LuceneSearchEngine.cs 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522
  1. using System.Text.RegularExpressions;
  2. using Lucene.Net.Analysis.Standard;
  3. using Lucene.Net.Documents;
  4. using Lucene.Net.Index;
  5. using Lucene.Net.QueryParsers;
  6. using Lucene.Net.Search;
  7. using Lucene.Net.Store;
  8. using MediaBrowser.Controller;
  9. using MediaBrowser.Controller.Entities;
  10. using MediaBrowser.Controller.Entities.Audio;
  11. using MediaBrowser.Controller.Library;
  12. using MediaBrowser.Model.Logging;
  13. using System;
  14. using System.Collections.Generic;
  15. using System.Linq;
  16. using System.Threading.Tasks;
  17. namespace MediaBrowser.Server.Implementations.Library
  18. {
  19. /// <summary>
  20. /// Class LuceneSearchEngine
  21. /// http://www.codeproject.com/Articles/320219/Lucene-Net-ultra-fast-search-for-MVC-or-WebForms
  22. /// </summary>
  23. public class LuceneSearchEngine : ILibrarySearchEngine, IDisposable
  24. {
  25. private readonly ILibraryManager _libraryManager;
  26. private readonly ILogger _logger;
  27. public LuceneSearchEngine(IServerApplicationPaths serverPaths, ILogManager logManager, ILibraryManager libraryManager)
  28. {
  29. _libraryManager = libraryManager;
  30. _logger = logManager.GetLogger("Lucene");
  31. //string luceneDbPath = serverPaths.DataPath + "\\SearchIndexDB";
  32. //if (!System.IO.Directory.Exists(luceneDbPath))
  33. // System.IO.Directory.CreateDirectory(luceneDbPath);
  34. //else if(File.Exists(luceneDbPath + "\\write.lock"))
  35. // File.Delete(luceneDbPath + "\\write.lock");
  36. //LuceneSearch.Init(luceneDbPath, _logger);
  37. //BaseItem.LibraryManager.LibraryChanged += LibraryChanged;
  38. }
  39. //public void LibraryChanged(object source, ChildrenChangedEventArgs changeInformation)
  40. //{
  41. // Task.Run(() =>
  42. // {
  43. // if (changeInformation.ItemsAdded.Count + changeInformation.ItemsUpdated.Count > 0)
  44. // {
  45. // LuceneSearch.AddUpdateLuceneIndex(changeInformation.ItemsAdded.Concat(changeInformation.ItemsUpdated));
  46. // }
  47. // if (changeInformation.ItemsRemoved.Count > 0)
  48. // {
  49. // LuceneSearch.RemoveFromLuceneIndex(changeInformation.ItemsRemoved);
  50. // }
  51. // });
  52. //}
  53. public void AddItemsToIndex(IEnumerable<BaseItem> items)
  54. {
  55. LuceneSearch.AddUpdateLuceneIndex(items);
  56. }
  57. /// <summary>
  58. /// Searches items and returns them in order of relevance.
  59. /// </summary>
  60. /// <param name="items">The items.</param>
  61. /// <param name="searchTerm">The search term.</param>
  62. /// <returns>IEnumerable{BaseItem}.</returns>
  63. /// <exception cref="System.ArgumentNullException">searchTerm</exception>
  64. public IEnumerable<BaseItem> Search(IEnumerable<BaseItem> items, string searchTerm)
  65. {
  66. if (string.IsNullOrEmpty(searchTerm))
  67. {
  68. throw new ArgumentNullException("searchTerm");
  69. }
  70. var hits = LuceneSearch.Search(searchTerm, items.Count());
  71. //return hits;
  72. return hits.Where(searchHit => items.Any(p => p.Id == searchHit.Id));
  73. }
  74. public void Dispose()
  75. {
  76. //BaseItem.LibraryManager.LibraryChanged -= LibraryChanged;
  77. //LuceneSearch.CloseAll();
  78. }
  79. /// <summary>
  80. /// Gets the search hints.
