Custom Parser Error

[MaximilianJHuber]

I am trying to parse a CSV and convert a column that holds the sex of a person into a Nullable{Bool}

My attempts failed:

sex_parser = CustomParser(Bool) do str, i, len, opts
    return (len == 0 ? Nullable{Bool}() : Nullable{Bool}(str[i] == 'M'), len == 0 ? i : i + 1)
end

I spent couple of hours reading the docs and the code, but I do not understand what the str is that the tryparsenext receives in a readcsv context, and whether my CustomParser must be aware of the delimiter symbol.

For example, consider a CSV with ';' delimiter:

123;F;6789;
124;M;6712;
125;;6716;

Is str a row and my custom parse receives pos=5, len=1 for the first row and pos=5, len=0 for the third row?
And after a successful parse in row one, do I return 6 or 7. The doc is saying "position the next token, if any, starts at", that sound like my parser needs to be aware of my delimiter!

[MaximilianJHuber]

I did some sleuthing and the str appears to be multiple (if not all) rows and len always is the same very big number.
However my new custom parser fails too:

sex_parser = CustomParser(Bool) do str, i, len, opts
    return (str[i] == ';' ? Nullable{Bool}(nothing) : Nullable{Bool}(str[i] == 'M'), str[i] == ';' ? i + 1: i + 2)
end

A JuliaDB loadtable call yields an error:

MethodError: no method matching isless(::TextParse.StrRange, ::TextParse.StrRange)
Closest candidates are:
  isless(::Missings.Missing, ::Any) at C:\Users\Max\AppData\Local\JuliaPro-0.6.2.1\pkgs-0.6.2.1\v0.6\Missings\src\Missings.jl:74
  isless(::DataValues.DataValue{Union{}}, ::Any) at C:\Users\Max\AppData\Local\JuliaPro-0.6.2.1\pkgs-0.6.2.1\v0.6\DataValues\src\scalar/core.jl:257
  isless(::DataValues.DataValue{S}, ::T) where {S, T} at C:\Users\Max\AppData\Local\JuliaPro-0.6.2.1\pkgs-0.6.2.1\v0.6\DataValues\src\scalar/core.jl:251
  ...
cmp at .\operators.jl:303 [inlined]
cmpelts at C:\Users\Max\AppData\Local\JuliaPro-0.6.2.1\pkgs-0.6.2.1\v0.6\IndexedTables\src\columns.jl:343 [inlined]
macro expansion at C:\Users\Max\AppData\Local\JuliaPro-0.6.2.1\pkgs-0.6.2.1\v0.6\IndexedTables\src\columns.jl:363 [inlined]
....

[MaximilianJHuber]

I found a solution:

sex_parser = CustomParser(Bool) do str, i, len, opts
    return (DataValues.DataValue{Bool}(str[i] == 'M'), i + 1)
end

This inner parser is wrapped in a TextParse.NAToken(sex_parser). This way an empty field is taken care of by NAToken and I can be sure that str[i] is not empty.

Might be worth adding to the docs!