Backport pull request #18096 from jellyfin/release-12.z

Decode HTML entities in OMDb responses

Original-merge: 8bd7306fc9aeff8996f474f65647ebde7d001577

Merged-by: crobibero <cody@robibe.ro>

Backported-by: Cody Robibero <cody@robibe.ro>

4 files changed, 54 insertions, 10 deletions

diff --git a/MediaBrowser.Providers/Plugins/Omdb/JsonOmdbNotAvailableStringConverter.cs b/MediaBrowser.Providers/Plugins/Omdb/JsonOmdbNotAvailableStringConverter.cs
index 7e9befa8c6..0278469807 100644
--- a/MediaBrowser.Providers/Plugins/Omdb/JsonOmdbNotAvailableStringConverter.cs
+++ b/MediaBrowser.Providers/Plugins/Omdb/JsonOmdbNotAvailableStringConverter.cs
@@ -1,4 +1,5 @@
 using System;
+using System.Net;
 using System.Text.Json;
 using System.Text.Json.Serialization;
 using Jellyfin.Extensions.Json;
@@ -6,7 +7,7 @@ using Jellyfin.Extensions.Json;
 namespace MediaBrowser.Providers.Plugins.Omdb
 {
     /// <summary>
-    /// Converts a string <c>N/A</c> to <c>string.Empty</c>.
+    /// Converts a string <c>N/A</c> to <c>string.Empty</c> and decodes HTML entities in every other string.
     /// </summary>
     public class JsonOmdbNotAvailableStringConverter : JsonConverter<string?>
     {
@@ -27,7 +28,9 @@ namespace MediaBrowser.Providers.Plugins.Omdb
                     return null;
                 }
 
-                return str;
+                // Some OMDb records are HTML encoded, e.g. the cast of tt0093058 lists "Vincent D&apos;Onofrio".
+                // Stored verbatim that name is a second person next to the correctly spelled one.
+                return WebUtility.HtmlDecode(str);
             }
 
             return JsonSerializer.Deserialize<string?>(ref reader, options);
diff --git a/MediaBrowser.Providers/Plugins/Omdb/OmdbItemProvider.cs b/MediaBrowser.Providers/Plugins/Omdb/OmdbItemProvider.cs
index 7b245ea5a7..c81ada3e00 100644
--- a/MediaBrowser.Providers/Plugins/Omdb/OmdbItemProvider.cs
+++ b/MediaBrowser.Providers/Plugins/Omdb/OmdbItemProvider.cs
@@ -13,7 +13,6 @@ using System.Text;
 using System.Text.Json;
 using System.Threading;
 using System.Threading.Tasks;
-using Jellyfin.Extensions.Json;
 using MediaBrowser.Common.Net;
 using MediaBrowser.Controller.Configuration;
 using MediaBrowser.Controller.Entities;
@@ -45,9 +44,7 @@ namespace MediaBrowser.Providers.Plugins.Omdb
             _libraryManager = libraryManager;
             _omdbProvider = new OmdbProvider(_httpClientFactory, fileSystem, configurationManager);
 
-            _jsonOptions = new JsonSerializerOptions(JsonDefaults.Options);
-            _jsonOptions.Converters.Add(new JsonOmdbNotAvailableStringConverter());
-            _jsonOptions.Converters.Add(new JsonOmdbNotAvailableInt32Converter());
+            _jsonOptions = OmdbProvider.CreateJsonOptions();
         }
 
         public string Name => "The Open Movie Database";
diff --git a/MediaBrowser.Providers/Plugins/Omdb/OmdbProvider.cs b/MediaBrowser.Providers/Plugins/Omdb/OmdbProvider.cs
index d51d913caa..60e4001189 100644
--- a/MediaBrowser.Providers/Plugins/Omdb/OmdbProvider.cs
+++ b/MediaBrowser.Providers/Plugins/Omdb/OmdbProvider.cs
@@ -45,10 +45,19 @@ namespace MediaBrowser.Providers.Plugins.Omdb
             _fileSystem = fileSystem;
             _configurationManager = configurationManager;
 