  81. /// </summary>
  82. /// <param name="inputItems">The input items.</param>
  83. /// <param name="searchTerm">The search term.</param>
  84. /// <returns>IEnumerable{SearchHintResult}.</returns>
  85. /// <exception cref="System.ArgumentNullException">searchTerm</exception>
  86. public async Task<IEnumerable<SearchHintInfo>> GetSearchHints(IEnumerable<BaseItem> inputItems, string searchTerm)
  87. {
  88. if (string.IsNullOrEmpty(searchTerm))
  89. {
  90. throw new ArgumentNullException("searchTerm");
  91. }
  92. var terms = GetWords(searchTerm);
  93. var hints = new List<Tuple<BaseItem, string, int>>();
  94. var items = inputItems.Where(i => !(i is MusicArtist)).ToList();
  95. hints.AddRange(items.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
  123. var genres = items.SelectMany(i => i.Genres)
  124. .Where(i => !string.IsNullOrEmpty(i))
  125. .Distinct(StringComparer.OrdinalIgnoreCase)
  126. .ToList();
  127. foreach (var item in genres)
  128. {
  129. var index = GetIndex(item, searchTerm, terms);
  130. if (index.Item2 != -1)
  131. {
  132. try
  133. {
  134. var genre = await _libraryManager.GetGenre(item).ConfigureAwait(false);
  135. hints.Add(new Tuple<BaseItem, string, int>(genre, index.Item1, index.Item2));
  136. }
  137. catch (Exception ex)
  138. {
  139. _logger.ErrorException("Error getting {0}", ex, item);
  140. }
  141. }
  142. }
  143. // Find studios
  144. var studios = items.SelectMany(i => i.Studios)
  145. .Where(i => !string.IsNullOrEmpty(i))
  146. .Distinct(StringComparer.OrdinalIgnoreCase)
  147. .ToList();
  148. foreach (var item in studios)
  149. {
  150. var index = GetIndex(item, searchTerm, terms);
  151. if (index.Item2 != -1)
  152. {
  153. try
  154. {
  155. var studio = await _libraryManager.GetStudio(item).ConfigureAwait(false);
  156. hints.Add(new Tuple<BaseItem, string, int>(studio, index.Item1, index.Item2));
  157. }
  158. catch (Exception ex)
  159. {
  160. _logger.ErrorException("Error getting {0}", ex, item);
  161. }
  162. }
  163. }
  164. // Find persons
  165. var persons = items.SelectMany(i => i.People)
  166. .Select(i => i.Name)
  167. .Where(i => !string.IsNullOrEmpty(i))
  168. .Distinct(StringComparer.OrdinalIgnoreCase)
  169. .ToList();
  170. foreach (var item in persons)
  171. {
  172. var index = GetIndex(item, searchTerm, terms);
  173. if (index.Item2 != -1)
  174. {
  175. try
  176. {
  177. var person = await _libraryManager.GetPerson(item).ConfigureAwait(false);
  178. hints.Add(new Tuple<BaseItem, string, int>(person, index.Item1, index.Item2));
  179. }
  180. catch (Exception ex)
  181. {
  182. _logger.ErrorException("Error getting {0}", ex, item);
  183. }
  184. }
  185. }
  186. return hints.Where(i => i.Item3 >= 0).OrderBy(i => i.Item3).Select(i => new SearchHintInfo
  187. {
  188. Item = i.Item1,
  189. MatchedTerm = i.Item2
  190. });
  191. }
  192. /// <summary>
  193. /// Gets the index.
  194. /// </summary>
  195. /// <param name="input">The input.</param>
  196. /// <param name="searchInput">The search input.</param>
  197. /// <param name="searchWords">The search input.</param>
  198. /// <returns>System.Int32.</returns>
  199. private Tuple<string, int> GetIndex(string input, string searchInput, string[] searchWords)
  200. {
  201. if (string.Equals(input, searchInput, StringComparison.OrdinalIgnoreCase))
  202. {
  203. return new Tuple<string, int>(searchInput, 0);
  204. }
  205. var index = input.IndexOf(searchInput, StringComparison.OrdinalIgnoreCase);
  206. if (index == 0)
  207. {
  208. return new Tuple<string, int>(searchInput, 1);
  209. }
  210. if (index > 0)
  211. {
  212. return new Tuple<string, int>(searchInput, 2);
  213. }
  214. var items = GetWords(input);
  215. for (var i = 0; i < searchWords.Length; i++)
  216. {
  217. var searchTerm = searchWords[i];
  218. for (var j = 0; j < items.Length; j++)
  219. {
  220. var item = items[j];
  221. if (string.Equals(item, searchTerm, StringComparison.OrdinalIgnoreCase))
  222. {
  223. return new Tuple<string, int>(searchTerm, 3 + (i + 1) * (j + 1));
  224. }
  225. index = item.IndexOf(searchTerm, StringComparison.OrdinalIgnoreCase);
  226. if (index == 0)
  227. {
  228. return new Tuple<string, int>(searchTerm, 4 + (i + 1) * (j + 1));
  229. }
  230. if (index > 0)
  231. {
  232. return new Tuple<string, int>(searchTerm, 5 + (i + 1) * (j + 1));
  233. }
  234. }
  235. }
  236. return new Tuple<string, int>(null, -1);
  237. }
  238. /// <summary>
  239. /// Gets the words.
