LuceneSearchEngine.cs 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554
  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<SearchHintInfo>> GetSearchHints(IEnumerable<BaseItem> inputItems, string searchTerm)
  86. {
  87. if (string.IsNullOrEmpty(searchTerm))
  88. {
  89. throw new ArgumentNullException("searchTerm");
  90. }
  91. var terms = GetWords(searchTerm);
  92. var hints = new List<Tuple<BaseItem, string, int>>();
  93. var items = inputItems.Where(i => !(i is MusicArtist)).ToList();
  94. // Add search hints based on item name
  95. hints.AddRange(items.Where(i => !string.IsNullOrEmpty(i.Name)).Select(item =>
  96. {
  97. var index = GetIndex(item.Name, searchTerm, terms);
  98. return new Tuple<BaseItem, string, int>(item, index.Item1, index.Item2);
  99. }));
  100. // Find artists
  101. var artists = items.OfType<Audio>()
  102. .SelectMany(i => new[] { i.Artist, i.AlbumArtist })
  103. .Where(i => !string.IsNullOrEmpty(i))
  104. .Distinct(StringComparer.OrdinalIgnoreCase)
  105. .ToList();
  106. foreach (var item in artists)
  107. {
  108. var index = GetIndex(item, searchTerm, terms);
  109. if (index.Item2 != -1)
  110. {
  111. try
  112. {
  113. var artist = await _libraryManager.GetArtist(item).ConfigureAwait(false);
  114. hints.Add(new Tuple<BaseItem, string, int>(artist, index.Item1, index.Item2));
  115. }
  116. catch (Exception ex)
  117. {
  118. _logger.ErrorException("Error getting {0}", ex, item);
  119. }
  120. }
  121. }
  122. // Find genres, from non-audio items
  123. var genres = items.Where(i => !(i is Audio) && !(i is MusicAlbum) && !(i is MusicAlbumDisc) && !(i is MusicArtist) && !(i is MusicVideo))
  124. .SelectMany(i => i.Genres)
  125. .Where(i => !string.IsNullOrEmpty(i))
  126. .Distinct(StringComparer.OrdinalIgnoreCase)
  127. .ToList();
  128. foreach (var item in genres)
  129. {
  130. var index = GetIndex(item, searchTerm, terms);
  131. if (index.Item2 != -1)
  132. {
  133. try
  134. {
  135. var genre = await _libraryManager.GetGenre(item).ConfigureAwait(false);
  136. hints.Add(new Tuple<BaseItem, string, int>(genre, index.Item1, index.Item2));
  137. }
  138. catch (Exception ex)
  139. {
  140. _logger.ErrorException("Error getting {0}", ex, item);
  141. }
  142. }
  143. }
  144. // Find music genres
  145. var musicGenres = items.Where(i => (i is Audio) || (i is MusicAlbum) || (i is MusicAlbumDisc) || (i is MusicArtist) || (i is MusicVideo))
  146. .SelectMany(i => i.Genres)
  147. .Where(i => !string.IsNullOrEmpty(i))
  148. .Distinct(StringComparer.OrdinalIgnoreCase)
  149. .ToList();
  150. foreach (var item in musicGenres)
  151. {
  152. var index = GetIndex(item, searchTerm, terms);
  153. if (index.Item2 != -1)
  154. {
  155. try
  156. {
  157. var genre = await _libraryManager.GetMusicGenre(item).ConfigureAwait(false);
  158. hints.Add(new Tuple<BaseItem, string, int>(genre, index.Item1, index.Item2));
  159. }
  160. catch (Exception ex)
  161. {
  162. _logger.ErrorException("Error getting {0}", ex, item);
  163. }
  164. }
  165. }
  166. // Find studios
  167. var studios = items.SelectMany(i => i.Studios)
  168. .Where(i => !string.IsNullOrEmpty(i))
  169. .Distinct(StringComparer.OrdinalIgnoreCase)
  170. .ToList();
  171. foreach (var item in studios)
  172. {
  173. var index = GetIndex(item, searchTerm, terms);
  174. if (index.Item2 != -1)
  175. {
  176. try
  177. {
  178. var studio = await _libraryManager.GetStudio(item).ConfigureAwait(false);
  179. hints.Add(new Tuple<BaseItem, string, int>(studio, index.Item1, index.Item2));
  180. }
  181. catch (Exception ex)
  182. {
  183. _logger.ErrorException("Error getting {0}", ex, item);
  184. }
  185. }
  186. }
  187. // Find persons
  188. var persons = items.SelectMany(i => i.People)
  189. .Select(i => i.Name)
  190. .Where(i => !string.IsNullOrEmpty(i))
  191. .Distinct(StringComparer.OrdinalIgnoreCase)
  192. .ToList();
  193. foreach (var item in persons)
  194. {
  195. var index = GetIndex(item, searchTerm, terms);
  196. if (index.Item2 != -1)
  197. {
  198. try
  199. {
  200. var person = await _libraryManager.GetPerson(item).ConfigureAwait(false);
  201. hints.Add(new Tuple<BaseItem, string, int>(person, index.Item1, index.Item2));
  202. }
  203. catch (Exception ex)
  204. {
  205. _logger.ErrorException("Error getting {0}", ex, item);
  206. }
  207. }
  208. }
  209. return hints.Where(i => i.Item3 >= 0).OrderBy(i => i.Item3).Select(i => new SearchHintInfo
  210. {
  211. Item = i.Item1,
  212. MatchedTerm = i.Item2
  213. });
  214. }
  215. /// <summary>
  216. /// Gets the index.
