CSV with empty values for integer fields throws

(Row: 1, Col: 3) Unexpected end of input when converting from type string to type int

Code that parses the CSV:

```
import std.algorithm;
import std.array;
import std.csv;
import std.stdio;
import std.conv;
import std.range;

private struct IdentifyResult
{
    string probe;
    string target;
    int rank;
    int score;
    bool vendorMatch;
}

void main(string[] args)
{
    if (args.length < 2)
    {
        writeln("Usage: ", args[0], " <csv-file-1> ... <csv-file-n>");
        return;
    }

    foreach (i; 1 .. args.length)
    {
        try
        {
            auto arr = File(args[i], "r").byLine.joiner("\n").csvReader!IdentifyResult.array;
            writeln(args[i], "\tValid");
        }
        catch (std.csv.CSVException e)
        {
            writeln(e);
            writeln(args[i], "\tInvalid");
        }
    }
}
```

Input CSV file below

CSV header (just for reference, but not part of the CSV)
Probe,Target,rank,score,match/nomatch,datetime,position,score
```
LIP_0905_1230.nist,,,,FALSE,2017-09-05 23:24:37,,
LIP_0905_1230.nist,,,,FALSE,2017-10-12 11:37:29,,
LIP_0905_1230.nist,,,,FALSE,2017-10-12 11:51:03,,
LIP_0905_1230.nist,,,,FALSE,2017-10-12 12:07:21,,
LIP_0905_1230.nist,CRM_1012_1920.nist,1,9999,true,2017-10-12 19:56:00,25,9999
LIP_0905_1230.nist,CRM_1012_1920.nist,1,9999,true,2017-10-13 00:55:00,25,9999
LIP_0905_1230.nist,CRM_1013_0005.nist,2,9999,true,2017-10-13 00:55:00,25,9999
LIP_0905_1230.nist,CRM_1012_1920.nist,1,9999,true,2017-10-18 18:27:22,25,9999
LIP_0905_1230.nist,CRM_1013_0005.nist,2,9999,true,2017-10-18 18:27:22,25,9999
LIP_0905_1230.nist,CRM_1013_0005.nist,1,9999,true,2017-10-20 11:04:31,25,9999
```


Is there anyway to overcome this without modifying the original CSV?


Cheers,
Arun

Not really you'll need to parse it out as a string and do the conversion later.

It probably would be good to support nullable!int pretty sure it doesn't currently.

On Sunday, 29 October 2017 at 15:45:23 UTC, Jesse Phillips wrote:
> Not really you'll need to parse it out as a string and do the conversion later.
>
> It probably would be good to support nullable!int pretty sure it doesn't currently.

Should I raise a ticket on Bugzilla to address this?