updates for PDF4LLM 1.28.0.1& MuPDF.NET 3.28.1.2

.gitignore

@@ -47,6 +47,9 @@ TestDocuments/Demo/_Output/
 # Visual Studio 2017 auto generated files
 Generated\ Files/
 
+# MuPDF.NativeAssets generation files
+MuPDF/packages
+
 # MSTest test Results
 [Tt]est[Rr]esult*/
 [Bb]uild[Ll]og.*

Demo/Samples/Llm/Program.Llm.TableExtract.cs

@@ -102,7 +102,8 @@ internal static void TestTableExtract3()
         internal static void TestOcr()
         {
             PdfExtractor.UseLayout = true;
-            PdfExtractor.ToMarkdown(@"..\..\..\..\TestDocuments\Demo\Ocr.pdf", useOcr: true, writeImages: false, embedImages: false);
+            string md = PdfExtractor.ToMarkdown(@"..\..\..\..\TestDocuments\Demo\Ocr.pdf", useOcr: true, writeImages: false, embedImages: false);
+            Console.WriteLine(md);
             string text = PdfExtractor.ToText(@"..\..\..\..\TestDocuments\Demo\Ocr.pdf", useOcr: true);
             Console.WriteLine(text);
         }

Demo/Samples/Regression/Program.Regression.cs

@@ -201,20 +201,19 @@ internal static void TestPixmapParallel()
 
         internal static void TestGetTablesParallel()
         {
-            const int iterations = 100;
-            const int degreeOfParallelism = 10;
-
             string testFilePath = DemoPaths.Input("err_table.pdf");
             if (!File.Exists(testFilePath))
             {
                 Console.WriteLine($"Test file not found: {testFilePath}");
                 return;
             }
 
+            const int iterations = 100;
+            const int degreeOfParallelism = 10;
             int failures = 0;
             long totalTables = 0;
 
-            Console.WriteLine("\n=== Parallel Utils.GetTables ===");
+            Console.WriteLine("Multi-thread Utils.GetTables test");
             Console.WriteLine($"PDF: {testFilePath}");
             Console.WriteLine($"Iterations: {iterations}");
             Console.WriteLine($"Degree of parallelism: {degreeOfParallelism}");
@@ -238,12 +237,12 @@ internal static void TestGetTablesParallel()
 
                         Interlocked.Add(ref totalTables, tables.Count);
                         Console.WriteLine(
-                            $"Iteration {iteration + 1} (thread {Environment.CurrentManagedThreadId}): {tables.Count} table(s)");
+                            $"Iteration {iteration + 1} (thread {Environment.CurrentManagedThreadId}): {totalTables} table(s)");
                     }
                     catch (Exception ex)
                     {
                         Interlocked.Increment(ref failures);
-                        Console.WriteLine($"Iteration {iteration + 1} FAILED: {ex}");
+                        Console.WriteLine($"Iteration {iteration + 1} FAILED: {ex.Message}");
                     }
                 });
 

Directory.Build.props

@@ -1,5 +1,6 @@
 <Project>
   <Import Project="Versions.props" />
+  <Import Project="MuPDF\NativeAssets.Dev.props" Condition="Exists('$(MSBuildThisFileDirectory)MuPDF\NativeAssets.Dev.props')" />
 
   <PropertyGroup>
     <!-- NuGet packages may target newer TFMs than PDF4LLM multi-targets (net461–net7). -->

Directory.Build.targets

@@ -15,20 +15,17 @@ namespace Artifex
     /// <summary>Central version constants for the MuPDF.NET solution.</summary>
     public static class Versions
     {
-        /// <summary>Native MuPDF library (FZ_VERSION).</summary>
+        /// <summary>Native MuPDF library (FZ_VERSION / MuPDF.NativeAssets).</summary>
         public const string MuPDF = "$(ArtifexMuPDFVersion)"%3B
 
