internal void Index(IList <Document> docs, int fieldIndex)
{
int orginalFieldIndex = fieldIndex;
Field field = _DBProvider.GetField(this.FieldName);
foreach (Document doc in docs)
{
if (doc.FieldValues[fieldIndex].FieldName.Trim().ToLower() != FieldName.Trim().ToLower())
{
//Field index does not match, find new field index again.
//Happen as input different field order of rows
fieldIndex = 0;
foreach (FieldValue fValue in doc.FieldValues)
{
if (doc.FieldValues[fieldIndex].FieldName.Trim().ToLower() == FieldName.Trim().ToLower())
{
break;
}
fieldIndex++;
}
if (fieldIndex >= doc.FieldValues.Count)
{
//no data of this field in this row
fieldIndex = orginalFieldIndex;
continue;
}
}
this.Index(doc.FieldValues[fieldIndex].Value, doc.DocId, field.GetAnalyzer());
}
}
unsafe private void TestFillPayloadRank(string tableName)
{
OutputMessage("TestGetDocIdReplaceFieldValue");
AddColumn("Times");
AddColumn("Elapse(ms)");
AddColumn("ElapseOneTime(ms)");
Data.DBProvider dbProvider = Data.DBProvider.GetDBProvider(Parameters[0], false);
if (dbProvider == null)
{
throw new DataException(string.Format("Table name {0} does not exist!", Parameters[0]));
}
Random rand = new Random();
int count = 1000000;
int lastDocId = dbProvider.LastDocId;
OriginalDocumentPositionList[] payloads = new OriginalDocumentPositionList[count];
for (int i = 0; i < payloads.Length; i++)
{
payloads[i] = new OriginalDocumentPositionList(i * 10);
payloads[i].CountAndWordCount = 1;
}
payloads[0].DocumentId = 8 * payloads.Length;
Data.Field rankField = dbProvider.GetField("Rank");
int tab = rankField.TabIndex;
Stopwatch sw = new Stopwatch();
//int docid = rand.Next(lastDocId);
sw.Start();
for (int j = 0; j < count / payloads.Length; j++)
{
dbProvider.FillPayloadRank(tab, count, payloads);
}
sw.Stop();
OutputValue("Times", count);
OutputValue("Elapse(ms)", sw.ElapsedMilliseconds);
OutputValue("ElapseOneTime(ms)", (double)sw.ElapsedMilliseconds / count);
}
public static DocId2LongComparer Generate(Data.DBProvider dbProvider,
OrderBy[] orderBys, out Data.Field[] orderByFields)
{
bool[] ascs = new bool[orderBys.Length];
orderByFields = new Hubble.Core.Data.Field[orderBys.Length];
int scoreFieldIndex = -1;
for (int i = 0; i < ascs.Length; i++)
{
if (orderBys[i].Order == null)
{
ascs[i] = true;
}
else
{
ascs[i] = !orderBys[i].Order.Equals("desc", StringComparison.CurrentCultureIgnoreCase);
}
if (orderBys[i].Name.Equals("docid", StringComparison.CurrentCultureIgnoreCase))
{
orderByFields[i] = new Hubble.Core.Data.Field("docid", Hubble.Core.Data.DataType.Int);
}
else if (orderBys[i].Name.Equals("score", StringComparison.CurrentCultureIgnoreCase))
{
scoreFieldIndex = i;
orderByFields[i] = new Hubble.Core.Data.Field("score", Hubble.Core.Data.DataType.BigInt);
}
else
{
orderByFields[i] = dbProvider.GetField(orderBys[i].Name);
}
}
return(new DocId2LongComparer(ascs, scoreFieldIndex));
}
/// <summary>
/// Get IdFields List from trigger table
/// </summary>
/// <param name="dbProvider">dbProvider that used to get field info</param>
/// <param name="dbAdapterName">DBAdapter name</param>
/// <param name="table">table that read from trigger table</param>
/// <param name="lastSerial">the last serial number of trigger table from which read</param>
/// <returns></returns>
internal static List <IdFields> GetIdFieldsList(Data.DBProvider dbProvider, string dbAdapterName,
System.Data.DataTable table, out long lastSerial)
{
lastSerial = -1;
HashSet <string> fieldsSetWithTokenizedFields = new HashSet <string>();
HashSet <string> fieldsSetWithoutTokenizedFields = new HashSet <string>();
List <string> tempFields = new List <string>(128);
List <IdFields> result = new List <IdFields>(table.Rows.Count);
