Convert IEnumerable<Reads> to CSV

Question

I have an IEnumerable of a view model that I need to convert to CSV and send back as a file response. I have written it out quite manually and I'm just seeking code review and suggestions on a more elegant solution, or alternatively if this looks acceptable, confirmation of that:

 public ActionResult ExportBasicReads(string nmi, DateTime fromDate, DateTime toDate)
 {
 var reads = _dataService.GetBasicReads(nmi, fromDate, toDate)
 .Where(r => r.CurrentReadQuality != ReadQuality.Estimate)
 .OrderBy(r => r.CurrentReadDate)
 .ThenBy(r => r.DeviceKey)
 .ThenBy(r => r.ValidationCode);
 StringBuilder csv = new StringBuilder();
 csv.AppendLine("Read Date,Device : Suffix,Read,Quality,Tran Code,Service Order,Previous Read Date,Previous Read,Previous Quality," +
 "Previous Tran Code,Dial Diff,Quantity,UOM,Dial Factor,Direction,Validation,From Date,To Date,Kwh,Avg Daily Load,Profile Name," +
 "Profle Area,Reading Timestamp,Update Date,ChangedBy,ChangeType,ChangeComment");
 reads.ForEach(x => csv.AppendLine(
 string.Format("{0},{1} : {2},{3},{4},{5},{6},{7},{8},{9},{10},{11},{12},{13},{14},{15},{16},{17},{18},{19},{20},{21},{22}," +
 "{23},{24},{25},{26},{27}",
 x.CurrentReadDate,
 x.DeviceKey,
 x.Suffix,
 x.CurrentReadRawValue,
 x.CurrentReadQualityDesc,
 x.CurrentReadTransactionCode,
 x.CurrentServiceOrder,
 x.PreviousReadDate,
 x.PreviousReadRawValue,
 x.PreviousReadQualityDesc,
 x.PreviousReadTransactionCode,
 x.DialDifference,
 x.Quantity,
 x.UOMDesc,
 x.Multiplier,
 x.DirectionDesc,
 x.ValidationDesc,
 x.EffectiveFromDateTime,
 x.EffectiveToDateTime,
 x.Kwh,
 x.AverageDailyLoad,
 x.ProfileName,
 x.ProfileArea,
 x.EffectiveFromDateTime,
 x.UpdateTimestamp,
 x.ChangedBy,
 x.ChangeType,
 x.ChangeComment
 )
 )); 
 return File(Encoding.ASCII.GetBytes(csv.ToString()), "text/csv", "BasicReadsExport.csv");
 }

• vasily.sib: You can use string interpolation instead of string.Format()?
• paparazzo: Looks pretty clean to me.
• Jesse C. Slicer: No need to reinvent - the popular ServiceStack.Text NuGet package has an extension method called .ToCsv(): docs.servicestack.net/csv-format
• Zer0: the string.Format is awfully hard to follow given the large number of parameters. I would suggest something like $"{x.CurrentReadDate},{x.DeviceKey}" .... This would also help avoid errors due to matching up parameter indexes if this code was later modified.
• Coda: I'd create an extention method such as ToCvsLine(...) and use string interpolation inside that instead of string.Format. Then you just need to call it this way: reads.ForEach(_ => csv.AppendLine(_.ToCsvLine()))

Answer 1

I modified the formatting part for better readability and maintainability, at the cost of lower performance:

var columns = new Dictionary<string, Func<Reads, object>>
{
 ["Read Date"] = x => x.CurrentReadDate,
 ["Device"] = x => x.DeviceKey,
 ["Suffix"] = x => x.Suffix,
 ["Read"] = x => x.CurrentReadRawValue,
 // add the rest of headers/property selectors below
};
//var format = "{0},{1} : {2},{3},{4},{5},{6},{7},{8},{9},{10},{11},{12},{13},{14},{15},{16},{17},{18},{19},{20},{21},{22},{23},{24},{25},{26},{27}";
var format = "{0}{1} : " + string.Join(",", columns.Select((_, i) => $"{{{i}}}").Skip(2));
var csv = new StringBuilder()
 .AppendFormat(format, columns.Keys.ToArray())
 .AppendLine();
foreach (var read in reads)
{
 var values = columns.Values.Select(selector => selector(read));
 csv
 .AppendFormat(format, values.ToArray())
 .AppendLine();
}

The advantage of this approach is that you can add/move/remove the columns without having to worry about updating the header and the format, and if they are kept in sync or not.

Note:

• You can also revert to the constant format if it becomes more complex.
• The property selector can be chained if you need special format on individual property, like: x => x.CurrentReadDate.ToString("yyyyMMddThhmm")