BaseItemXmlParser.cs 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692
  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. // Use XmlReader for best performance
  56. using (var reader = XmlReader.Create(metadataFile, settings))
  57. {
  58. reader.MoveToContent();
  59. // Loop through each element
  60. while (reader.Read())
  61. {
  62. cancellationToken.ThrowIfCancellationRequested();
  63. if (reader.NodeType == XmlNodeType.Element)
  64. {
  65. FetchDataFromXmlNode(reader, item);
  66. }
  67. }
  68. }
  69. }
  70. private readonly CultureInfo _usCulture = new CultureInfo("en-US");
  71. /// <summary>
  72. /// Fetches metadata from one Xml Element
  73. /// </summary>
  74. /// <param name="reader">The reader.</param>
  75. /// <param name="item">The item.</param>
  76. protected virtual void FetchDataFromXmlNode(XmlReader reader, T item)
  77. {
  78. switch (reader.Name)
  79. {
  80. // DateCreated
  81. case "Added":
  82. DateTime added;
  83. if (DateTime.TryParse(reader.ReadElementContentAsString() ?? string.Empty, out added))
  84. {
  85. item.DateCreated = added.ToUniversalTime();
  86. }
  87. break;
  88. case "LocalTitle":
  89. item.Name = reader.ReadElementContentAsString();
  90. break;
  91. case "Type":
  92. {
  93. var type = reader.ReadElementContentAsString();
  94. if (!string.IsNullOrWhiteSpace(type) && !type.Equals("none", StringComparison.OrdinalIgnoreCase))
  95. {
  96. item.DisplayMediaType = type;
  97. }
  98. break;
  99. }
  100. case "Budget":
  101. {
  102. var text = reader.ReadElementContentAsString();
  103. double value;
  104. if (double.TryParse(text, NumberStyles.Any, _usCulture, out value))
  105. {
  106. item.Budget = value;
  107. }
  108. break;
  109. }
  110. case "Revenue":
  111. {
  112. var text = reader.ReadElementContentAsString();
  113. double value;
  114. if (double.TryParse(text, NumberStyles.Any, _usCulture, out value))
  115. {
  116. item.Revenue = value;
  117. }
  118. break;
  119. }
  120. case "SortTitle":
  121. item.ForcedSortName = reader.ReadElementContentAsString();
  122. break;
  123. case "Overview":
  124. case "Description":
  125. item.Overview = reader.ReadElementContentAsString();
  126. break;
  127. case "TagLine":
  128. {
  129. var tagline = reader.ReadElementContentAsString();
  130. if (!string.IsNullOrWhiteSpace(tagline))
  131. {
  132. item.AddTagline(tagline);
  133. }
  134. break;
  135. }
  136. case "Website":
  137. {
  138. var val = reader.ReadElementContentAsString();
  139. if (!string.IsNullOrWhiteSpace(val))
  140. {
  141. item.HomePageUrl = val;
  142. }
  143. break;
  144. }
  145. case "TagLines":
  146. {
  147. FetchFromTaglinesNode(reader.ReadSubtree(), item);
  148. break;
  149. }
  150. case "ContentRating":
  151. case "certification":
  152. case "MPAARating":
  153. {
  154. var rating = reader.ReadElementContentAsString();
  155. if (!string.IsNullOrWhiteSpace(rating))
  156. {
  157. item.OfficialRating = rating;
  158. }
  159. break;
  160. }
  161. case "CustomRating":
  162. {
  163. var val = reader.ReadElementContentAsString();
  164. if (!string.IsNullOrWhiteSpace(val))
  165. {
  166. item.CustomRating = val;
  167. }
  168. break;
  169. }
  170. case "Runtime":
  171. case "RunningTime":
  172. {
  173. var text = reader.ReadElementContentAsString();
  174. if (!string.IsNullOrWhiteSpace(text))
  175. {
  176. int runtime;
  177. if (int.TryParse(text.Split(' ')[0], out runtime))
  178. {
  179. item.RunTimeTicks = TimeSpan.FromMinutes(runtime).Ticks;
  180. }
  181. }
  182. break;
  183. }
  184. case "Genre":
  185. {
  186. foreach (var name in SplitNames(reader.ReadElementContentAsString()))
  187. {
  188. if (string.IsNullOrWhiteSpace(name))
  189. {
  190. continue;
  191. }
  192. item.AddGenre(name);
  193. }
  194. break;
  195. }
  196. case "AspectRatio":
  197. {
  198. var val = reader.ReadElementContentAsString();
