public override int NextCodePoint()
{
int c;
for (; ;)
{
if (state == State.IterCheckFwd)
{
c = iter.Next();
if (c < 0)
{
return(c);
}
if (CollationFCD.HasTccc(c))
{
if (CollationFCD.MaybeTibetanCompositeVowel(c) ||
CollationFCD.HasLccc(iter.Current))
{
iter.Previous();
if (!NextSegment())
{
return(Collation.SENTINEL_CP);
}
continue;
}
}
if (IsLeadSurrogate(c))
{
int trail = iter.Next();
if (IsTrailSurrogate(trail))
{
return(Character.ToCodePoint((char)c, (char)trail));
}
else if (trail >= 0)
{
iter.Previous();
}
}
return(c);
}
else if (state == State.IterInFCDSegment && pos != limit)
{
c = iter.NextCodePoint();
pos += Character.CharCount(c);
Debug.Assert(c >= 0);
return(c);
}
else if (state.CompareTo(State.InNormIterAtLimit) >= 0 &&
pos != normalized.Length)
{
c = normalized.CodePointAt(pos);
pos += Character.CharCount(c);
return(c);
}
else
{
SwitchToForward();
}
}
}
/// <summary>
/// Extend the FCD text segment backward or normalize around pos.
/// To be called when checkDir < 0 && pos != start.
/// Returns with checkDir == 0 and pos != start.
/// </summary>
private void PreviousSegment()
{
Debug.Assert(checkDir < 0 && seq == rawSeq && pos != start);
// The input text [pos..segmentLimit[ passes the FCD check.
int p = pos;
int nextCC = 0;
for (; ;)
{
// Fetch the previous character's fcd16 value.
int q = p;
int c = Character.CodePointBefore(seq, p);
p -= Character.CharCount(c);
int fcd16 = nfcImpl.GetFCD16(c);
int trailCC = fcd16 & 0xff;
if (trailCC == 0 && q != pos)
{
// FCD boundary after the [p, q[ character.
start = segmentStart = q;
break;
}
if (trailCC != 0 && ((nextCC != 0 && trailCC > nextCC) ||
CollationFCD.IsFCD16OfTibetanCompositeVowel(fcd16)))
{
// Fails FCD check. Find the previous FCD boundary and normalize.
do
{
q = p;
if (fcd16 <= 0xff || p == rawStart)
{
break;
}
c = Character.CodePointBefore(seq, p);
p -= Character.CharCount(c);
} while ((fcd16 = nfcImpl.GetFCD16(c)) != 0);
Normalize(q, pos);
pos = limit;
break;
}
nextCC = fcd16 >> 8;
if (p == rawStart || nextCC == 0)
{
// FCD boundary before the following character.
start = segmentStart = p;
break;
}
}
Debug.Assert(pos != start);
checkDir = 0;
}
protected override long HandleNextCE32()
{
int c;
for (; ;)
{
if (state == State.IterCheckFwd)
{
c = iter.Next();
if (c < 0)
{
return(NO_CP_AND_CE32);
}
if (CollationFCD.HasTccc(c))
{
if (CollationFCD.MaybeTibetanCompositeVowel(c) ||
CollationFCD.HasLccc(iter.Current))
{
iter.Previous();
if (!NextSegment())
{
c = Collation.SENTINEL_CP;
return(Collation.FALLBACK_CE32);
}
continue;
}
}
break;
}
else if (state == State.IterInFCDSegment && pos != limit)
{
c = iter.Next();
++pos;
Debug.Assert(c >= 0);
break;
}
else if (state.CompareTo(State.InNormIterAtLimit) >= 0 &&
pos != normalized.Length)
{
c = normalized[pos++];
break;
}
else
{
SwitchToForward();
}
}
return(MakeCodePointAndCE32Pair(c, trie.GetFromU16SingleLead((char)c)));
}
/// <summary>
/// Extend the FCD text segment forward or normalize around pos.
/// To be called when checkDir > 0 && pos != limit.
/// Returns with checkDir == 0 and pos != limit.
