public void createGlobalInvertedIndex(IndexRoot indexRoot)
{
Placement wordPlacement = new Placement(Lexicon.PlaceInDatabase, 2);
BTreeSetOidShort <Word> wordSet = indexRoot.lexicon.WordSet;
BTreeSet <Document> docSet = indexRoot.repository.documentSet;
Word existingWord = null;
foreach (Document doc in docSet)
{
if (doc.Indexed == false)
{
foreach (Word word in doc.WordSet)
{
WordHit wordHit = doc.WordHit[word];
if (wordSet.TryGetKey(word, ref existingWord))
{
existingWord.GlobalCount = existingWord.GlobalCount + (uint)wordHit.Count;
}
else
{
existingWord = new WordGlobal(word.aWord, session, (uint)wordHit.Count);
existingWord.Persist(wordPlacement, session);
wordSet.Add(existingWord);
}
existingWord.DocumentHit.AddFast(doc);
}
doc.Indexed = true;
}
}
}
static void createTopLevelInvertedIndex()
{
Console.WriteLine(DateTime.Now.ToString() + ", start creating top level inverted index");
using (SessionNoServer session = new SessionNoServer(s_systemDir))
{
Placement wordPlacement = new Placement(Lexicon.PlaceInDatabase, 2, 1, 1000, 50000, true, false, UInt32.MaxValue, false);
session.BeginUpdate();
IndexRoot indexRoot = (IndexRoot)session.Open(Oid.Encode(IndexRoot.PlaceInDatabase, 1, 1));
BTreeSetOidShort <Word> wordSet = indexRoot.lexicon.WordSet;
BTreeSet <Document> documentSet = indexRoot.repository.documentSet;
Word existingWord = null;
foreach (Document doc in documentSet)
{
foreach (Word word in doc.WordSet)
{
WordHit wordHit = doc.WordHit[word];
if (wordSet.TryGetKey(word, ref existingWord))
{
existingWord.GlobalCount = existingWord.GlobalCount + (uint)wordHit.Count;
}
else
{
existingWord = new WordGlobal(word.aWord, session, (uint)wordHit.Count);
existingWord.Persist(wordPlacement, session);
indexRoot.lexicon.WordSet.Add(existingWord);
}
existingWord.DocumentHit.AddFast(doc);
}
doc.Indexed = true;
}
session.Commit();
Console.WriteLine(DateTime.Now.ToString() + ", done creating top level inverted index");
}
}
static void createDocumentInvertedIndex(SessionBase session, Database db, BTreeSet <Document> documentSet)
{
UInt32 dbNum = db.DatabaseNumber;
Document doc = null;
Document inputDoc = new Document(db.Id);
Placement wordPlacement = new Placement(inputDoc.DatabaseNumber, 20000, 1, 25000, 65000, true, false, 1, false);
Placement wordHitPlacement = new Placement(inputDoc.DatabaseNumber, 40000, 1, 25000, 65500, true, false, 1, false);
//session.SetTraceDbActivity(db.DatabaseNumber);
BTreeSetIterator <Document> iterator = documentSet.Iterator();
iterator.GoTo(inputDoc);
inputDoc = iterator.Current();
while (inputDoc != null && inputDoc.Page.Database.DatabaseNumber == dbNum)
{
doc = (Document)session.Open(inputDoc.Page.Database, inputDoc.Id); // if matching database is availeble, use it to speed up lookup
DocumentText docText = doc.Content;
string text = docText.Text.ToLower();
MatchCollection tagMatches = Regex.Matches(text, "[a-z][a-z.$]+");
UInt64 wordCt = 0;
WordHit wordHit;
Word word;
if (++s_docCountIndexed % 50000 == 0)
{
Console.WriteLine(DateTime.Now.ToString() + ", done indexing article: " + s_docCountIndexed);
}
BTreeSetOidShort <Word> wordSet = doc.WordSet;
foreach (Match m in tagMatches)
{
word = new Word(m.Value);
if (wordSet.TryGetKey(word, ref word))
{
//wordHit = doc.WordHit[word]; // to costly to add tight now - figure out a better way ?
//wordHit.Add(wordCt);
}
else
{
word = new Word(m.Value);
word.Persist(wordPlacement, session);
wordSet.Add(word);
wordHit = new WordHit(doc, wordCt++, session);
//wordHit.Persist(wordHitPlacement, session);
doc.WordHit.ValuePlacement = wordHitPlacement;
doc.WordHit.AddFast(word, wordHit);
}
}
inputDoc = iterator.Next();
}
session.FlushUpdates(db);
session.ClearCachedObjects(db); // free up memory for objects we no longer need to have cached
Console.WriteLine(DateTime.Now.ToString() + ", done indexing article: " + s_docCountIndexed + " Database: " + dbNum + " is completed.");
}
public void createLocalInvertedIndex(Document doc, Word word, UInt64 wordCt, Placement wordPlacement, Placement wordHitPlacement)
{
WordHit wordHit;
BTreeSetOidShort <Word> wordSet = doc.WordSet;
if (wordSet.TryGetKey(word, ref word))
{
wordHit = doc.WordHit[word];
wordHit.Add(wordCt);
}
else
{
word.Persist(wordPlacement, session);
wordSet.Add(word);
wordHit = new WordHit(doc, wordCt++, session);
doc.WordHit.ValuePlacement = wordHitPlacement;
doc.WordHit.AddFast(word, wordHit);
}
}