123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302 |
- #pragma warning disable CS1591
- using System;
- using System.Collections.Generic;
- using System.Globalization;
- using System.Linq;
- using System.Net;
- using System.Net.Http;
- using System.Net.Http.Headers;
- using System.Text.RegularExpressions;
- using System.Threading;
- using System.Threading.Tasks;
- using MediaBrowser.Controller.Library;
- using MediaBrowser.Controller.Providers;
- using MediaBrowser.Model.Entities;
- using MediaBrowser.Model.Providers;
- using MediaBrowser.Model.Serialization;
- using MediaBrowser.Providers.Plugins.Tmdb.Models.Search;
- using Microsoft.Extensions.Logging;
- namespace MediaBrowser.Providers.Plugins.Tmdb.Movies
- {
- public class TmdbSearch
- {
- private const string SearchUrl = TmdbUtils.BaseTmdbApiUrl + @"3/search/{3}?api_key={1}&query={0}&language={2}";
- private const string SearchUrlTvWithYear = TmdbUtils.BaseTmdbApiUrl + @"3/search/tv?api_key={1}&query={0}&language={2}&first_air_date_year={3}";
- private const string SearchUrlMovieWithYear = TmdbUtils.BaseTmdbApiUrl + @"3/search/movie?api_key={1}&query={0}&language={2}&primary_release_year={3}";
- private static readonly CultureInfo _usCulture = new CultureInfo("en-US");
- private static readonly Regex _cleanEnclosed = new Regex(@"\p{Ps}.*\p{Pe}", RegexOptions.Compiled);
- private static readonly Regex _cleanNonWord = new Regex(@"[\W_]+", RegexOptions.Compiled);
- private static readonly Regex _cleanStopWords = new Regex(
- @"\b( # Start at word boundary
- 19[0-9]{2}|20[0-9]{2}| # 1900-2099
- S[0-9]{2}| # Season
- E[0-9]{2}| # Episode
- (2160|1080|720|576|480)[ip]?| # Resolution
- [xh]?264| # Encoding
- (web|dvd|bd|hdtv|hd)rip| # *Rip
- web|hdtv|mp4|bluray|ktr|dl|single|imageset|internal|doku|dubbed|retail|xxx|flac
- ).* # Match rest of string",
- RegexOptions.Compiled | RegexOptions.IgnorePatternWhitespace | RegexOptions.IgnoreCase);
- private readonly ILogger _logger;
- private readonly IJsonSerializer _json;
- private readonly ILibraryManager _libraryManager;
- public TmdbSearch(ILogger logger, IJsonSerializer json, ILibraryManager libraryManager)
- {
- _logger = logger;
- _json = json;
- _libraryManager = libraryManager;
- }
- public Task<IEnumerable<RemoteSearchResult>> GetSearchResults(SeriesInfo idInfo, CancellationToken cancellationToken)
- {
- return GetSearchResults(idInfo, "tv", cancellationToken);
- }
- public Task<IEnumerable<RemoteSearchResult>> GetMovieSearchResults(ItemLookupInfo idInfo, CancellationToken cancellationToken)
- {
- return GetSearchResults(idInfo, "movie", cancellationToken);
- }
- public Task<IEnumerable<RemoteSearchResult>> GetSearchResults(BoxSetInfo idInfo, CancellationToken cancellationToken)
- {
- return GetSearchResults(idInfo, "collection", cancellationToken);
- }
- private async Task<IEnumerable<RemoteSearchResult>> GetSearchResults(ItemLookupInfo idInfo, string searchType, CancellationToken cancellationToken)
- {
- var name = idInfo.Name;
- var year = idInfo.Year;
- if (string.IsNullOrWhiteSpace(name))
- {
- return new List<RemoteSearchResult>();
- }
- var tmdbSettings = await TmdbMovieProvider.Current.GetTmdbSettings(cancellationToken).ConfigureAwait(false);
- var tmdbImageUrl = tmdbSettings.images.GetImageUrl("original");
- // ParseName is required here.
