/**************************************************************************/
private void ProcessText(MacroscopeDocument msDoc)
{
List <string> TextBlocks = new List <string> (16);
List <string> Terms = new List <string> (256);
bool CaseSensitive = MacroscopePreferencesManager.GetCaseSensitiveTextIndexing();
TextBlocks.Add(msDoc.GetTitle());
TextBlocks.Add(msDoc.GetDescription());
TextBlocks.Add(msDoc.GetKeywords());
TextBlocks.Add(msDoc.GetDocumentTextCleaned());
DebugMsg(string.Format("ProcessText: TextBlocks.Count: {0}", TextBlocks.Count));
if (TextBlocks.Count > 0)
{
for (int i = 0; i < TextBlocks.Count; i++)
{
string [] Chunk = TextBlocks[i].Split(' ');
if (Chunk.Length > 0)
{
for (int j = 0; j < Chunk.Length; j++)
{
if (Chunk[j].Length > 0)
{
if (!Terms.Contains(Chunk[j]))
{
Terms.Add(Chunk[j]);
}
}
}
}
}
}
DebugMsg(string.Format("ProcessText: Words :: {0}", Terms.Count));
for (int i = 0; i < Terms.Count; i++)
{
Dictionary <string, MacroscopeDocument> DocumentReference;
string Term = Terms[i];
if (!CaseSensitive)
{
Term = Term.ToLower();
}
DebugMsg(string.Format("ProcessText: Term :: {0}", Term));
if (InvertedIndex.ContainsKey(Term))
{
DocumentReference = this.InvertedIndex[Term];
}
else
{
DocumentReference = new Dictionary <string, MacroscopeDocument> ();
this.InvertedIndex.Add(Term, DocumentReference);
}
if (!DocumentReference.ContainsKey(msDoc.GetUrl()))
{
DocumentReference.Add(msDoc.GetUrl(), msDoc);
}
}
}