public void RecomputeSentenceAlignment()
{
var manualHighlights1 = Utils.GetAlignManual(AlignData, Text1, 0, Text1.Length, take2: false);
var sectionAll1 = new DocSection(0, Text1.Length);
var sectionsByManual1 = Utils.SplitSection(sectionAll1, manualHighlights1);
var manualHighlights2 = Utils.GetAlignManual(AlignData, Text2, 0, Text2.Length, take2: true);
var sectionAll2 = new DocSection(0, Text2.Length);
var sectionsByManual2 = Utils.SplitSection(sectionAll2, manualHighlights2);
sectionsByManual1.EnumerateWith(sectionsByManual2, (sectionM1, sectionM2, indexM) =>
{
var regexHighlights1 = Utils.GetAlignRegexes(AlignData, Text1, sectionM1.StartIndex, sectionM1.Length, take2: false)
.OrderBy(x => x.StartIndex)
.ToArray();
var regexHighlights2 = Utils.GetAlignRegexes(AlignData, Text2, sectionM2.StartIndex, sectionM2.Length, take2: true)
.OrderBy(x => x.StartIndex)
.ToArray();
var regexPositions1 = regexHighlights1.Select(x => x.StartIndex).ToArray();
var regexPositions2 = regexHighlights2.Select(x => x.StartIndex).ToArray();
var regexAlignmentIndices = algoArrayAlign.GetBestAlignment(regexPositions1, regexPositions2, Globals.CharThreshRegexAlign);
// TODO: use this data to align them store all sections `d21f`
});
//var regexHighlights = Utils.GetAlignRegexes(AlignData, text, 0, text.Length, take2);
}
public static IEnumerable <DocSection> SplitSection(DocSection sectionAll, IEnumerable <WordHighlight> manualHighlights)
{
var withinHighlights = manualHighlights
.Where(x => x.StartIndex >= sectionAll.StartIndex && x.StartIndex <= sectionAll.StartIndex + sectionAll.Length)
.OrderBy(x => x.StartIndex);
if (!manualHighlights.Any())
{
sectionAll.ToIEnumerable();
}
var results = new List <DocSection>();
var indexPrev = sectionAll.StartIndex;
foreach (var highlight in withinHighlights)
{
var indexCur = highlight.StartIndex;
if (highlight.StartIndex <= indexPrev)
{
continue;
}
results.Add(new DocSection(indexPrev, indexCur - indexPrev));
indexPrev = indexCur;
}
if (indexPrev < sectionAll.StartIndex + sectionAll.Length) // last one!
{
results.Add(new DocSection(indexPrev, sectionAll.StartIndex + sectionAll.Length - indexPrev));
}
return(results);
}
/// <summary>
/// Returns posB, B being this object
/// </summary>
public int GetRelativePosition(DocSection sectionA, int posA)
{
return((int)(StartIndex + (long)Length * posA / sectionA.Length));
}