LuceneSearchEngine.cs 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460
  1. using Lucene.Net.Analysis.Standard;
  2. using Lucene.Net.Documents;
  3. using Lucene.Net.Index;
  4. using Lucene.Net.QueryParsers;
  5. using Lucene.Net.Search;
  6. using Lucene.Net.Store;
  7. using MediaBrowser.Controller;
  8. using MediaBrowser.Controller.Entities;
  9. using MediaBrowser.Controller.Entities.Audio;
  10. using MediaBrowser.Controller.Library;
  11. using MediaBrowser.Model.Logging;
  12. using System;
  13. using System.Collections.Generic;
  14. using System.Linq;
  15. using System.Threading.Tasks;
  16. namespace MediaBrowser.Server.Implementations.Library
  17. {
  18. /// <summary>
  19. /// Class LuceneSearchEngine
  20. /// http://www.codeproject.com/Articles/320219/Lucene-Net-ultra-fast-search-for-MVC-or-WebForms
  21. /// </summary>
  22. public class LuceneSearchEngine : ILibrarySearchEngine, IDisposable
  23. {
  24. private readonly ILibraryManager _libraryManager;
  25. private readonly ILogger _logger;
  26. public LuceneSearchEngine(IServerApplicationPaths serverPaths, ILogManager logManager, ILibraryManager libraryManager)
  27. {
  28. _libraryManager = libraryManager;
  29. _logger = logManager.GetLogger("Lucene");
  30. //string luceneDbPath = serverPaths.DataPath + "\\SearchIndexDB";
  31. //if (!System.IO.Directory.Exists(luceneDbPath))
  32. // System.IO.Directory.CreateDirectory(luceneDbPath);
  33. //else if(File.Exists(luceneDbPath + "\\write.lock"))
  34. // File.Delete(luceneDbPath + "\\write.lock");
  35. //LuceneSearch.Init(luceneDbPath, _logger);
  36. //BaseItem.LibraryManager.LibraryChanged += LibraryChanged;
  37. }
  38. public void LibraryChanged(object source, ChildrenChangedEventArgs changeInformation)
  39. {
  40. Task.Run(() =>
  41. {
  42. if (changeInformation.ItemsAdded.Count + changeInformation.ItemsUpdated.Count > 0)
  43. {
  44. LuceneSearch.AddUpdateLuceneIndex(changeInformation.ItemsAdded.Concat(changeInformation.ItemsUpdated));
  45. }
  46. if (changeInformation.ItemsRemoved.Count > 0)
  47. {
  48. LuceneSearch.RemoveFromLuceneIndex(changeInformation.ItemsRemoved);
  49. }
  50. });
  51. }
  52. public void AddItemsToIndex(IEnumerable<BaseItem> items)
  53. {
  54. LuceneSearch.AddUpdateLuceneIndex(items);
  55. }
  56. /// <summary>
  57. /// Searches items and returns them in order of relevance.
  58. /// </summary>
  59. /// <param name="items">The items.</param>
  60. /// <param name="searchTerm">The search term.</param>
  61. /// <returns>IEnumerable{BaseItem}.</returns>
  62. /// <exception cref="System.ArgumentNullException">searchTerm</exception>
  63. public IEnumerable<BaseItem> Search(IEnumerable<BaseItem> items, string searchTerm)
  64. {
  65. if (string.IsNullOrEmpty(searchTerm))
  66. {
  67. throw new ArgumentNullException("searchTerm");
  68. }
  69. var hits = LuceneSearch.Search(searchTerm, items.Count());
  70. //return hits;
  71. return hits.Where(searchHit => items.Any(p => p.Id == searchHit.Id));
  72. }
  73. public void Dispose()
  74. {
  75. //BaseItem.LibraryManager.LibraryChanged -= LibraryChanged;
  76. //LuceneSearch.CloseAll();
  77. }
  78. /// <summary>
  79. /// Gets the search hints.
