/// <summary> </summary>
		/// <param name="field">The field to read in
		/// </param>
		/// <param name="tvfPointer">The pointer within the tvf file where we should start reading
		/// </param>
		/// <param name="mapper">The mapper used to map the TermVector
		/// </param>
		/// <throws>  IOException </throws>
		private void  ReadTermVector(System.String field, long tvfPointer, TermVectorMapper mapper)
		{
			
			// Now read the data from specified position
			//We don't need to offset by the FORMAT here since the pointer already includes the offset
			tvf.Seek(tvfPointer);
			
			int numTerms = tvf.ReadVInt();
			//System.out.println("Num Terms: " + numTerms);
			// If no terms - return a constant empty termvector. However, this should never occur!
			if (numTerms == 0)
				return ;
			
			bool storePositions;
			bool storeOffsets;
			
			if (format >= FORMAT_VERSION)
			{
				byte bits = tvf.ReadByte();
				storePositions = (bits & STORE_POSITIONS_WITH_TERMVECTOR) != 0;
				storeOffsets = (bits & STORE_OFFSET_WITH_TERMVECTOR) != 0;
			}
			else
			{
				tvf.ReadVInt();
				storePositions = false;
				storeOffsets = false;
			}
			mapper.SetExpectations(field, numTerms, storeOffsets, storePositions);
			int start = 0;
			int deltaLength = 0;
			int totalLength = 0;
			byte[] byteBuffer;
			char[] charBuffer;
			bool preUTF8 = format < FORMAT_UTF8_LENGTH_IN_BYTES;
			
			// init the buffers
			if (preUTF8)
			{
				charBuffer = new char[10];
				byteBuffer = null;
			}
			else
			{
				charBuffer = null;
				byteBuffer = new byte[20];
			}
			
			for (int i = 0; i < numTerms; i++)
			{
				start = tvf.ReadVInt();
				deltaLength = tvf.ReadVInt();
				totalLength = start + deltaLength;
				
				System.String term;
				
				if (preUTF8)
				{
					// Term stored as java chars
					if (charBuffer.Length < totalLength)
					{
						char[] newCharBuffer = new char[(int) (1.5 * totalLength)];
						Array.Copy(charBuffer, 0, newCharBuffer, 0, start);
						charBuffer = newCharBuffer;
					}
					tvf.ReadChars(charBuffer, start, deltaLength);
					term = new System.String(charBuffer, 0, totalLength);
				}
				else
				{
					// Term stored as utf8 bytes
					if (byteBuffer.Length < totalLength)
					{
						byte[] newByteBuffer = new byte[(int) (1.5 * totalLength)];
						Array.Copy(byteBuffer, 0, newByteBuffer, 0, start);
						byteBuffer = newByteBuffer;
					}
					tvf.ReadBytes(byteBuffer, start, deltaLength);
                    term = System.Text.Encoding.UTF8.GetString(byteBuffer, 0, totalLength);
				}
				int freq = tvf.ReadVInt();
				int[] positions = null;
				if (storePositions)
				{
					//read in the positions
					//does the mapper even care about positions?
					if (mapper.IsIgnoringPositions() == false)
					{
						positions = new int[freq];
						int prevPosition = 0;
						for (int j = 0; j < freq; j++)
						{
							positions[j] = prevPosition + tvf.ReadVInt();
							prevPosition = positions[j];
						}
					}
					else
					{
						//we need to skip over the positions.  Since these are VInts, I don't believe there is anyway to know for sure how far to skip
						//
						for (int j = 0; j < freq; j++)
						{
							tvf.ReadVInt();
						}
					}
				}
				TermVectorOffsetInfo[] offsets = null;
				if (storeOffsets)
				{
					//does the mapper even care about offsets?
					if (mapper.IsIgnoringOffsets() == false)
					{
						offsets = new TermVectorOffsetInfo[freq];
						int prevOffset = 0;
						for (int j = 0; j < freq; j++)
						{
							int startOffset = prevOffset + tvf.ReadVInt();
							int endOffset = startOffset + tvf.ReadVInt();
							offsets[j] = new TermVectorOffsetInfo(startOffset, endOffset);
							prevOffset = endOffset;
						}
					}
					else
					{
						for (int j = 0; j < freq; j++)
						{
							tvf.ReadVInt();
							tvf.ReadVInt();
						}
					}
				}
				mapper.Map(term, freq, offsets, positions);
			}
		}
		/// <summary> </summary>
		/// <param name="field">The field to read in
		/// </param>
		/// <param name="tvfPointer">The pointer within the tvf file where we should start reading
		/// </param>
		/// <param name="mapper">The mapper used to map the TermVector
		/// </param>
		/// <returns> The TermVector located at that position
		/// </returns>
		/// <throws>  IOException </throws>
		private void  ReadTermVector(System.String field, long tvfPointer, TermVectorMapper mapper)
		{
			