  199. if (!string.IsNullOrWhiteSpace(val))
  200. {
  201. item.AspectRatio = val;
  202. }
  203. break;
  204. }
  205. case "Network":
  206. {
  207. foreach (var name in SplitNames(reader.ReadElementContentAsString()))
  208. {
  209. if (string.IsNullOrWhiteSpace(name))
  210. {
  211. continue;
  212. }
  213. item.AddStudio(name);
  214. }
  215. break;
  216. }
  217. case "Director":
  218. {
  219. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v, Type = PersonType.Director }))
  220. {
  221. if (string.IsNullOrWhiteSpace(p.Name))
  222. {
  223. continue;
  224. }
  225. item.AddPerson(p);
  226. }
  227. break;
  228. }
  229. case "Writer":
  230. {
  231. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v, Type = PersonType.Writer }))
  232. {
  233. if (string.IsNullOrWhiteSpace(p.Name))
  234. {
  235. continue;
  236. }
  237. item.AddPerson(p);
  238. }
  239. break;
  240. }
  241. case "Actors":
  242. {
  243. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v.Trim(), Type = PersonType.Actor }))
  244. {
  245. if (string.IsNullOrWhiteSpace(p.Name))
  246. {
  247. continue;
  248. }
  249. item.AddPerson(p);
  250. }
  251. break;
  252. }
  253. case "GuestStars":
  254. {
  255. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v.Trim(), Type = PersonType.GuestStar }))
  256. {
  257. if (string.IsNullOrWhiteSpace(p.Name))
  258. {
  259. continue;
  260. }
  261. item.AddPerson(p);
  262. }
  263. break;
  264. }
  265. case "Trailer":
  266. {
  267. var val = reader.ReadElementContentAsString();
  268. if (!string.IsNullOrWhiteSpace(val))
  269. {
  270. item.AddTrailerUrl(val);
  271. }
  272. break;
  273. }
  274. case "ProductionYear":
  275. {
  276. var val = reader.ReadElementContentAsString();
  277. if (!string.IsNullOrWhiteSpace(val))
  278. {
  279. int ProductionYear;
  280. if (int.TryParse(val, out ProductionYear) && ProductionYear > 1850)
  281. {
  282. item.ProductionYear = ProductionYear;
  283. }
  284. }
  285. break;
  286. }
  287. case "Rating":
  288. case "IMDBrating":
  289. {
  290. var rating = reader.ReadElementContentAsString();
  291. if (!string.IsNullOrWhiteSpace(rating))
  292. {
  293. float val;
  294. if (float.TryParse(rating, out val))
  295. {
  296. item.CommunityRating = val;
  297. }
  298. }
  299. break;
  300. }
  301. case "FirstAired":
  302. {
  303. var firstAired = reader.ReadElementContentAsString();
  304. if (!string.IsNullOrWhiteSpace(firstAired))
  305. {
  306. DateTime airDate;
  307. if (DateTime.TryParse(firstAired, out airDate) && airDate.Year > 1850)
  308. {
  309. item.PremiereDate = airDate.ToUniversalTime();
  310. item.ProductionYear = airDate.Year;
  311. }
  312. }
  313. break;
  314. }
  315. case "TMDbId":
  316. var tmdb = reader.ReadElementContentAsString();
  317. if (!string.IsNullOrWhiteSpace(tmdb))
  318. {
  319. item.SetProviderId(MetadataProviders.Tmdb, tmdb);
  320. }
  321. break;
  322. case "TVcomId":
  323. var TVcomId = reader.ReadElementContentAsString();
  324. if (!string.IsNullOrWhiteSpace(TVcomId))
  325. {
  326. item.SetProviderId(MetadataProviders.Tvcom, TVcomId);
  327. }
  328. break;
  329. case "IMDB_ID":
  330. case "IMDB":
  331. case "IMDbId":
  332. var IMDbId = reader.ReadElementContentAsString();
  333. if (!string.IsNullOrWhiteSpace(IMDbId))
  334. {
  335. item.SetProviderId(MetadataProviders.Imdb, IMDbId);
  336. }
  337. break;
  338. case "Genres":
  339. FetchFromGenresNode(reader.ReadSubtree(), item);
  340. break;
  341. case "Persons":
  342. FetchDataFromPersonsNode(reader.ReadSubtree(), item);
  343. break;
  344. case "ParentalRating":
  345. FetchFromParentalRatingNode(reader.ReadSubtree(), item);
  346. break;
  347. case "Studios":
  348. FetchFromStudiosNode(reader.ReadSubtree(), item);
  349. break;
  350. default:
  351. reader.Skip();
  352. break;
  353. }
  354. }
  355. /// <summary>
  356. /// Fetches from taglines node.