  80. /// </summary>
  81. /// <param name="inputItems">The input items.</param>
  82. /// <param name="searchTerm">The search term.</param>
  83. /// <returns>IEnumerable{SearchHintResult}.</returns>
  84. /// <exception cref="System.ArgumentNullException">searchTerm</exception>
  85. public async Task<IEnumerable<BaseItem>> GetSearchHints(IEnumerable<BaseItem> inputItems, string searchTerm)
  86. {
  87. if (string.IsNullOrEmpty(searchTerm))
  88. {
  89. throw new ArgumentNullException("searchTerm");
  90. }
  91. var hints = new List<Tuple<BaseItem, int>>();
  92. var items = inputItems.Where(i => !(i is MusicArtist)).ToList();
  93. foreach (var item in items)
  94. {
  95. var index = IndexOf(item.Name, searchTerm);
  96. if (index != -1)
  97. {
  98. hints.Add(new Tuple<BaseItem, int>(item, index));
  99. }
  100. }
  101. // Find artists
  102. var artists = items.OfType<Audio>()
  103. .SelectMany(i => new[] { i.Artist, i.AlbumArtist })
  104. .Where(i => !string.IsNullOrEmpty(i))
  105. .Distinct(StringComparer.OrdinalIgnoreCase)
  106. .ToList();
  107. foreach (var item in artists)
  108. {
  109. var index = IndexOf(item, searchTerm);
  110. if (index != -1)
  111. {
  112. var artist = await _libraryManager.GetArtist(item).ConfigureAwait(false);
  113. hints.Add(new Tuple<BaseItem, int>(artist, index));
  114. }
  115. }
  116. // Find genres
  117. var genres = items.SelectMany(i => i.Genres)
  118. .Where(i => !string.IsNullOrEmpty(i))
  119. .Distinct(StringComparer.OrdinalIgnoreCase)
  120. .ToList();
  121. foreach (var item in genres)
  122. {
  123. var index = IndexOf(item, searchTerm);
  124. if (index != -1)
  125. {
  126. var genre = await _libraryManager.GetGenre(item).ConfigureAwait(false);
  127. hints.Add(new Tuple<BaseItem, int>(genre, index));
  128. }
  129. }
  130. // Find studios
  131. var studios = items.SelectMany(i => i.Studios)
  132. .Where(i => !string.IsNullOrEmpty(i))
  133. .Distinct(StringComparer.OrdinalIgnoreCase)
  134. .ToList();
  135. foreach (var item in studios)
  136. {
  137. var index = IndexOf(item, searchTerm);
  138. if (index != -1)
  139. {
  140. var studio = await _libraryManager.GetStudio(item).ConfigureAwait(false);
  141. hints.Add(new Tuple<BaseItem, int>(studio, index));
  142. }
  143. }
  144. // Find persons
  145. var persons = items.SelectMany(i => i.People)
  146. .Select(i => i.Name)
  147. .Where(i => !string.IsNullOrEmpty(i))
  148. .Distinct(StringComparer.OrdinalIgnoreCase)
  149. .ToList();
  150. foreach (var item in persons)
  151. {
  152. var index = IndexOf(item, searchTerm);
  153. if (index != -1)
  154. {
  155. var person = await _libraryManager.GetPerson(item).ConfigureAwait(false);
  156. hints.Add(new Tuple<BaseItem, int>(person, index));
  157. }
  158. }
  159. return hints.OrderBy(i => i.Item2).Select(i => i.Item1);
  160. }
  161. /// <summary>
  162. /// Gets the words.
  163. /// </summary>
  164. /// <param name="term">The term.</param>
  165. /// <returns>System.String[][].</returns>
  166. private string[] GetWords(string term)
  167. {
  168. // TODO: Improve this to be more accurate and respect culture
  169. var words = term.Split(' ');
  170. return words;
  171. }
  172. /// <summary>
  173. /// Indexes the of.