-        /// <summary>PyMuPDF Python package (VersionBind).</summary>
-        public const string PyMuPDF = "$(ArtifexPyMuPDFVersion)"%3B
-
-        /// <summary>PyMuPDF4LLM Python package.</summary>
-        public const string PyMuPDF4LLM = "$(ArtifexPyMuPDF4LLMVersion)"%3B
-
         /// <summary>MuPDF.NET NuGet package.</summary>
         public const string MuPDFNet = "$(ArtifexMuPDFNetVersion)"%3B
 
         /// <summary>PDF4LLM NuGet package.</summary>
         public const string PDF4LLM = "$(ArtifexPDF4LLMVersion)"%3B
+
+        /// <summary>pymupdf-layout PyPI package (PDF4LLM layout bridge).</summary>
+        public const string PyMuPDFLayout = "$(ArtifexPyMuPDFLayoutVersion)"%3B
     }
 }
 </_ArtifexVersionCs>

MuPDF.Fonts/FontRegistry.cs

@@ -4,10 +4,10 @@
 
 namespace MuPDF.Fonts
 {
-    /// <summary>Font descriptor and buffer loaders generated from pymupdf-fonts sources.</summary>
+    /// <summary>Font descriptor and buffer loaders generated from bundled font sources.</summary>
     public static class FontRegistry
     {
-        /// <summary>One pymupdf-fonts descriptor entry.</summary>
+        /// <summary>One bundled font descriptor entry.</summary>
         public sealed class Entry
         {
             /// <summary>True when the face is bold.</summary>

MuPDF.NET.Test/TestAnnots.cs

@@ -754,7 +754,7 @@ public void test_richtext()
         {
             var bullet = $"{(char)0x2610}{(char)0x2611}{(char)0x2612}";
             var text = $@"<p style=""text-align:justify;margin-top:-25px;"">
-    PyMuPDF <span style=""color: red;"">འདི་ ཡིག་ཆ་བཀྲམ་སྤེལ་གྱི་དོན་ལུ་ པའི་ཐོན་ཐུམ་སྒྲིལ་དྲག་ཤོས་དང་མགྱོགས་ཤོས་ཅིག་ཨིན།</span>
+    MuPDF.NET <span style=""color: red;"">འདི་ ཡིག་ཆ་བཀྲམ་སྤེལ་གྱི་དོན་ལུ་ པའི་ཐོན་ཐུམ་སྒྲིལ་དྲག་ཤོས་དང་མགྱོགས་ཤོས་ཅིག་ཨིན།</span>
     <span style=""color:blue;"">Here is some <b>bold</b> and <i>italic</i> text, followed by <b><i>bold-italic</i></b>. Text-based check boxes: {bullet}.</span>
     </p>";
             var doc = new Document();

MuPDF.NET.Test/TestFont.cs

@@ -33,7 +33,7 @@ private static string OptionalDocPath(string fileName) =>
         [Fact]
         public void test_font1()
         {
-            const string text = "PyMuPDF";
+            const string text = "MuPDF.NET";
             using var font = new Font("helv");
             Assert.Equal("Helvetica", font.Name);
             float tl = font.TextLength(text, fontSize: 20);
@@ -63,7 +63,7 @@ public void test_font1()
         public void test_font2()
         {
             using var font = new Font("helv");
-            const string text = "PyMuPDF";
+            const string text = "MuPDF.NET";
             Assert.Equal(font.TextLength(text), Utils.GetTextLength(text));
         }
 
@@ -72,7 +72,7 @@ public void test_font2()
         public void test_fontarchive()
         {
             var arch = new Archive();
-            string css = Utils.CssForPymupdfFont("notos", archive: arch, name: "sans-serif");
+            string css = Utils.CssForMupdfFont("notos", archive: arch, name: "sans-serif");
             Console.WriteLine(css);
             foreach (var entry in arch.EntryList)
                 Console.WriteLine($"fmt={entry.Fmt} entries={string.Join(", ", entry.Entries)} path={entry.Path}");