From acf4a68a4f6cd79cf3adce0d51017bd3dd8c4d44 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marko=20Ga=C4=87e=C5=A1a?= <marko.gacesa@harness.io>
Date: Tue, 19 Sep 2023 17:25:07 +0200
Subject: [PATCH] use gogit for get blob implementation

---
 gitrpc/internal/gitea/blob.go | 87 +++++++++++++----------------------
 1 file changed, 32 insertions(+), 55 deletions(-)

diff --git a/gitrpc/internal/gitea/blob.go b/gitrpc/internal/gitea/blob.go
index 04becf7c8..5066ebe49 100644
--- a/gitrpc/internal/gitea/blob.go
+++ b/gitrpc/internal/gitea/blob.go
@@ -5,89 +5,66 @@
 package gitea
 
 import (
-	"bytes"
 	"context"
 	"fmt"
 	"io"
 
 	"github.com/harness/gitness/gitrpc/internal/types"
 
-	"code.gitea.io/gitea/modules/git"
+	gogitplumbing "github.com/go-git/go-git/v5/plumbing"
 )
 
 // GetBlob returns the blob for the given object sha.
 func (g Adapter) GetBlob(ctx context.Context, repoPath string, sha string, sizeLimit int64) (*types.BlobReader, error) {
-	// Note: We are avoiding gitea blob implementation, as that is tied to the lifetime of the repository object.
-	// Instead, we just use the gitea helper methods ourselves.
-	stdIn, stdOut, cancel := git.CatFileBatch(ctx, repoPath)
-
-	_, err := stdIn.Write([]byte(sha + "\n"))
+	repo, err := g.repoProvider.Get(ctx, repoPath)
 	if err != nil {
-		cancel()
-		return nil, fmt.Errorf("failed to write blob sha to git stdin: %w", err)
+		return nil, fmt.Errorf("failed to open repository: %w", err)
 	}
 
-	objectSHA, objectType, objectSize, err := git.ReadBatchLine(stdOut)
+	blob, err := repo.BlobObject(gogitplumbing.NewHash(sha))
 	if err != nil {
-		cancel()
-		return nil, processGiteaErrorf(err, "failed to read cat-file batch line")
-	}
-
-	if string(objectSHA) != sha {
-		cancel()
-		return nil, fmt.Errorf("cat-file returned object sha '%s' but expected '%s'", objectSHA, sha)
-	}
-	if objectType != string(git.ObjectBlob) {
-		cancel()
-		return nil, fmt.Errorf("cat-file returned object type '%s' but expected '%s'", objectType, git.ObjectBlob)
+		if err == gogitplumbing.ErrObjectNotFound {
+			return nil, types.ErrNotFound
+		}
+		return nil, fmt.Errorf("failed to get blob object: %w", err)
 	}
 
+	objectSize := blob.Size
 	contentSize := objectSize
-	if sizeLimit > 0 && sizeLimit < contentSize {
+	if sizeLimit > 0 && contentSize > sizeLimit {
 		contentSize = sizeLimit
 	}
 
+	reader, err := blob.Reader()
+	if err != nil {
+		return nil, fmt.Errorf("failed to open blob object: %w", err)
+	}
+
 	return &types.BlobReader{
 		SHA:         sha,
 		Size:        objectSize,
 		ContentSize: contentSize,
-		Content: &exactLimitReader{
-			reader:         stdOut,
-			remainingBytes: contentSize,
-			close: func() error {
-				// TODO: is there a better (but short) way to clear the buffer?
-				// gitea is .Discard()'ing elements here until it's empty.
-				stdOut.Reset(bytes.NewBuffer([]byte{}))
-				cancel()
-				return nil
-			},
-		},
+		Content:     LimitReadCloser(reader, contentSize),
 	}, nil
 }
 
-// exactLimitReader reads the content of a reader and ensures no more than the specified bytes will be requested from
-// the underlaying reader. This is required for readers that don't ensure completion after reading all remaining bytes.
-// io.LimitReader doesn't work as it waits for bytes that never come, io.SectionReader would requrie an io.ReaderAt.
-type exactLimitReader struct {
-	reader         io.Reader
-	remainingBytes int64
-	close          func() error
-}
-
-func (r *exactLimitReader) Read(p []byte) (int, error) {
-	if r.remainingBytes <= 0 {
-		return 0, io.EOF
+func LimitReadCloser(r io.ReadCloser, n int64) io.ReadCloser {
+	return limitReadCloser{
+		r: io.LimitReader(r, n),
+		c: r,
 	}
-
-	if int64(len(p)) > r.remainingBytes {
-		p = p[0:r.remainingBytes]
-	}
-	n, err := r.reader.Read(p)
-	r.remainingBytes -= int64(n)
-
-	return n, err
 }
 
-func (r *exactLimitReader) Close() error {
-	return r.close()
+// limitReadCloser implements io.ReadCloser interface.
+type limitReadCloser struct {
+	r io.Reader
+	c io.Closer
+}
+
+func (l limitReadCloser) Read(p []byte) (n int, err error) {
+	return l.r.Read(p)
+}
+
+func (l limitReadCloser) Close() error {
+	return l.c.Close()
 }