BaseItemXmlParser.cs 24 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702
  1. using MediaBrowser.Controller.Entities;
  2. using MediaBrowser.Controller.Entities.Audio;
  3. using MediaBrowser.Model.Entities;
  4. using MediaBrowser.Model.Logging;
  5. using System;
  6. using System.Collections.Generic;
  7. using System.Globalization;
  8. using System.IO;
  9. using System.Linq;
  10. using System.Text;
  11. using System.Threading;
  12. using System.Xml;
  13. namespace MediaBrowser.Controller.Providers
  14. {
  15. /// <summary>
  16. /// Provides a base class for parsing metadata xml
  17. /// </summary>
  18. /// <typeparam name="T"></typeparam>
  19. public class BaseItemXmlParser<T>
  20. where T : BaseItem, new()
  21. {
  22. /// <summary>
  23. /// The logger
  24. /// </summary>
  25. protected ILogger Logger { get; private set; }
  26. /// <summary>
  27. /// Initializes a new instance of the <see cref="BaseItemXmlParser{T}" /> class.
  28. /// </summary>
  29. /// <param name="logger">The logger.</param>
  30. public BaseItemXmlParser(ILogger logger)
  31. {
  32. Logger = logger;
  33. }
  34. /// <summary>
  35. /// Fetches metadata for an item from one xml file
  36. /// </summary>
  37. /// <param name="item">The item.</param>
  38. /// <param name="metadataFile">The metadata file.</param>
  39. /// <param name="cancellationToken">The cancellation token.</param>
  40. /// <exception cref="System.ArgumentNullException"></exception>
  41. public void Fetch(T item, string metadataFile, CancellationToken cancellationToken)
  42. {
  43. if (item == null)
  44. {
  45. throw new ArgumentNullException();
  46. }
  47. if (string.IsNullOrEmpty(metadataFile))
  48. {
  49. throw new ArgumentNullException();
  50. }
  51. var settings = new XmlReaderSettings
  52. {
  53. CheckCharacters = false,
  54. IgnoreProcessingInstructions = true,
  55. IgnoreComments = true,
  56. ValidationType = ValidationType.None
  57. };
  58. item.Taglines.Clear();
  59. item.Studios.Clear();
  60. item.Genres.Clear();
  61. item.People.Clear();
  62. // Use european encoding as it will accept more characters
  63. using (var streamReader = new StreamReader(metadataFile, Encoding.GetEncoding("ISO-8859-1")))
  64. {
  65. // Use XmlReader for best performance
  66. using (var reader = XmlReader.Create(streamReader, settings))
  67. {
  68. reader.MoveToContent();
  69. // Loop through each element
  70. while (reader.Read())
  71. {
  72. cancellationToken.ThrowIfCancellationRequested();
  73. if (reader.NodeType == XmlNodeType.Element)
  74. {
  75. FetchDataFromXmlNode(reader, item);
  76. }
  77. }
  78. }
  79. }
  80. }
  81. private readonly CultureInfo _usCulture = new CultureInfo("en-US");
  82. /// <summary>
  83. /// Fetches metadata from one Xml Element
  84. /// </summary>
  85. /// <param name="reader">The reader.</param>
  86. /// <param name="item">The item.</param>
  87. protected virtual void FetchDataFromXmlNode(XmlReader reader, T item)
  88. {
  89. switch (reader.Name)
  90. {
  91. // DateCreated
  92. case "Added":
  93. DateTime added;
  94. if (DateTime.TryParse(reader.ReadElementContentAsString() ?? string.Empty, out added))
  95. {
  96. item.DateCreated = added.ToUniversalTime();
  97. }
  98. break;
  99. case "LocalTitle":
