public static void CheckFooter(ChecksumIndexInput input)
{
var scratch = new BytesRef();
var expectedChecksum = string.Format("{0:D}", input.Checksum);
ReadLine(input, scratch);
if (StringHelper.StartsWith(scratch, CHECKSUM) == false)
{
throw new CorruptIndexException("SimpleText failure: expected checksum line but got " +
scratch.Utf8ToString() + " (resource=" + input + ")");
}
var actualChecksum =
(new BytesRef(scratch.Bytes, CHECKSUM.Length, scratch.Length - CHECKSUM.Length)).Utf8ToString();
if (!expectedChecksum.Equals(actualChecksum))
{
throw new CorruptIndexException("SimpleText checksum failure: " + actualChecksum + " != " +
expectedChecksum + " (resource=" + input + ")");
}
if (input.Length() != input.FilePointer)
{
throw new CorruptIndexException(
"Unexpected stuff at the end of file, please be careful with your text editor! (resource=" + input +
")");
}
}
public override void LookupOrd(long ord, BytesRef result)
{
if (ord < 0 || ord >= _field.NumValues)
{
throw new IndexOutOfRangeException("ord must be 0 .. " + (_field.NumValues - 1) + "; got " + ord);
}
_input.Seek(_field.DataStartFilePointer + ord * (9 + _field.Pattern.Length + _field.MaxLength));
SimpleTextUtil.ReadLine(_input, _scratch);
Debug.Assert(StringHelper.StartsWith(_scratch, SimpleTextDocValuesWriter.LENGTH),
"got " + _scratch.Utf8ToString() + " in=" + _input);
int len;
try
{
len =
(int)
_decoderFormat.parse(_scratch.Bytes.SubList(
_scratch.Offset + SimpleTextDocValuesWriter.LENGTH.Length,
_scratch.Length - SimpleTextDocValuesWriter.LENGTH.Length).ToString());
}
catch (Exception pe)
{
var e = new CorruptIndexException("failed to parse int length (resource=" + _input + ")", pe);
throw e;
}
result.Bytes = new byte[len];
result.Offset = 0;
result.Length = len;
_input.ReadBytes(result.Bytes, 0, len);
}
public override long Get(int docId)
{
if (docId < 0 || docId >= _outerInstance.MAX_DOC)
{
throw new IndexOutOfRangeException("docID must be 0 .. " + (_outerInstance.MAX_DOC - 1) +
"; got " + docId);
}
_input.Seek(_field.DataStartFilePointer + (1 + _field.Pattern.Length + 2) * docId);
SimpleTextUtil.ReadLine(_input, _scratch);
long bd;
try
{
bd = long.Parse(_scratch.Utf8ToString());
}
catch (FormatException ex)
{
throw new CorruptIndexException("failed to parse long value (resource=" + _input + ")", ex);
}
SimpleTextUtil.ReadLine(_input, _scratch); // read the line telling us if its real or not
return(_field.MinValue + bd);
}
public override long Get(int docId)
{
if (docId < 0 || docId >= _outerInstance.MAX_DOC)
{
throw new IndexOutOfRangeException("docID must be 0 .. " + (_outerInstance.MAX_DOC - 1) +
"; got " + docId);
}
_input.Seek(_field.DataStartFilePointer + (1 + _field.Pattern.Length + 2) * docId);
SimpleTextUtil.ReadLine(_input, _scratch);
decimal bd;
try
{
// LUCNENENET: .NET doesn't have a way to specify a pattern with decimal, but all of the standard ones are built in.