/// </summary>
private void NextSegment()
{
Debug.Assert(checkDir > 0 && seq == rawSeq && pos != limit);
// The input text [segmentStart..pos[ passes the FCD check.
int p = pos;
int prevCC = 0;
for (; ;)
{
// Fetch the next character's fcd16 value.
int q = p;
int c = Character.CodePointAt(seq, p);
p += Character.CharCount(c);
int fcd16 = nfcImpl.GetFCD16(c);
int leadCC = fcd16 >> 8;
if (leadCC == 0 && q != pos)
{
// FCD boundary before the [q, p[ character.
limit = segmentLimit = q;
break;
}
if (leadCC != 0 && (prevCC > leadCC || CollationFCD.IsFCD16OfTibetanCompositeVowel(fcd16)))
{
// Fails FCD check. Find the next FCD boundary and normalize.
do
{
q = p;
if (p == rawLimit)
{
break;
}
c = Character.CodePointAt(seq, p);
p += Character.CharCount(c);
} while (nfcImpl.GetFCD16(c) > 0xff);
Normalize(pos, q);
pos = start;
break;
}
prevCC = fcd16 & 0xff;
if (p == rawLimit || prevCC == 0)
{
// FCD boundary after the last character.
limit = segmentLimit = p;
break;
}
}
Debug.Assert(pos != limit);
checkDir = 0;
}
public override int PreviousCodePoint()
{
char c;
for (; ;)
{
if (checkDir < 0)
{
if (pos == start)
{
return(Collation.SentinelCodePoint);
}
c = seq[--pos];
if (CollationFCD.HasLccc(c))
{
if (CollationFCD.MaybeTibetanCompositeVowel(c) ||
(pos != start && CollationFCD.HasTccc(seq[pos - 1])))
{
++pos;
PreviousSegment();
c = seq[--pos];
}
}
break;
}
else if (checkDir == 0 && pos != start)
{
c = seq[--pos];
break;
}
else
{
SwitchToBackward();
}
}
char lead;
if (char.IsLowSurrogate(c) && pos != start &&
char.IsHighSurrogate(lead = seq[pos - 1]))
{
--pos;
return(Character.ToCodePoint(lead, c));
}
else
{
return(c);
}
}
public override int NextCodePoint()
{
char c;
for (; ;)
{
if (checkDir > 0)
{
if (pos == limit)
{
return(Collation.SentinelCodePoint);
}
c = seq[pos++];
if (CollationFCD.HasTccc(c))
{
if (CollationFCD.MaybeTibetanCompositeVowel(c) ||
(pos != limit && CollationFCD.HasLccc(seq[pos])))
{
--pos;
NextSegment();
c = seq[pos++];
}
}
break;
}
else if (checkDir == 0 && pos != limit)
{
c = seq[pos++];
break;
}
else
{
SwitchToForward();
}
}
char trail;
if (char.IsHighSurrogate(c) && pos != limit &&
char.IsLowSurrogate(trail = seq[pos]))
{
++pos;
return(Character.ToCodePoint(c, trail));
}
else
{
return(c);
}
}
protected override long HandleNextCE32()
{
char c;
for (; ;)
{
if (checkDir > 0)
{
if (pos == limit)
{
return(NoCodePointAndCE32);
}
c = seq[pos++];
if (CollationFCD.HasTccc(c))
{
if (CollationFCD.MaybeTibetanCompositeVowel(c) ||
(pos != limit && CollationFCD.HasLccc(seq[pos])))
{
--pos;
NextSegment();
c = seq[pos++];
}
}
break;
}
else if (checkDir == 0 && pos != limit)
{
c = seq[pos++];
break;
}
else
{
SwitchToForward();
}
}
return(MakeCodePointAndCE32Pair(c, trie.GetFromU16SingleLead(c)));
}
/// <summary>
/// Extends the FCD text segment backward or normalizes around pos.
/// </summary>
/// <returns><c>true</c> if success.</returns>
private bool PreviousSegment()
{
Debug.Assert(state == State.IterCheckBwd);
// The input text [(iter index)..limit[ passes the FCD check.
pos = iter.Index;
// Collect the characters being checked, in case they need to be normalized.
if (s == null)
{
s = new StringBuilder();
}
else
{
s.Length = 0;
}
int nextCC = 0;
for (; ;)
{
// Fetch the previous character and its fcd16 value.
int c = iter.PreviousCodePoint();
if (c < 0)
{
break;
}
int fcd16 = nfcImpl.GetFCD16(c);
int trailCC = fcd16 & 0xff;
if (trailCC == 0 && s.Length != 0)
{
// FCD boundary after this character.
