internal TtsDocumentCyrillica(string docText, MultiLangStringRegioned multiLangText, string foreignLangCode,
bool highlightWordsOnly = true, bool invert = false)
{
Text = docText;
MultiLangText = multiLangText;
this.Lang = foreignLangCode;
if (highlightWordsOnly)
{
if (invert)
{
HighlightRegions.AddRange(Utils.FindAllCyrillicWords(docText));
}
else
{
HighlightRegions.AddRange(Utils.FindAllLatinWords(docText));
}
}
else
{
// NOTE: the following can be used for debugging purposes e.g. to check TTSer
foreach (var region in multiLangText.Regions)
{
var isRu = region.Lang == TtsGlobals.LangCodeRussian;
if (isRu == invert)
{
HighlightRegions.Add(new WordHighlight(region.Start, region.Length));
}
}
}
}
public static MultiLangStringRegioned ParseCyrillicaText(string docText, string latinaLangCode)
{
var result = new MultiLangStringRegioned(docText);
var splitIndices = Utils.FindAllNonCyrillics(docText);
var start = 0;
for (int i = 0; i < splitIndices.Length; i++)
{
var end = splitIndices[i];
result.AddRegion(start, end - start, (i % 2 == 0) ? TtsGlobals.LangCodeRussian : latinaLangCode);
start = end;
}
if (start < docText.Length - 1)
{
result.AddRegion(start, docText.Length - 1 - start, (result.Count % 2 == 0) ? TtsGlobals.LangCodeRussian : latinaLangCode);
}
return(result);
}