  240. /// </summary>
  241. /// <param name="term">The term.</param>
  242. /// <returns>System.String[][].</returns>
  243. private string[] GetWords(string term)
  244. {
  245. return term.Split().Where(i => !string.IsNullOrWhiteSpace(i)).ToArray();
  246. }
  247. }
  248. public static class LuceneSearch
  249. {
  250. private static ILogger logger;
  251. private static string path;
  252. private static object lockOb = new object();
  253. private static FSDirectory _directory;
  254. private static FSDirectory directory
  255. {
  256. get
  257. {
  258. if (_directory == null)
  259. {
  260. logger.Info("Opening new Directory: " + path);
  261. _directory = FSDirectory.Open(path);
  262. }
  263. return _directory;
  264. }
  265. set
  266. {
  267. _directory = value;
  268. }
  269. }
  270. private static IndexWriter _writer;
  271. private static IndexWriter writer
  272. {
  273. get
  274. {
  275. if (_writer == null)
  276. {
  277. logger.Info("Opening new IndexWriter");
  278. _writer = new IndexWriter(directory, analyzer, IndexWriter.MaxFieldLength.UNLIMITED);
  279. }
  280. return _writer;
  281. }
  282. set
  283. {
  284. _writer = value;
  285. }
  286. }
  287. private static Dictionary<string, float> bonusTerms;
  288. public static void Init(string path, ILogger logger)
  289. {
  290. logger.Info("Lucene: Init");
  291. bonusTerms = new Dictionary<string, float>();
  292. bonusTerms.Add("Name", 2);
  293. bonusTerms.Add("Overview", 1);
  294. // Optimize the DB on initialization
  295. // TODO: Test whether this has..
  296. // Any effect what-so-ever (apart from initializing the indexwriter on the mainthread context, which makes things a whole lot easier)
  297. // Costs too much time
  298. // Is heavy on the CPU / Memory
  299. LuceneSearch.logger = logger;
  300. LuceneSearch.path = path;
  301. writer.Optimize();
  302. }
  303. private static StandardAnalyzer analyzer = new StandardAnalyzer(Lucene.Net.Util.Version.LUCENE_30);
  304. private static Searcher searcher = null;
  305. private static Document createDocument(BaseItem data)
  306. {
  307. Document doc = new Document();
  308. doc.Add(new Field("Id", data.Id.ToString(), Field.Store.YES, Field.Index.NO));
  309. doc.Add(new Field("Name", data.Name, Field.Store.YES, Field.Index.ANALYZED) { Boost = 2 });
  310. doc.Add(new Field("Overview", data.Overview != null ? data.Overview : "", Field.Store.YES, Field.Index.ANALYZED));
  311. return doc;
  312. }
  313. private static void Create(BaseItem item)
  314. {
  315. lock (lockOb)
  316. {
  317. try
  318. {
  319. if (searcher != null)
  320. {
  321. try
  322. {
  323. searcher.Dispose();
  324. }
  325. catch (Exception e)
  326. {
  327. logger.ErrorException("Error in Lucene while creating index (disposing alive searcher)", e, item);
  328. }
  329. searcher = null;
  330. }
  331. _removeFromLuceneIndex(item);
  332. _addToLuceneIndex(item);
  333. }
  334. catch (Exception e)
  335. {
  336. logger.ErrorException("Error in Lucene while creating index", e, item);
  337. }
  338. }
  339. }
  340. private static void _addToLuceneIndex(BaseItem data)
  341. {
  342. // Prevent double entries
  343. var doc = createDocument(data);
  344. writer.AddDocument(doc);
  345. }
  346. private static void _removeFromLuceneIndex(BaseItem data)
