public static StringBuilder ScanText(string s)
{
if (s.Length < 20)
{
return(new StringBuilder());
}
s = FixOCRConfusing1L(s, "A[0-9l]+");
s = FixOCRConfusing1L(s, "B[0-9l]+");
// Define a regular expression for repeated words.
Regex rx = new Regex("[AB][0-9]+");
// Find matches.
MatchCollection matches = rx.Matches(s);
if (matches.Count == 0)
{
return(new StringBuilder("No match found."));
}
// Report on each match.
TTInfo[] vInfo = new TTInfo[matches.Count];
int i = 0;
foreach (Match match in matches)
{
GroupCollection groups = match.Groups;
vInfo[i] = new TTInfo();
vInfo[i].ID_idx = groups[0].Index;
vInfo[i].ID = groups[0].Value;
++i;
}
string[] residues = new string[vInfo.Length];
int a;
for (i = 0; i < vInfo.Length - 1; ++i)
{
a = vInfo[i].ID_idx + vInfo[i].ID.Length;
residues[i] = s.Substring(a, vInfo[i + 1].ID_idx - 1 - a);
}
a = vInfo.Length - 1;
residues[a] = s.Substring(vInfo[a].ID_idx + vInfo[a].ID.Length);
i = 0;
foreach (string t in residues)
{
SearchDate("[0-9]+-[0-9]+-[0-9]+", t, ref vInfo[i]);
if (vInfo[i].Birthday_idx < 0)
{
SearchDate("[0-9]+/[0-9]+/[0-9]+", t, ref vInfo[i]);
if (vInfo[i].Birthday_idx < 0)
{
SearchDate("[0-9]+-[0-9]+", t, ref vInfo[i]);
if (vInfo[i].Birthday_idx < 0)
{
SearchDate("[0-9]+/[0-9]+", t, ref vInfo[i]);
if (vInfo[i].Birthday_idx < 0)
{
SearchDate("[0-9]+", t, ref vInfo[i]);
}
}
}
}
++i;
}
i = 0;
foreach (string t in residues)
{
if (-1 < vInfo[i].Birthday_idx)
{
vInfo[i].Name = t.Substring(0, vInfo[i].Birthday_idx);
int birthPlaceX = vInfo[i].Birthday_idx + vInfo[i].Birthday.Length + 1;
if (birthPlaceX < t.Length)
{
vInfo[i].Birthplace = t.Substring(birthPlaceX);
}
}
else
{
vInfo[i].Name = t;
}
++i;
}
// Report the number of matches found.
StringBuilder sb = new StringBuilder();
foreach (TTInfo info in vInfo)
{
info.CleanUp();
sb.Append(info.ToString() + "\r\n");
}
sb.Append(matches.Count + " found.");
return(sb);
}
