoLinesWithTwoQuotedColumnsWithUnicode__TwoRowsTwoFields()
{
// --- Arrange
StringReader sr = new("""
"Hélló","Wórld"
"Hélló","Wórld"
""");
// --- Act
CsvFieldIndexer indexer = new();
indexer.GenerateIndex(sr);
// --- Assert
Assert.Equal(3, indexer.Index.Count);
Assert.Equal(0, indexer.Index[0]);
Assert.Equal(19, indexer.Index[1]);
Assert.Equal(38, indexer.Index[2]);
Assert.Equal(2, indexer.FieldIndex.Count);
Assert.Equal(4, indexer.FieldIndex[0].Count);
Assert.Equal(1, indexer.FieldIndex[0][0]);
Assert.Equal(7, indexer.FieldIndex[0][1]);
Assert.Equal(11, indexer.FieldIndex[0][2]);
Assert.Equal(16, indexer.FieldIndex[0][3]);
Assert.Equal(4, indexer.FieldIndex[1].Count);
Assert.Equal(20, indexer.FieldIndex[1][0]);
Assert.Equal(26, indexer.FieldIndex[1][1]);
Assert.Equal(30, indexer.FieldIndex[1][2]);
Assert.Equal(35, indexer.FieldIndex[1][3]);
}
#endregion GenerateIndex
}