BaseItemXmlParser.cs 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700
  1. using System.Globalization;
  2. using MediaBrowser.Controller.Entities;
  3. using MediaBrowser.Model.Entities;
  4. using MediaBrowser.Model.Logging;
  5. using System;
  6. using System.Collections.Generic;
  7. using System.Linq;
  8. using System.Threading;
  9. using System.Xml;
  10. namespace MediaBrowser.Controller.Providers
  11. {
  12. /// <summary>
  13. /// Provides a base class for parsing metadata xml
  14. /// </summary>
  15. /// <typeparam name="T"></typeparam>
  16. public class BaseItemXmlParser<T>
  17. where T : BaseItem, new()
  18. {
  19. /// <summary>
  20. /// The logger
  21. /// </summary>
  22. protected ILogger Logger { get; private set; }
  23. /// <summary>
  24. /// Initializes a new instance of the <see cref="BaseItemXmlParser{T}" /> class.
  25. /// </summary>
  26. /// <param name="logger">The logger.</param>
  27. public BaseItemXmlParser(ILogger logger)
  28. {
  29. Logger = logger;
  30. }
  31. /// <summary>
  32. /// Fetches metadata for an item from one xml file
  33. /// </summary>
  34. /// <param name="item">The item.</param>
  35. /// <param name="metadataFile">The metadata file.</param>
  36. /// <param name="cancellationToken">The cancellation token.</param>
  37. /// <exception cref="System.ArgumentNullException"></exception>
  38. public void Fetch(T item, string metadataFile, CancellationToken cancellationToken)
  39. {
  40. if (item == null)
  41. {
  42. throw new ArgumentNullException();
  43. }
  44. if (string.IsNullOrEmpty(metadataFile))
  45. {
  46. throw new ArgumentNullException();
  47. }
  48. var settings = new XmlReaderSettings
  49. {
  50. CheckCharacters = false,
  51. IgnoreProcessingInstructions = true,
  52. IgnoreComments = true,
  53. ValidationType = ValidationType.None
  54. };
  55. item.Taglines.Clear();
  56. item.Studios.Clear();
  57. item.Genres.Clear();
  58. item.People.Clear();
  59. // Use XmlReader for best performance
  60. using (var reader = XmlReader.Create(metadataFile, settings))
  61. {
  62. reader.MoveToContent();
  63. // Loop through each element
  64. while (reader.Read())
  65. {
  66. cancellationToken.ThrowIfCancellationRequested();
  67. if (reader.NodeType == XmlNodeType.Element)
  68. {
  69. FetchDataFromXmlNode(reader, item);
  70. }
  71. }
  72. }
  73. }
  74. private readonly CultureInfo _usCulture = new CultureInfo("en-US");
  75. /// <summary>
  76. /// Fetches metadata from one Xml Element
  77. /// </summary>
  78. /// <param name="reader">The reader.</param>
  79. /// <param name="item">The item.</param>
  80. protected virtual void FetchDataFromXmlNode(XmlReader reader, T item)
  81. {
  82. switch (reader.Name)
  83. {
  84. // DateCreated
  85. case "Added":
  86. DateTime added;
  87. if (DateTime.TryParse(reader.ReadElementContentAsString() ?? string.Empty, out added))
  88. {
  89. item.DateCreated = added.ToUniversalTime();
  90. }
  91. break;
  92. case "LocalTitle":
  93. item.Name = reader.ReadElementContentAsString();
  94. break;
  95. case "Type":
  96. {
  97. var type = reader.ReadElementContentAsString();
  98. if (!string.IsNullOrWhiteSpace(type) && !type.Equals("none", StringComparison.OrdinalIgnoreCase))
  99. {
  100. item.DisplayMediaType = type;
  101. }
  102. break;
  103. }
  104. case "Budget":
  105. {
  106. var text = reader.ReadElementContentAsString();
  107. double value;
  108. if (double.TryParse(text, NumberStyles.Any, _usCulture, out value))
  109. {
  110. item.Budget = value;
  111. }
  112. break;
  113. }
  114. case "Revenue":
  115. {
  116. var text = reader.ReadElementContentAsString();
  117. double value;
  118. if (double.TryParse(text, NumberStyles.Any, _usCulture, out value))
  119. {
  120. item.Revenue = value;
  121. }
  122. break;
  123. }
  124. case "SortTitle":
  125. item.ForcedSortName = reader.ReadElementContentAsString();
  126. break;