foreach (System.Data.DataRow row in table.Rows)
{
long id = long.Parse(row["id"].ToString());
lastSerial = long.Parse(row["Serial"].ToString());
string fields = row["Fields"].ToString();
tempFields.Clear();
bool hasTokenized = false;
//check fields
foreach (string field in fields.Split(new char[] { ',' }))
{
string f = field.Trim().ToLower();
if (f == "")
{
continue;
}
Data.Field dbField = dbProvider.GetField(f);
if (dbField == null)
{
continue;
}
if (dbField.IndexType == Hubble.Core.Data.Field.Index.Tokenized)
{
hasTokenized = true;
}
tempFields.Add(f);
}
//Fill hash set
if (hasTokenized)
{
foreach (string field in tempFields)
{
if (!fieldsSetWithTokenizedFields.Contains(field))
{
fieldsSetWithTokenizedFields.Add(field);
}
}
}
else
{
foreach (string field in tempFields)
{
if (!fieldsSetWithoutTokenizedFields.Contains(field))
{
fieldsSetWithoutTokenizedFields.Add(field);
}
}
}
result.Add(new IdFields(id, hasTokenized));
}
//Get new fields string
string fieldsWithTokenized = GetFieldsStringFromHashSet(fieldsSetWithTokenizedFields, dbAdapterName);
string fieldsWithoutTokenized = GetFieldsStringFromHashSet(fieldsSetWithoutTokenizedFields, dbAdapterName);
foreach (IdFields idFields in result)
{
if (idFields.HasTokenizedFields)
{
idFields.Fields = fieldsWithTokenized;
}
else
{
idFields.Fields = fieldsWithoutTokenized;
}
}
//Merge same Id
result.Sort();
if (result.Count > 0)
{
IdFields last = result[0];
for (int i = 1; i < result.Count; i++)
{
if (result[i].Equals(last))
{
result[i] = null;
continue;
}
else
{
last = result[i];
}
}
}
return(result);
}
unsafe internal void Sort(Query.DocumentResultForSort[] docResults, int top)
{
if (_OrderBys.Count <= 0)
{
return;
}
QueryResultHeapSort heapSort = new QueryResultHeapSort(_OrderBys, _DBProvider);
if (heapSort.CanDo)
{
heapSort.Prepare(docResults);
heapSort.TopSort(docResults, top);
return;
}
foreach (SyntaxAnalysis.Select.OrderBy orderBy in _OrderBys)
{
Data.Field field = _DBProvider.GetField(orderBy.Name);
Query.SortType sortType = Hubble.Core.Query.SortType.None;
bool isDocId = false;
bool isScore = false;
bool isAsc = orderBy.Order.Equals("ASC", StringComparison.CurrentCultureIgnoreCase);
if (field == null)
{
if (orderBy.Name.Equals("DocId", StringComparison.CurrentCultureIgnoreCase))
{
sortType = Hubble.Core.Query.SortType.Long;
isDocId = true;
}
else if (orderBy.Name.Equals("Score", StringComparison.CurrentCultureIgnoreCase))
{
sortType = Hubble.Core.Query.SortType.Long;
isScore = true;
}
else
{
throw new ParseException(string.Format("Unknown field name:{0}", orderBy.Name));
}
}
else
{
if (field.IndexType != Hubble.Core.Data.Field.Index.Untokenized)
{
throw new ParseException(string.Format("Order by field name:{0} is not Untokenized Index!", orderBy.Name));
}
}
for (int i = 0; i < docResults.Length; i++)
{
if (docResults[i].SortInfoList == null)
{
docResults[i].SortInfoList = new List <Hubble.Core.Query.SortInfo>(2);
}
if (isDocId)
{
docResults[i].SortInfoList.Add(new Hubble.Core.Query.SortInfo(isAsc, sortType, docResults[i].DocId));
}
else if (isScore)
{
docResults[i].SortInfoList.Add(new Hubble.Core.Query.SortInfo(isAsc, sortType, docResults[i].Score));
}
else
{
if (docResults[i].PayloadData == null)
{
int *payloadData = _DBProvider.GetPayloadData(docResults[i].DocId);
if (payloadData == null)
{
throw new ParseException(string.Format("DocId={0} has not payload!", docResults[i].DocId));
}
docResults[i].PayloadData = payloadData;
}
docResults[i].SortInfoList.Add(Data.DataTypeConvert.GetSortInfo(isAsc,
field.DataType, docResults[i].PayloadData, field.TabIndex, field.SubTabIndex, field.DataLength));
}
}
}
Array.Sort(docResults);
//Has a bug of partial sort, make comments on following codes until it fixed.