  100. item.Name = reader.ReadElementContentAsString();
  101. break;
  102. case "Type":
  103. {
  104. var type = reader.ReadElementContentAsString();
  105. if (!string.IsNullOrWhiteSpace(type) && !type.Equals("none", StringComparison.OrdinalIgnoreCase))
  106. {
  107. item.DisplayMediaType = type;
  108. }
  109. break;
  110. }
  111. case "Budget":
  112. {
  113. var text = reader.ReadElementContentAsString();
  114. double value;
  115. if (double.TryParse(text, NumberStyles.Any, _usCulture, out value))
  116. {
  117. item.Budget = value;
  118. }
  119. break;
  120. }
  121. case "Revenue":
  122. {
  123. var text = reader.ReadElementContentAsString();
  124. double value;
  125. if (double.TryParse(text, NumberStyles.Any, _usCulture, out value))
  126. {
  127. item.Revenue = value;
  128. }
  129. break;
  130. }
  131. case "SortTitle":
  132. item.ForcedSortName = reader.ReadElementContentAsString();
  133. break;
  134. case "Overview":
  135. case "Description":
  136. {
  137. var val = reader.ReadElementContentAsString();
  138. if (!string.IsNullOrWhiteSpace(val))
  139. {
  140. item.Overview = val;
  141. }
  142. break;
  143. }
  144. case "TagLine":
  145. {
  146. var tagline = reader.ReadElementContentAsString();
  147. if (!string.IsNullOrWhiteSpace(tagline))
  148. {
  149. item.AddTagline(tagline);
  150. }
  151. break;
  152. }
  153. case "Website":
  154. {
  155. var val = reader.ReadElementContentAsString();
  156. if (!string.IsNullOrWhiteSpace(val))
  157. {
  158. item.HomePageUrl = val;
  159. }
  160. break;
  161. }
  162. case "TagLines":
  163. {
  164. FetchFromTaglinesNode(reader.ReadSubtree(), item);
  165. break;
  166. }
  167. case "ContentRating":
  168. case "certification":
  169. case "MPAARating":
  170. {
  171. var rating = reader.ReadElementContentAsString();
  172. if (!string.IsNullOrWhiteSpace(rating))
  173. {
  174. item.OfficialRating = rating;
  175. }
  176. break;
  177. }
  178. case "CustomRating":
  179. {
  180. var val = reader.ReadElementContentAsString();
  181. if (!string.IsNullOrWhiteSpace(val))
  182. {
  183. item.CustomRating = val;
  184. }
  185. break;
  186. }
  187. case "Runtime":
  188. case "RunningTime":
  189. {
  190. var text = reader.ReadElementContentAsString();
  191. if (!string.IsNullOrWhiteSpace(text))
  192. {
  193. int runtime;
  194. if (int.TryParse(text.Split(' ')[0], NumberStyles.Integer, _usCulture, out runtime))
  195. {
  196. // For audio and video don't replace ffmpeg data
  197. if (item is Video || item is Audio)
  198. {
  199. item.OriginalRunTimeTicks = TimeSpan.FromMinutes(runtime).Ticks;
  200. }
  201. else
  202. {
  203. item.RunTimeTicks = TimeSpan.FromMinutes(runtime).Ticks;
  204. }
  205. }
  206. }
  207. break;
  208. }
  209. case "Genre":
  210. {
  211. foreach (var name in SplitNames(reader.ReadElementContentAsString()))
  212. {
  213. if (string.IsNullOrWhiteSpace(name))
  214. {
  215. continue;
  216. }
  217. item.AddGenre(name);
  218. }
  219. break;
  220. }
  221. case "AspectRatio":
  222. {
  223. var val = reader.ReadElementContentAsString();
  224. if (!string.IsNullOrWhiteSpace(val))
  225. {
  226. item.AspectRatio = val;
  227. }
  228. break;
  229. }
  230. case "Network":
  231. {
