public override async Task <IEnumerable <Entity> > RecognizeEntitiesAsync(DialogContext dialogContext, string text, string locale, IEnumerable <Entity> entities, CancellationToken cancellationToken = default)
{
List <Entity> newEntities = new List <Entity>();
// Get data from incoming entities
if (entities != null)
{
List <Entity> givenNames = new List <Entity>();
List <Entity> surnames = new List <Entity>();
var givenNameMap = await GetGivenNames(dialogContext);
var surnameMap = await GetSurnames(dialogContext);
var commonWords = defaultCommonWords.Value;
var entityTypes = EntityTypes.GetValue(dialogContext.State);
foreach (JObject entity in entities.Where(e => entityTypes.Contains(e.Type)).Select(e => JObject.FromObject(e, Serializer)))
{
text = ObjectPath.GetPathValue <string>(entity, "text");
if (!string.IsNullOrEmpty(text))
{
var tokens = TokenUtils.GetTokens(text);
Token prevToken = null;
foreach (var token in tokens)
{
if (prevToken?.Text == "'" && token.Text == "s")
{
continue;
}
string possibleGivenName = token.Text;
// if it starts with Cap letter and has no other cap letters, then it is highly likely it's a name
bool isNamePattern = char.IsUpper(token.Text[0]) &&
token.Text.Where(c => char.IsUpper(c)).Count() == 1;
if (givenNameMap.TryGetValue(possibleGivenName, out string gender))
{
// if it is not a common word, or it is name pattern
if (isNamePattern || !commonWords.Contains(possibleGivenName))
{
dynamic givenName = JObject.FromObject(entity, Serializer);
givenName.type = "givenName";
givenName.start = token.StartOffset;
givenName.end = token.EndOffset;
givenName.text = possibleGivenName;
givenName.source = entity;
givenName.resolution = new JObject();
givenName.resolution.value = possibleGivenName;
givenName.resolution.gender = gender;
var newEntity = ((JObject)givenName).ToObject <Entity>();
newEntities.Add(newEntity);
givenNames.Add(newEntity);
}
}
// if any of the surname parts are a surname, then it's probably a hyphenated surname
string possibleSurname = token.Text;
var parts = possibleSurname.Split('-');
if (parts.Any(part => surnameMap.Contains(part)))
{
// if it is not a common word, or it is name pattern
if (isNamePattern || !commonWords.Contains(possibleGivenName))
{
dynamic surname = JObject.FromObject(entity, Serializer);
surname.type = "surname";
surname.start = token.StartOffset;
surname.end = token.EndOffset;
surname.source = entity;
surname.text = token.Text;
surname.resolution = new JObject();
surname.resolution.value = token.Text;
var newEntity = ((JObject)surname).ToObject <Entity>();
newEntities.Add(newEntity);
surnames.Add(newEntity);
}
}
prevToken = token;
}
}
}
foreach (var givenName in givenNames)
{
var surname = surnames.Where(sn => (int)sn.Properties["start"] == (int)givenName.Properties["end"] + 1).FirstOrDefault();
if (surname != null)
{
dynamic fullName = new JObject();
fullName.type = "fullName";
fullName.text = $"{givenName.Properties["text"]} {surname.Properties["text"]}";
fullName.start = (int)givenName.Properties["start"];
fullName.end = (int)surname.Properties["end"];
fullName.resolution = new JObject();
fullName.resolution.givenName = JObject.FromObject(givenName);
fullName.resolution.surname = JObject.FromObject(surname);
fullName.resolution.value = fullName.text;
newEntities.Add(((JObject)fullName).ToObject <Entity>());
}
}
}
return(newEntities);
}