			// Now read the data from specified position
			//We don't need to offset by the FORMAT here since the pointer already includes the offset
			tvf.Seek(tvfPointer);
			
			int numTerms = tvf.ReadVInt();
			//System.out.println("Num Terms: " + numTerms);
			// If no terms - return a constant empty termvector. However, this should never occur!
			if (numTerms == 0)
				return ;
			
			bool storePositions;
			bool storeOffsets;
			
			if (tvfFormat == FORMAT_VERSION)
			{
				byte bits = tvf.ReadByte();
				storePositions = (bits & STORE_POSITIONS_WITH_TERMVECTOR) != 0;
				storeOffsets = (bits & STORE_OFFSET_WITH_TERMVECTOR) != 0;
			}
			else
			{
				tvf.ReadVInt();
				storePositions = false;
				storeOffsets = false;
			}
			mapper.SetExpectations(field, numTerms, storeOffsets, storePositions);
			int start = 0;
			int deltaLength = 0;
			int totalLength = 0;
			char[] buffer = new char[10]; // init the buffer with a length of 10 character
			char[] previousBuffer = new char[]{};
			
			for (int i = 0; i < numTerms; i++)
			{
				start = tvf.ReadVInt();
				deltaLength = tvf.ReadVInt();
				totalLength = start + deltaLength;
				if (buffer.Length < totalLength)
				{
					// increase buffer
					buffer = null; // give a hint to garbage collector
					buffer = new char[totalLength];
					
					if (start > 0)
					// just copy if necessary
						Array.Copy(previousBuffer, 0, buffer, 0, start);
				}
				
				tvf.ReadChars(buffer, start, deltaLength);
				System.String term = new System.String(buffer, 0, totalLength);
				previousBuffer = buffer;
				int freq = tvf.ReadVInt();
				int[] positions = null;
				if (storePositions)
				{
					//read in the positions
					//does the mapper even care about positions?
					if (mapper.IsIgnoringPositions() == false)
					{
						positions = new int[freq];
						int prevPosition = 0;
						for (int j = 0; j < freq; j++)
						{
							positions[j] = prevPosition + tvf.ReadVInt();
							prevPosition = positions[j];
						}
					}
					else
					{
						//we need to skip over the positions.  Since these are VInts, I don't believe there is anyway to know for sure how far to skip
						//
						for (int j = 0; j < freq; j++)
						{
							tvf.ReadVInt();
						}
					}
				}
				TermVectorOffsetInfo[] offsets = null;
				if (storeOffsets)
				{
					//does the mapper even care about offsets?
					if (mapper.IsIgnoringOffsets() == false)
					{
						offsets = new TermVectorOffsetInfo[freq];
						int prevOffset = 0;
						for (int j = 0; j < freq; j++)
						{
							int startOffset = prevOffset + tvf.ReadVInt();
							int endOffset = startOffset + tvf.ReadVInt();
							offsets[j] = new TermVectorOffsetInfo(startOffset, endOffset);
							prevOffset = endOffset;
						}
					}
					else
					{
						for (int j = 0; j < freq; j++)
						{
							tvf.ReadVInt();
							tvf.ReadVInt();
						}
					}
				}
				mapper.Map(term, freq, offsets, positions);
			}
		}
Beispiel #3
0
        /// <summary> </summary>
        /// <param name="field">The field to read in
        /// </param>
        /// <param name="tvfPointer">The pointer within the tvf file where we should start reading
        /// </param>
        /// <param name="mapper">The mapper used to map the TermVector
        /// </param>
        /// <throws>  IOException </throws>
        private void  ReadTermVector(System.String field, long tvfPointer, TermVectorMapper mapper)
        {
            // Now read the data from specified position
            //We don't need to offset by the FORMAT here since the pointer already includes the offset
            tvf.Seek(tvfPointer);

            int numTerms = tvf.ReadVInt();