  217. /// </summary>
  218. /// <param name="input">The input.</param>
  219. /// <param name="searchInput">The search input.</param>
  220. /// <param name="searchWords">The search input.</param>
  221. /// <returns>System.Int32.</returns>
  222. private Tuple<string, int> GetIndex(string input, string searchInput, string[] searchWords)
  223. {
  224. if (string.IsNullOrEmpty(input))
  225. {
  226. throw new ArgumentNullException("input");
  227. }
  228. if (string.Equals(input, searchInput, StringComparison.OrdinalIgnoreCase))
  229. {
  230. return new Tuple<string, int>(searchInput, 0);
  231. }
  232. var index = input.IndexOf(searchInput, StringComparison.OrdinalIgnoreCase);
  233. if (index == 0)
  234. {
  235. return new Tuple<string, int>(searchInput, 1);
  236. }
  237. if (index > 0)
  238. {
  239. return new Tuple<string, int>(searchInput, 2);
  240. }
  241. var items = GetWords(input);
  242. for (var i = 0; i < searchWords.Length; i++)
  243. {
  244. var searchTerm = searchWords[i];
  245. for (var j = 0; j < items.Length; j++)
  246. {
  247. var item = items[j];
  248. if (string.Equals(item, searchTerm, StringComparison.OrdinalIgnoreCase))
  249. {
  250. return new Tuple<string, int>(searchTerm, 3 + (i + 1) * (j + 1));
  251. }
  252. index = item.IndexOf(searchTerm, StringComparison.OrdinalIgnoreCase);
  253. if (index == 0)
  254. {
  255. return new Tuple<string, int>(searchTerm, 4 + (i + 1) * (j + 1));
  256. }
  257. if (index > 0)
  258. {
  259. return new Tuple<string, int>(searchTerm, 5 + (i + 1) * (j + 1));
  260. }
  261. }
  262. }
  263. return new Tuple<string, int>(null, -1);
  264. }
  265. /// <summary>
  266. /// Gets the words.
  267. /// </summary>
  268. /// <param name="term">The term.</param>
  269. /// <returns>System.String[][].</returns>
  270. private string[] GetWords(string term)
  271. {
  272. return term.Split().Where(i => !string.IsNullOrWhiteSpace(i)).ToArray();
  273. }
  274. }
  275. public static class LuceneSearch
  276. {
  277. private static ILogger logger;
  278. private static string path;
  279. private static object lockOb = new object();
  280. private static FSDirectory _directory;
  281. private static FSDirectory directory
  282. {
  283. get
  284. {
  285. if (_directory == null)
  286. {
  287. logger.Info("Opening new Directory: " + path);
  288. _directory = FSDirectory.Open(path);
  289. }
  290. return _directory;
  291. }
  292. set
  293. {
  294. _directory = value;
  295. }
  296. }
  297. private static IndexWriter _writer;
  298. private static IndexWriter writer
  299. {
  300. get
  301. {
  302. if (_writer == null)
  303. {
  304. logger.Info("Opening new IndexWriter");
  305. _writer = new IndexWriter(directory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED);
  306. }
  307. return _writer;
  308. }
  309. set
  310. {
  311. _writer = value;
  312. }
  313. }
  314. private static Dictionary<string, float> bonusTerms;
  315. public static void Init(string path, ILogger logger)
  316. {
  317. logger.Info("Lucene: Init");
  318. bonusTerms = new Dictionary<string, float>();
  319. bonusTerms.Add("Name", 2);
  320. bonusTerms.Add("Overview", 1);
  321. // Optimize the DB on initialization
  322. // TODO: Test whether this has..
