private void LoadIndex(string dir, ulong collection)
{
var timer = new Stopwatch();
var batchTimer = new Stopwatch();
timer.Start();
var files = Directory.GetFiles(dir, "*.docs");
_log.Log(string.Format("index scan found {0} document files", files.Length));
foreach (var docFileName in files)
{
var name = Path.GetFileNameWithoutExtension(docFileName)
.Split(".", StringSplitOptions.RemoveEmptyEntries);
var collectionId = ulong.Parse(name[0]);
if (collectionId == collection)
{
using (var readSession = new DocumentReadSession(collectionId, _sessionFactory))
{
var docs = readSession.ReadDocs();
var job = new IndexJob(collectionId, docs);
using (var writeSession = _sessionFactory.CreateWriteSession(collectionId))
{
writeSession.WriteToIndex(job);
}
_log.Log(string.Format("loaded batch into {0} in {1}",
collectionId, batchTimer.Elapsed));
}
break;
}
}
_log.Log(string.Format("loaded {0} indexes in {1}", files.Length, timer.Elapsed));
}
private static void Reindex(string dir, int batchSize)
{
var timer = new Stopwatch();
var batchTimer = new Stopwatch();
timer.Start();
var files = Directory.GetFiles(dir, "*.docs");
Console.WriteLine("re-indexing process found {0} document files", files.Length);
foreach (var docFileName in files)
{
var name = Path.GetFileNameWithoutExtension(docFileName)
.Split(".", StringSplitOptions.RemoveEmptyEntries);
var collectionId = ulong.Parse(name[0]);
using (var readSession = new DocumentReadSession(collectionId, new LocalStorageSessionFactory(dir, new LatinTokenizer())))
{
foreach (var batch in readSession.ReadDocs().Batch(batchSize))
{
batchTimer.Restart();
using (var writeSession = new LocalStorageSessionFactory(dir, new LatinTokenizer()).CreateWriteSession(collectionId))
{
var job = new IndexJob(collectionId, batch);
writeSession.WriteToIndex(job);
}
Console.WriteLine("wrote batch to {0} in {1}", collectionId, batchTimer.Elapsed);
}
}
}
Console.WriteLine("rebuilt {0} indexes in {1}", files.Length, timer.Elapsed);
}