  174. /// </summary>
  175. /// <param name="input">The input.</param>
  176. /// <param name="term">The term.</param>
  177. /// <returns>System.Int32.</returns>
  178. private int IndexOf(string input, string term)
  179. {
  180. var index = 0;
  181. foreach (var word in GetWords(input))
  182. {
  183. if (word.IndexOf(term, StringComparison.OrdinalIgnoreCase) != -1)
  184. {
  185. return index;
  186. }
  187. index++;
  188. }
  189. return -1;
  190. }
  191. }
  192. public static class LuceneSearch
  193. {
  194. private static ILogger logger;
  195. private static string path;
  196. private static object lockOb = new object();
  197. private static FSDirectory _directory;
  198. private static FSDirectory directory
  199. {
  200. get
  201. {
  202. if (_directory == null)
  203. {
  204. logger.Info("Opening new Directory: " + path);
  205. _directory = FSDirectory.Open(path);
  206. }
  207. return _directory;
  208. }
  209. set
  210. {
  211. _directory = value;
  212. }
  213. }
  214. private static IndexWriter _writer;
  215. private static IndexWriter writer
  216. {
  217. get
  218. {
  219. if (_writer == null)
  220. {
  221. logger.Info("Opening new IndexWriter");
  222. _writer = new IndexWriter(directory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED);
  223. }
  224. return _writer;
  225. }
  226. set
  227. {
  228. _writer = value;
  229. }
  230. }
  231. private static Dictionary<string, float> bonusTerms;
  232. public static void Init(string path, ILogger logger)
  233. {
  234. logger.Info("Lucene: Init");
  235. bonusTerms = new Dictionary<string, float>();
  236. bonusTerms.Add("Name", 2);
  237. bonusTerms.Add("Overview", 1);
  238. // Optimize the DB on initialization
  239. // TODO: Test whether this has..
  240. // Any effect what-so-ever (apart from initializing the indexwriter on the mainthread context, which makes things a whole lot easier)
  241. // Costs too much time
  242. // Is heavy on the CPU / Memory
  243. LuceneSearch.logger = logger;
  244. LuceneSearch.path = path;
  245. writer.Optimize();
  246. }
  247. private static StandardAnalyzer analyzer = new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_30);
  248. private static Searcher searcher = null;
  249. private static Document createDocument(BaseItem data)
  250. {
  251. Document doc = new Document();
  252. doc.Add(new Field("Id", data.Id.ToString(), Field.Store.YES, Field.Index.NO));
  253. doc.Add(new Field("Name", data.Name, Field.Store.YES, Field.Index.ANALYZED) { Boost = 2 });
  254. doc.Add(new Field("Overview", data.Overview != null ? data.Overview : "", Field.Store.YES, Field.Index.ANALYZED));
  255. return doc;
  256. }
  257. private static void Create(BaseItem item)
  258. {
  259. lock (lockOb)
  260. {
  261. try
  262. {
  263. if (searcher != null)
  264. {
  265. try
  266. {
  267. searcher.Dispose();
  268. }
  269. catch (Exception e)
  270. {
  271. logger.ErrorException("Error in Lucene while creating index (disposing alive searcher)", e, item);
  272. }
  273. searcher = null;
  274. }
  275. _removeFromLuceneIndex(item);
  276. _addToLuceneIndex(item);
  277. }
  278. catch (Exception e)
  279. {
  280. logger.ErrorException("Error in Lucene while creating index", e, item);
  281. }
  282. }
  283. }
  284. private static void _addToLuceneIndex(BaseItem data)
  285. {
  286. // Prevent double entries
  287. var doc = createDocument(data);