  323. // Any effect what-so-ever (apart from initializing the indexwriter on the mainthread context, which makes things a whole lot easier)
  324. // Costs too much time
  325. // Is heavy on the CPU / Memory
  326. LuceneSearch.logger = logger;
  327. LuceneSearch.path = path;
  328. writer.Optimize();
  329. }
  330. private static StandardAnalyzer analyzer = new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_30);
  331. private static Searcher searcher = null;
  332. private static Document createDocument(BaseItem data)
  333. {
  334. Document doc = new Document();
  335. doc.Add(new Field("Id", data.Id.ToString(), Field.Store.YES, Field.Index.NO));
  336. doc.Add(new Field("Name", data.Name, Field.Store.YES, Field.Index.ANALYZED) { Boost = 2 });
  337. doc.Add(new Field("Overview", data.Overview != null ? data.Overview : "", Field.Store.YES, Field.Index.ANALYZED));
  338. return doc;
  339. }
  340. private static void Create(BaseItem item)
  341. {
  342. lock (lockOb)
  343. {
  344. try
  345. {
  346. if (searcher != null)
  347. {
  348. try
  349. {
  350. searcher.Dispose();
  351. }
  352. catch (Exception e)
  353. {
  354. logger.ErrorException("Error in Lucene while creating index (disposing alive searcher)", e, item);
  355. }
  356. searcher = null;
  357. }
  358. _removeFromLuceneIndex(item);
  359. _addToLuceneIndex(item);
  360. }
  361. catch (Exception e)
  362. {
  363. logger.ErrorException("Error in Lucene while creating index", e, item);
  364. }
  365. }
  366. }
  367. private static void _addToLuceneIndex(BaseItem data)
  368. {
  369. // Prevent double entries
  370. var doc = createDocument(data);
  371. writer.AddDocument(doc);
  372. }
  373. private static void _removeFromLuceneIndex(BaseItem data)
  374. {
  375. var query = new TermQuery(new Term("Id", data.Id.ToString()));
  376. writer.DeleteDocuments(query);
  377. }
  378. public static void AddUpdateLuceneIndex(IEnumerable<BaseItem> items)
  379. {
  380. foreach (var item in items)
  381. {
  382. logger.Info("Adding/Updating BaseItem " + item.Name + "(" + item.Id.ToString() + ") to/on Lucene Index");
  383. Create(item);
  384. }
  385. writer.Commit();
  386. writer.Flush(true, true, true);
  387. }
  388. public static void RemoveFromLuceneIndex(IEnumerable<BaseItem> items)
  389. {
  390. foreach (var item in items)
  391. {
  392. logger.Info("Removing BaseItem " + item.Name + "(" + item.Id.ToString() + ") from Lucene Index");
  393. _removeFromLuceneIndex(item);
  394. }
  395. writer.Commit();
  396. writer.Flush(true, true, true);
  397. }
  398. public static IEnumerable<BaseItem> Search(string searchQuery, int maxHits)
  399. {
  400. var results = new List<BaseItem>();
  401. lock (lockOb)
  402. {
  403. try
  404. {
  405. if (searcher == null)
  406. {
  407. searcher = new IndexSearcher(directory, true);
  408. }
  409. BooleanQuery finalQuery = new BooleanQuery();
  410. MultiFieldQueryParser parser = new MultiFieldQueryParser(Lucene.Net.Util.Version.LUCENE_30, new string[] { "Name", "Overview" }, analyzer, bonusTerms);
  411. string[] terms = searchQuery.Split(new[] { " " }, StringSplitOptions.RemoveEmptyEntries);
  412. foreach (string term in terms)
  413. finalQuery.Add(parser.Parse(term.Replace("~", "") + "~0.75"), Occur.SHOULD);
  414. foreach (string term in terms)
  415. finalQuery.Add(parser.Parse(term.Replace("*", "") + "*"), Occur.SHOULD);
  416. logger.Debug("Querying Lucene with query: " + finalQuery.ToString());
  417. long start = DateTime.Now.Ticks;
  418. var searchResult = searcher.Search(finalQuery, maxHits);
  419. foreach (var searchHit in searchResult.ScoreDocs)
  420. {
  421. Document hit = searcher.Doc(searchHit.Doc);
  422. results.Add(BaseItem.LibraryManager.GetItemById(Guid.Parse(hit.Get("Id"))));
  423. }
  424. long total = DateTime.Now.Ticks - start;
  425. float msTotal = (float)total / TimeSpan.TicksPerMillisecond;
  426. logger.Debug(searchResult.ScoreDocs.Length + " result" + (searchResult.ScoreDocs.Length == 1 ? "" : "s") + " in " + msTotal + " ms.");
  427. }
  428. catch (Exception e)
  429. {
  430. logger.ErrorException("Error while searching Lucene index", e);
  431. }
  432. }
  433. return results;
  434. }
  435. public static void CloseAll()
  436. {
  437. logger.Debug("Lucene: CloseAll");
  438. if (writer != null)
  439. {
  440. logger.Debug("Lucene: CloseAll - Writer is alive");
  441. writer.Flush(true, true, true);
  442. writer.Commit();
  443. writer.WaitForMerges();
  444. writer.Dispose();
  445. writer = null;
  446. }
  447. if (analyzer != null)
  448. {
  449. logger.Debug("Lucene: CloseAll - Analyzer is alive");
  450. analyzer.Close();
  451. analyzer.Dispose();
  452. analyzer = null;
  453. }
  454. if (searcher != null)
  455. {
  456. logger.Debug("Lucene: CloseAll - Searcher is alive");
  457. searcher.Dispose();
  458. searcher = null;
  459. }
  460. if (directory != null)
  461. {
  462. logger.Debug("Lucene: CloseAll - Directory is alive");
  463. directory.Dispose();
  464. directory = null;
  465. }
  466. }
  467. }
  468. }