private HashSet <string> GetKeywordsFor(string taisho)
{
var taishoLines = _lines.FindAll(l => l.Substring(0, 8).Equals(taisho));
var keywords = new HashSet <string>();
string[] lineSplit;
for (int i = 0; i < taishoLines.Count; i++)
{
string line = taishoLines[i];
string nextLine = "";
if (i + 1 <= taishoLines.Count - 1)
{
nextLine = taishoLines[i + 1];
}
lineSplit = line.Split('║');
if (lineSplit.Length < 2)
{
continue;
}
string[] content = lineSplit[1].Split(' ');
foreach (string cjk in content)
{
if (cjk.Length == 0)
{
continue;
}
string cleanCjk;
if (cjk[cjk.Length - 1].Equals("-"))
{
cleanCjk = cjk.Substring(0, cjk.Length - 1);
string firstWord = GetFirstWordOf(nextLine);
nextLine = RemoveFirstWordFrom(nextLine, firstWord);
cleanCjk += firstWord;
}
cleanCjk = RemoveNonCjkCharsFrom(cjk);
if (!_dbTerms.Contains(cleanCjk))
{
foreach (string c in Utf8.AsCodePoints(cleanCjk))
{
if (!_nonCjk.Contains(c))
{
keywords.Add(c);
}
}
}
else
{
keywords.Add(cleanCjk);
}
}
}
return(keywords);
}
private Dictionary <string, List <string> > BuildConcordanceOfHeadwords()
{
var concordance = new Dictionary <string, List <string> >();
string[] lineSplit;
foreach (string line in _lines)
{
lineSplit = line.Split('║');
if (lineSplit.Length < 2)
{
continue;
}
string[] content = lineSplit[1].Split(' ');
foreach (string cjk in content)
{
if (!_dbTerms.Contains(cjk))
{
foreach (string c in Utf8.AsCodePoints(cjk))
{
if (!_nonCjk.Contains(c))
{
if (!concordance.ContainsKey(c))
{
concordance[c] = new List <string>();
}
concordance[c].Add(line);
}
}
}
else
{
if (!_nonCjk.Contains(cjk))
{
if (!concordance.ContainsKey(cjk))
{
concordance[cjk] = new List <string>();
}
concordance[cjk].Add(line);
}
}
}
}
return(concordance);
}
private Dictionary <string, List <string> > BuildConcordanceOfHeadwords(HashSet <string> keywords)
{
var concordance = new Dictionary <string, List <string> >();
string[] lineSplit;
var lines = new List <string>(_lines);
for (int i = 0; i < lines.Count; i++)
{
string line = lines[i];
string nextLine = "";
if (i + 1 <= lines.Count - 1)
{
nextLine = lines[i + 1];
}
lineSplit = line.Split('║');
if (lineSplit.Length < 2)
{
continue;
}
string[] content = lineSplit[1].Split(' ');
bool straddle;
string cleanCjk;
foreach (string cjk in content)
{
if (cjk.Length < 1)
{
continue;
}
cleanCjk = cjk;
straddle = false;
if (cjk[cjk.Length - 1].Equals('-'))
{
straddle = true;
cleanCjk = cjk.TrimEnd('-');
string firstWord = GetFirstWordOf(nextLine);
lines[i + 1] = RemoveFirstWordFrom(nextLine, firstWord);
cleanCjk += firstWord;
}
cleanCjk = RemoveNonCjkCharsFrom(cleanCjk);
if (!_dbTerms.Contains(cleanCjk))
{
foreach (string c in Utf8.AsCodePoints(cleanCjk))
{
if (!_nonCjk.Contains(c) && keywords.Contains(c))
{
if (!concordance.ContainsKey(c))
{
concordance[c] = new List <string>();
}
concordance[c].Add(_lines[i]);
}
}
}
else
{
if (keywords.Contains(cleanCjk))
{
if (!concordance.ContainsKey(cleanCjk))
{
concordance[cleanCjk] = new List <string>();
}
concordance[cleanCjk].Add(_lines[i]);
if (straddle)
{
concordance[cleanCjk].Add(_lines[i + 1]);
}
}
}
}
}
return(concordance);
}