- // Caller provides the filename with extension stripped and NOT the parsed filename
- var parsedName = _libraryManager.ParseName(name);
- var yearInName = parsedName.Year;
- name = parsedName.Name;
- year ??= yearInName;
- var language = idInfo.MetadataLanguage.ToLowerInvariant();
- // Replace sequences of non-word characters with space
- // TMDB expects a space separated list of words make sure that is the case
- name = _cleanNonWord.Replace(name, " ").Trim();
- _logger.LogInformation("TmdbSearch: Finding id for item: {0} ({1})", name, year);
- var results = await GetSearchResults(name, searchType, year, language, tmdbImageUrl, cancellationToken).ConfigureAwait(false);
- if (results.Count == 0)
- {
- // try in english if wasn't before
- if (!string.Equals(language, "en", StringComparison.OrdinalIgnoreCase))
- {
- results = await GetSearchResults(name, searchType, year, "en", tmdbImageUrl, cancellationToken).ConfigureAwait(false);
- }
- }
- // TODO: retrying alternatives should be done outside the search
- // provider so that the retry logic can be common for all search
- // providers
- if (results.Count == 0)
- {
- var name2 = parsedName.Name;
- // Remove things enclosed in []{}() etc
- name2 = _cleanEnclosed.Replace(name2, string.Empty);
- // Replace sequences of non-word characters with space
- name2 = _cleanNonWord.Replace(name2, " ");
- // Clean based on common stop words / tokens
- name2 = _cleanStopWords.Replace(name2, string.Empty);
- // Trim whitespace
- name2 = name2.Trim();
- // Search again if the new name is different
- if (!string.Equals(name2, name, StringComparison.Ordinal) && !string.IsNullOrWhiteSpace(name2))
- {
- _logger.LogInformation("TmdbSearch: Finding id for item: {0} ({1})", name2, year);
- results = await GetSearchResults(name2, searchType, year, language, tmdbImageUrl, cancellationToken).ConfigureAwait(false);
- if (results.Count == 0 && !string.Equals(language, "en", StringComparison.OrdinalIgnoreCase))
- {
- // one more time, in english
- results = await GetSearchResults(name2, searchType, year, "en", tmdbImageUrl, cancellationToken).ConfigureAwait(false);
- }
- }
- }
- return results.Where(i =>
- {
- if (year.HasValue && i.ProductionYear.HasValue)
- {
- // Allow one year tolerance
- return Math.Abs(year.Value - i.ProductionYear.Value) <= 1;
- }
- return true;
- });
- }
- private Task<List<RemoteSearchResult>> GetSearchResults(string name, string type, int? year, string language, string baseImageUrl, CancellationToken cancellationToken)
- {
- switch (type)
- {
- case "tv":
- return GetSearchResultsTv(name, year, language, baseImageUrl, cancellationToken);
- default:
- return GetSearchResultsGeneric(name, type, year, language, baseImageUrl, cancellationToken);
- }
- }
- private async Task<List<RemoteSearchResult>> GetSearchResultsGeneric(string name, string type, int? year, string language, string baseImageUrl, CancellationToken cancellationToken)
- {
- if (string.IsNullOrWhiteSpace(name))
- {
- throw new ArgumentException("String can't be null or empty.", nameof(name));
- }
- string url3;
- if (year != null && string.Equals(type, "movie", StringComparison.OrdinalIgnoreCase))
- {
- url3 = string.Format(
- CultureInfo.InvariantCulture,
- SearchUrlMovieWithYear,
- WebUtility.UrlEncode(name),
- TmdbUtils.ApiKey,
- language,
- year);
- }
- else
- {
- url3 = string.Format(
- CultureInfo.InvariantCulture,
- SearchUrl,
- WebUtility.UrlEncode(name),