bd = decimal.Parse(_scratch.Utf8ToString(), NumberStyles.Float, CultureInfo.InvariantCulture);
}
catch (FormatException ex)
{
throw new CorruptIndexException("failed to parse long value (resource=" + _input + ")", ex);
}
SimpleTextUtil.ReadLine(_input, _scratch); // read the line telling us if its real or not
return((long)BigInteger.Add(new BigInteger(_field.MinValue), new BigInteger(bd)));
}
public override void LookupOrd(long ord, BytesRef result)
{
if (ord < 0 || ord >= field.NumValues)
{
throw new IndexOutOfRangeException("ord must be 0 .. " + (field.NumValues - 1) + "; got " + ord);
}
@in.Seek(field.DataStartFilePointer + ord * (9 + field.Pattern.Length + field.MaxLength));
SimpleTextUtil.ReadLine(@in, scratch);
Debug.Assert(StringHelper.StartsWith(scratch, SimpleTextDocValuesWriter.LENGTH),
"got " + scratch.Utf8ToString() + " in=" + @in);
int len;
try
{
len =
(int)
decoder.parse(scratch.Bytes.SubList(
scratch.Offset + SimpleTextDocValuesWriter.LENGTH.Length,
scratch.Length - SimpleTextDocValuesWriter.LENGTH.Length).ToString());
}
catch (ParseException pe)
{
CorruptIndexException e =
new CorruptIndexException("failed to parse int length (resource=" + @in + ")");
e.initCause(pe);
throw e;
}
result.Bytes = new sbyte[len];
result.Offset = 0;
result.Length = len;
@in.ReadBytes(result.Bytes, 0, len);
}
public override void LookupOrd(long ord, BytesRef result)
{
if (ord < 0 || ord >= _field.NumValues)
{
throw new IndexOutOfRangeException("ord must be 0 .. " + (_field.NumValues - 1) + "; got " + ord);
}
_input.Seek(_field.DataStartFilePointer + ord * (9 + _field.Pattern.Length + _field.MaxLength));
SimpleTextUtil.ReadLine(_input, _scratch);
Debug.Assert(StringHelper.StartsWith(_scratch, SimpleTextDocValuesWriter.LENGTH),
"got " + _scratch.Utf8ToString() + " in=" + _input);
int len;
try
{
// LUCNENENET: .NET doesn't have a way to specify a pattern with integer, but all of the standard ones are built in.
len = int.Parse(Encoding.UTF8.GetString(_scratch.Bytes, _scratch.Offset + SimpleTextDocValuesWriter.LENGTH.Length,
_scratch.Length - SimpleTextDocValuesWriter.LENGTH.Length), NumberStyles.Integer, CultureInfo.InvariantCulture);
}
catch (Exception pe)
{
var e = new CorruptIndexException("failed to parse int length (resource=" + _input + ")", pe);
throw e;
}
result.Bytes = new byte[len];
result.Offset = 0;
result.Length = len;
_input.ReadBytes(result.Bytes, 0, len);
}
public static void CheckFooter(ChecksumIndexInput input)
{
var scratch = new BytesRef();
var expectedChecksum = string.Format("{0:D}", input.Checksum);
ReadLine(input, scratch);
if (StringHelper.StartsWith(scratch, CHECKSUM) == false)
{
throw new CorruptIndexException("SimpleText failure: expected checksum line but got " +
scratch.Utf8ToString() + " (resource=" + input + ")");
}
var actualChecksum =
(new BytesRef(scratch.Bytes, CHECKSUM.Length, scratch.Length - CHECKSUM.Length)).Utf8ToString();
if (!expectedChecksum.Equals(actualChecksum))
{
throw new CorruptIndexException("SimpleText checksum failure: " + actualChecksum + " != " +
expectedChecksum + " (resource=" + input + ")");
}
if (input.Length() != input.FilePointer)
{
throw new CorruptIndexException(
"Unexpected stuff at the end of file, please be careful with your text editor! (resource=" + input +
")");
}
}
public override int GetOrd(int docId)
{
if (docId < 0 || docId >= _outerInstance.MAX_DOC)
{
throw new IndexOutOfRangeException("docID must be 0 .. " + (_outerInstance.MAX_DOC - 1) + "; got " +
docId);
}
_input.Seek(_field.DataStartFilePointer + _field.NumValues * (9 + _field.Pattern.Length + _field.MaxLength) +
docId * (1 + _field.OrdPattern.Length));
SimpleTextUtil.ReadLine(_input, _scratch);
try
{
return(_scratch.Utf8ToString().ToString(_ordDecoderFormat) - 1);
}
catch (Exception pe)
{
var e = new CorruptIndexException("failed to parse ord (resource=" + _input + ")", pe);
throw e;
}
}
public override int GetOrd(int docId)
{
if (docId < 0 || docId >= _outerInstance.MAX_DOC)
{
throw new IndexOutOfRangeException("docID must be 0 .. " + (_outerInstance.MAX_DOC - 1) + "; got " +
docId);
}
_input.Seek(_field.DataStartFilePointer + _field.NumValues * (9 + _field.Pattern.Length + _field.MaxLength) +
docId * (1 + _field.OrdPattern.Length));
SimpleTextUtil.ReadLine(_input, _scratch);
try
{
// LUCNENENET: .NET doesn't have a way to specify a pattern with integer, but all of the standard ones are built in.
