/// <summary>
/// Construtor padrão.
/// </summary>
/// <param name="analyzer"></param>
/// <param name="dataRepository"></param>
/// <param name="sortTerms">Termos que serão usados na ordenação.</param>
/// <param name="numHits">Total de itens do resultado.</param>
/// <param name="field">Nome do campos.</param>
internal FullTextFieldComparator(global::Lucene.Net.Analysis.Standard.StandardAnalyzer analyzer, IDataRepository dataRepository, IList <string> sortTerms, int numHits, string field)
{
_analyzer = analyzer;
_dataRepository = dataRepository;
_sortTerms = sortTerms;
_values = new List <string> [numHits];
_field = field;
}
/// <summary>
/// Get searchterms from examineindex
/// </summary>
/// <param name="rootId">Where to start from</param>
/// <param name="minLength">Min length of words to find</param>
/// <returns></returns>
public TermExtractorWrapper GetTerms(int rootId = -1, int minLength = 5)
{
var debugTimeStart = DateTime.Now;
var debugTimeEnd = DateTime.MinValue;
var termsList = new List<global::Lucene.Net.Index.Term>(); // data to return
var indexPath = GetIndexPath(); // path to index
var directory = FSDirectory.Open(indexPath); //
var indexReader = IndexReader.Open(directory, true); //
var searcher = new IndexSearcher(indexReader); //searcher for checking if the term is valid
var analyzer = new global::Lucene.Net.Analysis.Standard.StandardAnalyzer(global::Lucene.Net.Util.Version.LUCENE_29);
foreach (var fieldName in FieldNames)
{
var tEnum = indexReader.Terms(new global::Lucene.Net.Index.Term(fieldName));
do
{
var t = tEnum.Term();
// filter in fieldname and minlength og term
if (t.Field() == fieldName && t.Text().Length >= minLength)
{
var searchPhrase = t.Text();
var parser = new global::Lucene.Net.QueryParsers.QueryParser(global::Lucene.Net.Util.Version.LUCENE_29, fieldName, analyzer);
var bq = new BooleanQuery();
var collector = TopScoreDocCollector.create(1, true);
var query = parser.Parse(searchPhrase);
bq.Add(query, BooleanClause.Occur.MUST);
searcher.Search(bq, collector);
var hits = collector.TopDocs().ScoreDocs;
if (hits.Any())
{
var docId = hits[0].doc;
var doc = searcher.Doc(docId);
if (doc.Get("path").Contains("," + rootId + ","))
{
termsList.Add(t);
}
}
}
} while (tEnum.Next());
}
debugTimeEnd = DateTime.Now;
return new TermExtractorWrapper
{
QueryTime = (debugTimeEnd - debugTimeStart).Milliseconds.ToString(),
Terms = termsList
};
}
/// <summary>
/// Construtor padrão.
/// </summary>
/// <param name="analyzer"></param>
/// <param name="dataRepository">Repositorio de dados associado.</param>
/// <param name="sortTerms">Termos que serão usados para a ordenação.</param>
public FullTextFieldComparatorSource(global::Lucene.Net.Analysis.Standard.StandardAnalyzer analyzer, IDataRepository dataRepository, IList <string> sortTerms)
{
_analyzer = analyzer;
_dataRepository = dataRepository;
_sortTerms = sortTerms;
}