LuceneSearchEngine.cs 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524
  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. foreach (var item in items)
  95. {
  96. var index = GetIndex(item.Name, searchTerm, terms);
  97. if (index.Item2 != -1)
  98. {
  99. hints.Add(new Tuple<BaseItem, string, int>(item, index.Item1, index.Item2));
  100. }
  101. }
  102. // Find artists
  103. var artists = items.OfType<Audio>()
  104. .SelectMany(i => new[] { i.Artist, i.AlbumArtist })
  105. .Where(i => !string.IsNullOrEmpty(i))
  106. .Distinct(StringComparer.OrdinalIgnoreCase)
  107. .ToList();
  108. foreach (var item in artists)
  109. {
  110. var index = GetIndex(item, searchTerm, terms);
  111. if (index.Item2 != -1)
  112. {
  113. try
  114. {
  115. var artist = await _libraryManager.GetArtist(item).ConfigureAwait(false);
  116. hints.Add(new Tuple<BaseItem, string, int>(artist, index.Item1, index.Item2));
  117. }
  118. catch (Exception ex)
  119. {
  120. _logger.ErrorException("Error getting {0}", ex, item);
  121. }
  122. }
  123. }
  124. // Find genres
  125. var genres = items.SelectMany(i => i.Genres)
  126. .Where(i => !string.IsNullOrEmpty(i))
  127. .Distinct(StringComparer.OrdinalIgnoreCase)
  128. .ToList();
  129. foreach (var item in genres)
  130. {
  131. var index = GetIndex(item, searchTerm, terms);
  132. if (index.Item2 != -1)
  133. {
  134. try
  135. {
  136. var genre = await _libraryManager.GetGenre(item).ConfigureAwait(false);
  137. hints.Add(new Tuple<BaseItem, string, int>(genre, index.Item1, index.Item2));
  138. }
  139. catch (Exception ex)
  140. {
  141. _logger.ErrorException("Error getting {0}", ex, item);
  142. }
  143. }
  144. }
  145. // Find studios
  146. var studios = items.SelectMany(i => i.Studios)
  147. .Where(i => !string.IsNullOrEmpty(i))
  148. .Distinct(StringComparer.OrdinalIgnoreCase)
  149. .ToList();
  150. foreach (var item in studios)
  151. {
  152. var index = GetIndex(item, searchTerm, terms);
  153. if (index.Item2 != -1)
  154. {
  155. try
  156. {
  157. var studio = await _libraryManager.GetStudio(item).ConfigureAwait(false);
  158. hints.Add(new Tuple<BaseItem, string, int>(studio, index.Item1, index.Item2));
  159. }
  160. catch (Exception ex)
  161. {
  162. _logger.ErrorException("Error getting {0}", ex, item);
  163. }
  164. }
  165. }
  166. // Find persons
  167. var persons = items.SelectMany(i => i.People)
  168. .Select(i => i.Name)
  169. .Where(i => !string.IsNullOrEmpty(i))
  170. .Distinct(StringComparer.OrdinalIgnoreCase)
  171. .ToList();
  172. foreach (var item in persons)
  173. {
  174. var index = GetIndex(item, searchTerm, terms);
  175. if (index.Item2 != -1)
  176. {
  177. try
  178. {
  179. var person = await _libraryManager.GetPerson(item).ConfigureAwait(false);
  180. hints.Add(new Tuple<BaseItem, string, int>(person, index.Item1, index.Item2));
  181. }
  182. catch (Exception ex)
  183. {
  184. _logger.ErrorException("Error getting {0}", ex, item);
  185. }
  186. }
  187. }
  188. return hints.OrderBy(i => i.Item3).Select(i => new SearchHintInfo
  189. {
  190. Item = i.Item1,
  191. MatchedTerm = i.Item2
  192. });
  193. }
  194. /// <summary>
  195. /// Gets the index.