//if (top <= 0 || top >= docResults.Length/2)
//{
// Array.Sort(docResults);
//}
//else
//{
// QuickSort<Query.DocumentResult>.TopSort(docResults, top, new Query.DocumentResultComparer());
//}
}
public QueryResultHeapSort(List <SyntaxAnalysis.Select.OrderBy> orderBys, Data.DBProvider dbProvider)
{
_DBProvider = dbProvider;
_OrderBys = orderBys;
_CanDo = true;
if (orderBys == null)
{
_CanDo = false;
}
else if (orderBys.Count <= 0 || orderBys.Count > 2)
{
_CanDo = false;
}
else
{
_SortFieldsCount = orderBys.Count;
for (int i = 0; i < orderBys.Count; i++)
{
bool asc = orderBys[i].Order.Equals("ASC", StringComparison.CurrentCultureIgnoreCase);
if (orderBys[i].Name.Equals("DocId", StringComparison.CurrentCultureIgnoreCase))
{
if (i == 0)
{
_Asc1 = asc;
}
else
{
_Asc2 = asc;
}
}
else if (orderBys[i].Name.Equals("Score", StringComparison.CurrentCultureIgnoreCase))
{
if (i == 0)
{
_Asc1 = asc;
}
else
{
_Asc2 = asc;
}
}
else
{
if (i == 0)
{
_Asc1 = asc;
}
else
{
_Asc2 = asc;
}
Data.Field field = _DBProvider.GetField(orderBys[i].Name);
if (field != null)
{
if (field.IndexType != Hubble.Core.Data.Field.Index.Untokenized)
{
throw new ParseException(string.Format("Order by field name:{0} is not Untokenized Index!", orderBys[i].Name));
}
switch (field.DataType)
{
case Hubble.Core.Data.DataType.Date:
case Hubble.Core.Data.DataType.SmallDateTime:
case Hubble.Core.Data.DataType.Int:
case Hubble.Core.Data.DataType.SmallInt:
case Hubble.Core.Data.DataType.TinyInt:
case Hubble.Core.Data.DataType.BigInt:
case Hubble.Core.Data.DataType.DateTime:
case Hubble.Core.Data.DataType.Float:
break;
default:
_CanDo = false;
break;
}
}
}
if (!_CanDo)
{
break;
}
}
}
}
unsafe public void Prepare(DocumentResultForSort[] docResults)
{
for (int index = 0; index < _SortFieldsCount; index++)
{
bool asc;
if (index == 0)
{
asc = _Asc1;
}
else
{
asc = _Asc2;
}
if (_OrderBys[index].Name.Equals("DocId", StringComparison.CurrentCultureIgnoreCase))
{
for (int i = 0; i < docResults.Length; i++)
{
if (index == 0)
{
docResults[i].SortValue = docResults[i].DocId;
}
else
{
docResults[i].SortValue1 = docResults[i].DocId;
}
}
}
else if (_OrderBys[index].Name.Equals("Score", StringComparison.CurrentCultureIgnoreCase))
{
for (int i = 0; i < docResults.Length; i++)
{
if (index == 0)
{
docResults[i].SortValue = docResults[i].Score;
}
else
{
docResults[i].SortValue1 = docResults[i].Score;
}
}
}
else
{
Data.Field field = _DBProvider.GetField(_OrderBys[index].Name);
if (field != null)
{
if (field.IndexType != Hubble.Core.Data.Field.Index.Untokenized)
{
throw new ParseException(string.Format("Order by field name:{0} is not Untokenized Index!", _OrderBys[index].Name));
}
switch (field.DataType)
{
case Hubble.Core.Data.DataType.Date:
case Hubble.Core.Data.DataType.SmallDateTime:
case Hubble.Core.Data.DataType.Int:
case Hubble.Core.Data.DataType.SmallInt:
case Hubble.Core.Data.DataType.TinyInt:
{
_DBProvider.FillPayloadData(docResults);
for (int i = 0; i < docResults.Length; i++)
