LuceneSearchEngine.cs 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521
  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. hints.AddRange(items.Select(item =>
  95. {
  96. var index = GetIndex(item.Name, searchTerm, terms);
  97. return new Tuple<BaseItem, string, int>(item, index.Item1, index.Item2);
  98. }));
  99. // Find artists
  100. var artists = items.OfType<Audio>()
  101. .SelectMany(i => new[] { i.Artist, i.AlbumArtist })
  102. .Where(i => !string.IsNullOrEmpty(i))
  103. .Distinct(StringComparer.OrdinalIgnoreCase)
  104. .ToList();
  105. foreach (var item in artists)
  106. {
  107. var index = GetIndex(item, searchTerm, terms);
  108. if (index.Item2 != -1)
  109. {
  110. try
  111. {
  112. var artist = await _libraryManager.GetArtist(item).ConfigureAwait(false);
  113. hints.Add(new Tuple<BaseItem, string, int>(artist, index.Item1, index.Item2));
  114. }
  115. catch (Exception ex)
  116. {
  117. _logger.ErrorException("Error getting {0}", ex, item);
  118. }
  119. }
  120. }
  121. // Find genres
  122. var genres = items.SelectMany(i => i.Genres)
  123. .Where(i => !string.IsNullOrEmpty(i))
  124. .Distinct(StringComparer.OrdinalIgnoreCase)
  125. .ToList();
  126. foreach (var item in genres)
  127. {
  128. var index = GetIndex(item, searchTerm, terms);
  129. if (index.Item2 != -1)
  130. {
  131. try
  132. {
  133. var genre = await _libraryManager.GetGenre(item).ConfigureAwait(false);
  134. hints.Add(new Tuple<BaseItem, string, int>(genre, index.Item1, index.Item2));
  135. }
  136. catch (Exception ex)
  137. {
  138. _logger.ErrorException("Error getting {0}", ex, item);
  139. }
  140. }
  141. }
  142. // Find studios
  143. var studios = items.SelectMany(i => i.Studios)
  144. .Where(i => !string.IsNullOrEmpty(i))
  145. .Distinct(StringComparer.OrdinalIgnoreCase)
  146. .ToList();
  147. foreach (var item in studios)
  148. {
  149. var index = GetIndex(item, searchTerm, terms);
  150. if (index.Item2 != -1)
  151. {
  152. try
  153. {
  154. var studio = await _libraryManager.GetStudio(item).ConfigureAwait(false);
  155. hints.Add(new Tuple<BaseItem, string, int>(studio, index.Item1, index.Item2));
  156. }
  157. catch (Exception ex)
  158. {
  159. _logger.ErrorException("Error getting {0}", ex, item);
  160. }
  161. }
  162. }
  163. // Find persons
  164. var persons = items.SelectMany(i => i.People)
  165. .Select(i => i.Name)
  166. .Where(i => !string.IsNullOrEmpty(i))
  167. .Distinct(StringComparer.OrdinalIgnoreCase)
  168. .ToList();
  169. foreach (var item in persons)
  170. {
  171. var index = GetIndex(item, searchTerm, terms);
  172. if (index.Item2 != -1)
  173. {
  174. try
  175. {
  176. var person = await _libraryManager.GetPerson(item).ConfigureAwait(false);
  177. hints.Add(new Tuple<BaseItem, string, int>(person, index.Item1, index.Item2));
  178. }
  179. catch (Exception ex)
  180. {
  181. _logger.ErrorException("Error getting {0}", ex, item);
  182. }
  183. }
  184. }
  185. return hints.Where(i => i.Item3 >= 0).OrderBy(i => i.Item3).Select(i => new SearchHintInfo
  186. {
  187. Item = i.Item1,
  188. MatchedTerm = i.Item2
  189. });
  190. }
  191. /// <summary>
  192. /// Gets the index.
