//======================================================//
public static void BuildIndexInt32OnDisk(string dataFileFullName, Encoding dataFileEncoding, uint hashTableSize, NormlizeTextFunction normlizeTextFunction = null)
{
//create header-of-disk-hashtable
var indexHeader = new HashtableIndexFileHeader
(
dataFileFullName,
dataFileEncoding,
hashTableSize,
HashtableIndexFileHeader.SlotType.Int32
);
//create index-file
using (var indexFileStream = IndexFileHelper.CreateFileStreamReadWrite(indexHeader.GetIndexFileFullName()))
using (var diskSlotInt32ReadBuffer = new ReadBuffer <DiskSlotInt32>())
using (var diskTagInt32ReadBuffer = new ReadBuffer <DiskTagInt32>())
{
//Write header in index file
indexFileStream.SeekFromBegin(0);
indexHeader.SerializeIndexHeader(indexFileStream);
Int64 tagAreaOffset = (indexHeader.SizeOf + indexHeader.HashtableSize * DiskSlotInt32.SizeOf);
//Guard from tagAreaOffset > MemorySlotInt32.MaxValue
if (tagAreaOffset > MemorySlotInt32.MaxValue)
{
throw (new InvalidOperationException("tag area offset more then allow possible value => tag area offset: " + tagAreaOffset + ", possible value: " + MemorySlotInt32.MaxValue));
}
//fill disk-hashtable empty slot's
byte[] emptyDiskSlotBytes = DiskSlotInt32.GetEmptyDiskSlot().StructureToByteArray();
for (var i = 0U; i < hashTableSize; i++)
{
indexFileStream.WriteBytes(emptyDiskSlotBytes);
}
DiskTagInt32 diskTag = new DiskTagInt32();
//open data-file
using (var dataFileTextLineReader = new TextLineReader(indexHeader.DataFileFullName, indexHeader.DataFileEncoding))
{
#region [.build index on disk.]
int dataRecordCount = 0;
int dataRecordMaxLenght = 0;
while (!dataFileTextLineReader.EndOfStream)
{
Int64 _posInt64 = dataFileTextLineReader.StreamPosition;
//Guard from file-pointer > MemorySlotInt32.MaxValue
if (_posInt64 > MemorySlotInt32.MaxValue)
{
throw (new InvalidOperationException("file-pointer more then allow possible value => file-pointer: " + _posInt64 + ", possible value: " + MemorySlotInt32.MaxValue));
}
Int32 position = (Int32)_posInt64;
var text = dataFileTextLineReader.ReadLine4Indexing();
if (text.IsEmptyOrNull())
{
continue;
}
//Normlize text if allowed
if (normlizeTextFunction != null)
{
text = normlizeTextFunction(text);
if (text.IsEmptyOrNull())
{
continue;
}
}
uint hashCode = IndexFileHelper.HashFunction(text, indexHeader.HashtableSize);
#region [.code.]
DiskSlotInt32 diskSlot = indexFileStream.SeekAndReadDiskSlotInt32
(
indexHeader.SizeOf + hashCode * DiskSlotInt32.SizeOf,
diskSlotInt32ReadBuffer
);
//Free slot => search text not found
if (diskSlot.PositionInDataFile == Consts.FREE_INT32)
{
//Calc data-slot properties
diskSlot.PositionInDataFile = position;
diskSlot.FirstTagOffset = 0;
//Write in index-file data-slot
indexFileStream.SeekFromCurrentAndWrite
(
-diskSlotInt32ReadBuffer.Size,
diskSlot.StructureToByteArray()
);
}
else
{
if (diskSlot.FirstTagOffset == 0)
{
diskSlot.FirstTagOffset = (Int32)tagAreaOffset;
//ReWrite in index-file data-slot
indexFileStream.SeekFromCurrentAndWrite
(
-diskSlotInt32ReadBuffer.Size,
diskSlot.StructureToByteArray()
);
}
else //if ( diskSlot.FirstTagOffset != 0 )
{
//Search in hashtable tag chain's
diskTag.NextTagOffset = diskSlot.FirstTagOffset;
while (0 < diskTag.NextTagOffset)
{
diskTag = indexFileStream.SeekAndReadDiskTagInt32(diskTag.NextTagOffset, diskTagInt32ReadBuffer);
}
//Calc disk-tag properties
diskTag.PositionInDataFile = position;
diskTag.NextTagOffset = (Int32)tagAreaOffset;
//ReWrite previous tag in chain
indexFileStream.SeekFromCurrentAndWrite
(
-diskTagInt32ReadBuffer.Size,
diskTag.StructureToByteArray()
);
}
//Calc disk-tag properties
diskTag.PositionInDataFile = position;
diskTag.NextTagOffset = 0;
//Write in index-file
indexFileStream.SeekAndWrite
(
tagAreaOffset,
diskTag.StructureToByteArray()
);
//Calc disk-tag offset
tagAreaOffset += DiskTagInt32.SizeOf;
//Guard from tagAreaOffset > MemorySlotInt32.MaxValue
if (tagAreaOffset > MemorySlotInt32.MaxValue)
{
throw (new InvalidOperationException("tag area offset more then allow possible value => tag area offset: " + tagAreaOffset + ", possible value: " + MemorySlotInt32.MaxValue));
}
}
#endregion
dataRecordCount++;
if (dataRecordMaxLenght < text.Length)
{
dataRecordMaxLenght = text.Length; //dataRecordMaxLenght = Math.Max( dataRecordMaxLenght, text.Length );
}
}
indexHeader.SetDataRecordCount(dataRecordCount);
