public static void RunApplicaiton(IndexerConfig config)
{
ProgressBar progress = null;
IIndexer indexer;
if (config.UseLucene)
{
indexer = new LuceneIndexer();
}
else
{
indexer = new ElasticIndexer();
}
var indexResult = indexer.IndexFiles(config, result =>
{
if (progress == null)
{
progress = new ProgressBar(PbStyle.DoubleLine, result.TotalFile, 200);
}
progress.Refresh(result.ReadedFiles, result.ToString());
});
Console.WriteLine(indexResult);
}
public IndexResult IndexFiles(IndexerConfig config, IndexerCallback callback = null)
{
var totalWatch = new Stopwatch();
var directory = FSDirectory.Open(config.IndexPath);
var analyzer = new WhitespaceAnalyzer();
var files = TextExtractor.GetFilesFromFolder(
config.FolderPath,
config.Extensions,
config.IsRecusive);
var result = new IndexResult
{
TotalFile = files.Count
};
totalWatch.Start();
foreach (var file in files)
{
result.CurrentFile = file.FullName;
callback?.Invoke(result);
using (var writer = new IndexWriter(directory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED))
{
var stopwatch = new Stopwatch();
var document = new Document();
var txtInfo = TextExtractor.ReadText(file.FullName);
var fields = GetFields(txtInfo, config.Excludes);
foreach (var field in fields)
{
document.Add(field);
}
stopwatch.Start();
writer.AddDocument(document, analyzer);
writer.Optimize();
stopwatch.Stop();
result.Time = stopwatch.Elapsed;
result.ReadedFiles = TextExtractor.ReadedFile;
result.Size = txtInfo.FileSize;
callback?.Invoke(result);
}
}
totalWatch.Stop();
result.Time = totalWatch.Elapsed;
result.ReadedFiles = TextExtractor.ReadedFile;
result.Size = TextExtractor.TotalSize;
return(result);
}
public IndexResult IndexFiles(IndexerConfig config, IndexerCallback callback = null)
{
var totalWatch = new Stopwatch();
var settings = new ConnectionSettings(new Uri(config.IndexPath)).DefaultIndex("text");
var files = TextExtractor.GetFilesFromFolder(
config.FolderPath,
config.Extensions,
config.IsRecusive);
var client = new ElasticClient(settings);
var result = new IndexResult
{
TotalFile = files.Count
};
ConfigMapping(client);
totalWatch.Start();
foreach (var file in files)
{
result.CurrentFile = file.FullName;
callback?.Invoke(result);
var stopwatch = new Stopwatch();
var txtInfo = TextExtractor.ReadText(file.FullName);
var request = new IndexRequest <TextInfo>
{
Document = txtInfo,
Refresh = Elasticsearch.Net.Refresh.True
};
stopwatch.Start();
client.Index(request);
stopwatch.Stop();
result.Time = stopwatch.Elapsed;
result.ReadedFiles = TextExtractor.ReadedFile;
result.Size = txtInfo.FileSize;
callback?.Invoke(result);
}
totalWatch.Stop();
result.Time = totalWatch.Elapsed;
result.ReadedFiles = TextExtractor.ReadedFile;
result.Size = TextExtractor.TotalSize;
return(result);
}