  127. case "Overview":
  128. case "Description":
  129. item.Overview = reader.ReadElementContentAsString();
  130. break;
  131. case "TagLine":
  132. {
  133. var tagline = reader.ReadElementContentAsString();
  134. if (!string.IsNullOrWhiteSpace(tagline))
  135. {
  136. item.AddTagline(tagline);
  137. }
  138. break;
  139. }
  140. case "Website":
  141. {
  142. var val = reader.ReadElementContentAsString();
  143. if (!string.IsNullOrWhiteSpace(val))
  144. {
  145. item.HomePageUrl = val;
  146. }
  147. break;
  148. }
  149. case "TagLines":
  150. {
  151. FetchFromTaglinesNode(reader.ReadSubtree(), item);
  152. break;
  153. }
  154. case "ContentRating":
  155. case "certification":
  156. case "MPAARating":
  157. {
  158. var rating = reader.ReadElementContentAsString();
  159. if (!string.IsNullOrWhiteSpace(rating))
  160. {
  161. item.OfficialRating = rating;
  162. }
  163. break;
  164. }
  165. case "CustomRating":
  166. {
  167. var val = reader.ReadElementContentAsString();
  168. if (!string.IsNullOrWhiteSpace(val))
  169. {
  170. item.CustomRating = val;
  171. }
  172. break;
  173. }
  174. case "Runtime":
  175. case "RunningTime":
  176. {
  177. var text = reader.ReadElementContentAsString();
  178. if (!string.IsNullOrWhiteSpace(text))
  179. {
  180. int runtime;
  181. if (int.TryParse(text.Split(' ')[0], out runtime))
  182. {
  183. item.RunTimeTicks = TimeSpan.FromMinutes(runtime).Ticks;
  184. }
  185. }
  186. break;
  187. }
  188. case "Genre":
  189. {
  190. foreach (var name in SplitNames(reader.ReadElementContentAsString()))
  191. {
  192. if (string.IsNullOrWhiteSpace(name))
  193. {
  194. continue;
  195. }
  196. item.AddGenre(name);
  197. }
  198. break;
  199. }
  200. case "AspectRatio":
  201. {
  202. var val = reader.ReadElementContentAsString();
  203. if (!string.IsNullOrWhiteSpace(val))
  204. {
  205. item.AspectRatio = val;
  206. }
  207. break;
  208. }
  209. case "Network":
  210. {
  211. foreach (var name in SplitNames(reader.ReadElementContentAsString()))
  212. {
  213. if (string.IsNullOrWhiteSpace(name))
  214. {
  215. continue;
  216. }
  217. item.AddStudio(name);
  218. }
  219. break;
  220. }
  221. case "Director":
  222. {
  223. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v, Type = PersonType.Director }))
  224. {
  225. if (string.IsNullOrWhiteSpace(p.Name))
  226. {
  227. continue;
  228. }
  229. item.AddPerson(p);
  230. }
  231. break;
  232. }
  233. case "Writer":
  234. {
  235. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v, Type = PersonType.Writer }))
  236. {
  237. if (string.IsNullOrWhiteSpace(p.Name))
  238. {
  239. continue;
  240. }
  241. item.AddPerson(p);
  242. }
  243. break;
  244. }
  245. case "Actors":
  246. {
  247. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v.Trim(), Type = PersonType.Actor }))
  248. {
  249. if (string.IsNullOrWhiteSpace(p.Name))
  250. {
  251. continue;
  252. }
  253. item.AddPerson(p);
  254. }
  255. break;
  256. }
  257. case "GuestStars":
  258. {
  259. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v.Trim(), Type = PersonType.GuestStar }))
  260. {
  261. if (string.IsNullOrWhiteSpace(p.Name))
  262. {
  263. continue;
  264. }
  265. item.AddPerson(p);
  266. }
  267. break;
  268. }
  269. case "Trailer":
  270. {
  271. var val = reader.ReadElementContentAsString();
  272. if (!string.IsNullOrWhiteSpace(val))
  273. {
  274. item.AddTrailerUrl(val);
  275. }
  276. break;
  277. }
  278. case "ProductionYear":
  279. {
  280. var val = reader.ReadElementContentAsString();
  281. if (!string.IsNullOrWhiteSpace(val))
  282. {
  283. int ProductionYear;
  284. if (int.TryParse(val, out ProductionYear) && ProductionYear > 1850)
  285. {
  286. item.ProductionYear = ProductionYear;
  287. }
  288. }
  289. break;
  290. }
  291. case "Rating":
  292. case "IMDBrating":
  293. {
  294. var rating = reader.ReadElementContentAsString();
