/// <summary>
/// 添加索引
/// </summary>
/// <param name="recnum">文档唯一编号</param>
/// <param name="text">文档内容</param>
private void AddtoIndex(int recnum, string text)
{
if (text == "" || text == null)
{
return;
}
text = text.ToLowerInvariant(); // lowercase index 转小写
string[] keys;
//分词
if (_docMode)
{
//_log.Debug("text size = " + text.Length);
//分词器
//Dictionary<string, int> wordfreq = Tokenizer.GenerateWordFreq(text);
Dictionary <string, int> wordfreq = _tokenizer.GenerateWordFreq(text);
//_log.Debug("word count = " + wordfreq.Count);
var kk = wordfreq.Keys;
keys = new string[kk.Count];
kk.CopyTo(keys, 0);
}
else
{
keys = text.Split(' ');
}
//------------------------------
//将分词出来的词进行判断
//已经存在则得到该词对应的位图索引
//不存在则创建新的索引编号
//------------------------------
foreach (string key in keys)
{
if (key == "")
{
continue;
}
int bmp;
if (_words.TryGetValue(key, out bmp))//存在
{
_bitmaps.GetBitmap(bmp).Set(recnum, true);
}
else
{
//索引编号
bmp = _bitmaps.GetFreeRecordNumber();
_bitmaps.SetDuplicate(bmp, recnum);
_words.Add(key, bmp);
}
}
_wordschanged = true;
}
/// <summary>
/// Add Text to the index
/// </summary>
/// <param name="recnum"></param>
/// <param name="text"></param>
private void AddtoIndex(int recnum, string text)
{
if (text == "" || text == null)
{
return;
}
text = text.ToLowerInvariant(); // lowercase index
string[] keys;
if (HootConfOptions.DocMode)
{
//_log.Debug("text size = " + text.Length);
Dictionary <string, int> wordfreq = _tokenizer.GenerateWordFreq(text, HootConfOptions);
//_log.Debug("word count = " + wordfreq.Count);
var kk = wordfreq.Keys;
keys = new string[kk.Count];
kk.CopyTo(keys, 0);
}
else
{
keys = text.Split(' ');
}
foreach (string key in keys)
{
if (key == "")
{
continue;
}
int bmp;
if (_words.TryGetValue(key, out bmp))
{
_bitmaps.GetBitmap(bmp).Set(recnum, true);
}
else
{
bmp = _bitmaps.GetFreeRecordNumber();
_bitmaps.SetDuplicate(bmp, recnum);
_words.Add(key, bmp);
}
}
_wordschanged = true;
}