  193. /// </summary>
  194. /// <param name="input">The input.</param>
  195. /// <param name="searchInput">The search input.</param>
  196. /// <param name="searchWords">The search input.</param>
  197. /// <returns>System.Int32.</returns>
  198. private Tuple<string, int> GetIndex(string input, string searchInput, string[] searchWords)
  199. {
  200. if (string.Equals(input, searchInput, StringComparison.OrdinalIgnoreCase))
  201. {
  202. return new Tuple<string, int>(searchInput, 0);
  203. }
  204. var index = input.IndexOf(searchInput, StringComparison.OrdinalIgnoreCase);
  205. if (index == 0)
  206. {
  207. return new Tuple<string, int>(searchInput, 1);
  208. }
  209. if (index > 0)
  210. {
  211. return new Tuple<string, int>(searchInput, 2);
  212. }
  213. var items = GetWords(input);
  214. for (var i = 0; i < searchWords.Length; i++)
  215. {
  216. var searchTerm = searchWords[i];
  217. for (var j = 0; j < items.Length; j++)
  218. {
  219. var item = items[j];
  220. if (string.Equals(item, searchTerm, StringComparison.OrdinalIgnoreCase))
  221. {
  222. return new Tuple<string, int>(searchTerm, 3 + (i + 1) * (j + 1));
  223. }
  224. index = item.IndexOf(searchTerm, StringComparison.OrdinalIgnoreCase);
  225. if (index == 0)
  226. {
  227. return new Tuple<string, int>(searchTerm, 4 + (i + 1) * (j + 1));
  228. }
  229. if (index > 0)
  230. {
  231. return new Tuple<string, int>(searchTerm, 5 + (i + 1) * (j + 1));
  232. }
  233. }
  234. }
  235. return new Tuple<string, int>(null, -1);
  236. }
  237. /// <summary>
  238. /// Gets the words.
  239. /// </summary>
  240. /// <param name="term">The term.</param>
  241. /// <returns>System.String[][].</returns>
  242. private string[] GetWords(string term)
  243. {
  244. return term.Split().Where(i => !string.IsNullOrWhiteSpace(i)).ToArray();
  245. }
  246. }
  247. public static class LuceneSearch
  248. {
  249. private static ILogger logger;
  250. private static string path;
  251. private static object lockOb = new object();
  252. private static FSDirectory _directory;
  253. private static FSDirectory directory
  254. {
  255. get
  256. {
  257. if (_directory == null)
  258. {
  259. logger.Info("Opening new Directory: " + path);
  260. _directory = FSDirectory.Open(path);
  261. }
  262. return _directory;
  263. }
  264. set
  265. {
  266. _directory = value;
  267. }
  268. }
  269. private static IndexWriter _writer;
  270. private static IndexWriter writer
  271. {
  272. get
  273. {
  274. if (_writer == null)
  275. {
  276. logger.Info("Opening new IndexWriter");
  277. _writer = new IndexWriter(directory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED);
  278. }
  279. return _writer;
  280. }
  281. set
  282. {
  283. _writer = value;
  284. }
  285. }
  286. private static Dictionary<string, float> bonusTerms;
  287. public static void Init(string path, ILogger logger)
  288. {
  289. logger.Info("Lucene: Init");
  290. bonusTerms = new Dictionary<string, float>();
  291. bonusTerms.Add("Name", 2);
  292. bonusTerms.Add("Overview", 1);
  293. // Optimize the DB on initialization
  294. // TODO: Test whether this has..
  295. // Any effect what-so-ever (apart from initializing the indexwriter on the mainthread context, which makes things a whole lot easier)
  296. // Costs too much time
  297. // Is heavy on the CPU / Memory
  298. LuceneSearch.logger = logger;
  299. LuceneSearch.path = path;
  300. writer.Optimize();
  301. }
  302. private static StandardAnalyzer analyzer = new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_30);
  303. private static Searcher searcher = null;
  304. private static Document createDocument(BaseItem data)
  305. {
  306. Document doc = new Document();
  307. doc.Add(new Field("Id", data.Id.ToString(), Field.Store.YES, Field.Index.NO));
  308. doc.Add(new Field("Name", data.Name, Field.Store.YES, Field.Index.ANALYZED) { Boost = 2 });
  309. doc.Add(new Field("Overview", data.Overview != null ? data.Overview : "", Field.Store.YES, Field.Index.ANALYZED));
  310. return doc;
  311. }
  312. private static void Create(BaseItem item)
