protected AlgoBase(AsyncSaver saver,
IWikidumpReader wikiReader,
IMongoCollection <Triplet> triplets,
string positionsPath)
{
_saver = saver;
_wikiReader = wikiReader;
_triplets = triplets;
_positions = PrepareCsvReader(positionsPath);
}
protected bool ProcessTriplet(ObjectId id,
PositionLine object_,
PositionLine subject,
IWikidumpReader reader)
{
var position = new AnotherArticlePosition
{
ArticleTitle = object_.WikiTitle,
ArticleId = object_.PageId,
ObjectPosition = object_.ToPosition(),
SubjectPosition = subject.ToPosition()
};
var text = reader.ExtractArticleText(object_.PageId);
if (text == null)
{
return(false);
}
var startPosition = object_.Start < subject.Start ? object_.Start : subject.Start;
var endPosition = object_.End > subject.End ? object_.End : subject.End;
int newStart;
int newEnd;
position.Text = TextHelper.ExtractTextWithSentenceWindow(text, startPosition, endPosition, out newStart,
out newEnd);
// do not save with line break
if (position.Text.Contains('\n') || position.Text.Contains('\r'))
{
return(false);
}
position.Start = newStart;
position.End = newEnd;
position.Distance = newEnd - newStart;
_saver.Save(id, position);
return(true);
}
public AlgoInMemory(AsyncSaver saver, IWikidumpReader wikiReader, IMongoCollection <Triplet> triplets, string positionsPath)
: base(saver, wikiReader, triplets, positionsPath)
{
_tripletsInMemory = GetTripletsCache();
}
public AlgoInDb(AsyncSaver saver, IWikidumpReader wikiReader, IMongoCollection <Triplet> triplets, string positionsPath)
: base(saver, wikiReader, triplets, positionsPath)
{
}
