private static LuceneQuery HandleFuzzyYngpingQuery(string queryText)
{
var queryParser = new MultiFieldQueryParser(LuceneUtils.AppLuceneVersion,
new string[] { LuceneUtils.Fields.Yngping },
LuceneUtils.GetAnalyzer(),
new Dictionary <string, float> {
{ LuceneUtils.Fields.Yngping, 100 },
});
return(queryParser.Parse(queryText));
}
private void CreateLuceneIndex(YngdiengIndex index)
{
var dirInfo = Path.GetFullPath(Path.Join(outputFolder, "lucene"));
Console.WriteLine($"Writing to {dirInfo}");
using (var dir = FSDirectory.Open(new DirectoryInfo(dirInfo)))
{
var indexConfig = new IndexWriterConfig(LuceneUtils.AppLuceneVersion,
LuceneUtils.GetAnalyzer());
using (var writer = new IndexWriter(dir, indexConfig))
{
writer.DeleteAll();
foreach (var yDoc in index.YngdiengDocuments)
{
var doc = new Lucene.Net.Documents.Document {
new Int32Field(LuceneUtils.Fields.IsSourceless, yDoc.Sources.Count == 0?1:0, Field.Store.YES),
new StringField(LuceneUtils.Fields.DocId, yDoc.DocId, Field.Store.YES),
new TextField(LuceneUtils.Fields.Yngping, yDoc.YngpingSandhi, Field.Store.NO),
new TextField(LuceneUtils.Fields.Hanzi, yDoc.HanziCanonical.Regular, Field.Store.NO),
new StringField(LuceneUtils.Fields.YngpingSandhiTonePattern, GetTonePattern(yDoc.YngpingSandhi), Field.Store.NO)
};
foreach (var m in yDoc.IndexingExtension.MandarinWords)
{
doc.Add(new TextField(LuceneUtils.Fields.Mandarin, m, Field.Store.NO));
doc.Add(new TextField(LuceneUtils.Fields.Mandarin, openCcClient.SimplifyMandarinText(m), Field.Store.NO));
}
foreach (var a in yDoc.HanziAlternatives)
{
doc.Add(new TextField(LuceneUtils.Fields.HanziAlternative, a.Regular, Field.Store.NO));
}
// Simplify Hanzi for search
doc.Add(new TextField(LuceneUtils.Fields.Hanzi, openCcClient.SimplifyHukziuText(yDoc.HanziCanonical.Regular), Field.Store.NO));
foreach (var e in yDoc.IndexingExtension.ExplanationText)
{
doc.Add(new TextField(LuceneUtils.Fields.Explanation, e, Field.Store.NO));
}
// TODO: encapsulate this in a YngpingAnalyzer
foreach (var yp in yDoc.IndexingExtension.YngpingPermutations)
{
doc.Add(new TextField(LuceneUtils.Fields.Yngping, yp, Field.Store.NO));
}
writer.AddDocument(doc);
}
writer.Flush(triggerMerge: false, applyAllDeletes: false);
}
}
}
private static LuceneQuery HandleHanziQuery(string queryText)
{
var queryParser = new MultiFieldQueryParser(
LuceneUtils.AppLuceneVersion,
new string[] {
LuceneUtils.Fields.Mandarin,
LuceneUtils.Fields.Hanzi,
LuceneUtils.Fields.HanziAlternative,
LuceneUtils.Fields.Explanation
},
LuceneUtils.GetAnalyzer(),
new Dictionary <string, float> {
{ LuceneUtils.Fields.Mandarin, 200 },
{ LuceneUtils.Fields.Hanzi, 100 },
{ LuceneUtils.Fields.HanziAlternative, 50 },
{ LuceneUtils.Fields.Explanation, 1 }
});
return(queryParser.Parse(queryText));
}