private void AddFanoutHanzi(DfdDto d)
{
var allHanziList = new List <string>();
allHanziList.Add(HanziToString(d.HanziCanonical));
allHanziList.AddRange(d.HanziAlternatives.Select(h => HanziToString(h)).ToList());
var fanOutHanziList = _hanziVariantsUtil.GetFanoutVariants(allHanziList.ToArray());
d.HanziMatchable.AddRange(fanOutHanziList);
}
public IEnumerable <DfdDto> Run()
{
var jsonOutput = new List <string>();
var documents = new List <DfdDto>();
using (var reader = new StreamReader(_dfdCharactersFile))
{
using (var csv = new CsvReader(reader, CultureInfo.InvariantCulture))
{
var records = csv.GetRecords <DFDRow>();
foreach (var r in records)
{
try
{
(var sInitial, var sFinal, var sTone) = Parse(r.Buc);
var document = new DfdDto
{
HanziCanonical = StringToHanziProto(r.Hanzi),
Initial = sInitial,
Final = sFinal,
Tone = sTone,
Buc = r.Buc,
DFDSourceInfo = new DFDSourceInfo()
{
PageNumber = r.PageNumber,
ColumnNumber = r.ColumnNumber,
LineNumber = r.LineNumber,
RadicalId = r.RadicalId
}
};
if (r.HanziAlt.Length > 0)
{
document.HanziAlternatives.Add(StringToHanziProto(r.HanziAlt));
}
AddFanoutHanzi(document);
documents.Add(document);
jsonOutput.Add(document.ToString());
}
catch (Exception)
{
Console.WriteLine($"Skipping {r.Id} {r.Hanzi}");
continue;
}
}
}
}
File.WriteAllLines(Path.Combine(_outputFolder, "dfd_index_debug.txt"), jsonOutput);
return(documents);
}