  357. /// </summary>
  358. /// <param name="reader">The reader.</param>
  359. /// <param name="item">The item.</param>
  360. private void FetchFromTaglinesNode(XmlReader reader, T item)
  361. {
  362. reader.MoveToContent();
  363. while (reader.Read())
  364. {
  365. if (reader.NodeType == XmlNodeType.Element)
  366. {
  367. switch (reader.Name)
  368. {
  369. case "Tagline":
  370. {
  371. var val = reader.ReadElementContentAsString();
  372. if (!string.IsNullOrWhiteSpace(val))
  373. {
  374. item.AddTagline(val);
  375. }
  376. break;
  377. }
  378. default:
  379. reader.Skip();
  380. break;
  381. }
  382. }
  383. }
  384. }
  385. /// <summary>
  386. /// Fetches from genres node.
  387. /// </summary>
  388. /// <param name="reader">The reader.</param>
  389. /// <param name="item">The item.</param>
  390. private void FetchFromGenresNode(XmlReader reader, T item)
  391. {
  392. reader.MoveToContent();
  393. while (reader.Read())
  394. {
  395. if (reader.NodeType == XmlNodeType.Element)
  396. {
  397. switch (reader.Name)
  398. {
  399. case "Genre":
  400. {
  401. var genre = reader.ReadElementContentAsString();
  402. if (!string.IsNullOrWhiteSpace(genre))
  403. {
  404. item.AddGenre(genre);
  405. }
  406. break;
  407. }
  408. default:
  409. reader.Skip();
  410. break;
  411. }
  412. }
  413. }
  414. }
  415. /// <summary>
  416. /// Fetches the data from persons node.
  417. /// </summary>
  418. /// <param name="reader">The reader.</param>
  419. /// <param name="item">The item.</param>
  420. private void FetchDataFromPersonsNode(XmlReader reader, T item)
  421. {
  422. reader.MoveToContent();
  423. while (reader.Read())
  424. {
  425. if (reader.NodeType == XmlNodeType.Element)
  426. {
  427. switch (reader.Name)
  428. {
  429. case "Person":
  430. {
  431. item.AddPeople(GetPersonsFromXmlNode(reader.ReadSubtree()));
  432. break;
  433. }
  434. default:
  435. reader.Skip();
  436. break;
  437. }
  438. }
  439. }
  440. }
  441. /// <summary>
  442. /// Fetches from studios node.
  443. /// </summary>
  444. /// <param name="reader">The reader.</param>
  445. /// <param name="item">The item.</param>
  446. private void FetchFromStudiosNode(XmlReader reader, T item)
  447. {
  448. reader.MoveToContent();
  449. while (reader.Read())
  450. {
  451. if (reader.NodeType == XmlNodeType.Element)
  452. {
  453. switch (reader.Name)
  454. {
  455. case "Studio":
  456. {
  457. var studio = reader.ReadElementContentAsString();
  458. if (!string.IsNullOrWhiteSpace(studio))
  459. {
  460. item.AddStudio(studio);
  461. }
  462. break;
  463. }
  464. default:
  465. reader.Skip();
  466. break;
  467. }
  468. }
  469. }
  470. }
  471. /// <summary>
  472. /// Fetches from parental rating node.
