private int GetUnitIndexTour(string value, int segmentStart, byte[] buffer)
{
string word;
long position;
long length;
for (int i = 0; i < SegmentSize / UnitBlockSize; i++)
{
if (UnicodeString.Decode(buffer, segmentStart + i * UnitBlockSize, 0, out word, "",
out position, out length) < 0)
{
return(i - 1);
}
int c = UnicodeString.Comparer(word, value);
if (c > 0)
{
return(i - 1);
}
else if (c == 0)
{
return(i);
}
}
return(SegmentSize / UnitBlockSize - 1);
}
public DDXUnit GetNext()
{
if (_Mode != Mode.Enum)
{
throw new System.IO.IOException(string.Format("DDX file: {0} is not enum mode. Can't be read!", _FilePath));
}
if (_SegmentEnum == null)
{
EnumReset();
}
string word;
long position;
long length;
if (_CurrentPositionInUnitBlockEnum < 0)
{
return(null);
}
_CurrentPositionInUnitBlockEnum = UnicodeString.Decode(_SegmentEnum, _CurrentUnitIndexEnum * UnitBlockSize,
_CurrentPositionInUnitBlockEnum,
out word, _PreWordEnum, out position, out length);
if (_CurrentPositionInUnitBlockEnum < 0)
{
_CurrentPositionInUnitBlockEnum = 0;
_CurrentUnitIndexEnum++;
if (_CurrentUnitIndexEnum * UnitBlockSize >= _SegmentEnum.Length)
{
if (!Hubble.Framework.IO.File.ReadToBuffer(_File, _SegmentEnum))
{
return(null);
}
else
{
_CurrentUnitIndexEnum = 0;
}
}
_CurrentPositionInUnitBlockEnum = UnicodeString.Decode(_SegmentEnum, _CurrentUnitIndexEnum * UnitBlockSize,
_CurrentPositionInUnitBlockEnum,
out word, "", out position, out length);
if (_CurrentPositionInUnitBlockEnum < 0)
{
return(null);
}
}
_PreWordEnum = word;
return(new DDXUnit(word, position, length));
}
private void GetSegmentStartStrings()
{
List <string> strList = new List <string>(256);
_File.Seek(0, SeekOrigin.Begin);
byte[] segment = new byte[SegmentSize];
if (Hubble.Framework.IO.File.ReadToBuffer(_File, segment))
{
do
{
string word;
long position;
long length;
if (UnicodeString.Decode(segment, 0, out word, "", out position, out length) >= 0)
{
strList.Add(word);
}
} while (Hubble.Framework.IO.File.ReadToBuffer(_File, segment));
}
_SegmentStartStrings = strList.ToArray();
}
public DDXUnit Find(string word)
{
lock (_ReadLockObj)
{
int segmentIndex = GetSegmentIndex(word);
if (segmentIndex < 0)
{
return(null);
}
else
{
if (_DDXFileBuffer == null)
{
if (segmentIndex * SegmentSize >= _File.Length)
{
return(null);
}
}
else
{
if (segmentIndex * SegmentSize >= _DDXFileBuffer.Length)
{
return(null);
}
}
byte[] segmentBuffer;
int segmentStart;
if (_DDXFileBuffer == null)
{
_File.Seek(segmentIndex * SegmentSize, SeekOrigin.Begin);
segmentBuffer = new byte[SegmentSize]; //One segment include 32 unit blocks .
segmentStart = 0;
}
else
{
segmentBuffer = _DDXFileBuffer;
segmentStart = segmentIndex * SegmentSize;
}
if (_DDXFileBuffer == null)
{
if (!Hubble.Framework.IO.File.ReadToBuffer(_File, segmentBuffer))
{
throw new StoreException(string.Format("DDX Find fail, Word:{0} FilePostion = {1}",
word, _File.Position));
}
}
int unitIndex;
if (segmentIndex == _SegmentStartStrings.Length - 1)
{
//Last segment
unitIndex = GetUnitIndexTour(word, segmentStart, segmentBuffer);
}
else
{
unitIndex = GetUnitIndexBin(word, segmentStart, segmentBuffer);
}
if (unitIndex < 0)
{
return(null);
}
string enumWord;
string preWord = "";
long position;
long length;
int positionInUnit = 0;
positionInUnit = UnicodeString.Decode(segmentBuffer, segmentStart + unitIndex * UnitBlockSize, positionInUnit,
out enumWord, preWord, out position, out length);
preWord = enumWord;
if (positionInUnit < 0)
{
return(null);
}
int c = UnicodeString.Comparer(enumWord, word);
while (c < 0)
{
positionInUnit = UnicodeString.Decode(segmentBuffer, segmentStart + unitIndex * UnitBlockSize, positionInUnit,
out enumWord, preWord, out position, out length);
preWord = enumWord;
if (positionInUnit < 0)
{
return(null);
}
c = UnicodeString.Comparer(enumWord, word);
}
if (c == 0)
{
return(new DDXUnit(word, position, length));
}
else
{
return(null);
}
}
}
}
private int GetUnitIndexBin(string value, int segmentStart, byte[] buffer)
{
byte[] objArray = buffer;
if (objArray.Length <= 0)
{
return(-1);
}
int lo = 0;
int hi = (SegmentSize / UnitBlockSize) - 1;
string word;
long position;
long length;
while (lo <= hi)
{
// i might overflow if lo and hi are both large positive numbers.
int i = lo + ((hi - lo) >> 1);
if (UnicodeString.Decode(buffer, segmentStart + i * UnitBlockSize, 0, out word, "",
out position, out length) < 0)
{
throw new StoreException(string.Format("GetUnitIndexBin fail, Word:{0} SegmentStart:{1} index={2}",
word, segmentStart, i));
}
int c = UnicodeString.Comparer(word, value);
if (c == 0)
{
return(i);
}
if (c < 0)
{
lo = i + 1;
}
else
{
hi = i - 1;
}
}
lo--;
if (lo >= objArray.Length)
{
lo = objArray.Length - 1;
}
else if (lo < 0)
{
lo = 0;
}
if (UnicodeString.Decode(buffer, segmentStart + lo * UnitBlockSize, 0, out word, "",
out position, out length) < 0)
{
throw new StoreException(string.Format("GetUnitIndexBin fail, Word:{0} SegmentStart:{1} index={2}",
word, segmentStart, lo));
}
if (UnicodeString.Comparer(word, value) <= 0)
{
return(lo);
}
else
{
return(-1);
}
}