private IndexResult AddtoIndex(int recnum, string text /*, bool optimizeIndexes*/)
{
_log("text size = " + text.Length);
Dictionary <string, int> wordFrequences = GenerateWordFreq(text);
_log("word count = " + wordFrequences.Count);
var result = new IndexResult {
DocNumber = recnum
};
using (var bmp = CreateBitmapStream())
{
foreach (string word in wordFrequences.Keys)
{
Cache cache;
if (_index.TryGetValue(word, out cache))
{
LoadCacheIfNotLoaded(cache, bmp);
cache.SetBit(recnum, true);
}
else
{
cache = new Cache {
isLoaded = true
};
cache.SetBit(recnum, true);
_index.Add(word, cache);
}
result.WordsAdded.Add(word, wordFrequences[word]);
}
}
return(result);
}
public IndexResult UpdateExistingIndex(int recnum, string text)
{
Log("UpdateExistingIndex started");
LogIndex();
Log(string.Format("text size = {0}", text.Length));
Dictionary <string, int> wordFrequencies = GenerateWordFreq(text);
Log(string.Format("word count = {0}", wordFrequencies.Count));
var result = new IndexResult {
DocNumber = recnum
};
using (var bmp = CreateBitmapStream())
{
foreach (KeyValuePair <string, Cache> vc in _index)
{
string indexedWord = vc.Key;
Cache cache = vc.Value;
LoadCacheIfNotLoaded(cache, bmp);
bool isWordIndexed = cache.GetBitmap().Get(recnum);
bool isWordInText = wordFrequencies.ContainsKey(indexedWord);
if (isWordIndexed && isWordInText)
{
wordFrequencies.Remove(indexedWord);
}
else if (isWordIndexed)
{
result.WordsRemoved.Add(indexedWord);
cache.SetBit(recnum, false);
}
else if (isWordInText)
{
result.WordsAdded.Add(indexedWord, wordFrequencies[indexedWord]);
wordFrequencies.Remove(indexedWord);
cache.SetBit(recnum, true);
}
}
}
foreach (var wordFrequency in wordFrequencies)
{
result.WordsAdded.Add(wordFrequency.Key, wordFrequency.Value);
var cache = new Cache {
isLoaded = true
};
cache.SetBit(recnum, true);
_index.Add(wordFrequency.Key, cache);
}
LogIndex();
Log("UpdateExistingIndex ended");
return(result);
}