{
int *payLoadData = docResults[i].PayloadData;
Query.SortInfo sortInfo = Data.DataTypeConvert.GetSortInfo(asc, field.DataType,
payLoadData, field.TabIndex, field.SubTabIndex, field.DataLength);
if (index == 0)
{
docResults[i].SortValue = sortInfo.IntValue;
}
else
{
docResults[i].SortValue1 = sortInfo.IntValue;
}
}
}
break;
case Hubble.Core.Data.DataType.BigInt:
case Hubble.Core.Data.DataType.DateTime:
{
_DBProvider.FillPayloadData(docResults);
for (int i = 0; i < docResults.Length; i++)
{
int *payLoadData = docResults[i].PayloadData;
Query.SortInfo sortInfo = Data.DataTypeConvert.GetSortInfo(asc, field.DataType,
payLoadData, field.TabIndex, field.SubTabIndex, field.DataLength);
if (index == 0)
{
docResults[i].SortValue = sortInfo.LongValue;
}
else
{
docResults[i].SortValue1 = sortInfo.LongValue;
}
}
}
break;
case Hubble.Core.Data.DataType.Float:
{
_DBProvider.FillPayloadData(docResults);
for (int i = 0; i < docResults.Length; i++)
{
int *payLoadData = docResults[i].PayloadData;
Query.SortInfo sortInfo = Data.DataTypeConvert.GetSortInfo(asc, field.DataType,
payLoadData, field.TabIndex, field.SubTabIndex, field.DataLength);
if (index == 0)
{
docResults[i].SortValue = (long)(sortInfo.DoubleValue * 1000);
}
else
{
docResults[i].SortValue1 = (long)(sortInfo.DoubleValue * 1000);
}
}
}
break;
}
}
}
}
}
public void Run()
{
if (Parameters.Count < 3)
{
throw new ArgumentException("Parameter 1 is table name, Parameter 2 is field name, Parameter 3 is a text for test, Parameter 4 is analyzer type(optional)");
}
Data.DBProvider dbProvider = Data.DBProvider.GetDBProvider(Parameters[0]);
if (dbProvider == null)
{
throw new Data.DataException(string.Format("Can't find table name : {0}", Parameters[0]));
}
Data.Field field = dbProvider.GetField(Parameters[1]);
if (field == null)
{
throw new Data.DataException(string.Format("Can't find field name : {0}", Parameters[2]));
}
if (field.AnalyzerName == null)
{
throw new Data.DataException(string.Format("Field: {0} in {1} is not tokenized", Parameters[2],
dbProvider.Table.Name));
}
Analysis.IAnalyzer analyzer = Data.DBProvider.GetAnalyzer(field.AnalyzerName);
if (analyzer == null)
{
throw new Data.DataException(string.Format("Can't find analyzer name : {0}", field.AnalyzerName));
}
bool clientAnalyzer = false;
if (Parameters.Count == 4)
{
if (Parameters[3].Equals("sqlclient", StringComparison.CurrentCultureIgnoreCase))
{
clientAnalyzer = true;
}
}
AddColumn("Word");
AddColumn("Position");
AddColumn("Rank");
if (clientAnalyzer)
{
foreach (Entity.WordInfo word in analyzer.TokenizeForSqlClient(Parameters[2]))
{
NewRow();
OutputValue("Word", word.Word);
OutputValue("Position", word.Position.ToString());
OutputValue("Rank", word.Rank.ToString());
}
}
else
{
foreach (Entity.WordInfo word in analyzer.Tokenize(Parameters[2]))
{
NewRow();
OutputValue("Word", word.Word);
OutputValue("Position", word.Position.ToString());
OutputValue("Rank", word.Rank.ToString());
}
}
}
