public static IList <string> SegmentToWord(string text, bool ignoreCase)
{
if (s_HasSegmentInit == false)
{
lock (s_SyncSegmentObj)
{
if (s_HasSegmentInit == false)
{
string configPath = ConfigurationManager.AppSettings["PanGu_ConfigFilePath"];
if (string.IsNullOrWhiteSpace(configPath) == false)
{
string p = Path.GetPathRoot(configPath);
if (p == null || p.Trim().Length <= 0) // 说明是相对路径
{
configPath = Path.Combine(AppDomain.CurrentDomain.SetupInformation.ApplicationBase, configPath);
}
}
else
{
configPath = Path.Combine(AppDomain.CurrentDomain.SetupInformation.ApplicationBase, "pangu.xml");
}
PanGu.Segment.Init(configPath);
s_HasSegmentInit = true;
}
}
}
List <string> dicts = new List <string>(s_Dicts.Length);
foreach (var w in s_Dicts)
{
if (text.IndexOf(w, StringComparison.CurrentCultureIgnoreCase) >= 0)
{
dicts.Add(w);
text = text.Replace(w, string.Empty);
}
}
PanGu.Segment segment = new PanGu.Segment();
var list = segment.DoSegment(text);
List <string> rstList = new List <string>(list.Count + dicts.Count);
rstList.AddRange(dicts);
foreach (var w in list)
{
if (rstList.Contains(w.Word, ignoreCase ? StringComparer.CurrentCultureIgnoreCase : StringComparer.CurrentCulture) == false)
{
rstList.Add(w.Word);
}
}
return(rstList);
}
/// <summary>
/// Analysises the word.
/// </summary>
/// <returns>The word.</returns>
/// <param name="text">Text.</param>
public Dictionary <string, int> AnalysisWord(string text)
{
PanGu.Segment segment = new PanGu.Segment();
ICollection <PanGu.WordInfo> words = segment.DoSegment(text);
Dictionary <string, int> result = new Dictionary <string, int>();
foreach (PanGu.WordInfo w in words)
{
if (w == null)
{
continue;
}
result[w.Word] = w.Rank;
}
return(result);
}