- TmdbUtils.ApiKey,
- language,
- type);
- }
- using var requestMessage = new HttpRequestMessage(HttpMethod.Get, url3);
- foreach (var header in TmdbUtils.AcceptHeaders)
- {
- requestMessage.Headers.Accept.Add(new MediaTypeWithQualityHeaderValue(header));
- }
- using var response = await TmdbMovieProvider.Current.GetMovieDbResponse(requestMessage, cancellationToken).ConfigureAwait(false);
- await using var stream = await response.Content.ReadAsStreamAsync().ConfigureAwait(false);
- var searchResults = await _json.DeserializeFromStreamAsync<TmdbSearchResult<MovieResult>>(stream).ConfigureAwait(false);
- var results = searchResults.Results ?? new List<MovieResult>();
- return results
- .Select(i =>
- {
- var remoteResult = new RemoteSearchResult
- {
- SearchProviderName = TmdbMovieProvider.Current.Name,
- Name = i.Title ?? i.Name ?? i.Original_Title,
- ImageUrl = string.IsNullOrWhiteSpace(i.Poster_Path) ? null : baseImageUrl + i.Poster_Path
- };
- if (!string.IsNullOrWhiteSpace(i.Release_Date))
- {
- // These dates are always in this exact format
- if (DateTime.TryParseExact(i.Release_Date, "yyyy-MM-dd", _usCulture, DateTimeStyles.None, out var r))
- {
- remoteResult.PremiereDate = r.ToUniversalTime();
- remoteResult.ProductionYear = remoteResult.PremiereDate.Value.Year;
- }
- }
- remoteResult.SetProviderId(MetadataProvider.Tmdb, i.Id.ToString(_usCulture));
- return remoteResult;
- })
- .ToList();
- }
- private async Task<List<RemoteSearchResult>> GetSearchResultsTv(string name, int? year, string language, string baseImageUrl, CancellationToken cancellationToken)
- {
- if (string.IsNullOrWhiteSpace(name))
- {
- throw new ArgumentException("String can't be null or empty.", nameof(name));
- }
- string url3;
- if (year == null)
- {
- url3 = string.Format(
- CultureInfo.InvariantCulture,
- SearchUrl,
- WebUtility.UrlEncode(name),
- TmdbUtils.ApiKey,
- language,
- "tv");
- }
- else
- {
- url3 = string.Format(
- CultureInfo.InvariantCulture,
- SearchUrlTvWithYear,
- WebUtility.UrlEncode(name),
- TmdbUtils.ApiKey,
- language,
- year);
- }
- using var requestMessage = new HttpRequestMessage(HttpMethod.Get, url3);
- foreach (var header in TmdbUtils.AcceptHeaders)
- {
- requestMessage.Headers.Accept.Add(new MediaTypeWithQualityHeaderValue(header));
- }
- using var response = await TmdbMovieProvider.Current.GetMovieDbResponse(requestMessage, cancellationToken).ConfigureAwait(false);
- await using var stream = await response.Content.ReadAsStreamAsync().ConfigureAwait(false);
- var searchResults = await _json.DeserializeFromStreamAsync<TmdbSearchResult<TvResult>>(stream).ConfigureAwait(false);
- var results = searchResults.Results ?? new List<TvResult>();
- return results
- .Select(i =>
- {
- var remoteResult = new RemoteSearchResult
- {
- SearchProviderName = TmdbMovieProvider.Current.Name,
- Name = i.Name ?? i.Original_Name,
- ImageUrl = string.IsNullOrWhiteSpace(i.Poster_Path) ? null : baseImageUrl + i.Poster_Path
- };
- if (!string.IsNullOrWhiteSpace(i.First_Air_Date))
- {
- // These dates are always in this exact format
- if (DateTime.TryParseExact(i.First_Air_Date, "yyyy-MM-dd", _usCulture, DateTimeStyles.None, out var r))
- {
- remoteResult.PremiereDate = r.ToUniversalTime();
- remoteResult.ProductionYear = remoteResult.PremiereDate.Value.Year;
- }
- }
- remoteResult.SetProviderId(MetadataProvider.Tmdb, i.Id.ToString(_usCulture));
- return remoteResult;
- })
- .ToList();
- }
- }
- }
|