/// <summary>
/// Reads the next character from the text reader and advances the character position by one character.
/// </summary>
/// <returns>The next character from the text reader, or -1 if no more characters are available.</returns>
public override int Read()
{
int ic = buffer.Get(bufferPosition);
// End of input
if (ic == -1)
{
buffer.FreeBefore(bufferPosition);
return(ic);
}
char c = (char)ic;
// Skip surrogate pair characters
if (char.IsHighSurrogate(c) || char.IsLowSurrogate(c))
{
iterationMarkSpanEndPosition = bufferPosition + 1;
}
// Free rolling buffer on full stop
if (c == FULL_STOP_PUNCTUATION)
{
buffer.FreeBefore(bufferPosition);
iterationMarkSpanEndPosition = bufferPosition + 1;
}
// Normalize iteration mark
if (IsIterationMark(c))
{
c = NormalizeIterationMark(c);
}
bufferPosition++;
return(c);
}
public override int Read()
{
//System.out.println("\nread");
while (true)
{
if (replacement != null && replacementPointer < replacement.Length)
{
//System.out.println(" return repl[" + replacementPointer + "]=" + replacement.chars[replacement.offset + replacementPointer]);
return(replacement.Chars[replacement.Offset + replacementPointer++]);
}
// TODO: a more efficient approach would be Aho/Corasick's
// algorithm
// (http://en.wikipedia.org/wiki/Aho%E2%80%93Corasick_string_matching_algorithm)
// or this generalizatio: www.cis.uni-muenchen.de/people/Schulz/Pub/dictle5.ps
//
// I think this would be (almost?) equivalent to 1) adding
// epsilon arcs from all final nodes back to the init
// node in the FST, 2) adding a .* (skip any char)
// loop on the initial node, and 3) determinizing
// that. Then we would not have to Restart matching
// at each position.
int lastMatchLen = -1;
CharsRef lastMatch = null;
int firstCH = buffer.Get(inputOff);
if (firstCH != -1)
{
// LUCENENET fix: Check the dictionary to ensure it contains a key before reading it.
char key = Convert.ToChar((char)firstCH);
if (cachedRootArcs.TryGetValue(key, out FST.Arc <CharsRef> arc) && arc != null)
{
if (!FST.TargetHasArcs(arc))
{
// Fast pass for single character match:
if (Debugging.AssertsEnabled)
{
Debugging.Assert(arc.IsFinal);
}
lastMatchLen = 1;
lastMatch = arc.Output;
}
else
{
int lookahead = 0;
CharsRef output = arc.Output;
while (true)
{
lookahead++;
if (arc.IsFinal)
{
// Match! (to node is final)
lastMatchLen = lookahead;
lastMatch = outputs.Add(output, arc.NextFinalOutput);
// Greedy: keep searching to see if there's a
// longer match...
}
if (!FST.TargetHasArcs(arc))
{
break;
}
int ch = buffer.Get(inputOff + lookahead);
if (ch == -1)
{
break;
}
if ((arc = map.FindTargetArc(ch, arc, scratchArc, fstReader)) == null)
{
// Dead end
break;
}
output = outputs.Add(output, arc.Output);
}
}
}
}
if (lastMatch != null)
{
inputOff += lastMatchLen;
//System.out.println(" match! len=" + lastMatchLen + " repl=" + lastMatch);
int diff = lastMatchLen - lastMatch.Length;
if (diff != 0)
{
int prevCumulativeDiff = LastCumulativeDiff;
if (diff > 0)
{
// Replacement is shorter than matched input:
AddOffCorrectMap(inputOff - diff - prevCumulativeDiff, prevCumulativeDiff + diff);
}
else
{
// Replacement is longer than matched input: remap
// the "extra" chars all back to the same input
// offset:
int outputStart = inputOff - prevCumulativeDiff;
for (int extraIDX = 0; extraIDX < -diff; extraIDX++)
{
AddOffCorrectMap(outputStart + extraIDX, prevCumulativeDiff - extraIDX - 1);
}
}
}
replacement = lastMatch;
replacementPointer = 0;
}
else
{
int ret = buffer.Get(inputOff);
if (ret != -1)
{
inputOff++;
buffer.FreeBefore(inputOff);
}
return(ret);
}
}
}
public virtual void Test()
{
var ITERS = AtLeast(1000);
var buffer = new RollingCharBuffer();
var random = Random();
for (var iter = 0; iter < ITERS; iter++)
{
var stringLen = random.NextBoolean() ? random.Next(50) : random.Next(20000);
string s;
if (stringLen == 0)
{
s = "";
}
else
{
s = TestUtil.RandomUnicodeString(random, stringLen);
}
if (VERBOSE)
{
Console.WriteLine("\nTEST: iter=" + iter + " s.length()=" + s.Length);
}
buffer.Reset(new StringReader(s));
var nextRead = 0;
var availCount = 0;
while (nextRead < s.Length)
{
if (VERBOSE)
{
Console.WriteLine(" cycle nextRead=" + nextRead + " avail=" + availCount);
}
if (availCount == 0 || random.NextBoolean())
{
// Read next char
if (VERBOSE)
{
Console.WriteLine(" new char");
}
assertEquals(s[nextRead], buffer.Get(nextRead));
nextRead++;
availCount++;
}
else if (random.NextBoolean())
{
// Read previous char
var pos = TestUtil.NextInt(random, nextRead - availCount, nextRead - 1);
if (VERBOSE)
{
Console.WriteLine(" old char pos=" + pos);
}
assertEquals(s[pos], buffer.Get(pos));
}
else
{
// Read slice
int length;
if (availCount == 1)
{
length = 1;
}
else
{
length = TestUtil.NextInt(random, 1, availCount);
}
int start;
if (length == availCount)
{
start = nextRead - availCount;
}
else
{
start = nextRead - availCount + random.Next(availCount - length);
}
if (VERBOSE)
{
Console.WriteLine(" slice start=" + start + " length=" + length);
}
assertEquals(s.Substring(start, length), new string(buffer.Get(start, length)));
}
if (availCount > 0 && random.Next(20) == 17)
{
var toFree = random.Next(availCount);
if (VERBOSE)
{
Console.WriteLine(" free " + toFree + " (avail=" + (availCount - toFree) + ")");
}
buffer.FreeBefore(nextRead - (availCount - toFree));
availCount -= toFree;
}
}
}
}