  288. writer.AddDocument(doc);
  289. }
  290. private static void _removeFromLuceneIndex(BaseItem data)
  291. {
  292. var query = new TermQuery(new Term("Id", data.Id.ToString()));
  293. writer.DeleteDocuments(query);
  294. }
  295. public static void AddUpdateLuceneIndex(IEnumerable<BaseItem> items)
  296. {
  297. foreach (var item in items)
  298. {
  299. logger.Info("Adding/Updating BaseItem " + item.Name + "(" + item.Id.ToString() + ") to/on Lucene Index");
  300. Create(item);
  301. }
  302. writer.Commit();
  303. writer.Flush(true, true, true);
  304. }
  305. public static void RemoveFromLuceneIndex(IEnumerable<BaseItem> items)
  306. {
  307. foreach (var item in items)
  308. {
  309. logger.Info("Removing BaseItem " + item.Name + "(" + item.Id.ToString() + ") from Lucene Index");
  310. _removeFromLuceneIndex(item);
  311. }
  312. writer.Commit();
  313. writer.Flush(true, true, true);
  314. }
  315. public static IEnumerable<BaseItem> Search(string searchQuery, int maxHits)
  316. {
  317. var results = new List<BaseItem>();
  318. lock (lockOb)
  319. {
  320. try
  321. {
  322. if (searcher == null)
  323. {
  324. searcher = new IndexSearcher(directory, true);
  325. }
  326. BooleanQuery finalQuery = new BooleanQuery();
  327. MultiFieldQueryParser parser = new MultiFieldQueryParser(Lucene.Net.Util.Version.LUCENE_30, new string[] { "Name", "Overview" }, analyzer, bonusTerms);
  328. string[] terms = searchQuery.Split(new[] { " " }, StringSplitOptions.RemoveEmptyEntries);
  329. foreach (string term in terms)
  330. finalQuery.Add(parser.Parse(term.Replace("~", "") + "~0.75"), Occur.SHOULD);
  331. foreach (string term in terms)
  332. finalQuery.Add(parser.Parse(term.Replace("*", "") + "*"), Occur.SHOULD);
  333. logger.Debug("Querying Lucene with query: " + finalQuery.ToString());
  334. long start = DateTime.Now.Ticks;
  335. var searchResult = searcher.Search(finalQuery, maxHits);
  336. foreach (var searchHit in searchResult.ScoreDocs)
  337. {
  338. Document hit = searcher.Doc(searchHit.Doc);
  339. results.Add(BaseItem.LibraryManager.GetItemById(Guid.Parse(hit.Get("Id"))));
  340. }
  341. long total = DateTime.Now.Ticks - start;
  342. float msTotal = (float)total / TimeSpan.TicksPerMillisecond;
  343. logger.Debug(searchResult.ScoreDocs.Length + " result" + (searchResult.ScoreDocs.Length == 1 ? "" : "s") + " in " + msTotal + " ms.");
  344. }
  345. catch (Exception e)
  346. {
  347. logger.ErrorException("Error while searching Lucene index", e);
  348. }
  349. }
  350. return results;
  351. }
  352. public static void CloseAll()
  353. {
  354. logger.Debug("Lucene: CloseAll");
  355. if (writer != null)
  356. {
  357. logger.Debug("Lucene: CloseAll - Writer is alive");
  358. writer.Flush(true, true, true);
  359. writer.Commit();
  360. writer.WaitForMerges();
  361. writer.Dispose();
  362. writer = null;
  363. }
  364. if (analyzer != null)
  365. {
  366. logger.Debug("Lucene: CloseAll - Analyzer is alive");
  367. analyzer.Close();
  368. analyzer.Dispose();
  369. analyzer = null;
  370. }
  371. if (searcher != null)
  372. {
  373. logger.Debug("Lucene: CloseAll - Searcher is alive");
  374. searcher.Dispose();
  375. searcher = null;
  376. }
  377. if (directory != null)
  378. {
  379. logger.Debug("Lucene: CloseAll - Directory is alive");
  380. directory.Dispose();
  381. directory = null;
  382. }
  383. }
  384. }
  385. }