indexHeader.SetDataRecordMaxBytesLenght(indexHeader.DataFileEncoding.GetMaxByteCount(dataRecordMaxLenght));
#endregion
}
//one more time - Write header in index file
indexFileStream.SeekFromBegin(0);
indexHeader.SerializeIndexHeader(indexFileStream);
}
}
/// <summary>
/// Int32 file-pointer version
/// </summary>
/// <param name="searchText"></param>
/// <returns></returns>
private bool IsExistsDiskRoutineIn32(string searchText)
{
//Normlize text if allowed
if (_NormlizeTextFunction != null)
{
searchText = _NormlizeTextFunction(searchText);
}
//Calculate hash-code
uint hashCode = IndexFileHelper.HashFunction(searchText, this.IndexHeader.HashtableSize);
//Search in hashtable
//Read structure [DiskHtRecord_t] from index file
DiskSlotInt32 diskSlot = _IndexFileBinaryReader.SeekAndReadDiskSlotInt32
(
this.IndexHeader.SizeOf + hashCode * DiskSlotInt32.SizeOf,
_DiskSlotInt32ReadBuffer
);
//Free slot => search text not found
if (diskSlot.PositionInDataFile == Consts.FREE_INT32)
{
// == NOT FOUND
return(false); /*"ни хуяшки нету :("*/
}
//Set file-pointer in position into data-file
//Read text from data-file
var text = _DataFileTextLineReader.SeekAndRead4Searching(diskSlot.PositionInDataFile);
//Normlize text if allowed
if (_NormlizeTextFunction != null)
{
text = _NormlizeTextFunction(text);
}
//Input text equal text in data-file => search text found (with first step)
if (text == searchText)
{
// == FOUND ! first step
return(true);
}
//Search in hashtable tag chain's
DiskTagInt32 diskTag = new DiskTagInt32()
{
NextTagOffset = diskSlot.FirstTagOffset
};
while (0 < diskTag.NextTagOffset)
{
diskTag = _IndexFileBinaryReader.SeekAndReadDiskTagInt32(diskTag.NextTagOffset, _DiskTagInt32ReadBuffer);
text = _DataFileTextLineReader.SeekAndRead4Searching(diskTag.PositionInDataFile);
//Normlize text if allowed
if (_NormlizeTextFunction != null)
{
text = _NormlizeTextFunction(text);
}
if (text == searchText)
{
// == FOUND !
return(true);
}
}
// == NOT FOUND
return(false);
}
/// <summary>
/// Int32 file-pointer version
/// </summary>
/// <param name="dataFileFullName"></param>
/// <param name="dataFileEncoding"></param>
/// <param name="hashTableSize"></param>
private static void BuildIndexInt32InMemory(string dataFileFullName, Encoding dataFileEncoding, uint hashTableSize, NormlizeTextFunction normlizeTextFunction)
{
//
var indexHeader = new HashtableIndexFileHeader
(
dataFileFullName,
dataFileEncoding,
hashTableSize,
HashtableIndexFileHeader.SlotType.Int32
);
//
var hashTable = new MemorySlotInt32[indexHeader.HashtableSize];
//
using (var dataFileTextLineReader = new TextLineReader(indexHeader.DataFileFullName, indexHeader.DataFileEncoding))
{
#region [.2 calc index in memory.]
int dataRecordCount = 0;
int dataRecordMaxLenght = 0;
while (!dataFileTextLineReader.EndOfStream)
{
Int64 _posInt64 = dataFileTextLineReader.StreamPosition;
//Guard from file-pointer > MemorySlotInt32.MaxValue
if (_posInt64 > MemorySlotInt32.MaxValue)
{
throw (new InvalidOperationException("file-pointer more then allow possible value => file-pointer: " + _posInt64 + ", possible value: " + MemorySlotInt32.MaxValue));
}
Int32 position = (Int32)_posInt64;
var text = dataFileTextLineReader.ReadLine4Indexing();
if (text.IsEmptyOrNull())
{
continue;
}
//Normlize text if allowed
if (normlizeTextFunction != null)
{
text = normlizeTextFunction(text);
if (text.IsEmptyOrNull())
{
continue;
}
}
uint hashCode = IndexFileHelper.HashFunction(text, indexHeader.HashtableSize);
MemorySlotInt32 memorySlot = hashTable[hashCode];
if (memorySlot == null)
{
hashTable[hashCode] = new MemorySlotInt32(position);
}
else
{
MemoryTagInt32 newMemoryTag = new MemoryTagInt32(position);
if (memorySlot.FirstMemoryTag == null)
{
memorySlot.FirstMemoryTag = newMemoryTag;
}
else
{
MemoryTagInt32 memoryTag = memorySlot.FirstMemoryTag;
while (memoryTag.NextMemoryTag != null)
{
memoryTag = memoryTag.NextMemoryTag;
}
memoryTag.NextMemoryTag = newMemoryTag;
}
}
dataRecordCount++;
if (dataRecordMaxLenght < text.Length)
{
dataRecordMaxLenght = text.Length; //dataRecordMaxLenght = Math.Max( dataRecordMaxLenght, text.Length );
}
}
indexHeader.SetDataRecordCount(dataRecordCount);
indexHeader.SetDataRecordMaxBytesLenght(indexHeader.DataFileEncoding.GetMaxByteCount(dataRecordMaxLenght));
#endregion
}
#region [.Calulate Tag collision statistica.]
