Introduce blob FilteringOptions for content stream

Get the content of a blob as it would be checked out to the
working directory via the filters.

Author: Edward Thomson

LibGit2Sharp.Tests/BlobFixture.cs

@@ -1,4 +1,5 @@
-using System.IO;
+using System;
+using System.IO;
 using System.Linq;
 using System.Text;
 using LibGit2Sharp.Tests.TestHelpers;
@@ -22,6 +23,28 @@ public void CanGetBlobAsText()
             }
         }
 
+        [Fact]
+        public void CanGetBlobAsFilteredText()
+        {
+            using (var repo = new Repository(BareTestRepoPath))
+            {
+                var blob = repo.Lookup<Blob>("a8233120f6ad708f843d861ce2b7228ec4e3dec6");
+
+                var text = blob.ContentAsText(new FilteringOptions("foo.txt"));
+
+                ConfigurationEntry<bool> autocrlf = repo.Config.Get<bool>("core.autocrlf");
+
+                if (autocrlf != null && autocrlf.Value)
+                {
+                    Assert.Equal("hey there\r\n", text);
+                }
+                else
+                {
+                    Assert.Equal("hey there\n", text);
+                }
+            }
+        }
+
         [Theory]
         [InlineData("ascii", 4, "31 32 33 34")]
         [InlineData("utf-7", 4, "31 32 33 34")]
@@ -99,14 +122,59 @@ public void CanReadBlobStream()
             {
                 var blob = repo.Lookup<Blob>("a8233120f6ad708f843d861ce2b7228ec4e3dec6");
 
-                using (var tr = new StreamReader(blob.ContentStream, Encoding.UTF8))
+                using (var tr = new StreamReader(blob.ContentStream(), Encoding.UTF8))
                 {
                     string content = tr.ReadToEnd();
                     Assert.Equal("hey there\n", content);
                 }
             }
         }
 
+        [Fact]
+        public void CanReadBlobFilteredStream()
+        {
+            using (var repo = new Repository(BareTestRepoPath))
+            {
+                var blob = repo.Lookup<Blob>("a8233120f6ad708f843d861ce2b7228ec4e3dec6");
+
+                using (var tr = new StreamReader(blob.ContentStream(new FilteringOptions("foo.txt")), Encoding.UTF8))
+                {
+                    string content = tr.ReadToEnd();
+
+                    ConfigurationEntry<bool> autocrlf = repo.Config.Get<bool>("core.autocrlf");
+
+                    if (autocrlf != null && autocrlf.Value)
+                    {
+                        Assert.Equal("hey there\r\n", content);
+                    }
+                    else
+                    {
+                        Assert.Equal("hey there\n", content);
+                    }
+                }
+            }
+        }
+
+        [Fact]
+        public void CanReadBlobFilteredStreamOfUnmodifiedBinary()
+        {
+            var binaryContent = new byte[] { 0, 1, 2, 3, 4, 5 };
+
+            string path = CloneBareTestRepo();
+            using (var repo = new Repository(path))
+            {
+                using (var stream = new MemoryStream(binaryContent))
+                {
+                    Blob blob = repo.ObjectDatabase.CreateBlob(stream);
+
+                    using (var filtered = blob.ContentStream(new FilteringOptions("foo.txt")))
+                    {
+                        Assert.True(StreamEquals(stream, filtered));
+                    }
+                }
+            }
+        }
+
         public static void CopyStream(Stream input, Stream output)
         {
             // Reused from the following Stack Overflow post with permission
@@ -120,6 +188,19 @@ public static void CopyStream(Stream input, Stream output)
             }
         }
 
+        public static bool StreamEquals(Stream one, Stream two)
+        {
+            int onebyte, twobyte;
+
+            while ((onebyte = one.ReadByte()) >= 0 && (twobyte = two.ReadByte()) >= 0)
+            {
+                if (onebyte != twobyte)
+                    return false;
+            }
+
+            return true;
+        }
+
         [Fact]
         public void CanStageAFileGeneratedFromABlobContentStream()
         {
@@ -143,7 +224,7 @@ public void CanStageAFileGeneratedFromABlobContentStream()
 
                 var blob = repo.Lookup<Blob>(entry.Id.Sha);
 
-                using (Stream stream = blob.ContentStream)
+                using (Stream stream = blob.ContentStream())
                 using (Stream file = File.OpenWrite(Path.Combine(repo.Info.WorkingDirectory, "small.fromblob.txt")))
                 {
                     CopyStream(stream, file);

LibGit2Sharp/Blob.cs

@@ -49,12 +49,20 @@ public virtual byte[] Content
         /// <summary>
         /// Gets the blob content in a <see cref="Stream"/>.
         /// </summary>
-        public virtual Stream ContentStream
+        public virtual Stream ContentStream()
         {
-            get
-            {
-                return Proxy.git_blob_rawcontent_stream(repo.Handle, Id, Size);
-            }
+            return Proxy.git_blob_rawcontent_stream(repo.Handle, Id, Size);
+        }
+
+        /// <summary>
+        /// Gets the blob content in a <see cref="Stream"/> as it would be
+        /// checked out to the working directory.
+        /// <param name="filteringOptions">Parameter controlling content filtering behavior</param>
+        /// </summary>
+        public virtual Stream ContentStream(FilteringOptions filteringOptions)
+        {
+            Ensure.ArgumentNotNull(filteringOptions, "filteringOptions");
+            return Proxy.git_blob_filtered_content_stream(repo.Handle, Id, filteringOptions.HintPath, false);
         }
     }
 }

