/// <summary>
/// Получение текста из xml
/// </summary>
/// <param name="xml">xml</param>
/// <returns>текст</returns>
private string GetText(string xml, IEnumerable <SimpleEntity> entities)
{
_shiftTable = new StairTableShift();
_shift = 0;
var replace = new List <Tuple <TextPart, string> >();
foreach (var match in _tagRegex.Matches(xml).Cast <Match>())
{
string result = " ";
if ((match.Value == "</p>") || (match.Value == "</claim>"))
{
result = "\n";
}
else if ((match.Value == "<sub>") || (match.Value == "</sub>") || (match.Value == "<sup>") || (match.Value == "</sup>"))
{
result = "";
}
replace.Add(new Tuple <TextPart, string>(new TextPart(match.Index, match.Length), result));
}
foreach (var entity in entities)
{
if ((entity.Type == RegexEntityType.Diacritical) ||
(entity.Type == RegexEntityType.NumberWithDiacritical) ||
(entity.Type == RegexEntityType.SimpleDiacritical))
{
int start = entity.Text.IndexOf("<sup>") + 5;
int end = entity.Text.IndexOf("</sup>");
string replaced = entity.Text.Substring(start, end - start).IndexOf(' ') >= 0 ? " " : "";
replace.Add(new Tuple <TextPart, string>(new TextPart(entity.StartPosition + start, end - start), replaced));
}
}
AddSimpleReplace(replace, xml);
replace = replace.OrderBy(_ => _.Item1.Start).ToList();
StringBuilder text = new StringBuilder();
int position = 0;
int shift = 0;
foreach (var replaced in replace)
{
shift += replaced.Item1.End - replaced.Item1.Start - replaced.Item2.Length;
_shiftTable.UpdateShift(replaced.Item1.Start, shift);
text.Append(xml.Substring(position, replaced.Item1.Start - position));
text.Append(replaced.Item2);
position = replaced.Item1.End;
}
text.Append(xml.Substring(position));
return(text.ToString());
}
private void Initialization()
{
_builder = new StringBuilder();
_tableShift = new StairTableShift();
_isSpace = true;
}
