I'm reading csv file:
string line;
StreamReader sr = new StreamReader(file.ToString());
while ((line = sr.ReadLine()) != null)
{
string col1 = line.Split(',')[10]; //old value
col1 = "my value"; //new value
}
sr.Close();
sr.Dispose();
I want to replace old value by the new.
Then I need to save the file with the changes.
How can I do that?
I suggest using File class instead of Streams and Readers. Linq is very convenient when querying data:
var modifiedData = File
.ReadLines(file.ToString())
.Select(line => line.Split(','))
.Select(items => {
//TODO: put relevant logic here: given items we should return csv line
items[10] = "my value";
return string.Join(",", items);
})
.ToList(); // <- we have to store modified data in memory
File.WriteAllLines(file.ToString(), modifiedData);
Another possibility (say, when initial file is too long to fit memory) is to save the modified data into a temporary file and then Move it:
var modifiedData = File
.ReadLines(file.ToString())
.Select(line => line.Split(','))
.Select(items => {
//TODO: put relevant logic here: given items we should return csv line
items[10] = "my value";
return string.Join(",", items);
});
string tempFile = Path.Combine(Path.GetTempPath(), $"{Guid.NewGuid()}.tmp");
File.WriteAllLines(tempFile, modifiedData);
File.Delete(file.ToString());
File.Move(tempFile, file.ToString());
modifiedData is IEnumerable<string>, not a single string. You can put a loop, e.g. foreach (var line in modifiedData) {sw.WriteLine(line);}Last() or LastOrDefault(), e.g. string lastLine = modifiedData.Last();Reading an entire file at once is memory-expensive. Not to mention creating its parallel copy. Using streams can fix it. Try this:
void Modify()
{
using (var fs = new FileStream(file, FileMode.Open, FileAccess.ReadWrite))
{
string line;
long position;
while ((line = fs.ReadLine(out position)) != null)
{
var tmp = line.Split(',');
tmp[1] = "00"; // new value
var newLine = string.Join(",", tmp);
fs.WriteLine(position, newLine);
}
}
}
with extensions:
static class FileStreamExtensions
{
private static readonly char[] newLine = Environment.NewLine.ToCharArray();
private static readonly int length = Environment.NewLine.Length;
private static readonly char eof = '\uFFFF';
public static string ReadLine(this FileStream fs, out long position)
{
position = fs.Position;
var chars = new List<char>();
char c;
while ((c = (char)fs.ReadByte()) != eof && (chars.Count < length || !chars.Skip(chars.Count - 2).SequenceEqual(newLine)))
{
chars.Add(c);
}
fs.Position--;
if (chars.Count == 0)
return null;
return new string(chars.ToArray());
}
public static void WriteLine(this FileStream fs, long position, string line)
{
var bytes = line.ToCharArray().Concat(newLine).Select(c => (byte)c).ToArray();
fs.Position = position;
fs.Write(bytes, 0, bytes.Length);
}
}
The shortcoming is you must keep your values the same length. E.g. 999 and __9 are both of length 3. Fixing this makes things much more complicated, so I'd leave it this way.
using (StreamReader sr = new StreamReader(file.ToString())) {...}is far better (no resource leakage on exception, more readable and maintable) then explitDisposecall