static SentenceMergingTranslationProcessor()
{
var wordSplitEngine = new StringSplitEngine(WordDelimiterAfterChars);
StringSplitsChunkAssigners.Add(new GreedyStringSplitsChunkAssigner(wordSplitEngine));
StringSplitsChunkAssigners.Add(new RandomizedStringSplitsChunkAssigner(wordSplitEngine));
}
/**
* divides the full sentence into multiple chunks and assigns them to the source paragraphs.
* It tries to split by percentage equivalent to fit the length of the source paragraph
*/
public void SetTranslation(string targetText)
{
List <int> sourceChunksTextLength = SentenceParagraphs.ConvertAll(x => x.Text.Length);
var breakPositions = EvaluateBreakPositions(targetText, sourceChunksTextLength);
string[] targetChunks = StringSplitEngine.SplitAt(targetText, breakPositions);
for (int i = 0; i < targetChunks.Length; i++)
{
SentenceParagraphs[i].Translation = targetChunks[i];
}
}
public override int[] GetSplitPositions(List <int> sourceChunksTextLength, string targetText)
{
_sourceChunksTextLength = sourceChunksTextLength;
_targetSplits = StringSplitEngine.Split(targetText);
_targetSplitsLength = _targetSplits.ConvertAll(x => x.Length);
var rateResults = new List <RateResult>();
for (int i = 0; i < _iterations; i++)
{
int[] breakPositions = GetRandomBreakPositions(targetText, _sourceChunksTextLength);
var rateResult = CalculateRateResult(_sourceChunksTextLength, targetText.Length, breakPositions);
rateResults.Add(rateResult);
}
rateResults = rateResults.OrderBy(x => x.Fitness).ToList();
return(rateResults.First().BreakPositions);
}
public override int[] GetSplitPositions(List <int> sourceChunksTextLength, string targetText)
{
int overallSourceLength = sourceChunksTextLength.Sum();
int currentSentenceParagraph = 0;
int currentSourceChunkEndPosition = sourceChunksTextLength[0];
int[] splitPositions = new int[sourceChunksTextLength.Count - 1];
int splitPositionCount = 0;
for (int i = 0; i < targetText.Length; i++)
{
var c = targetText[i];
if (StringSplitEngine.IsSplittable(targetText, i))
{
double currentTargetPositionPercentage = (double)i / targetText.Length;
double currentSourceChunkEndPositionPercentage = (double)currentSourceChunkEndPosition / overallSourceLength;
if (currentTargetPositionPercentage > currentSourceChunkEndPositionPercentage)
{
currentSentenceParagraph++;
currentSourceChunkEndPosition += sourceChunksTextLength[currentSentenceParagraph];
splitPositions[splitPositionCount] = i;
splitPositionCount++;
}
}
}
//ensure that there is always to correct amount of resulting splitPositions
while (splitPositionCount < sourceChunksTextLength.Count - 1)
{
splitPositions[splitPositionCount] = splitPositionCount > 0 ? splitPositions[splitPositionCount - 1] : 0;
splitPositionCount++;
}
return(splitPositions);
}
public GreedyStringSplitsChunkAssigner(StringSplitEngine stringSplitEngine) : base(stringSplitEngine)
{
}
public AbstractStringSplitsChunkAssigner(StringSplitEngine stringSplitEngine)
{
StringSplitEngine = stringSplitEngine;
}
public RandomizedStringSplitsChunkAssigner(StringSplitEngine stringSplitEngine, int iterations = DefaultIterations) : base(stringSplitEngine)
{
_iterations = iterations;
}