LibGit2Sharp/BlobExtensions.cs

@@ -22,7 +22,28 @@ public static string ContentAsText(this Blob blob, Encoding encoding = null)
         {
             Ensure.ArgumentNotNull(blob, "blob");
 
-            using (var reader = new StreamReader(blob.ContentStream, encoding ?? Encoding.UTF8, encoding == null))
+            using (var reader = new StreamReader(blob.ContentStream(), encoding ?? Encoding.UTF8, encoding == null))
+            {
+                return reader.ReadToEnd();
+            }
+        }
+
+        /// <summary>
+        /// Gets the blob content as it would be checked out to the
+        /// working directory, decoded with the specified encoding,
+        /// or according to byte order marks, with UTF8 as fallback,
+        /// if <paramref name="encoding"/> is null.
+        /// </summary>
+        /// <param name="blob">The blob for which the content will be returned.</param>
+        /// <param name="filteringOptions">Parameter controlling content filtering behavior</param>
+        /// <param name="encoding">The encoding of the text. (default: detected or UTF8)</param>
+        /// <returns>Blob content as text.</returns>
+        public static string ContentAsText(this Blob blob, FilteringOptions filteringOptions, Encoding encoding = null)
+        {
+            Ensure.ArgumentNotNull(blob, "blob");
+            Ensure.ArgumentNotNull(filteringOptions, "filteringOptions");
+
+            using(var reader = new StreamReader(blob.ContentStream(filteringOptions), encoding ?? Encoding.UTF8, encoding == null))
             {
                 return reader.ReadToEnd();
             }

LibGit2Sharp/Core/GitBuf.cs

@@ -0,0 +1,18 @@
+using System;
+using System.Runtime.InteropServices;
+
+namespace LibGit2Sharp.Core.Handles
+{
+    [StructLayout(LayoutKind.Sequential)]
+    internal class GitBuf : IDisposable
+    {
+        public IntPtr ptr;
+        public UIntPtr asize;
+        public UIntPtr size;
+
+        public void Dispose()
+        {
+            Proxy.git_buf_free(this);
+        }
+    }
+}

LibGit2Sharp/Core/NativeMethods.cs

@@ -131,6 +131,13 @@ internal static extern int git_blob_create_fromchunks(
             source_callback fileCallback,
             IntPtr data);
 
+        [DllImport(libgit2)]
+        internal static extern int git_blob_filtered_content(
+            GitBuf buf,
+            GitObjectSafeHandle blob,
+            [MarshalAs(UnmanagedType.CustomMarshaler, MarshalCookie = UniqueId.UniqueIdentifier, MarshalTypeRef = typeof(StrictFilePathMarshaler))] FilePath as_path,
+            [MarshalAs(UnmanagedType.Bool)] bool check_for_binary_data);
+
         [DllImport(libgit2)]
         internal static extern IntPtr git_blob_rawcontent(GitObjectSafeHandle blob);
 
@@ -182,6 +189,9 @@ internal static extern int git_branch_upstream_name(
             RepositorySafeHandle repo,
             [MarshalAs(UnmanagedType.CustomMarshaler, MarshalCookie = UniqueId.UniqueIdentifier, MarshalTypeRef = typeof(StrictUtf8Marshaler))] string referenceName);
 
+        [DllImport(libgit2)]
+        internal static extern void git_buf_free(GitBuf buf);
+
         [DllImport(libgit2)]
         internal static extern int git_checkout_tree(
             RepositorySafeHandle repo,

LibGit2Sharp/Core/Proxy.cs

@@ -73,6 +73,20 @@ public static ObjectId git_blob_create_fromfile(RepositorySafeHandle repo, FileP
             }
         }
 
+        public static UnmanagedMemoryStream git_blob_filtered_content_stream(RepositorySafeHandle repo, ObjectId id, FilePath path, bool check_for_binary_data)
+        {
+            var buf = new GitBuf();
+            var handle = new ObjectSafeWrapper(id, repo).ObjectPtr;
+
+            return new RawContentStream(handle, h =>
+            {
+                Ensure.ZeroResult(NativeMethods.git_blob_filtered_content(buf, h, path, check_for_binary_data));
+                return buf.ptr;
+            },
+            h => (long)buf.size,
+            new[] { buf });
+        }
+
         public static byte[] git_blob_rawcontent(RepositorySafeHandle repo, ObjectId id, int size)
         {
             using (var obj = new ObjectSafeWrapper(id, repo))
@@ -85,7 +99,8 @@ public static byte[] git_blob_rawcontent(RepositorySafeHandle repo, ObjectId id,
 
         public static UnmanagedMemoryStream git_blob_rawcontent_stream(RepositorySafeHandle repo, ObjectId id, Int64 size)
         {
-            return new RawContentStream(id, repo, NativeMethods.git_blob_rawcontent, size);
+            var handle = new ObjectSafeWrapper(id, repo).ObjectPtr;
+            return new RawContentStream(handle, NativeMethods.git_blob_rawcontent, h => size);
         }
 
         public static Int64 git_blob_rawsize(GitObjectSafeHandle obj)
@@ -188,6 +203,15 @@ public static string git_branch_upstream_name(RepositorySafeHandle handle, strin
 