  196. /// </summary>
  197. /// <param name="input">The input.</param>
  198. /// <param name="searchInput">The search input.</param>
  199. /// <param name="searchWords">The search input.</param>
  200. /// <returns>System.Int32.</returns>
  201. private Tuple<string, int> GetIndex(string input, string searchInput, string[] searchWords)
  202. {
  203. if (string.Equals(input, searchInput, StringComparison.OrdinalIgnoreCase))
  204. {
  205. return new Tuple<string, int>(searchInput, 0);
  206. }
  207. var index = input.IndexOf(searchInput, StringComparison.OrdinalIgnoreCase);
  208. if (index == 0)
  209. {
  210. return new Tuple<string, int>(searchInput, 1);
  211. }
  212. if (index > 0)
  213. {
  214. return new Tuple<string, int>(searchInput, 2);
  215. }
  216. var items = GetWords(input);
  217. for (var i = 0; i < searchWords.Length; i++)
  218. {
  219. var searchTerm = searchWords[i];
  220. for (var j = 0; j < items.Length; j++)
  221. {
  222. var item = items[j];
  223. if (string.Equals(item, searchTerm, StringComparison.OrdinalIgnoreCase))
  224. {
  225. return new Tuple<string, int>(searchTerm, 3 + (i + 1) * (j + 1));
  226. }
  227. index = item.IndexOf(searchTerm, StringComparison.OrdinalIgnoreCase);
  228. if (index == 0)
  229. {
  230. return new Tuple<string, int>(searchTerm, 4 + (i + 1) * (j + 1));
  231. }
  232. if (index > 0)
  233. {
  234. return new Tuple<string, int>(searchTerm, 5 + (i + 1) * (j + 1));
  235. }
  236. }
  237. }
  238. return new Tuple<string, int>(null, -1);
  239. }
  240. /// <summary>
  241. /// Gets the words.
  242. /// </summary>
  243. /// <param name="term">The term.</param>
  244. /// <returns>System.String[][].</returns>
  245. private string[] GetWords(string term)
  246. {
  247. return term.Split().Where(i => !string.IsNullOrWhiteSpace(i)).ToArray();
  248. }
  249. }
  250. public static class LuceneSearch
  251. {
  252. private static ILogger logger;
  253. private static string path;
  254. private static object lockOb = new object();
  255. private static FSDirectory _directory;
  256. private static FSDirectory directory
  257. {
  258. get
  259. {
  260. if (_directory == null)
  261. {
  262. logger.Info("Opening new Directory: " + path);
  263. _directory = FSDirectory.Open(path);
  264. }
  265. return _directory;
  266. }
  267. set
  268. {
  269. _directory = value;
  270. }
  271. }
  272. private static IndexWriter _writer;
  273. private static IndexWriter writer
  274. {
  275. get
  276. {
  277. if (_writer == null)
  278. {
  279. logger.Info("Opening new IndexWriter");
  280. _writer = new IndexWriter(directory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED);
  281. }
  282. return _writer;
  283. }
  284. set
  285. {
  286. _writer = value;
  287. }
  288. }
  289. private static Dictionary<string, float> bonusTerms;
  290. public static void Init(string path, ILogger logger)
  291. {
  292. logger.Info("Lucene: Init");
  293. bonusTerms = new Dictionary<string, float>();
  294. bonusTerms.Add("Name", 2);
  295. bonusTerms.Add("Overview", 1);
  296. // Optimize the DB on initialization
  297. // TODO: Test whether this has..
  298. // Any effect what-so-ever (apart from initializing the indexwriter on the mainthread context, which makes things a whole lot easier)
  299. // Costs too much time
  300. // Is heavy on the CPU / Memory
  301. LuceneSearch.logger = logger;
  302. LuceneSearch.path = path;
  303. writer.Optimize();
  304. }
  305. private static StandardAnalyzer analyzer = new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_30);
  306. private static Searcher searcher = null;
  307. private static Document createDocument(BaseItem data)
  308. {
  309. Document doc = new Document();
  310. doc.Add(new Field("Id", data.Id.ToString(), Field.Store.YES, Field.Index.NO));
  311. doc.Add(new Field("Name", data.Name, Field.Store.YES, Field.Index.ANALYZED) { Boost = 2 });
  312. doc.Add(new Field("Overview", data.Overview != null ? data.Overview : "", Field.Store.YES, Field.Index.ANALYZED));
  313. return doc;
  314. }
  315. private static void Create(BaseItem item)
