extract page count from archives and PDFs

5 files changed, 57 insertions, 6 deletions

diff --git a/Directory.Packages.props b/Directory.Packages.props
index f0a655d488..17add2f9c9 100644
--- a/Directory.Packages.props
+++ b/Directory.Packages.props
@@ -53,6 +53,7 @@
     <PackageVersion Include="Moq" Version="4.18.4" />
     <PackageVersion Include="NEbml" Version="1.1.0.5" />
     <PackageVersion Include="Newtonsoft.Json" Version="13.0.4" />
+    <PackageVersion Include="PDFtoImage" Version="5.2.1" />
     <PackageVersion Include="PlaylistsNET" Version="1.4.1" />
     <PackageVersion Include="prometheus-net.AspNetCore" Version="8.2.1" />
     <PackageVersion Include="prometheus-net.DotNetRuntime" Version="4.4.1" />
diff --git a/MediaBrowser.Controller/Entities/Book.cs b/MediaBrowser.Controller/Entities/Book.cs
index 5187669373..8559681bdc 100644
--- a/MediaBrowser.Controller/Entities/Book.cs
+++ b/MediaBrowser.Controller/Entities/Book.cs
@@ -13,11 +13,6 @@ namespace MediaBrowser.Controller.Entities
     [Common.RequiresSourceSerialisation]
     public class Book : BaseItem, IHasLookupInfo<BookInfo>, IHasSeries
     {
-        public Book()
-        {
-            this.RunTimeTicks = TimeSpan.TicksPerSecond;
-        }
-
         [JsonIgnore]
         public override MediaType MediaType => MediaType.Book;
 
diff --git a/MediaBrowser.Providers/Manager/MetadataService.cs b/MediaBrowser.Providers/Manager/MetadataService.cs
index 118ccf8679..62827c07b9 100644
--- a/MediaBrowser.Providers/Manager/MetadataService.cs
+++ b/MediaBrowser.Providers/Manager/MetadataService.cs
@@ -1116,7 +1116,7 @@ namespace MediaBrowser.Providers.Manager
             {
                 if (replaceData || !target.RunTimeTicks.HasValue)
                 {
-                    if (target is not Audio && target is not Video)
+                    if (target is not Audio && target is not Video && target is not Book)
                     {
                         target.RunTimeTicks = source.RunTimeTicks;
                     }
diff --git a/MediaBrowser.Providers/MediaBrowser.Providers.csproj b/MediaBrowser.Providers/MediaBrowser.Providers.csproj
index df51dd8421..2b0f480b1c 100644
--- a/MediaBrowser.Providers/MediaBrowser.Providers.csproj
+++ b/MediaBrowser.Providers/MediaBrowser.Providers.csproj
@@ -21,6 +21,7 @@
     <PackageReference Include="Microsoft.Extensions.Caching.Abstractions" />
     <PackageReference Include="Microsoft.Extensions.Http" />
     <PackageReference Include="Newtonsoft.Json" />
+    <PackageReference Include="PDFtoImage" />
     <PackageReference Include="PlaylistsNET" />
     <PackageReference Include="SharpCompress" />
     <PackageReference Include="z440.atl.core" />
diff --git a/MediaBrowser.Providers/MediaInfo/ProbeProvider.cs b/MediaBrowser.Providers/MediaInfo/ProbeProvider.cs
index 789df8f061..221c6bff5e 100644
--- a/MediaBrowser.Providers/MediaInfo/ProbeProvider.cs
+++ b/MediaBrowser.Providers/MediaInfo/ProbeProvider.cs
@@ -24,6 +24,8 @@ using MediaBrowser.Model.Globalization;
 using MediaBrowser.Model.IO;
 using MediaBrowser.Model.MediaInfo;
 using Microsoft.Extensions.Logging;
+using PDFtoImage;
+using SharpCompress.Archives;
 
 namespace MediaBrowser.Providers.MediaInfo
 {
@@ -37,6 +39,7 @@ namespace MediaBrowser.Providers.MediaInfo
         ICustomMetadataProvider<Video>,
         ICustomMetadataProvider<Audio>,
         ICustomMetadataProvider<AudioBook>,
+        ICustomMetadataProvider<Book>,
         IHasOrder,
         IForcedProvider,
         IPreRefreshProvider,
@@ -214,6 +217,57 @@ namespace MediaBrowser.Providers.MediaInfo
             return FetchAudioInfo(item, options, cancellationToken);
         }
 
+        /// <inheritdoc />
+        public Task<ItemUpdateType> FetchAsync(Book item, MetadataRefreshOptions options, CancellationToken cancellationToken)
+        {
+            if (item.IsVirtualItem || !item.IsFileProtocol)
+            {
+                return _cachedTask;
+            }
+
+            long pageCount;
+            switch (Path.GetExtension(item.Path).ToLowerInvariant())
+            {
+                case ".cb7":
+                case ".cbr":
+                case ".cbt":
+                case ".cbz":
+                    using (var stream = File.OpenRead(item.Path))
+                    using (var archive = ArchiveFactory.OpenArchive(stream))
+                    {
+                        pageCount = archive.Entries.Count(e => !e.IsDirectory);
+                    }
+
+                    break;
+
+#pragma warning disable CA1416
+                case ".pdf":
+                    using (var stream = File.OpenRead(item.Path))
+                    {
+                        pageCount = Conversion.GetPageCount(stream);
+                    }
+
+                    break;
+#pragma warning restore CA1416
+
+                case ".epub":
+                    // TODO process CFI and store as a string when multiple progress types are supported
+                    // current progress value is percentage stored as a proportion of one second worth of ticks
+                    item.RunTimeTicks = TimeSpan.TicksPerSecond;
+
+                    return Task.FromResult(ItemUpdateType.MetadataImport);
+
+                default:
+                    return _cachedTask;
+            }
+
+            // TODO use page count without modification when multiple progress types are supported
+            // book players report page count and the web client multiplies that value by 10000 to convert the expected milliseconds into ticks
+            item.RunTimeTicks = pageCount * 10000;
+
+            return Task.FromResult(ItemUpdateType.MetadataImport);
+        }
+
         /// <summary>
         /// Fetches video information for an item.
         /// </summary>