return(int.Parse(_scratch.Utf8ToString(), NumberStyles.Integer, CultureInfo.InvariantCulture) - 1);
}
catch (Exception pe)
{
var e = new CorruptIndexException("failed to parse ord (resource=" + _input + ")", pe);
throw e;
}
}
public override int GetOrd(int docID)
{
if (docID < 0 || docID >= outerInstance.MAX_DOC)
{
throw new IndexOutOfRangeException("docID must be 0 .. " + (outerInstance.MAX_DOC - 1) + "; got " +
docID);
}
@in.Seek(field.DataStartFilePointer + field.NumValues * (9 + field.Pattern.Length + field.MaxLength) +
docID * (1 + field.OrdPattern.Length));
SimpleTextUtil.ReadLine(@in, scratch);
try
{
return((long)(int)ordDecoder.Parse(scratch.Utf8ToString()) - 1);
}
catch (ParseException pe)
{
CorruptIndexException e = new CorruptIndexException("failed to parse ord (resource=" + @in + ")");
e.initCause(pe);
throw e;
}
}
public override int NextPosition()
{
int pos;
if (_readPositions)
{
SimpleTextUtil.ReadLine(_in, _scratch);
Debug.Assert(StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.POS), "got line=" + _scratch.Utf8ToString());
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.POS.Length, _scratch.Length - SimpleTextFieldsWriter.POS.Length,
_scratchUtf162);
pos = ArrayUtil.ParseInt(_scratchUtf162.Chars, 0, _scratchUtf162.Length);
}
else
{
pos = -1;
}
if (_readOffsets)
{
SimpleTextUtil.ReadLine(_in, _scratch);
Debug.Assert(StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.START_OFFSET), "got line=" + _scratch.Utf8ToString());
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.START_OFFSET.Length,
_scratch.Length - SimpleTextFieldsWriter.START_OFFSET.Length, _scratchUtf162);
_startOffset = ArrayUtil.ParseInt(_scratchUtf162.Chars, 0, _scratchUtf162.Length);
SimpleTextUtil.ReadLine(_in, _scratch);
Debug.Assert(StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.END_OFFSET), "got line=" + _scratch.Utf8ToString());
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.END_OFFSET.Length,
_scratch.Length - SimpleTextFieldsWriter.END_OFFSET.Length, _scratchUtf162);
_endOffset = ArrayUtil.ParseInt(_scratchUtf162.Chars, 0, _scratchUtf162.Length);
}
long fp = _in.FilePointer;
SimpleTextUtil.ReadLine(_in, _scratch);
if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.PAYLOAD))
{
int len = _scratch.Length - SimpleTextFieldsWriter.PAYLOAD.Length;
if (_scratch2.Bytes.Length < len)
{
_scratch2.Grow(len);
}
Array.Copy(_scratch.Bytes, SimpleTextFieldsWriter.PAYLOAD.Length, _scratch2.Bytes, 0, len);
_scratch2.Length = len;
_payload = _scratch2;
}
else
{
_payload = null;
_in.Seek(fp);
}
return(pos);
}
public override int NextDoc()
{
if (_docId == NO_MORE_DOCS)
{
return(_docId);
}
bool first = true;
int termFreq = 0;
while (true)
{
long lineStart = _in.FilePointer;
SimpleTextUtil.ReadLine(_in, _scratch);
if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.DOC))
{
if (!first && (_liveDocs == null || _liveDocs.Get(_docId)))
{
_in.Seek(lineStart);
if (!_omitTf)
{
_tf = termFreq;
}
return(_docId);
}
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.DOC.Length, _scratch.Length - SimpleTextFieldsWriter.DOC.Length,
_scratchUtf16);
_docId = ArrayUtil.ParseInt(_scratchUtf16.Chars, 0, _scratchUtf16.Length);
termFreq = 0;
first = false;
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.FREQ))
{
UnicodeUtil.UTF8toUTF16(_scratch.Bytes, _scratch.Offset + SimpleTextFieldsWriter.FREQ.Length,
_scratch.Length - SimpleTextFieldsWriter.FREQ.Length, _scratchUtf16);
termFreq = ArrayUtil.ParseInt(_scratchUtf16.Chars, 0, _scratchUtf16.Length);
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.POS))
{
// skip termFreq++;
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.START_OFFSET))
{
// skip
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.END_OFFSET))
{
// skip
}
else if (StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.PAYLOAD))
{
// skip
}
else
{
Debug.Assert(
StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.TERM) || StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.FIELD) ||
StringHelper.StartsWith(_scratch, SimpleTextFieldsWriter.END), "scratch=" + _scratch.Utf8ToString());
if (!first && (_liveDocs == null || _liveDocs.Get(_docId)))
{
_in.Seek(lineStart);
if (!_omitTf)
{
_tf = termFreq;
}
return(_docId);
}
return(_docId = NO_MORE_DOCS);
}
}
}
