public OcrConfig(string text, OCR_DATA_TYPE type, string textSplit = "")
{
Type = type;
Input = text;
}
string[] ___extract_startWith(string t, string startWith_,
int numberLineOutput = 1, int numberWordOutput = -1, OCR_DATA_TYPE type = OCR_DATA_TYPE.NONE)
{
if (string.IsNullOrEmpty(t))
{
return new string[] { string.Empty, string.Empty, string.Empty }
}
;
string v = "", err = "";
try
{
int k = t.ToLower().IndexOf(startWith_);
if (k != -1)
{
k = k + startWith_.Length;
t = t.Substring(k, t.Length - k).Trim();
t = ___remove_startWith(t, ".");
t = ___remove_startWith(t, ":");
v = string.Join(" ", t.Split('\n').Where((x, i) => i < numberLineOutput)).Trim();
k = v.Length;
t = t.Substring(k, t.Length - k).Trim();
t = ___remove_startWith(t, ".");
t = ___remove_startWith(t, ":");
}
}
catch (Exception ex)
{
err = ex.Message;
}
v = ___remove_endWith(v, ".").Trim();
switch (type)
{
case OCR_DATA_TYPE.DATE_TIME_BIRTHDAY:
case OCR_DATA_TYPE.DATE_TIME_EXPIRY:
//v = Regex.Replace(v, @"[^\d]", " ").Trim();
v = Regex.Replace(v, @"[^0-9]", " ").Trim();
break;
}
v = ___remove_endWith(v, ".").Trim();
if (numberWordOutput > 0)
{
v = string.Join(" ", v.Split(' ').Where((x, i) => i < numberWordOutput).ToArray());
}
// Replace multi space
v = Regex.Replace(v, @"\s+", " ").Trim();
switch (type)
{
case OCR_DATA_TYPE.DATE_TIME_BIRTHDAY:
case OCR_DATA_TYPE.DATE_TIME_EXPIRY:
v = v.Replace(' ', '-');
break;
}
return(new string[] { v, t, err });
}