            //System.out.println("Num Terms: " + numTerms);
            // If no terms - return a constant empty termvector. However, this should never occur!
            if (numTerms == 0)
            {
                return;
            }

            bool storePositions;
            bool storeOffsets;

            if (format >= FORMAT_VERSION)
            {
                byte bits = tvf.ReadByte();
                storePositions = (bits & STORE_POSITIONS_WITH_TERMVECTOR) != 0;
                storeOffsets   = (bits & STORE_OFFSET_WITH_TERMVECTOR) != 0;
            }
            else
            {
                tvf.ReadVInt();
                storePositions = false;
                storeOffsets   = false;
            }
            mapper.SetExpectations(field, numTerms, storeOffsets, storePositions);
            int start       = 0;
            int deltaLength = 0;
            int totalLength = 0;

            byte[] byteBuffer;
            char[] charBuffer;
            bool   preUTF8 = format < FORMAT_UTF8_LENGTH_IN_BYTES;

            // init the buffers
            if (preUTF8)
            {
                charBuffer = new char[10];
                byteBuffer = null;
            }
            else
            {
                charBuffer = null;
                byteBuffer = new byte[20];
            }

            for (int i = 0; i < numTerms; i++)
            {
                start       = tvf.ReadVInt();
                deltaLength = tvf.ReadVInt();
                totalLength = start + deltaLength;

                System.String term;

                if (preUTF8)
                {
                    // Term stored as java chars
                    if (charBuffer.Length < totalLength)
                    {
                        char[] newCharBuffer = new char[(int)(1.5 * totalLength)];
                        Array.Copy(charBuffer, 0, newCharBuffer, 0, start);
                        charBuffer = newCharBuffer;
                    }
                    tvf.ReadChars(charBuffer, start, deltaLength);
                    term = new System.String(charBuffer, 0, totalLength);
                }
                else
                {
                    // Term stored as utf8 bytes
                    if (byteBuffer.Length < totalLength)
                    {
                        byte[] newByteBuffer = new byte[(int)(1.5 * totalLength)];
                        Array.Copy(byteBuffer, 0, newByteBuffer, 0, start);
                        byteBuffer = newByteBuffer;
                    }
                    tvf.ReadBytes(byteBuffer, start, deltaLength);
                    term = System.Text.Encoding.UTF8.GetString(byteBuffer, 0, totalLength);
                }
                int   freq      = tvf.ReadVInt();
                int[] positions = null;
                if (storePositions)
                {
                    //read in the positions
                    //does the mapper even care about positions?
                    if (mapper.IsIgnoringPositions() == false)
                    {
                        positions = new int[freq];
                        int prevPosition = 0;
                        for (int j = 0; j < freq; j++)
                        {
                            positions[j] = prevPosition + tvf.ReadVInt();
                            prevPosition = positions[j];
                        }
                    }
                    else
                    {
                        //we need to skip over the positions.  Since these are VInts, I don't believe there is anyway to know for sure how far to skip
                        //
                        for (int j = 0; j < freq; j++)
                        {
                            tvf.ReadVInt();
                        }
                    }
                }
                TermVectorOffsetInfo[] offsets = null;
                if (storeOffsets)
                {
                    //does the mapper even care about offsets?
                    if (mapper.IsIgnoringOffsets() == false)
                    {
                        offsets = new TermVectorOffsetInfo[freq];
                        int prevOffset = 0;
                        for (int j = 0; j < freq; j++)
                        {
                            int startOffset = prevOffset + tvf.ReadVInt();
                            int endOffset   = startOffset + tvf.ReadVInt();
                            offsets[j] = new TermVectorOffsetInfo(startOffset, endOffset);
                            prevOffset = endOffset;
                        }
                    }
                    else
                    {
                        for (int j = 0; j < freq; j++)
                        {
                            tvf.ReadVInt();
                            tvf.ReadVInt();
                        }
                    }
                }
                mapper.Map(term, freq, offsets, positions);
            }
        }
        /// <summary> </summary>
        /// <param name="field">The field to read in
        /// </param>
        /// <param name="tvfPointer">The pointer within the tvf file where we should start reading
        /// </param>
        /// <param name="mapper">The mapper used to map the TermVector
        /// </param>
        /// <returns> The TermVector located at that position
        /// </returns>
        /// <throws>  IOException </throws>
        private void  ReadTermVector(System.String field, long tvfPointer, TermVectorMapper mapper)
        {
            // Now read the data from specified position
            //We don't need to offset by the FORMAT here since the pointer already includes the offset
            tvf.Seek(tvfPointer);