  232. foreach (var name in SplitNames(reader.ReadElementContentAsString()))
  233. {
  234. if (string.IsNullOrWhiteSpace(name))
  235. {
  236. continue;
  237. }
  238. item.AddStudio(name);
  239. }
  240. break;
  241. }
  242. case "Director":
  243. {
  244. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v, Type = PersonType.Director }))
  245. {
  246. if (string.IsNullOrWhiteSpace(p.Name))
  247. {
  248. continue;
  249. }
  250. item.AddPerson(p);
  251. }
  252. break;
  253. }
  254. case "Writer":
  255. {
  256. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v, Type = PersonType.Writer }))
  257. {
  258. if (string.IsNullOrWhiteSpace(p.Name))
  259. {
  260. continue;
  261. }
  262. item.AddPerson(p);
  263. }
  264. break;
  265. }
  266. case "Actors":
  267. {
  268. var actors = reader.ReadInnerXml();
  269. if (actors.Contains("<"))
  270. {
  271. // This is one of the mis-named "Actors" full nodes created by MB2
  272. // Create a reader and pass it to the persons node processor
  273. FetchDataFromPersonsNode(new XmlTextReader(new StringReader("<Persons>" + actors + "</Persons>")), item);
  274. }
  275. else
  276. {
  277. // Old-style piped string
  278. foreach (var p in SplitNames(actors).Select(v => new PersonInfo { Name = v.Trim(), Type = PersonType.Actor }))
  279. {
  280. if (string.IsNullOrWhiteSpace(p.Name))
  281. {
  282. continue;
  283. }
  284. item.AddPerson(p);
  285. }
  286. }
  287. break;
  288. }
  289. case "GuestStars":
  290. {
  291. foreach (var p in SplitNames(reader.ReadElementContentAsString()).Select(v => new PersonInfo { Name = v.Trim(), Type = PersonType.GuestStar }))
  292. {
  293. if (string.IsNullOrWhiteSpace(p.Name))
  294. {
  295. continue;
  296. }
  297. item.AddPerson(p);
  298. }
  299. break;
  300. }
  301. case "Trailer":
  302. {
  303. var val = reader.ReadElementContentAsString();
  304. if (!string.IsNullOrWhiteSpace(val))
  305. {
  306. //item.AddTrailerUrl(val);
  307. }
  308. break;
  309. }
  310. case "ProductionYear":
  311. {
  312. var val = reader.ReadElementContentAsString();
  313. if (!string.IsNullOrWhiteSpace(val))
  314. {
  315. int productionYear;
  316. if (int.TryParse(val, out productionYear) && productionYear > 1850)
  317. {
  318. item.ProductionYear = productionYear;
  319. }
  320. }
  321. break;
  322. }
  323. case "Rating":
  324. case "IMDBrating":
  325. {
  326. var rating = reader.ReadElementContentAsString();
  327. if (!string.IsNullOrWhiteSpace(rating))
  328. {
  329. float val;
  330. // All external meta is saving this as '.' for decimal I believe...but just to be sure
  331. if (float.TryParse(rating.Replace(',','.'), NumberStyles.AllowDecimalPoint, CultureInfo.InvariantCulture, out val))
  332. {
  333. item.CommunityRating = val;
  334. }
  335. }
  336. break;
  337. }
  338. case "FirstAired":
  339. {
  340. var firstAired = reader.ReadElementContentAsString();
  341. if (!string.IsNullOrWhiteSpace(firstAired))
  342. {
  343. DateTime airDate;
  344. if (DateTime.TryParse(firstAired, out airDate) && airDate.Year > 1850)
  345. {
  346. item.PremiereDate = airDate.ToUniversalTime();
  347. item.ProductionYear = airDate.Year;
  348. }
  349. }
  350. break;
  351. }
  352. case "TMDbId":
  353. var tmdb = reader.ReadElementContentAsString();
