static List <TextClassificationTag> Merge(List <TextClassificationTag> list)
{
if (list.Count <= 1)
{
return(list);
}
var prev = list[0];
int read = 1, write = 0;
for (; read < list.Count; read++)
{
var a = list[read];
if (prev.ClassificationType == a.ClassificationType && prev.Span.End == a.Span.Start)
{
list[write] = prev = new TextClassificationTag(Span.FromBounds(prev.Span.Start, a.Span.End), prev.ClassificationType);
}
else
{
prev = a;
list[++write] = a;
}
}
write++;
if (list.Count != write)
{
list.RemoveRange(write, list.Count - write);
}
return(list);
}
public IEnumerable <TextClassificationTag> GetTags(TextClassifierContext context)
{
var list = new List <TextClassificationTag>();
int textLength = context.Text.Length;
foreach (var classifier in textClassifiers)
{
foreach (var tagTmp in classifier.GetTags(context))
{
var tag = tagTmp;
if (tag.Span.End > textLength)
{
tag = new TextClassificationTag(Span.FromBounds(Math.Min(textLength, tag.Span.Start), Math.Min(textLength, tag.Span.End)), tag.ClassificationType);
}
if (tag.Span.Length == 0)
{
continue;
}
list.Add(tag);
}
}
if (list.Count <= 1)
{
return(list);
}
list.Sort(TextClassificationTagComparer.Instance);
// Common case
if (!HasOverlaps(list))
{
return(Merge(list));
}
int min = 0;
int minOffset = 0;
var newList = new List <TextClassificationTag>();
var ctList = new List <IClassificationType>();
while (min < list.Count)
{
while (min < list.Count && minOffset >= list[min].Span.End)
{
min++;
}
if (min >= list.Count)
{
break;
}
var cspan = list[min];
minOffset = Math.Max(minOffset, cspan.Span.Start);
int end = cspan.Span.End;
ctList.Clear();
ctList.Add(cspan.ClassificationType);
for (int i = min + 1; i < list.Count; i++)
{
cspan = list[i];
int cspanStart = cspan.Span.Start;
if (cspanStart > minOffset)
{
if (cspanStart < end)
{
end = cspanStart;
}
break;
}
int cspanEnd = cspan.Span.End;
if (minOffset >= cspanEnd)
{
continue;
}
if (cspanEnd < end)
{
end = cspanEnd;
}
if (!ctList.Contains(cspan.ClassificationType))
{
ctList.Add(cspan.ClassificationType);
}
}
Debug.Assert(minOffset < end);
var newSpan = new Span(minOffset, end - minOffset);
var ct = ctList.Count == 1 ? ctList[0] : classificationTypeRegistryService.CreateTransientClassificationType(ctList);
newList.Add(new TextClassificationTag(newSpan, ct));
minOffset = end;
}
Debug.Assert(!HasOverlaps(newList));
return(Merge(newList));
}