            int numTerms = tvf.ReadVInt();

            //System.out.println("Num Terms: " + numTerms);
            // If no terms - return a constant empty termvector. However, this should never occur!
            if (numTerms == 0)
            {
                return;
            }

            bool storePositions;
            bool storeOffsets;

            if (tvfFormat == FORMAT_VERSION)
            {
                byte bits = tvf.ReadByte();
                storePositions = (bits & STORE_POSITIONS_WITH_TERMVECTOR) != 0;
                storeOffsets   = (bits & STORE_OFFSET_WITH_TERMVECTOR) != 0;
            }
            else
            {
                tvf.ReadVInt();
                storePositions = false;
                storeOffsets   = false;
            }
            mapper.SetExpectations(field, numTerms, storeOffsets, storePositions);
            int start       = 0;
            int deltaLength = 0;
            int totalLength = 0;

            char[] buffer         = new char[10];     // init the buffer with a length of 10 character
            char[] previousBuffer = new char[] {};

            for (int i = 0; i < numTerms; i++)
            {
                start       = tvf.ReadVInt();
                deltaLength = tvf.ReadVInt();
                totalLength = start + deltaLength;
                if (buffer.Length < totalLength)
                {
                    // increase buffer
                    buffer = null;                     // give a hint to garbage collector
                    buffer = new char[totalLength];

                    if (start > 0)
                    {
                        // just copy if necessary
                        Array.Copy(previousBuffer, 0, buffer, 0, start);
                    }
                }

                tvf.ReadChars(buffer, start, deltaLength);
                System.String term = new System.String(buffer, 0, totalLength);
                previousBuffer = buffer;
                int   freq      = tvf.ReadVInt();
                int[] positions = null;
                if (storePositions)
                {
                    //read in the positions
                    //does the mapper even care about positions?
                    if (mapper.IsIgnoringPositions() == false)
                    {
                        positions = new int[freq];
                        int prevPosition = 0;
                        for (int j = 0; j < freq; j++)
                        {
                            positions[j] = prevPosition + tvf.ReadVInt();
                            prevPosition = positions[j];
                        }
                    }
                    else
                    {
                        //we need to skip over the positions.  Since these are VInts, I don't believe there is anyway to know for sure how far to skip
                        //
                        for (int j = 0; j < freq; j++)
                        {
                            tvf.ReadVInt();
                        }
                    }
                }
                TermVectorOffsetInfo[] offsets = null;
                if (storeOffsets)
                {
                    //does the mapper even care about offsets?
                    if (mapper.IsIgnoringOffsets() == false)
                    {
                        offsets = new TermVectorOffsetInfo[freq];
                        int prevOffset = 0;
                        for (int j = 0; j < freq; j++)
                        {
                            int startOffset = prevOffset + tvf.ReadVInt();
                            int endOffset   = startOffset + tvf.ReadVInt();
                            offsets[j] = new TermVectorOffsetInfo(startOffset, endOffset);
                            prevOffset = endOffset;
                        }
                    }
                    else
                    {
                        for (int j = 0; j < freq; j++)
                        {
                            tvf.ReadVInt();
                            tvf.ReadVInt();
                        }
                    }
                }
                mapper.Map(term, freq, offsets, positions);
            }
        }