Reputation: 2292
I've got a text file full of records where each field in each record is a fixed width. My first approach would be to parse each record simply using string.Substring(). Is there a better way?
For example, the format could be described as:
<Field1(8)><Field2(16)><Field3(12)>
And an example file with two records could look like:
SomeData0000000000123456SomeMoreData
Data2 0000000000555555MoreData
I just want to make sure I'm not overlooking a more elegant way than Substring().
Update: I ultimately went with a regex like Killersponge suggested:
private readonly Regex reLot = new Regex(REGEX_LOT, RegexOptions.Compiled);
const string REGEX_LOT = "^(?<Field1>.{6})" +
"(?<Field2>.{16})" +
"(?<Field3>.{12})";
I then use the following to access the fields:
Match match = reLot.Match(record);
string field1 = match.Groups["Field1"].Value;
Upvotes: 22
Views: 34435
Reputation: 4632
As mentioned in Colonel Panic's answer, you can use .NET's TextFieldParser class. The sample code provided in the docs is for VB. Here is sample code for C#:
using var fileReader = new Microsoft.VisualBasic.FileIO.TextFieldParser("myFile.txt");
fileReader.TextFieldType = Microsoft.VisualBasic.FileIO.FieldType.FixedWidth;
// In this example, the columns widths are 5, 10, and 2 characters long
fileReader.FieldWidths = [5, 10, 2];
var lines = new List<string[]>();
while(!fileReader.EndOfData)
{
try
{
var fields = fileReader.ReadFields();
lines.Add(fields ?? []);
}
catch(Exception)
{
// Ignore for this sample code
}
}
fileReader.Close();
// Write the output to a csv file
// Note that a proper csv writer library should be used
// This is for demonstration purposes only
File.WriteAllLines(fullFilePath + ".csv", lines.Select(line => String.Join(',', line)));
Upvotes: 0
Reputation: 2509
There is open-source .Net lib for this:
https://github.com/borisdj/FixedWidthParserWriter/
PS I'm the author.
Upvotes: 0
Reputation: 137594
Why reinvent the wheel? Use .NET's TextFieldParser class per this how-to for Visual Basic: How to read from fixed-width text files.
Upvotes: 9
Reputation: 577
Use FileHelpers.
Example:
[FixedLengthRecord()]
public class MyData
{
[FieldFixedLength(8)]
public string someData;
[FieldFixedLength(16)]
public int SomeNumber;
[FieldFixedLength(12)]
[FieldTrim(TrimMode.Right)]
public string someMoreData;
}
Then, it's as simple as this:
var engine = new FileHelperEngine<MyData>();
// To Read Use:
var res = engine.ReadFile("FileIn.txt");
// To Write Use:
engine.WriteFile("FileOut.txt", res);
Upvotes: 34
Reputation: 6746
You could set up an ODBC data source for the fixed format file, and then access it as any other database table. This has the added advantage that specific knowledge of the file format is not compiled into your code for that fateful day that someone decides to stick an extra field in the middle.
Upvotes: 0
Reputation: 95432
Unfortunately out of the box the CLR only provides Substring for this.
Someone over at CodeProject made a custom parser using attributes to define fields, you might wanna look at that.
Upvotes: 1
Reputation: 4479
You may have to watch out, if the end of the lines aren't padded out with spaces to fill the field, your substring won't work without a bit of fiddling to work out how much more of the line there is to read. This of course only applies to the last field :)
Upvotes: 2
Reputation: 1500923
Substring sounds good to me. The only downside I can immediately think of is that it means copying the data each time, but I wouldn't worry about that until you prove it's a bottleneck. Substring is simple :)
You could use a regex to match a whole record at a time and capture the fields, but I think that would be overkill.
Upvotes: 8