/// <summary>
/// Luceneインデックスに登録されているドキュメントの辞書を作成
/// </summary>
/// <param name="idxDir"></param>
/// <returns></returns>
public Dictionary <string, DocInfo> CreateDocumentDic(FSDirectory idxDir)
{
Dictionary <string, DocInfo> dic = new Dictionary <string, DocInfo>();
IndexReader ir = DirectoryReader.Open(idxDir);
try {
int max = ir.MaxDoc();
for (int i = 0; i < max; i++)
{
var doc = ir.Document(i);
var docInfo = new DocInfo();
docInfo.Id = i;
docInfo.Path = doc.GetField(LuceneIndexBuilder.Path).StringValue();
docInfo.UpdateDate = DateTime.FromBinary(long.Parse(doc.GetField(LuceneIndexBuilder.UpdateDate).StringValue()));
//docInfo.UpdateDate = DateTime.FromBinary(long.Parse(doc.GetBinaryValue(LuceneIndexBuilder.UpdateDate).ToString()));
docInfo.Exists = false;
if (!dic.ContainsKey(docInfo.Path))
{
dic.Add(docInfo.Path, docInfo);
}
}
} finally {
ir.Close();
}
return(dic);
}
/// <summary>
/// 指定したテキスト抽出器でテキスト化したものをインデックス化
/// テキスト抽出器の種類は以下のとおり
/// ・Apache Tika
/// ・IFilter
/// </summary>
/// <param name="path"></param>
/// <param name="indexWriter"></param>
private bool AddDocument(string path, IndexWriter indexWriter, string threadName, Dictionary <string, DocInfo> docDic)
{
string filename = System.IO.Path.GetFileName(path);
string extension = System.IO.Path.GetExtension(path);
FileInfo fi = new FileInfo(path);
if (extension == "" ||
!_targetExtensionDic.ContainsKey(extension.ToLower()))
{
//拡張子なし or 対象拡張子外
AppObject.Logger.Info(threadName + ":" + "Out of target extension. Skipped: " + path);
Interlocked.Increment(ref _skippedCount);
return(false);
}
if (extension.ToLower() != ".mp4" && fi.Length > this.FileSizeLimit)
{
//サイズオーバー(mp4は対象外)
AppObject.Logger.Info(threadName + ":" + "File size over. Skipped: " + path);
Interlocked.Increment(ref _skippedCount);
return(false);
}
//存在するドキュメントか？
if (docDic != null && docDic.ContainsKey(path))
{
DocInfo di = docDic[path];
di.Exists = true;
docDic[path] = di;
//更新日時チェック(秒単位で比較)
if (di.UpdateDate < DateTimeUtil.Truncate(fi.LastWriteTime, TimeSpan.FromSeconds(1)))
{
//更新されている場合Delete+Insert
Term t = new Term(LuceneIndexBuilder.Path, di.Path);
indexWriter.DeleteDocuments(t);
}
else
{
//更新されていない。
AppObject.Logger.Info(threadName + ":" + "No updated. Skipped: " + path);
Interlocked.Increment(ref _skippedCount);
return(false);
}
}
//ドキュメント追加
Document doc = new Document();
if (extension.ToLower() == ".md")
{
//Markdown形式
string content = ReadToString(path);
doc.Add(new Field(Content, content, _hilightFieldType));
}
else if (extension.ToLower() == ".txt")
{
//TXTファイル
var sjis = Encoding.GetEncoding("Shift_JIS");
if (FileUtil.GetTextEncoding(path) == sjis)
{
string content = "";
using (var reader = new StreamReader(path, sjis)) {
content = reader.ReadToEnd();
}
doc.Add(new Field(Content, content, _hilightFieldType));
}
else
{
if (_txtExtractMode == TextExtractModes.Tika)
{
var content = _txtExtractor.Extract(path);
doc.Add(new Field(Content, content.Text, _hilightFieldType));
}
else
{
doc.Add(new Field(Content, IFilterParser.Parse(path), _hilightFieldType));
}
}
}
else
{
if (_txtExtractMode == TextExtractModes.Tika)
{
var content = _txtExtractor.Extract(path);
doc.Add(new Field(Content, content.Text, _hilightFieldType));
}
else
{
doc.Add(new Field(Content, IFilterParser.Parse(path), _hilightFieldType));
}
}
doc.Add(new StringField(Path, path, FieldStore.YES));
doc.Add(new StringField(Title, filename.ToLower(), FieldStore.YES));
doc.Add(new StringField(Extension, extension.ToLower(), FieldStore.YES));
//NOTE:Date型のFieldは存在しないのでlongで保持
long l = long.Parse(fi.LastWriteTime.ToString("yyyyMMddHHmmss"));
doc.Add(new LongPoint(UpdateDate, l));
doc.Add(new StoredField(UpdateDate, l));
//doc.Add(new StringField(UpdateDate,
// DateTools.DateToString(_sdf.parse(fi.LastWriteTime.ToString("yyyy/MM/dd")), DateToolsResolution.DAY),
// FieldStore.YES));
indexWriter.AddDocument(doc);
return(true);
}