  295. if (!string.IsNullOrWhiteSpace(rating))
  296. {
  297. float val;
  298. if (float.TryParse(rating, out val))
  299. {
  300. item.CommunityRating = val;
  301. }
  302. }
  303. break;
  304. }
  305. case "FirstAired":
  306. {
  307. var firstAired = reader.ReadElementContentAsString();
  308. if (!string.IsNullOrWhiteSpace(firstAired))
  309. {
  310. DateTime airDate;
  311. if (DateTime.TryParse(firstAired, out airDate) && airDate.Year > 1850)
  312. {
  313. item.PremiereDate = airDate.ToUniversalTime();
  314. item.ProductionYear = airDate.Year;
  315. }
  316. }
  317. break;
  318. }
  319. case "TMDbId":
  320. var tmdb = reader.ReadElementContentAsString();
  321. if (!string.IsNullOrWhiteSpace(tmdb))
  322. {
  323. item.SetProviderId(MetadataProviders.Tmdb, tmdb);
  324. }
  325. break;
  326. case "TVcomId":
  327. var TVcomId = reader.ReadElementContentAsString();
  328. if (!string.IsNullOrWhiteSpace(TVcomId))
  329. {
  330. item.SetProviderId(MetadataProviders.Tvcom, TVcomId);
  331. }
  332. break;
  333. case "IMDB_ID":
  334. case "IMDB":
  335. case "IMDbId":
  336. var IMDbId = reader.ReadElementContentAsString();
  337. if (!string.IsNullOrWhiteSpace(IMDbId))
  338. {
  339. item.SetProviderId(MetadataProviders.Imdb, IMDbId);
  340. }
  341. break;
  342. case "Genres":
  343. FetchFromGenresNode(reader.ReadSubtree(), item);
  344. break;
  345. case "Persons":
  346. FetchDataFromPersonsNode(reader.ReadSubtree(), item);
  347. break;
  348. case "ParentalRating":
  349. FetchFromParentalRatingNode(reader.ReadSubtree(), item);
  350. break;
  351. case "Studios":
  352. FetchFromStudiosNode(reader.ReadSubtree(), item);
  353. break;
  354. default:
  355. reader.Skip();
  356. break;
  357. }
  358. }
  359. /// <summary>
  360. /// Fetches from taglines node.
  361. /// </summary>
  362. /// <param name="reader">The reader.</param>
  363. /// <param name="item">The item.</param>
  364. private void FetchFromTaglinesNode(XmlReader reader, T item)
  365. {
  366. reader.MoveToContent();
  367. while (reader.Read())
  368. {
  369. if (reader.NodeType == XmlNodeType.Element)
  370. {
  371. switch (reader.Name)
  372. {
  373. case "Tagline":
  374. {
  375. var val = reader.ReadElementContentAsString();
  376. if (!string.IsNullOrWhiteSpace(val))
  377. {
  378. item.AddTagline(val);
  379. }
  380. break;
  381. }
  382. default:
  383. reader.Skip();
  384. break;
  385. }
  386. }
  387. }
  388. }
  389. /// <summary>
  390. /// Fetches from genres node.
  391. /// </summary>
  392. /// <param name="reader">The reader.</param>
  393. /// <param name="item">The item.</param>
  394. private void FetchFromGenresNode(XmlReader reader, T item)
  395. {
  396. reader.MoveToContent();
  397. while (reader.Read())
  398. {
  399. if (reader.NodeType == XmlNodeType.Element)
  400. {
  401. switch (reader.Name)
  402. {
  403. case "Genre":
  404. {
  405. var genre = reader.ReadElementContentAsString();
  406. if (!string.IsNullOrWhiteSpace(genre))
  407. {
  408. item.AddGenre(genre);
  409. }
  410. break;
  411. }
  412. default:
  413. reader.Skip();
  414. break;
  415. }
  416. }
  417. }
  418. }
  419. /// <summary>
  420. /// Fetches the data from persons node.
  421. /// </summary>
  422. /// <param name="reader">The reader.</param>
  423. /// <param name="item">The item.</param>
  424. private void FetchDataFromPersonsNode(XmlReader reader, T item)
  425. {
  426. reader.MoveToContent();
  427. while (reader.Read())
  428. {
  429. if (reader.NodeType == XmlNodeType.Element)
  430. {
  431. switch (reader.Name)
  432. {
  433. case "Person":
  434. {
  435. foreach (var person in GetPersonsFromXmlNode(reader.ReadSubtree()))
  436. {
  437. item.AddPerson(person);
  438. }
  439. break;
  440. }
  441. default:
  442. reader.Skip();
  443. break;
  444. }
  445. }
  446. }
  447. }
  448. /// <summary>
  449. /// Fetches from studios node.