  473. /// </summary>
  474. /// <param name="reader">The reader.</param>
  475. /// <param name="item">The item.</param>
  476. private void FetchFromParentalRatingNode(XmlReader reader, T item)
  477. {
  478. reader.MoveToContent();
  479. while (reader.Read())
  480. {
  481. if (reader.NodeType == XmlNodeType.Element)
  482. {
  483. switch (reader.Name)
  484. {
  485. case "Value":
  486. {
  487. var ratingString = reader.ReadElementContentAsString();
  488. int rating = 7;
  489. if (!string.IsNullOrWhiteSpace(ratingString))
  490. {
  491. int.TryParse(ratingString, out rating);
  492. }
  493. switch (rating)
  494. {
  495. case -1:
  496. item.OfficialRating = "NR";
  497. break;
  498. case 0:
  499. item.OfficialRating = "UR";
  500. break;
  501. case 1:
  502. item.OfficialRating = "G";
  503. break;
  504. case 3:
  505. item.OfficialRating = "PG";
  506. break;
  507. case 4:
  508. item.OfficialRating = "PG-13";
  509. break;
  510. case 5:
  511. item.OfficialRating = "NC-17";
  512. break;
  513. case 6:
  514. item.OfficialRating = "R";
  515. break;
  516. }
  517. break;
  518. }
  519. default:
  520. reader.Skip();
  521. break;
  522. }
  523. }
  524. }
  525. }
  526. /// <summary>
  527. /// Gets the persons from XML node.
  528. /// </summary>
  529. /// <param name="reader">The reader.</param>
  530. /// <returns>IEnumerable{PersonInfo}.</returns>
  531. private IEnumerable<PersonInfo> GetPersonsFromXmlNode(XmlReader reader)
  532. {
  533. var names = new List<string>();
  534. var type = string.Empty;
  535. var role = string.Empty;
  536. reader.MoveToContent();
  537. while (reader.Read())
  538. {
  539. if (reader.NodeType == XmlNodeType.Element)
  540. {
  541. switch (reader.Name)
  542. {
  543. case "Name":
  544. names.AddRange(SplitNames(reader.ReadElementContentAsString()));
  545. break;
  546. case "Type":
  547. {
  548. var val = reader.ReadElementContentAsString();
  549. if (!string.IsNullOrWhiteSpace(val))
  550. {
  551. type = val;
  552. }
  553. break;
  554. }
  555. case "Role":
  556. {
  557. var val = reader.ReadElementContentAsString();
  558. if (!string.IsNullOrWhiteSpace(val))
  559. {
  560. role = val;
  561. }
  562. break;
  563. }
  564. default:
  565. reader.Skip();
  566. break;
  567. }
  568. }
  569. }
  570. return names.Select(n => new PersonInfo { Name = n, Role = role, Type = type });
  571. }
  572. /// <summary>
  573. /// Used to split names of comma or pipe delimeted genres and people
  574. /// </summary>
  575. /// <param name="value">The value.</param>
  576. /// <returns>IEnumerable{System.String}.</returns>
  577. private IEnumerable<string> SplitNames(string value)
  578. {
  579. value = value ?? string.Empty;
  580. // Only split by comma if there is no pipe in the string
  581. // We have to be careful to not split names like Matthew, Jr.
  582. var separator = value.IndexOf('|') == -1 ? ',' : '|';
  583. value = value.Trim().Trim(separator);
  584. return string.IsNullOrWhiteSpace(value) ? new string[] { } : Split(value, separator, StringSplitOptions.RemoveEmptyEntries);
  585. }
  586. /// <summary>
  587. /// Provides an additional overload for string.split
  588. /// </summary>
  589. /// <param name="val">The val.</param>
  590. /// <param name="separator">The separator.</param>
  591. /// <param name="options">The options.</param>
  592. /// <returns>System.String[][].</returns>
  593. private static string[] Split(string val, char separator, StringSplitOptions options)
  594. {
  595. return val.Split(new[] { separator }, options);
  596. }
  597. }
  598. }