foreach (var memorySlot in hashTable)
{
if (memorySlot == null)
{
//empty slot's in hash-table
indexHeader.TagCollisionStatistica.IncremetByKey(-1);
}
else
{
//Current tag chain depth
int currentTagChainDepth = 0;
//Get first memory-tag
MemoryTagInt32 memoryTag = memorySlot.FirstMemoryTag;
while (memoryTag != null)
{
//Get next tag
memoryTag = memoryTag.NextMemoryTag;
//Current tag chain depth
currentTagChainDepth++;
}
//
indexHeader.TagCollisionStatistica.IncremetByKey(currentTagChainDepth);
}
}
#endregion
//
using (var indexFileBinaryWriter = IndexFileHelper.CreateBinaryWriterRandomAccess(indexHeader.GetIndexFileFullName()))
{
#region [.3 write hash table on disk.]
var diskSlot = new DiskSlotInt32();
var diskTag = new DiskTagInt32();
byte[] emptyDiskSlotBytes = DiskSlotInt32.GetEmptyDiskSlot().StructureToByteArray();
//Write header in index file
indexFileBinaryWriter.SeekFromBegin(0);
indexHeader.SerializeIndexHeader(indexFileBinaryWriter);
Int64 tagAreaOffset = (indexHeader.SizeOf + indexHeader.HashtableSize * DiskSlotInt32.SizeOf);
//Guard from tagAreaOffset > MemorySlotInt32.MaxValue
if (tagAreaOffset > MemorySlotInt32.MaxValue)
{
throw (new InvalidOperationException("tag area offset more then allow possible value => tag area offset: " + tagAreaOffset + ", possible value: " + MemorySlotInt32.MaxValue));
}
#region [.iterate over memory-slot's.]
uint hashTableItemIndex = 0;
foreach (var memorySlot in hashTable)
{
//Empty hash-table slot
if (memorySlot == null)
{
//Write in index-file empty slot
indexFileBinaryWriter.SeekAndWrite
(
indexHeader.SizeOf + hashTableItemIndex * DiskSlotInt32.SizeOf,
emptyDiskSlotBytes
);
}
//Have a data hash-table slot
else
{
//Calc data-slot properties
diskSlot.PositionInDataFile = memorySlot.PositionInDataFile;
diskSlot.FirstTagOffset = (memorySlot.FirstMemoryTag != null) ? (Int32)tagAreaOffset : 0;
//Write in index-file data-slot
indexFileBinaryWriter.SeekAndWrite
(
indexHeader.SizeOf + hashTableItemIndex * DiskSlotInt32.SizeOf,
diskSlot.StructureToByteArray()
);
//Calc Tag collision statistica
int currentTagChainDepth = 0;
//Get first memory-tag
MemoryTagInt32 memoryTag = memorySlot.FirstMemoryTag;
while (memoryTag != null)
{
//Calc disk-tag properties
diskTag.PositionInDataFile = memoryTag.PositionInDataFile;
diskTag.NextTagOffset = (memoryTag.NextMemoryTag != null) ? (Int32)tagAreaOffset + DiskTagInt32.SizeOf : 0;
//Write in index-file
indexFileBinaryWriter.SeekAndWrite
(
tagAreaOffset,
diskTag.StructureToByteArray()
);
//Calc disk-tag offset
tagAreaOffset += DiskTagInt32.SizeOf;
//Guard from tagAreaOffset > MemorySlotInt32.MaxValue
if (tagAreaOffset > MemorySlotInt32.MaxValue)
{
throw (new InvalidOperationException("tag area offset more then allow possible value => tag area offset: " + tagAreaOffset + ", possible value: " + MemorySlotInt32.MaxValue));
}
//Get next tag
memoryTag = memoryTag.NextMemoryTag;
//Calc Tag collision statistica
currentTagChainDepth++;
}
/*
* //Calulate Tag collision statistica
* //---indexHeader.TagCollisionStatistica.IncremetByKey( currentTagChainDepth /*memoryHash.GetTagChainDepth()* / );
*/
}
hashTableItemIndex++;
}
#endregion
#endregion
}
//Free memory
hashTable = null;
GC.Collect();
}