  354. if (!string.IsNullOrWhiteSpace(tmdb))
  355. {
  356. item.SetProviderId(MetadataProviders.Tmdb, tmdb);
  357. }
  358. break;
  359. case "TVcomId":
  360. var TVcomId = reader.ReadElementContentAsString();
  361. if (!string.IsNullOrWhiteSpace(TVcomId))
  362. {
  363. item.SetProviderId(MetadataProviders.Tvcom, TVcomId);
  364. }
  365. break;
  366. case "IMDB_ID":
  367. case "IMDB":
  368. case "IMDbId":
  369. var imDbId = reader.ReadElementContentAsString();
  370. if (!string.IsNullOrWhiteSpace(imDbId))
  371. {
  372. item.SetProviderId(MetadataProviders.Imdb, imDbId);
  373. }
  374. break;
  375. case "Genres":
  376. FetchFromGenresNode(reader.ReadSubtree(), item);
  377. break;
  378. case "Persons":
  379. FetchDataFromPersonsNode(reader.ReadSubtree(), item);
  380. break;
  381. case "ParentalRating":
  382. FetchFromParentalRatingNode(reader.ReadSubtree(), item);
  383. break;
  384. case "Studios":
  385. FetchFromStudiosNode(reader.ReadSubtree(), item);
  386. break;
  387. default:
  388. reader.Skip();
  389. break;
  390. }
  391. }
  392. /// <summary>
  393. /// Fetches from taglines node.
  394. /// </summary>
  395. /// <param name="reader">The reader.</param>
  396. /// <param name="item">The item.</param>
  397. private void FetchFromTaglinesNode(XmlReader reader, T item)
  398. {
  399. reader.MoveToContent();
  400. while (reader.Read())
  401. {
  402. if (reader.NodeType == XmlNodeType.Element)
  403. {
  404. switch (reader.Name)
  405. {
  406. case "Tagline":
  407. {
  408. var val = reader.ReadElementContentAsString();
  409. if (!string.IsNullOrWhiteSpace(val))
  410. {
  411. item.AddTagline(val);
  412. }
  413. break;
  414. }
  415. default:
  416. reader.Skip();
  417. break;
  418. }
  419. }
  420. }
  421. }
  422. /// <summary>
  423. /// Fetches from genres node.
  424. /// </summary>
  425. /// <param name="reader">The reader.</param>
  426. /// <param name="item">The item.</param>
  427. private void FetchFromGenresNode(XmlReader reader, T item)
  428. {
  429. reader.MoveToContent();
  430. while (reader.Read())
  431. {
  432. if (reader.NodeType == XmlNodeType.Element)
  433. {
  434. switch (reader.Name)
  435. {
  436. case "Genre":
  437. {
  438. var genre = reader.ReadElementContentAsString();
  439. if (!string.IsNullOrWhiteSpace(genre))
  440. {
  441. item.AddGenre(genre);
  442. }
  443. break;
  444. }
  445. default:
  446. reader.Skip();
  447. break;
  448. }
  449. }
  450. }
  451. }
  452. /// <summary>
  453. /// Fetches the data from persons node.
  454. /// </summary>
  455. /// <param name="reader">The reader.</param>
  456. /// <param name="item">The item.</param>
  457. private void FetchDataFromPersonsNode(XmlReader reader, T item)
  458. {
  459. reader.MoveToContent();
  460. while (reader.Read())
  461. {
  462. if (reader.NodeType == XmlNodeType.Element)
  463. {
  464. switch (reader.Name)
  465. {
  466. case "Person":
  467. case "Actor":
  468. {
  469. foreach (var person in GetPersonsFromXmlNode(reader.ReadSubtree()))
  470. {
  471. item.AddPerson(person);
  472. }
  473. break;
  474. }
  475. default:
  476. reader.Skip();
  477. break;
  478. }
  479. }
  480. }
  481. }
  482. /// <summary>
  483. /// Fetches from studios node.
