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 IndexSample(bool useServerSession) { string brandName = "Toyota"; string color = "Blue"; int maxPassengers = 5; int fuelCapacity = 40; double litresPer100Kilometers = 5; DateTime modelYear = new DateTime(2003, 1, 1); string modelName = "Highlander"; int maxSpeed = 200; int odometer = 100000; string registrationState = "Texas"; string registrationPlate = "TX343434"; string insurancePolicy = "CAA7878787"; DriversLicense license = new DriversLicense("California", "B7788888", DateTime.Now + new TimeSpan(1825, 0, 0, 0)); Person person = new Person("Mats Persson", license); InsuranceCompany insuranceCompany = new InsuranceCompany("Allstate", "858727878"); Car car = new Car(color, maxPassengers, fuelCapacity, litresPer100Kilometers, modelYear, brandName, modelName, maxSpeed, odometer, registrationState, registrationPlate, insuranceCompany, insurancePolicy); using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); foreach (Database db in session.OpenAllDatabases(true)) { if (db.DatabaseNumber >= 10 || db.DatabaseNumber == SessionBase.IndexDescriptorDatabaseNumber) { session.DeleteDatabase(db); } } session.Commit(); session.BeginUpdate(); DatabaseLocation defaultLocation = session.DatabaseLocations.Default(); List <Database> dbList = session.OpenLocationDatabases(defaultLocation, true); foreach (Database db in dbList) { if (db.DatabaseNumber > Database.InitialReservedDatabaseNumbers) { session.DeleteDatabase(db); } } session.DeleteLocation(defaultLocation); session.Commit(); CreateDirectoryAndCopyLicenseDb(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginRead(); int ct = 0; var mIndex = session.Index <Motorcycle>(); if (mIndex != null) { foreach (Motorcycle mc in session.Index <Motorcycle>()) { Assert.NotNull(mc); ++ct; } } Assert.AreEqual(ct, 0); session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); for (int i = 0; i < 10000; i++) { Motorcycle mc = new Motorcycle(); session.Persist(mc); } session.Commit(); session.BeginUpdate(); int ct = 0; var mIndex = session.Index <Car>(); if (mIndex != null) { foreach (Car c in mIndex) { Assert.NotNull(c); ++ct; } } Assert.AreEqual(ct, 0); ct = 0; session.RegisterClass(typeof(Person)); foreach (Person p in session.AllObjects <Person>(true, false)) { Assert.NotNull(p); ++ct; } session.Commit(); session.BeginRead(); ct = 0; foreach (Motorcycle mc in session.AllObjects <Motorcycle>(false, true)) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 10000); session.Commit(); session.BeginRead(); ct = 0; foreach (Motorcycle mc in session.AllObjects <Motorcycle>(false, true)) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 10000); session.Abort(); session.BeginRead(); ct = 0; foreach (Motorcycle mc in session.AllObjects <Motorcycle>(false, true)) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 10000); session.Commit(); session.BeginRead(); ct = 0; foreach (Motorcycle mc in session.Index <Motorcycle>()) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 10000); session.Abort(); session.BeginRead(); ct = 0; foreach (Motorcycle mc in session.Index <Motorcycle>()) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 10000); session.Commit(); try { ct = 0; foreach (Motorcycle mc in session.Index <Motorcycle>()) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 10000); } catch (NotInTransactionException ex) { Console.WriteLine(ex.Message); } session.BeginUpdate(); ct = 0; foreach (Motorcycle mc in session.AllObjects <Motorcycle>(false, true)) { if (++ct % 2 == 0) { mc.Unpersist(session); } } ct = 0; foreach (Motorcycle mc in session.Index <Motorcycle>()) { ++ct; } Assert.AreEqual(ct, 5000); session.Abort(); session.BeginUpdate(); ct = 0; foreach (Motorcycle mc in session.AllObjects <Motorcycle>(false, true)) { if (++ct % 2 == 0) { mc.Unpersist(session); } } ct = 0; foreach (Motorcycle mc in session.Index <Motorcycle>()) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 5000); ct = 0; foreach (Motorcycle mc in session.Index <Motorcycle>("cc")) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 5000); ct = 0; try { foreach (Motorcycle mc in session.Index <Motorcycle>("ccx")) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 5000); } catch (FieldDoesNotExistException) { } session.Commit(); session.BeginUpdate(); ct = 0; double prior = -44.0; foreach (Motorcycle mc in session.Index <Motorcycle>("cc")) { Assert.NotNull(mc); mc.CC = mc.CC - prior; prior = mc.CC; ++ct; } Assert.AreEqual(ct, 