//
// Find out what we should generate next. It can be a symbol, a distance/length pair
// or a symbol followed by distance/length pair
//
internal bool GetNextSymbolOrMatch(Match match)
{
Debug.Assert(_bufPos >= FastEncoderWindowSize && _bufPos < (2 * FastEncoderWindowSize),
"Invalid Buffer Position!");
// initialise the value of the hash, no problem if locations bufPos, bufPos+1
// are invalid (not enough data), since we will never insert using that hash value
var hash = HashValue(0, _window[_bufPos]);
hash = HashValue(hash, _window[_bufPos + 1]);
int matchLen;
var matchPos = 0;
this.VerifyHashes(); // Debug only code
if (_bufEnd - _bufPos <= 3)
{
// The hash value becomes corrupt when we get within 3 characters of the end of the
// input window, since the hash value is based on 3 characters. We just stop
// inserting into the hash table at this point, and allow no matches.
matchLen = 0;
}
else
{
// insert string into hash table and return most recent location of same hash value
var search = (int)InsertString(ref hash);
// did we find a recent location of this hash value?
if (search != 0)
{
// yes, now find a match at what we'll call position X
matchLen = FindMatch(search, out matchPos, SearchDepth);
// truncate match if we're too close to the end of the input window
if (_bufPos + matchLen > _bufEnd)
matchLen = _bufEnd - _bufPos;
}
else
{
// no most recent location found
matchLen = 0;
}
}
if (matchLen < MinMatch)
{
// didn't find a match, so output unmatched char
match.State = MatchState.HasSymbol;
match.Symbol = _window[_bufPos];
_bufPos++;
}
else
{
// bufPos now points to X+1
_bufPos++;
// is this match so good (long) that we should take it automatically without
// checking X+1 ?
if (matchLen <= LazyMatchThreshold)
{
int nextMatchLen;
var nextMatchPos = 0;
// search at position X+1
var search = (int)InsertString(ref hash);
// no, so check for a better match at X+1
if (search != 0)
{
nextMatchLen = FindMatch(search, out nextMatchPos,
matchLen < GoodLength ? SearchDepth : (SearchDepth >> 2));
// truncate match if we're too close to the end of the window
// note: nextMatchLen could now be < MinMatch
if (_bufPos + nextMatchLen > _bufEnd)
{
nextMatchLen = _bufEnd - _bufPos;
}
}
else
{
nextMatchLen = 0;
}
// right now X and X+1 are both inserted into the search tree
if (nextMatchLen > matchLen)
{
// since nextMatchLen > matchLen, it can't be < MinMatch here
// match at X+1 is better, so output unmatched char at X
match.State = MatchState.HasSymbolAndMatch;
match.Symbol = _window[_bufPos - 1];
match.Position = nextMatchPos;
match.Length = nextMatchLen;
// insert remainder of second match into search tree
// example: (*=inserted already)
//
// X X+1 X+2 X+3 X+4
// * *
// nextmatchlen=3
// bufPos
//
// If nextMatchLen == 3, we want to perform 2
// insertions (at X+2 and X+3). However, first we must
// inc bufPos.
//
_bufPos++; // now points to X+2
matchLen = nextMatchLen;
InsertStrings(ref hash, matchLen);
}
else
{
// match at X is better, so take it
match.State = MatchState.HasMatch;
match.Position = matchPos;
match.Length = matchLen;
// Insert remainder of first match into search tree, minus the first
// two locations, which were inserted by the FindMatch() calls.
//
// For example, if matchLen == 3, then we've inserted at X and X+1
// already (and bufPos is now pointing at X+1), and now we need to insert
// only at X+2.
//
matchLen--;
_bufPos++; // now bufPos points to X+2
InsertStrings(ref hash, matchLen);
}
}
else
{
// match_length >= good_match
// in assertion: bufPos points to X+1, location X inserted already
// first match is so good that we're not even going to check at X+1
match.State = MatchState.HasMatch;
match.Position = matchPos;
match.Length = matchLen;
// insert remainder of match at X into search tree
InsertStrings(ref hash, matchLen);
}
}
if (_bufPos == 2 * FastEncoderWindowSize)
{
MoveWindows();
}
return true;
}
private readonly Match _currentMatch; // current match in history window
public FastEncoder()
{
_inputWindow = new FastEncoderWindow();
_currentMatch = new Match();
}