iter.NextCodePoint();
break;
}
s.AppendCodePoint(c);
if (trailCC != 0 && ((nextCC != 0 && trailCC > nextCC) ||
CollationFCD.IsFCD16OfTibetanCompositeVowel(fcd16)))
{
// Fails FCD check. Find the previous FCD boundary and normalize.
while (fcd16 > 0xff)
{
c = iter.PreviousCodePoint();
if (c < 0)
{
break;
}
fcd16 = nfcImpl.GetFCD16(c);
if (fcd16 == 0)
{
iter.NextCodePoint();
break;
}
s.AppendCodePoint(c);
}
s.Reverse();
Normalize(s);
limit = pos;
start = pos - s.Length;
state = State.InNormIterAtStart;
pos = normalized.Length;
return(true);
}
nextCC = fcd16 >> 8;
if (nextCC == 0)
{
// FCD boundary before the following character.
break;
}
}
start = pos - s.Length;
Debug.Assert(pos != start);
iter.MoveIndex(s.Length);
state = State.IterInFCDSegment;
return(true);
}
/// <summary>
/// Extends the FCD text segment forward or normalizes around pos.
/// </summary>
/// <returns><c>true</c> if success.</returns>
private bool NextSegment()
{
Debug.Assert(state == State.IterCheckFwd);
// The input text [start..(iter index)[ passes the FCD check.
pos = iter.Index;
// Collect the characters being checked, in case they need to be normalized.
if (s == null)
{
s = new StringBuilder();
}
else
{
s.Length = 0;
}
int prevCC = 0;
for (; ;)
{
// Fetch the next character and its fcd16 value.
int c = iter.NextCodePoint();
if (c < 0)
{
break;
}
int fcd16 = nfcImpl.GetFCD16(c);
int leadCC = fcd16 >> 8;
if (leadCC == 0 && s.Length != 0)
{
// FCD boundary before this character.
iter.PreviousCodePoint();
break;
}
s.AppendCodePoint(c);
if (leadCC != 0 && (prevCC > leadCC || CollationFCD.IsFCD16OfTibetanCompositeVowel(fcd16)))
{
// Fails FCD check. Find the next FCD boundary and normalize.
for (; ;)
{
c = iter.NextCodePoint();
if (c < 0)
{
break;
}
if (nfcImpl.GetFCD16(c) <= 0xff)
{
iter.PreviousCodePoint();
break;
}
s.AppendCodePoint(c);
}
Normalize(s);
start = pos;
limit = pos + s.Length;
state = State.InNormIterAtLimit;
pos = 0;
return(true);
}
prevCC = fcd16 & 0xff;
if (prevCC == 0)
{
// FCD boundary after the last character.
break;
}
}
limit = pos + s.Length;
Debug.Assert(pos != limit);
iter.MoveIndex(-s.Length);
state = State.IterInFCDSegment;
return(true);
}
public override int PreviousCodePoint()
{
int c;
for (; ;)
{
if (state == State.IterCheckBwd)
{
c = iter.Previous();
if (c < 0)
{
start = pos = 0;
state = State.IterInFCDSegment;
return(Collation.SENTINEL_CP);
}
if (CollationFCD.HasLccc(c))
{
int prev = Collation.SENTINEL_CP;
if (CollationFCD.MaybeTibetanCompositeVowel(c) ||
CollationFCD.HasTccc(prev = iter.Previous()))
{
iter.Next();
if (prev >= 0)
{
iter.Next();
}
if (!PreviousSegment())
{
return(Collation.SENTINEL_CP);
}
continue;
}
// hasLccc(trail)=true for all trail surrogates
if (IsTrailSurrogate(c))
{
if (prev < 0)
{
prev = iter.Previous();
}
if (IsLeadSurrogate(prev))
{
return(Character.ToCodePoint((char)prev, (char)c));
}
}
if (prev >= 0)
{
iter.Next();
}
}
return(c);
}
else if (state == State.IterInFCDSegment && pos != start)
{
c = iter.PreviousCodePoint();
pos -= Character.CharCount(c);
Debug.Assert(c >= 0);
return(c);
}
else if (state.CompareTo(State.InNormIterAtLimit) >= 0 && pos != 0)
{
c = normalized.CodePointBefore(pos);
pos -= Character.CharCount(c);
return(c);
}
else
{
SwitchToBackward();
}
}
}