public static uint?NthCodepoint(AbstractString str, int codepointIndex, int relativeToCharacterIndex = 0)
{
foreach (var cp in str.Codepoints(relativeToCharacterIndex))
{
if (codepointIndex == 0)
{
return(cp.Codepoint);
}
codepointIndex--;
}
return(null);
}
public static Pair <int> FindWordBoundary(AbstractString str, int?searchFromCodepointIndex = null, int?searchFromCharacterIndex = null)
{
int firstWhitespaceCharacter = -1,
lastWhitespaceCharacter = -1,
firstWordCharacter = -1,
lastWordCharacter = -1;
if ((searchFromCharacterIndex == null) && (searchFromCodepointIndex == null))
{
throw new ArgumentException("Either a starting codepoint index or character index must be provided");
}
bool searchStartedInWhiteSpace = false, inWord = false;
foreach (var cp in str.Codepoints())
{
bool transitioned = false;
var isWhiteSpace = IsWhiteSpace(cp.Codepoint);
if (
(cp.CodepointIndex == searchFromCodepointIndex) ||
(cp.CharacterIndex == searchFromCharacterIndex)
)
{
searchStartedInWhiteSpace = isWhiteSpace;
}
if (isWhiteSpace)
{
if (inWord || firstWhitespaceCharacter < 0)
{
transitioned = inWord;
inWord = false;
firstWhitespaceCharacter = cp.CharacterIndex;
}
lastWhitespaceCharacter = cp.CharacterIndex;
}
else
{
if (!inWord || firstWordCharacter < 0)
{
transitioned = !inWord;
inWord = true;
firstWordCharacter = cp.CharacterIndex;
}
lastWordCharacter = cp.CharacterIndex;
}
if (transitioned &&
(
(searchFromCodepointIndex.HasValue && (cp.CodepointIndex > searchFromCodepointIndex)) ||
(searchFromCharacterIndex.HasValue && (cp.CharacterIndex > searchFromCharacterIndex))
)
)
{
break;
}
}
if (searchStartedInWhiteSpace)
{
return(new Pair <int>(firstWhitespaceCharacter, lastWhitespaceCharacter + 1));
}
else
{
if ((lastWordCharacter > 0) && char.IsHighSurrogate(str[lastWordCharacter]))
{
lastWordCharacter++;
}
return(new Pair <int>(firstWordCharacter, lastWordCharacter + 1));
}
}