  484. /// </summary>
  485. /// <param name="reader">The reader.</param>
  486. /// <param name="item">The item.</param>
  487. private void FetchFromStudiosNode(XmlReader reader, T item)
  488. {
  489. reader.MoveToContent();
  490. while (reader.Read())
  491. {
  492. if (reader.NodeType == XmlNodeType.Element)
  493. {
  494. switch (reader.Name)
  495. {
  496. case "Studio":
  497. {
  498. var studio = reader.ReadElementContentAsString();
  499. if (!string.IsNullOrWhiteSpace(studio))
  500. {
  501. item.AddStudio(studio);
  502. }
  503. break;
  504. }
  505. default:
  506. reader.Skip();
  507. break;
  508. }
  509. }
  510. }
  511. }
  512. /// <summary>
  513. /// Fetches from parental rating node.
  514. /// </summary>
  515. /// <param name="reader">The reader.</param>
  516. /// <param name="item">The item.</param>
  517. private void FetchFromParentalRatingNode(XmlReader reader, T item)
  518. {
  519. reader.MoveToContent();
  520. while (reader.Read())
  521. {
  522. if (reader.NodeType == XmlNodeType.Element)
  523. {
  524. switch (reader.Name)
  525. {
  526. // Removed support for "Value" tag as it conflicted with MPAA rating but leaving this function for possible
  527. // future support of "Description" -ebr
  528. default:
  529. reader.Skip();
  530. break;
  531. }
  532. }
  533. }
  534. }
  535. /// <summary>
  536. /// Gets the persons from XML node.
  537. /// </summary>
  538. /// <param name="reader">The reader.</param>
  539. /// <returns>IEnumerable{PersonInfo}.</returns>
  540. private IEnumerable<PersonInfo> GetPersonsFromXmlNode(XmlReader reader)
  541. {
  542. var names = new List<string>();
  543. var type = "Actor"; // If type is not specified assume actor
  544. var role = string.Empty;
  545. reader.MoveToContent();
  546. while (reader.Read())
  547. {
  548. if (reader.NodeType == XmlNodeType.Element)
  549. {
  550. switch (reader.Name)
  551. {
  552. case "Name":
  553. names.AddRange(SplitNames(reader.ReadElementContentAsString()));
  554. break;
  555. case "Type":
  556. {
  557. var val = reader.ReadElementContentAsString();
  558. if (!string.IsNullOrWhiteSpace(val))
  559. {
  560. type = val;
  561. }
  562. break;
  563. }
  564. case "Role":
  565. {
  566. var val = reader.ReadElementContentAsString();
  567. if (!string.IsNullOrWhiteSpace(val))
  568. {
  569. role = val;
  570. }
  571. break;
  572. }
  573. default:
  574. reader.Skip();
  575. break;
  576. }
  577. }
  578. }
  579. return names.Select(n => new PersonInfo { Name = n, Role = role, Type = type });
  580. }
  581. /// <summary>
  582. /// Used to split names of comma or pipe delimeted genres and people
  583. /// </summary>
  584. /// <param name="value">The value.</param>
  585. /// <returns>IEnumerable{System.String}.</returns>
  586. private IEnumerable<string> SplitNames(string value)
  587. {
  588. value = value ?? string.Empty;
  589. // Only split by comma if there is no pipe in the string
  590. // We have to be careful to not split names like Matthew, Jr.
  591. var separator = value.IndexOf('|') == -1 ? ',' : '|';
  592. value = value.Trim().Trim(separator);
  593. return string.IsNullOrWhiteSpace(value) ? new string[] { } : Split(value, separator, StringSplitOptions.RemoveEmptyEntries);
  594. }
  595. /// <summary>
  596. /// Provides an additional overload for string.split
  597. /// </summary>
  598. /// <param name="val">The val.</param>
  599. /// <param name="separator">The separator.</param>
  600. /// <param name="options">The options.</param>
  601. /// <returns>System.String[][].</returns>
  602. private static string[] Split(string val, char separator, StringSplitOptions options)
  603. {
  604. return val.Split(new[] { separator }, options);
  605. }
  606. }
  607. }