public static IEnumerable <ISearchItem> Search(string pattern, DirectoryInfo dataFolder, int page)
{
using (Analyzer analyzer = new SimpleAnalyzer(LuceneVersion.LUCENE_48))
using (IndexReader reader = new PagesReader(dataFolder))
{
Query query = new QueryParser(LuceneVersion.LUCENE_48, string.Empty, analyzer).Parse(pattern);
IndexSearcher searcher = new IndexSearcher(reader);
TopFieldCollector collector =
TopFieldCollector.Create(Sort.INDEXORDER, NumHits, false, false, false, false);
searcher.Search(query, collector);
/*
* IFormatter formatter = new SimpleHTMLFormatter();
* IScorer scorer = new QueryScorer(query);
* Highlighter highlighter = new Highlighter(formatter, scorer)
* {
* TextFragmenter = new SimpleFragmenter(3)
* };
*/
ScoreDoc[] docs = collector.GetTopDocs(page * PageSize, PageSize).ScoreDocs;
return(docs.Select(doc => new SearchItem(doc.Doc.ToString(), doc.Doc)));
/*
* Document document = searcher.Doc(doc.Doc);
* string body = document.Get("body");
* TokenStream stream = TokenSources.GetAnyTokenStream(reader, doc.Doc, "body", analyzer);
* //TokenStream stream = analyzer.GetTokenStream("test123", new StringReader("test456"));
* string best = highlighter.GetBestFragments(stream, body, 1, " ");
*/
}
}
static void Main(string[] args)
{
ConsoleKey keyPressed;
var pageReader = new PagesReader();
pageReader.Read();
var pageNumber = pageReader.GetNextPageNumber();
do
{
Console.Clear();
var pageModel = pageReader.GetPage(pageNumber);
Console.Write("{0,50}", "-");
Console.Write(pageModel.PageNumber);
Console.Write("-");
Console.WriteLine();
Console.WriteLine();
Console.WriteLine();
Console.WriteLine(pageModel.Content);
Console.WriteLine();
Console.WriteLine();
Console.WriteLine();
Console.WriteLine();
Console.Write("P - Poprzedni");
Console.Write("{0,100}", "N - Nastepny");
Console.WriteLine();
keyPressed = Console.ReadKey().Key;
switch (keyPressed)
{
case ConsoleKey.N:
pageNumber = pageReader.GetNextPageNumber();
break;
case ConsoleKey.P:
pageNumber = pageReader.GetPrevPageNumber();
break;
}
} while (keyPressed != ConsoleKey.Enter);
Console.ReadKey();
}
public static ISet <int> HighlightedOffsets(int pagenumber, string pattern, DirectoryInfo dataFolder)
{
using (Analyzer analyzer = new SimpleAnalyzer(LuceneVersion.LUCENE_48))
using (IndexReader reader = new PagesReader(dataFolder))
{
Query query = new QueryParser(LuceneVersion.LUCENE_48, string.Empty, analyzer).Parse(pattern);
TermsEnum termsEnum = reader.GetTermVector(pagenumber, string.Empty).GetIterator(TermsEnum.EMPTY);
IEnumerable <int> highlightOffsets = ExtractTerms(query).SelectMany(term =>
{
termsEnum.SeekExact(term.Bytes);
DocsAndPositionsEnum docsAndPositionsEnum =
termsEnum.DocsAndPositions(null, null, DocsAndPositionsFlags.OFFSETS);
docsAndPositionsEnum.Advance(pagenumber);
return(Enumerable.Range(0, docsAndPositionsEnum.Freq).Select(i =>
{
docsAndPositionsEnum.NextPosition();
return docsAndPositionsEnum.StartOffset;
}));
});
return(new HashSet <int>(highlightOffsets));
}
}
