public virtual FieldFragList CreateFieldFragList(FieldPhraseList fieldPhraseList,
int fragCharSize)
{
FieldFragList ffl = new SimpleFieldFragList(fragCharSize);
IList <WeightedPhraseInfo> wpil = new JCG.List <WeightedPhraseInfo>();
using IEnumerator <WeightedPhraseInfo> ite = fieldPhraseList.PhraseList.GetEnumerator();
WeightedPhraseInfo phraseInfo = null;
while (true)
{
if (!ite.MoveNext())
{
break;
}
phraseInfo = ite.Current;
if (phraseInfo is null)
{
break;
}
wpil.Add(phraseInfo);
}
if (wpil.Count > 0)
{
ffl.Add(0, int.MaxValue, wpil);
}
return(ffl);
}
/// <summary>
/// A predicate to decide if the given <see cref="WeightedPhraseInfo"/> should be
/// accepted as a highlighted phrase or if it should be discarded.
/// <para/>
/// The default implementation discards phrases that are composed of more than one term
/// and where the matchLength exceeds the fragment character size.
/// </summary>
/// <param name="info">the phrase info to accept</param>
/// <param name="matchLength">the match length of the current phrase</param>
/// <param name="fragCharSize">the configured fragment character size</param>
/// <returns><c>true</c> if this phrase info should be accepted as a highligh phrase</returns>
protected virtual bool AcceptPhrase(WeightedPhraseInfo info, int matchLength, int fragCharSize)
{
// LUCENENET specific - added guard clause to check for null
if (info is null)
{
throw new ArgumentNullException(nameof(WeightedPhraseInfo));
}
return(info.TermsOffsets.Count <= 1 || matchLength <= fragCharSize);
}
protected virtual FieldFragList CreateFieldFragList(FieldPhraseList fieldPhraseList, FieldFragList fieldFragList, int fragCharSize)
{
if (fragCharSize < minFragCharSize)
{
throw new ArgumentException("fragCharSize(" + fragCharSize + ") is too small. It must be " + minFragCharSize + " or higher.");
}
List <WeightedPhraseInfo> wpil = new List <WeightedPhraseInfo>();
using (IteratorQueue <WeightedPhraseInfo> queue = new IteratorQueue <WeightedPhraseInfo>(fieldPhraseList.PhraseList.GetEnumerator()))
{
WeightedPhraseInfo phraseInfo = null;
int startOffset = 0;
while ((phraseInfo = queue.Top()) != null)
{
// if the phrase violates the border of previous fragment, discard it and try next phrase
if (phraseInfo.StartOffset < startOffset)
{
queue.RemoveTop();
continue;
}
wpil.Clear();
int currentPhraseStartOffset = phraseInfo.StartOffset;
int currentPhraseEndOffset = phraseInfo.EndOffset;
int spanStart = Math.Max(currentPhraseStartOffset - margin, startOffset);
int spanEnd = Math.Max(currentPhraseEndOffset, spanStart + fragCharSize);
if (AcceptPhrase(queue.RemoveTop(), currentPhraseEndOffset - currentPhraseStartOffset, fragCharSize))
{
wpil.Add(phraseInfo);
}
while ((phraseInfo = queue.Top()) != null)
{ // pull until we crossed the current spanEnd
if (phraseInfo.EndOffset <= spanEnd)
{
currentPhraseEndOffset = phraseInfo.EndOffset;
if (AcceptPhrase(queue.RemoveTop(), currentPhraseEndOffset - currentPhraseStartOffset, fragCharSize))
{
wpil.Add(phraseInfo);
}
}
else
{
break;
}
}
if (wpil.Count == 0)
{
continue;
}
int matchLen = currentPhraseEndOffset - currentPhraseStartOffset;
// now recalculate the start and end position to "center" the result
int newMargin = Math.Max(0, (fragCharSize - matchLen) / 2); // matchLen can be > fragCharSize prevent IAOOB here
spanStart = currentPhraseStartOffset - newMargin;
if (spanStart < startOffset)
{
spanStart = startOffset;
}
// whatever is bigger here we grow this out
spanEnd = spanStart + Math.Max(matchLen, fragCharSize);
startOffset = spanEnd;
fieldFragList.Add(spanStart, spanEnd, wpil);
}
}
return(fieldFragList);
}
/// <summary>
/// A predicate to decide if the given <see cref="WeightedPhraseInfo"/> should be
/// accepted as a highlighted phrase or if it should be discarded.
/// <para/>
/// The default implementation discards phrases that are composed of more than one term
/// and where the matchLength exceeds the fragment character size.
/// </summary>
/// <param name="info">the phrase info to accept</param>
/// <param name="matchLength">the match length of the current phrase</param>
/// <param name="fragCharSize">the configured fragment character size</param>
/// <returns><c>true</c> if this phrase info should be accepted as a highligh phrase</returns>
protected virtual bool AcceptPhrase(WeightedPhraseInfo info, int matchLength, int fragCharSize)
{
return(info.TermsOffsets.Count <= 1 || matchLength <= fragCharSize);
}