2500); for (int i = 0; i < 95000; i++) { Motorcycle mc = new Motorcycle(); session.Persist(mc); } session.FlushUpdates(); ct = 0; foreach (Motorcycle mc in session.Index <Motorcycle>()) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 100000); session.Abort(); session.BeginUpdate(); ct = 0; prior = double.MinValue; foreach (Motorcycle mc in session.Index <Motorcycle>("cc")) { Assert.NotNull(mc); Assert.GreaterOrEqual(mc.CC, prior); prior = mc.CC; ++ct; if (ct < 25) { Console.Write(prior.ToString() + " "); } } ct = 0; prior = -44.0; foreach (Motorcycle mc in session.Index <Motorcycle>("cc")) { Assert.NotNull(mc); mc.CC = mc.CC - prior; prior = mc.CC; ++ct; } Assert.AreEqual(ct, 2500); session.Commit(); session.BeginUpdate(); ct = 0; prior = double.MinValue; foreach (Motorcycle mc in session.Index <Motorcycle>("cc")) { Assert.NotNull(mc); Assert.GreaterOrEqual(mc.CC, prior); prior = mc.CC; ++ct; } for (int i = 0; i < 95000; i++) { Motorcycle mc = new Motorcycle(); session.Persist(mc); DataBaseFileEntry dbEntry = new DataBaseFileEntry { Something = "Something" }; session.Persist(dbEntry); mc.AddChild(dbEntry); mc.AddChild(dbEntry); } session.FlushUpdates(); ct = 0; foreach (Motorcycle mc in session.Index <Motorcycle>()) { Assert.NotNull(mc); ++ct; } Assert.AreEqual(ct, 100000); session.Commit(); session.BeginRead(); session.Abort(); session.BeginUpdate(); foreach (Motorcycle mc in session.Index <Motorcycle>()) { Assert.NotNull(mc); VelocityDbList <DataBaseFileEntry> children = mc.Children; if (children != null && children.Count > 0) { mc.RemoveChild(children[0]); } ++ct; } session.Commit(); session.BeginRead(); ct = 0; prior = double.MinValue; foreach (Motorcycle mc in session.Index <Motorcycle>("cc")) { Assert.NotNull(mc); Assert.GreaterOrEqual(mc.CC, prior); prior = mc.CC; ++ct; } Assert.AreEqual(ct, 100000); session.Commit(); Console.WriteLine("Motorcycle index Test OK"); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); session.Persist(car); registrationState = "Maine"; car = new Car(color, maxPassengers, fuelCapacity, litresPer100Kilometers, modelYear, brandName, modelName, maxSpeed, odometer, registrationState, registrationPlate, insuranceCompany, insurancePolicy); session.Persist(car); color = "Red"; maxPassengers = 5; fuelCapacity = 50; litresPer100Kilometers = 8; modelYear = new DateTime(2006, 1, 1); brandName = "Toyota"; modelName = "Tacoma"; maxSpeed = 210; odometer = 50000; registrationState = "Texas"; registrationPlate = "TX343433"; insurancePolicy = "CAA7878777"; car = new Car(color, maxPassengers, fuelCapacity, litresPer100Kilometers, modelYear, brandName, modelName, maxSpeed, odometer, registrationState, registrationPlate, insuranceCompany, insurancePolicy); session.Persist(car); color = "Black"; maxPassengers = 5; fuelCapacity = 60; litresPer100Kilometers = 3; modelYear = new DateTime(2001, 1, 1); brandName = "Lincoln"; modelName = "Town Car"; maxSpeed = 220; odometer = 250000; registrationState = "Texas"; registrationPlate = "TX543433"; insurancePolicy = "CAA7878775"; for (int i = 0; i < 1; i++) { registrationState = RandomString(2); registrationPlate = RandomString(12); car = new Car(color, maxPassengers, fuelCapacity, litresPer100Kilometers, modelYear, brandName, modelName, maxSpeed, odometer, registrationState, registrationPlate, insuranceCompany, insurancePolicy); session.Persist(car); color = null; maxPassengers = i; fuelCapacity = 60; litresPer100Kilometers = 3; modelYear = new DateTime(2001, 1, 1); brandName = "Null Car"; modelName = null; maxSpeed = 220; odometer = 250000; insurancePolicy = null; } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginRead(); Console.WriteLine("Blue Cars"); BTreeSet <Car> bTree = session.Index <Car>("color"); foreach (Car c in (from aCar in bTree where aCar.Color == "Blue" select aCar)) { Console.WriteLine(c.ToStringDetails(session)); } Console.WriteLine("Cars in fuel efficierncy order"); foreach (Car c in session.Index <Car>("litresPer100Kilometers")) { Console.WriteLine(c.ToStringDetails(session)); } Console.WriteLine("Vehicles ordered modelYear, brandName, modelName, color"); foreach (Vehicle v in session.Index <Vehicle>()) { Console.WriteLine(v.ToStringDetails(session)); } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); // these LINQ statements will trigger a binary search lookup (not a linear serach) of the matching Car objects in the BTreeSet Car c = (from aCar in session.Index <Car>("color") where aCar.Color == "Blue" select aCar).First(); c.Color = "Green"; session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); // these LINQ statements will trigger a binary search lookup (not a linear serach) of the matching Car objects in the BTreeSet Car c = (from aCar in session.Index <Car>("color") where aCar.Color == "Green" select aCar).First(); UInt64 id = c.Id; session.DeleteObject(id); session.Abort(); session.BeginUpdate(); session.DeleteObject(id); session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { Stopwatch sw = new Stopwatch(); sw.Start(); session.BeginRead(); // these LINQ statements will trigger a binary search lookup (not a linear serach) of the matching Car objects in the BTreeSet Console.WriteLine("Blue Cars"); foreach (Car c in (from aCar in session.Index <Car>("color") where aCar.Color == "Blue" select aCar)) { Console.WriteLine(c.ToStringDetails(session)); } session.Commit(); sw.Stop(); Console.WriteLine(sw.Elapsed); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); for (int i = 0; i < 10000; i++) { // add some junk to make search harder car = new Car(color, maxPassengers, fuelCapacity, litresPer100Kilometers, modelYear, brandName, modelName, i, odometer, registrationState, registrationPlate + i, insuranceCompany, insurancePolicy); session.Persist(car); } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginRead(); // these LINQ statements will trigger a binary search lookup (not a linear serach) of the matching Car objects in the BTreeSet Console.WriteLine("Blue Cars"); foreach (Car c in (from aCar in session.Index <Car>("color") where aCar.Color == "Blue" select aCar)) { Console.WriteLine(c.ToStringDetails(session)); } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); Car c = (from aCar in session.Index <Car>("color") where aCar.Color == "Blue" select aCar).First(); c.Unpersist(session); session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginRead(); foreach (Car c in session.Index <Car>()) { Console.WriteLine(c.ToStringDetails(session)); } Console.WriteLine("Blue Cars"); foreach (Car c in (from aCar in session.Index <Car>() where aCar.Color == "Blue" select aCar)) { Console.WriteLine(c.ToStringDetails(session)); } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); InsuranceCompany prior = insuranceCompany; try { for (int i = 0; i < 100000; i++) { insuranceCompany = new InsuranceCompany("AAA", "858787878"); insuranceCompany.Persist(session, prior); prior = insuranceCompany; } Assert.IsTrue(false); // should not get here } catch (UniqueConstraintException) { } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginRead(); Database db = session.OpenDatabase(session.DatabaseNumberOf(typeof(InsuranceCompany))); var q = from company in session.Index <InsuranceCompany>("name", db) where company.Name == "AAA" select company; foreach (InsuranceCompany company in q) { Console.WriteLine(company.ToStringDetails(session)); // only one will match } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); InsuranceCompany prior = insuranceCompany; try { for (int i = 0; i < 100000; i++) { insuranceCompany = new InsuranceCompany("AAA", "858787878"); session.Persist(insuranceCompany); } Assert.IsTrue(false); // should not get here } catch (UniqueConstraintException) { } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginRead(); Database db = session.OpenDatabase(session.DatabaseNumberOf(typeof(InsuranceCompany))); var q = from company in session.Index <InsuranceCompany>("name", db) where company.Name == "AAA" select company; foreach (InsuranceCompany company in q) { Console.WriteLine(company.ToStringDetails(session)); // only one will match } bool exists = (from anEntry in session.Index <InsuranceCompany>("name", db) where anEntry.Name == "AAA" select 0).Any(); Assert.IsTrue(exists); session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); try { for (int i = 0; i < 100000; i++) { insuranceCompany = new InsuranceCompanySpecial("AAA", "858787878"); session.Persist(insuranceCompany); } Assert.IsTrue(false); // should not get here } catch (UniqueConstraintException) { } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginUpdate(); try { for (int i = 0; i < 100000; i++) { insuranceCompany = new InsuranceCompanySpecial2("AAA", "858787878"); session.Persist(insuranceCompany); } Assert.IsTrue(false); // should not get here } catch (UniqueConstraintException) { } session.Commit(); } using (SessionBase session = useServerSession ? (SessionBase) new ServerClientSession(systemDir) : (SessionBase) new SessionNoServer(systemDir)) { session.BeginRead(); Database db = session.OpenDatabase(session.DatabaseNumberOf(typeof(InsuranceCompanySpecial))); var q = from company in session.Index <InsuranceCompany>("name", db) where company.Name == "AAA" select company; foreach (InsuranceCompany company in q) { Console.WriteLine(company.ToStringDetails(session)); // only one will match } bool exists = (from anEntry in session.Index <InsuranceCompanySpecial>("name", db) where anEntry.Name == "AAA" select 0).Any(); Assert.IsTrue(exists); session.Commit(); } }
