From 067afedb600485af20e2ec7b70bb41055499a365 Mon Sep 17 00:00:00 2001
From: Diogo Trindade <diogotr7@gmail.com>
Date: Thu, 17 Oct 2024 17:39:01 +0100
Subject: [PATCH] file processors initial work

---
 .../FileProcessing/CryXmlProcessor.cs         | 12 +++----
 .../FileProcessing/IFileProcessor.cs          | 31 +++++++++++++++++++
 .../FileProcessing/ZipFileProcessor.cs        |  6 ++--
 src/StarBreaker.P4k/P4kFile.cs                |  8 +++++
 src/StarBreaker.Sandbox/Program.cs            |  4 +--
 5 files changed, 50 insertions(+), 11 deletions(-)
 create mode 100644 src/StarBreaker.P4k/FileProcessing/IFileProcessor.cs

diff --git a/src/StarBreaker.P4k/FileProcessing/CryXmlProcessor.cs b/src/StarBreaker.P4k/FileProcessing/CryXmlProcessor.cs
index b80aaa5..1cf6b7a 100644
--- a/src/StarBreaker.P4k/FileProcessing/CryXmlProcessor.cs
+++ b/src/StarBreaker.P4k/FileProcessing/CryXmlProcessor.cs
@@ -6,17 +6,17 @@ public sealed class CryXmlProcessor : IFileProcessor
 {
     public bool CanProcess(string entryName, Stream stream)
     {
-        if (stream.Length < 4)
-            return false;
-
         Span<byte> test = stackalloc byte[4];
-        if (stream.Read(test) != 4)
+        var read = stream.Read(test);
+        stream.Seek(0, SeekOrigin.Current);
+
+        if (read != 4)
             return false;
-        
+
         return CryXml.IsCryXmlB(test);
     }
 
-    public void ProcessEntry(string outputRootFolder, string entryName, Stream stream)
+    public void ProcessEntry(string outputRootFolder, string entryName, Stream entryStream)
     {
         throw new NotImplementedException();
     }
diff --git a/src/StarBreaker.P4k/FileProcessing/IFileProcessor.cs b/src/StarBreaker.P4k/FileProcessing/IFileProcessor.cs
new file mode 100644
index 0000000..e10c2c2
--- /dev/null
+++ b/src/StarBreaker.P4k/FileProcessing/IFileProcessor.cs
@@ -0,0 +1,31 @@
+namespace StarBreaker.P4k;
+
+public interface IFileProcessor
+{
+    bool CanProcess(string entryName, Stream stream);
+    void ProcessEntry(string outputRootFolder, string entryName, Stream entryStream);
+}
+
+public static class FileProcessors
+{
+    private static readonly GenericFileProcessor _genericFileProcessor = new();
+    public static List<IFileProcessor> Processors { get; }
+    
+    static FileProcessors()
+    {
+        //reflection would be nicer but native aot doesn't support it
+        Processors =
+        [
+            new CryXmlProcessor(),
+            new ZipFileProcessor(),
+        ];
+    }
+    
+    public static IFileProcessor GetProcessor(string entryName, Stream entryStream)
+    {
+        if (Processors.FirstOrDefault(p => p.CanProcess(entryName, entryStream)) is { } processor)
+            return processor;
+        
+        return _genericFileProcessor;
+    }
+}
\ No newline at end of file
diff --git a/src/StarBreaker.P4k/FileProcessing/ZipFileProcessor.cs b/src/StarBreaker.P4k/FileProcessing/ZipFileProcessor.cs
index 9e07c34..f352aef 100644
--- a/src/StarBreaker.P4k/FileProcessing/ZipFileProcessor.cs
+++ b/src/StarBreaker.P4k/FileProcessing/ZipFileProcessor.cs
@@ -10,11 +10,11 @@ public bool CanProcess(string entryName, Stream stream)
         return entryName.EndsWith(".zip", StringComparison.OrdinalIgnoreCase);
     }
 
-    public void ProcessEntry(string outputRootFolder, string entryName, Stream stream)
+    public void ProcessEntry(string outputRootFolder, string entryName, Stream entryStream)
     {
         var entryPath = Path.Combine(outputRootFolder, Path.ChangeExtension(entryName, "unzipped"));
-        
-        using var archive = new ZipArchive(stream, ZipArchiveMode.Read);
+
+        using var archive = new ZipArchive(entryStream, ZipArchiveMode.Read, leaveOpen: true);
         foreach (var childEntry in archive.Entries)
         {
             if (childEntry.Length == 0)
diff --git a/src/StarBreaker.P4k/P4kFile.cs b/src/StarBreaker.P4k/P4kFile.cs
index 753f77b..1512bac 100644
--- a/src/StarBreaker.P4k/P4kFile.cs
+++ b/src/StarBreaker.P4k/P4kFile.cs
@@ -132,6 +132,14 @@ public void Extract(string outputDir, string? filter = null, IProgress<double>?
         var processedEntries = 0;
 
         progress?.Report(0);
+        
+        //TODO: Preprocessing step:
+        // 1. start with the list of total files
+        // 2. run the following according to the filter:
+        // 3. find one-shot single file procesors
+        // 4. find file -> multiple file processors
+        // 5. find multiple file -> single file unsplit processors - remove from the list so we don't double process
+        // run it!
 
         Parallel.ForEach(filteredEntries, entry =>
             {
diff --git a/src/StarBreaker.Sandbox/Program.cs b/src/StarBreaker.Sandbox/Program.cs
index 823a8c1..6dadee0 100644
--- a/src/StarBreaker.Sandbox/Program.cs
+++ b/src/StarBreaker.Sandbox/Program.cs
@@ -5,7 +5,7 @@
 //ExtractChunkFiles.Run();
 //ExtractSocPak.Run();
 //await GrpcClient.RunAsync();
-//TimeP4kExtract.Run();
+TimeP4kExtract.Run();
 //TimeZipNode.Run();
 
-StringCrc32c.Run();
+//StringCrc32c.Run();