  313. {
  314. lock (lockOb)
  315. {
  316. try
  317. {
  318. if (searcher != null)
  319. {
  320. try
  321. {
  322. searcher.Dispose();
  323. }
  324. catch (Exception e)
  325. {
  326. logger.ErrorException("Error in Lucene while creating index (disposing alive searcher)", e, item);
  327. }
  328. searcher = null;
  329. }
  330. _removeFromLuceneIndex(item);
  331. _addToLuceneIndex(item);
  332. }
  333. catch (Exception e)
  334. {
  335. logger.ErrorException("Error in Lucene while creating index", e, item);
  336. }
  337. }
  338. }
  339. private static void _addToLuceneIndex(BaseItem data)
  340. {
  341. // Prevent double entries
  342. var doc = createDocument(data);
  343. writer.AddDocument(doc);
  344. }
  345. private static void _removeFromLuceneIndex(BaseItem data)
  346. {
  347. var query = new TermQuery(new Term("Id", data.Id.ToString()));
  348. writer.DeleteDocuments(query);
  349. }
  350. public static void AddUpdateLuceneIndex(IEnumerable<BaseItem> items)
  351. {
  352. foreach (var item in items)
  353. {
  354. logger.Info("Adding/Updating BaseItem " + item.Name + "(" + item.Id.ToString() + ") to/on Lucene Index");
  355. Create(item);
  356. }
  357. writer.Commit();
  358. writer.Flush(true, true, true);
  359. }
  360. public static void RemoveFromLuceneIndex(IEnumerable<BaseItem> items)
  361. {
  362. foreach (var item in items)
  363. {
  364. logger.Info("Removing BaseItem " + item.Name + "(" + item.Id.ToString() + ") from Lucene Index");
  365. _removeFromLuceneIndex(item);
  366. }
  367. writer.Commit();
  368. writer.Flush(true, true, true);
  369. }
  370. public static IEnumerable<BaseItem> Search(string searchQuery, int maxHits)
  371. {
  372. var results = new List<BaseItem>();
  373. lock (lockOb)
  374. {
  375. try
  376. {
  377. if (searcher == null)
  378. {
  379. searcher = new IndexSearcher(directory, true);
  380. }
  381. BooleanQuery finalQuery = new BooleanQuery();
  382. MultiFieldQueryParser parser = new MultiFieldQueryParser(Lucene.Net.Util.Version.LUCENE_30, new string[] { "Name", "Overview" }, analyzer, bonusTerms);
  383. string[] terms = searchQuery.Split(new[] { " " }, StringSplitOptions.RemoveEmptyEntries);
  384. foreach (string term in terms)
  385. finalQuery.Add(parser.Parse(term.Replace("~", "") + "~0.75"), Occur.SHOULD);
  386. foreach (string term in terms)
  387. finalQuery.Add(parser.Parse(term.Replace("*", "") + "*"), Occur.SHOULD);
  388. logger.Debug("Querying Lucene with query: " + finalQuery.ToString());
  389. long start = DateTime.Now.Ticks;
  390. var searchResult = searcher.Search(finalQuery, maxHits);
  391. foreach (var searchHit in searchResult.ScoreDocs)
  392. {
  393. Document hit = searcher.Doc(searchHit.Doc);
  394. results.Add(BaseItem.LibraryManager.GetItemById(Guid.Parse(hit.Get("Id"))));
  395. }
  396. long total = DateTime.Now.Ticks - start;
  397. float msTotal = (float)total / TimeSpan.TicksPerMillisecond;
  398. logger.Debug(searchResult.ScoreDocs.Length + " result" + (searchResult.ScoreDocs.Length == 1 ? "" : "s") + " in " + msTotal + " ms.");
  399. }
  400. catch (Exception e)
  401. {
  402. logger.ErrorException("Error while searching Lucene index", e);
  403. }
  404. }
  405. return results;
  406. }
  407. public static void CloseAll()
  408. {
  409. logger.Debug("Lucene: CloseAll");
  410. if (writer != null)
  411. {
  412. logger.Debug("Lucene: CloseAll - Writer is alive");
  413. writer.Flush(true, true, true);
  414. writer.Commit();
  415. writer.WaitForMerges();
  416. writer.Dispose();
  417. writer = null;
  418. }
  419. if (analyzer != null)
  420. {
  421. logger.Debug("Lucene: CloseAll - Analyzer is alive");
  422. analyzer.Close();
  423. analyzer.Dispose();
  424. analyzer = null;
  425. }
  426. if (searcher != null)
  427. {
  428. logger.Debug("Lucene: CloseAll - Searcher is alive");
  429. searcher.Dispose();
  430. searcher = null;
  431. }
  432. if (directory != null)
  433. {
  434. logger.Debug("Lucene: CloseAll - Directory is alive");
  435. directory.Dispose();
  436. directory = null;
  437. }
  438. }
  439. }
  440. }