  347. {
  348. var query = new TermQuery(new Term("Id", data.Id.ToString()));
  349. writer.DeleteDocuments(query);
  350. }
  351. public static void AddUpdateLuceneIndex(IEnumerable<BaseItem> items)
  352. {
  353. foreach (var item in items)
  354. {
  355. logger.Info("Adding/Updating BaseItem " + item.Name + "(" + item.Id.ToString() + ") to/on Lucene Index");
  356. Create(item);
  357. }
  358. writer.Commit();
  359. writer.Flush(true, true, true);
  360. }
  361. public static void RemoveFromLuceneIndex(IEnumerable<BaseItem> items)
  362. {
  363. foreach (var item in items)
  364. {
  365. logger.Info("Removing BaseItem " + item.Name + "(" + item.Id.ToString() + ") from Lucene Index");
  366. _removeFromLuceneIndex(item);
  367. }
  368. writer.Commit();
  369. writer.Flush(true, true, true);
  370. }
  371. public static IEnumerable<BaseItem> Search(string searchQuery, int maxHits)
  372. {
  373. var results = new List<BaseItem>();
  374. lock (lockOb)
  375. {
  376. try
  377. {
  378. if (searcher == null)
  379. {
  380. searcher = new IndexSearcher(directory, true);
  381. }
  382. BooleanQuery finalQuery = new BooleanQuery();
  383. MultiFieldQueryParser parser = new MultiFieldQueryParser(Lucene.Net.Util.Version.LUCENE_30, new string[] { "Name", "Overview" }, analyzer, bonusTerms);
  384. string[] terms = searchQuery.Split(new[] { " " }, StringSplitOptions.RemoveEmptyEntries);
  385. foreach (string term in terms)
  386. finalQuery.Add(parser.Parse(term.Replace("~", "") + "~0.75"), Occur.SHOULD);
  387. foreach (string term in terms)
  388. finalQuery.Add(parser.Parse(term.Replace("*", "") + "*"), Occur.SHOULD);
  389. logger.Debug("Querying Lucene with query: " + finalQuery.ToString());
  390. long start = DateTime.Now.Ticks;
  391. var searchResult = searcher.Search(finalQuery, maxHits);
  392. foreach (var searchHit in searchResult.ScoreDocs)
  393. {
  394. Document hit = searcher.Doc(searchHit.Doc);
  395. results.Add(BaseItem.LibraryManager.GetItemById(Guid.Parse(hit.Get("Id"))));
  396. }
  397. long total = DateTime.Now.Ticks - start;
  398. float msTotal = (float)total / TimeSpan.TicksPerMillisecond;
  399. logger.Debug(searchResult.ScoreDocs.Length + " result" + (searchResult.ScoreDocs.Length == 1 ? "" : "s") + " in " + msTotal + " ms.");
  400. }
  401. catch (Exception e)
  402. {
  403. logger.ErrorException("Error while searching Lucene index", e);
  404. }
  405. }
  406. return results;
  407. }
  408. public static void CloseAll()
  409. {
  410. logger.Debug("Lucene: CloseAll");
  411. if (writer != null)
  412. {
  413. logger.Debug("Lucene: CloseAll - Writer is alive");
  414. writer.Flush(true, true, true);
  415. writer.Commit();
  416. writer.WaitForMerges();
  417. writer.Dispose();
  418. writer = null;
  419. }
  420. if (analyzer != null)
  421. {
  422. logger.Debug("Lucene: CloseAll - Analyzer is alive");
  423. analyzer.Close();
  424. analyzer.Dispose();
  425. analyzer = null;
  426. }
  427. if (searcher != null)
  428. {
  429. logger.Debug("Lucene: CloseAll - Searcher is alive");
  430. searcher.Dispose();
  431. searcher = null;
  432. }
  433. if (directory != null)
  434. {
  435. logger.Debug("Lucene: CloseAll - Directory is alive");
  436. directory.Dispose();
  437. directory = null;
  438. }
  439. }
  440. }
  441. }