public void Recover1(SessionBase session) { Database db = null; session.BeginUpdate(); db = session.OpenDatabase(88, true, false); if (db != null) session.DeleteDatabase(db); db = session.OpenDatabase(89, true, false); if (db != null) session.DeleteDatabase(db); session.Commit(); session.BeginUpdate(); db = session.NewDatabase(88); session.FlushUpdates(); session.Abort(); session.BeginUpdate(); db = session.NewDatabase(88); SortedSetAny<float> floatSet; Oid floatSetOid; string dbPath = System.IO.Path.Combine(systemDir, "89.odb"); Placement place = new Placement(88); for (int i = 0; i < 10; i++) { floatSet = new SortedSetAny<float>(); floatSet.Persist(place, session); floatSetOid = floatSet.Oid; } db = session.NewDatabase(89); session.Commit(); File.Delete(dbPath); session.BeginUpdate(); db = session.NewDatabase(89); session.Commit(); FileInfo info = new FileInfo(dbPath); info.CopyTo(dbPath + "X"); session.BeginUpdate(); SortedSetAny<int> intSet; place = new Placement(89); for (int i = 0; i < 10; i++) { intSet = new SortedSetAny<int>(); intSet.Persist(place, session); } db = session.OpenDatabase(88); var list = db.AllObjects<SortedSetAny<float>>(); foreach (SortedSetAny<float> set in list) set.Unpersist(session); db = session.OpenDatabase(89); session.Commit(); intSet = null; db = null; // release refs so that cached data isn't stale File.Delete(dbPath); info = new FileInfo(dbPath + "X"); info.MoveTo(dbPath); session.BeginUpdate(); intSet = (SortedSetAny<int>)session.Open(89, 1, 1, false); Debug.Assert(intSet == null); object o = session.Open(88, 1, 1, false); floatSet = (SortedSetAny<float>)o; Debug.Assert(floatSet != null); session.Commit(); session.BeginUpdate(); db = session.OpenDatabase(88); session.DeleteDatabase(db); db = session.OpenDatabase(89); session.DeleteDatabase(db); session.Commit(); }
public void Recover1(SessionBase session) { Database db = null; session.BeginUpdate(); session.RegisterClass(typeof(SortedSetAny <int>)); session.RegisterClass(typeof(SortedSetAny <float>)); db = session.OpenDatabase(88, true, false); if (db != null) { session.DeleteDatabase(db); } db = session.OpenDatabase(89, true, false); if (db != null) { session.DeleteDatabase(db); } session.Commit(); session.BeginUpdate(); db = session.NewDatabase(88); session.FlushUpdates(); session.Abort(); session.BeginUpdate(); db = session.NewDatabase(88); SortedSetAny <float> floatSet; Oid floatSetOid; string dbPath = System.IO.Path.Combine(systemDir, "89.odb"); Placement place = new Placement(88); for (int i = 0; i < 10; i++) { floatSet = new SortedSetAny <float>(); floatSet.Persist(place, session); floatSetOid = floatSet.Oid; } db = session.NewDatabase(89); session.Commit(); File.Delete(dbPath); session.BeginUpdate(); db = session.NewDatabase(89); session.Commit(); FileInfo info = new FileInfo(dbPath); info.CopyTo(dbPath + "X"); session.BeginUpdate(); SortedSetAny <int> intSet; place = new Placement(89); for (int i = 0; i < 10; i++) { intSet = new SortedSetAny <int>(); intSet.Persist(place, session); } db = session.OpenDatabase(88); var list = db.AllObjects <SortedSetAny <float> >(); foreach (SortedSetAny <float> set in list) { set.Unpersist(session); } db = session.OpenDatabase(89); session.Commit(); intSet = null; db = null; // release refs so that cached data isn't stale File.Delete(dbPath); info = new FileInfo(dbPath + "X"); info.MoveTo(dbPath); session.BeginUpdate(); intSet = (SortedSetAny <int>)session.Open(89, 1, 1, false); Debug.Assert(intSet == null); object o = session.Open(88, 1, 1, false); floatSet = (SortedSetAny <float>)o; Debug.Assert(floatSet != null); session.Checkpoint(); db = session.OpenDatabase(88); session.DeleteDatabase(db); db = session.OpenDatabase(89); session.DeleteDatabase(db); session.Commit(); }
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."); }