  316. {
  317. lock (lockOb)
  318. {
  319. try
  320. {
  321. if (searcher != null)
  322. {
  323. try
  324. {
  325. searcher.Dispose();
  326. }
  327. catch (Exception e)
  328. {
  329. logger.ErrorException("Error in Lucene while creating index (disposing alive searcher)", e, item);
  330. }
  331. searcher = null;
  332. }
  333. _removeFromLuceneIndex(item);
  334. _addToLuceneIndex(item);
  335. }
  336. catch (Exception e)
  337. {
  338. logger.ErrorException("Error in Lucene while creating index", e, item);
  339. }
  340. }
  341. }
  342. private static void _addToLuceneIndex(BaseItem data)
  343. {
  344. // Prevent double entries
  345. var doc = createDocument(data);
  346. writer.AddDocument(doc);
  347. }
  348. private static void _removeFromLuceneIndex(BaseItem data)
  349. {
  350. var query = new TermQuery(new Term("Id", data.Id.ToString()));
  351. writer.DeleteDocuments(query);
  352. }
  353. public static void AddUpdateLuceneIndex(IEnumerable<BaseItem> items)
  354. {
  355. foreach (var item in items)
  356. {
  357. logger.Info("Adding/Updating BaseItem " + item.Name + "(" + item.Id.ToString() + ") to/on Lucene Index");
  358. Create(item);
  359. }
  360. writer.Commit();
  361. writer.Flush(true, true, true);
  362. }
  363. public static void RemoveFromLuceneIndex(IEnumerable<BaseItem> items)
  364. {
  365. foreach (var item in items)
  366. {
  367. logger.Info("Removing BaseItem " + item.Name + "(" + item.Id.ToString() + ") from Lucene Index");
  368. _removeFromLuceneIndex(item);
  369. }
  370. writer.Commit();
  371. writer.Flush(true, true, true);
  372. }
  373. public static IEnumerable<BaseItem> Search(string searchQuery, int maxHits)
  374. {
  375. var results = new List<BaseItem>();
  376. lock (lockOb)
  377. {
  378. try
  379. {
  380. if (searcher == null)
  381. {
  382. searcher = new IndexSearcher(directory, true);
  383. }
  384. BooleanQuery finalQuery = new BooleanQuery();
  385. MultiFieldQueryParser parser = new MultiFieldQueryParser(Lucene.Net.Util.Version.LUCENE_30, new string[] { "Name", "Overview" }, analyzer, bonusTerms);
  386. string[] terms = searchQuery.Split(new[] { " " }, StringSplitOptions.RemoveEmptyEntries);
  387. foreach (string term in terms)
  388. finalQuery.Add(parser.Parse(term.Replace("~", "") + "~0.75"), Occur.SHOULD);
  389. foreach (string term in terms)
  390. finalQuery.Add(parser.Parse(term.Replace("*", "") + "*"), Occur.SHOULD);
  391. logger.Debug("Querying Lucene with query: " + finalQuery.ToString());
  392. long start = DateTime.Now.Ticks;
  393. var searchResult = searcher.Search(finalQuery, maxHits);
  394. foreach (var searchHit in searchResult.ScoreDocs)
  395. {
  396. Document hit = searcher.Doc(searchHit.Doc);
  397. results.Add(BaseItem.LibraryManager.GetItemById(Guid.Parse(hit.Get("Id"))));
  398. }
  399. long total = DateTime.Now.Ticks - start;
  400. float msTotal = (float)total / TimeSpan.TicksPerMillisecond;
  401. logger.Debug(searchResult.ScoreDocs.Length + " result" + (searchResult.ScoreDocs.Length == 1 ? "" : "s") + " in " + msTotal + " ms.");
  402. }
  403. catch (Exception e)
  404. {
  405. logger.ErrorException("Error while searching Lucene index", e);
  406. }
  407. }
  408. return results;
  409. }
  410. public static void CloseAll()
  411. {
  412. logger.Debug("Lucene: CloseAll");
  413. if (writer != null)
  414. {
  415. logger.Debug("Lucene: CloseAll - Writer is alive");
  416. writer.Flush(true, true, true);
  417. writer.Commit();
  418. writer.WaitForMerges();
  419. writer.Dispose();
  420. writer = null;
  421. }
  422. if (analyzer != null)
  423. {
  424. logger.Debug("Lucene: CloseAll - Analyzer is alive");
  425. analyzer.Close();
  426. analyzer.Dispose();
  427. analyzer = null;
  428. }
  429. if (searcher != null)
  430. {
  431. logger.Debug("Lucene: CloseAll - Searcher is alive");
  432. searcher.Dispose();
  433. searcher = null;
  434. }
  435. if (directory != null)
  436. {
  437. logger.Debug("Lucene: CloseAll - Directory is alive");
  438. directory.Dispose();
  439. directory = null;
  440. }
  441. }
  442. }
  443. }