-            _jsonOptions = new JsonSerializerOptions(JsonDefaults.Options);
-            // These converters need to take priority
-            _jsonOptions.Converters.Insert(0, new JsonOmdbNotAvailableStringConverter());
-            _jsonOptions.Converters.Insert(0, new JsonOmdbNotAvailableInt32Converter());
+            _jsonOptions = CreateJsonOptions();
+        }
+
+        /// <summary>Creates the <see cref="JsonSerializerOptions"/> to read OMDB responses with.</summary>
+        /// <returns>The options.</returns>
+        internal static JsonSerializerOptions CreateJsonOptions()
+        {
+            var options = new JsonSerializerOptions(JsonDefaults.Options);
+            // These have to be inserted, appending leaves the default string converter ahead of them
+            options.Converters.Insert(0, new JsonOmdbNotAvailableStringConverter());
+            options.Converters.Insert(0, new JsonOmdbNotAvailableInt32Converter());
+
+            return options;
         }
 
         /// <summary>Fetches data from OMDB service.</summary>
diff --git a/tests/Jellyfin.Providers.Tests/Omdb/JsonOmdbConverterTests.cs b/tests/Jellyfin.Providers.Tests/Omdb/JsonOmdbConverterTests.cs
index 3062cb7b42..0275d13242 100644
--- a/tests/Jellyfin.Providers.Tests/Omdb/JsonOmdbConverterTests.cs
+++ b/tests/Jellyfin.Providers.Tests/Omdb/JsonOmdbConverterTests.cs
@@ -63,6 +63,41 @@ namespace Jellyfin.Providers.Tests.Omdb
             Assert.Equal(expected, result);
         }
 
+        [Theory]
+        [InlineData("\"Vincent D&apos;Onofrio\"", "Vincent D'Onofrio")]
+        [InlineData("\"Alan Moore &#39;Swamp Thing&#39;\"", "Alan Moore 'Swamp Thing'")]
+        [InlineData("\"Tom &amp; Jerry\"", "Tom & Jerry")]
+        [InlineData("\"Mail &lt;a@b.com&gt; for a translation\"", "Mail <a@b.com> for a translation")]
+        [InlineData("\"Elisa Mont\\u00e9s\"", "Elisa Montés")]
+        [InlineData("\"R&D and 100% & more\"", "R&D and 100% & more")]
+        public void Deserialize_Html_Encoded_String_Is_Decoded(string input, string expected)
+        {
+            var result = JsonSerializer.Deserialize<string?>(input, _options);
+            Assert.Equal(expected, result);
+        }
+
+        [Fact]
+        public void Deserialize_Html_Encoded_Credits_Are_Decoded()
+        {
+            const string Input = "{\"Title\":\"Full Metal Jacket\",\"Actors\":\"Matthew Modine, R. Lee Ermey, Vincent D&apos;Onofrio\",\"Response\":\"True\"}";
+            var rootObject = JsonSerializer.Deserialize<OmdbProvider.RootObject>(Input, _options);
+            Assert.NotNull(rootObject);
+            Assert.Equal("Matthew Modine, R. Lee Ermey, Vincent D'Onofrio", rootObject!.Actors);
+        }
+
+        [Fact]
+        public void ProviderJsonOptions_Take_Priority_Over_The_Defaults()
+        {
+            // The default string converter handles string as well, so appending the OMDb ones would never run them
+            var options = OmdbProvider.CreateJsonOptions();
+
+            const string Input = "{\"Title\":\"Full Metal Jacket\",\"Actors\":\"Vincent D&apos;Onofrio\",\"Awards\":\"N/A\",\"Response\":\"True\"}";
+            var rootObject = JsonSerializer.Deserialize<OmdbProvider.RootObject>(Input, options);
+            Assert.NotNull(rootObject);
+            Assert.Equal("Vincent D'Onofrio", rootObject!.Actors);
+            Assert.Null(rootObject.Awards);
+        }
+
         [Fact]
         public void Roundtrip_Valid_Success()
         {