         #endregion
 
+        #region git_buf_
+
+        public static void git_buf_free(GitBuf buf)
+        {
+            NativeMethods.git_buf_free(buf);
+        }
+
+        #endregion
+
         #region git_checkout_
 
         public static void git_checkout_tree(

LibGit2Sharp/Core/RawContentStream.cs

@@ -1,30 +1,68 @@
 using System;
+using System.Collections.Generic;
 using System.IO;
 using LibGit2Sharp.Core.Handles;
 
 namespace LibGit2Sharp.Core
 {
     internal class RawContentStream : UnmanagedMemoryStream
     {
-        private readonly ObjectSafeWrapper wrapper;
+        private readonly GitObjectSafeHandle handle;
+        private readonly ICollection<IDisposable> linkedResources;
 
-        internal RawContentStream(ObjectId id, RepositorySafeHandle repo,
-            Func<GitObjectSafeHandle, IntPtr> bytePtrProvider, long length)
-            : this(new ObjectSafeWrapper(id, repo), bytePtrProvider, length)
+        internal unsafe RawContentStream(
+            GitObjectSafeHandle handle,
+            Func<GitObjectSafeHandle, IntPtr> bytePtrProvider,
+            Func<GitObjectSafeHandle, long> sizeProvider,
+            ICollection<IDisposable> linkedResources = null)
+            : base((byte*)Wrap(handle, bytePtrProvider, linkedResources).ToPointer(),
+            Wrap(handle, sizeProvider, linkedResources))
         {
+            this.handle = handle;
+            this.linkedResources = linkedResources;
         }
 
-        unsafe RawContentStream(ObjectSafeWrapper wrapper,
-            Func<GitObjectSafeHandle, IntPtr> bytePtrProvider, long length)
-            : base((byte*)bytePtrProvider(wrapper.ObjectPtr).ToPointer(), length)
+        private static T Wrap<T>(
+            GitObjectSafeHandle handle,
+            Func<GitObjectSafeHandle, T> provider,
+            IEnumerable<IDisposable> linkedResources)
         {
-            this.wrapper = wrapper;
+            T value;
+
+            try
+            {
+                value = provider(handle);
+            }
+            catch
+            {
+                Dispose(handle, linkedResources);
+                throw;
+            }
+
+            return value;
+        }
+
+        private static void Dispose(
+            GitObjectSafeHandle handle,
+            IEnumerable<IDisposable> linkedResources)
+        {
+            handle.SafeDispose();
+
+            if (linkedResources == null)
+            {
+                return;
+            }
+
+            foreach (IDisposable linkedResource in linkedResources)
+            {
+                linkedResource.Dispose();
+            }
         }
 
         protected override void Dispose(bool disposing)
         {
             base.Dispose(disposing);
-            wrapper.SafeDispose();
+            Dispose(handle, linkedResources);
         }
-    }
+   }
 }

LibGit2Sharp/FilteringOptions.cs

@@ -0,0 +1,27 @@
+using LibGit2Sharp.Core;
+
+namespace LibGit2Sharp
+{
+    /// <summary>
+    /// Allows callers to specify how blob content filters will be applied.
+    /// </summary>
+    public sealed class FilteringOptions
+    {
+        /// <summary>
+        /// Initializes a new instance of the <see cref="FilteringOptions"/> class.
+        /// </summary>
+        /// <param name="hintPath">The path that a file would be checked out as</param>
+        public FilteringOptions(string hintPath)
+        {
+            Ensure.ArgumentNotNull(hintPath, "hintPath");
+
+            this.HintPath = hintPath;
+        }
+
+        /// <summary>
+        /// The path to "hint" to the filters will be used to apply
+        /// attributes.
+        /// </summary>
+        public string HintPath { get; private set;  }
+    }
+}

LibGit2Sharp/LibGit2Sharp.csproj

@@ -73,6 +73,8 @@
     <Compile Include="CompareOptions.cs" />
     <Compile Include="Core\EncodingMarshaler.cs" />
     <Compile Include="PushOptions.cs" />
+    <Compile Include="Core\GitBuf.cs" />
+    <Compile Include="FilteringOptions.cs" />
     <Compile Include="UnbornBranchException.cs" />
     <Compile Include="LockedFileException.cs" />
     <Compile Include="Core\GitRepositoryInitOptions.cs" />