  450. /// </summary>
  451. /// <param name="reader">The reader.</param>
  452. /// <param name="item">The item.</param>
  453. private void FetchFromStudiosNode(XmlReader reader, T item)
  454. {
  455. reader.MoveToContent();
  456. while (reader.Read())
  457. {
  458. if (reader.NodeType == XmlNodeType.Element)
  459. {
  460. switch (reader.Name)
  461. {
  462. case "Studio":
  463. {
  464. var studio = reader.ReadElementContentAsString();
  465. if (!string.IsNullOrWhiteSpace(studio))
  466. {
  467. item.AddStudio(studio);
  468. }
  469. break;
  470. }
  471. default:
  472. reader.Skip();
  473. break;
  474. }
  475. }
  476. }
  477. }
  478. /// <summary>
  479. /// Fetches from parental rating node.
  480. /// </summary>
  481. /// <param name="reader">The reader.</param>
  482. /// <param name="item">The item.</param>
  483. private void FetchFromParentalRatingNode(XmlReader reader, T item)
  484. {
  485. reader.MoveToContent();
  486. while (reader.Read())
  487. {
  488. if (reader.NodeType == XmlNodeType.Element)
  489. {
  490. switch (reader.Name)
  491. {
  492. case "Value":
  493. {
  494. var ratingString = reader.ReadElementContentAsString();
  495. int rating = 7;
  496. if (!string.IsNullOrWhiteSpace(ratingString))
  497. {
  498. int.TryParse(ratingString, out rating);
  499. }
  500. switch (rating)
  501. {
  502. case -1:
  503. item.OfficialRating = "NR";
  504. break;
  505. case 0:
  506. item.OfficialRating = "UR";
  507. break;
  508. case 1:
  509. item.OfficialRating = "G";
  510. break;
  511. case 3:
  512. item.OfficialRating = "PG";
  513. break;
  514. case 4:
  515. item.OfficialRating = "PG-13";
  516. break;
  517. case 5:
  518. item.OfficialRating = "NC-17";
  519. break;
  520. case 6:
  521. item.OfficialRating = "R";
  522. break;
  523. }
  524. break;
  525. }
  526. default:
  527. reader.Skip();
  528. break;
  529. }
  530. }
  531. }
  532. }
  533. /// <summary>
  534. /// Gets the persons from XML node.
  535. /// </summary>
  536. /// <param name="reader">The reader.</param>
  537. /// <returns>IEnumerable{PersonInfo}.</returns>
  538. private IEnumerable<PersonInfo> GetPersonsFromXmlNode(XmlReader reader)
  539. {
  540. var names = new List<string>();
  541. var type = string.Empty;
  542. var role = string.Empty;
  543. reader.MoveToContent();
  544. while (reader.Read())
  545. {
  546. if (reader.NodeType == XmlNodeType.Element)
  547. {
  548. switch (reader.Name)
  549. {
  550. case "Name":
  551. names.AddRange(SplitNames(reader.ReadElementContentAsString()));
  552. break;
  553. case "Type":
  554. {
  555. var val = reader.ReadElementContentAsString();
  556. if (!string.IsNullOrWhiteSpace(val))
  557. {
  558. type = val;
  559. }
  560. break;
  561. }
  562. case "Role":
  563. {
  564. var val = reader.ReadElementContentAsString();
  565. if (!string.IsNullOrWhiteSpace(val))
  566. {
  567. role = val;
  568. }
  569. break;
  570. }
  571. default:
  572. reader.Skip();
  573. break;
  574. }
  575. }
  576. }
  577. return names.Select(n => new PersonInfo { Name = n, Role = role, Type = type });
  578. }
  579. /// <summary>
  580. /// Used to split names of comma or pipe delimeted genres and people
  581. /// </summary>
  582. /// <param name="value">The value.</param>
  583. /// <returns>IEnumerable{System.String}.</returns>
  584. private IEnumerable<string> SplitNames(string value)
  585. {
  586. value = value ?? string.Empty;
  587. // Only split by comma if there is no pipe in the string
  588. // We have to be careful to not split names like Matthew, Jr.
  589. var separator = value.IndexOf('|') == -1 ? ',' : '|';
  590. value = value.Trim().Trim(separator);
  591. return string.IsNullOrWhiteSpace(value) ? new string[] { } : Split(value, separator, StringSplitOptions.RemoveEmptyEntries);
  592. }
  593. /// <summary>
  594. /// Provides an additional overload for string.split
  595. /// </summary>
  596. /// <param name="val">The val.</param>
  597. /// <param name="separator">The separator.</param>
  598. /// <param name="options">The options.</param>
  599. /// <returns>System.String[][].</returns>
  600. private static string[] Split(string val, char separator, StringSplitOptions options)
  601. {
  602. return val.Split(new[] { separator }, options);
  603. }
  604. }
  605. }