linguist: Move individual file handling into separate file

As an extra step to prep for more features, this change moves the
determining the language of a single file into a separate file and
struct.

2 files changed, 43 insertions, 24 deletions

diff --git a/internal/gitaly/linguist/file_instance.go b/internal/gitaly/linguist/file_instance.go
new file mode 100644
index 000000000..5c01da23e
--- /dev/null
+++ b/internal/gitaly/linguist/file_instance.go
@@ -0,0 +1,42 @@
+package linguist
+
+import (
+	"fmt"
+	"io"
+
+	"github.com/go-enry/go-enry/v2"
+	"gitlab.com/gitlab-org/gitaly/v15/internal/git/gitpipe"
+)
+
+type fileInstance struct {
+	filename string
+}
+
+func newFileInstance(filename string) fileInstance {
+	return fileInstance{
+		filename: filename,
+	}
+}
+
+// DetermineStats determines the size and language of the given file. The
+// language will be an empty string when the stats should be omitted from the
+// count.
+func (f fileInstance) DetermineStats(object gitpipe.CatfileObjectResult) (string, uint64, error) {
+	// Read arbitrary number of bytes considered enough to determine language
+	content, err := io.ReadAll(io.LimitReader(object, 2048))
+	if err != nil {
+		return "", 0, fmt.Errorf("determine stats read blob: %w", err)
+	}
+
+	lang := enry.GetLanguage(f.filename, content)
+
+	// Ignore anything that's neither markup nor a programming language,
+	// similar to what the linguist gem does:
+	// https://github.com/github/linguist/blob/v7.20.0/lib/linguist/blob_helper.rb#L378-L387
+	if enry.GetLanguageType(lang) != enry.Programming &&
+		enry.GetLanguageType(lang) != enry.Markup {
+		return "", 0, nil
+	}
+
+	return lang, uint64(object.Object.ObjectSize()), nil
+}
diff --git a/internal/gitaly/linguist/linguist.go b/internal/gitaly/linguist/linguist.go
index 0eef5e4f4..c17cad698 100644
--- a/internal/gitaly/linguist/linguist.go
+++ b/internal/gitaly/linguist/linguist.go
@@ -162,7 +162,7 @@ func (inst *Instance) enryStats(ctx context.Context, commitID string) (ByteCount
 		object := objectIt.Result()
 		filename := string(object.ObjectName)
 
-		lang, size, err := inst.determineStats(filename, object)
+		lang, size, err := newFileInstance(filename).DetermineStats(object)
 		if err != nil {
 			return nil, fmt.Errorf("linguist determine stats: %w", err)
 		}
@@ -191,26 +191,3 @@ func (inst *Instance) enryStats(ctx context.Context, commitID string) (ByteCount
 
 	return stats.Totals, nil
 }
-
-// determineStats determines the size and language of the given file. The
-// language will be an empty string when the stats should be omitted from the
-// count.
-func (inst *Instance) determineStats(filename string, object gitpipe.CatfileObjectResult) (string, uint64, error) {
-	// Read arbitrary number of bytes considered enough to determine language
-	content, err := io.ReadAll(io.LimitReader(object, 2048))
-	if err != nil {
-		return "", 0, fmt.Errorf("determine stats read blob: %w", err)
-	}
-
-	lang := enry.GetLanguage(filename, content)
-
-	// Ignore anything that's neither markup nor a programming language,
-	// similar to what the linguist gem does:
-	// https://github.com/github/linguist/blob/v7.20.0/lib/linguist/blob_helper.rb#L378-L387
-	if enry.GetLanguageType(lang) != enry.Programming &&
-		enry.GetLanguageType(lang) != enry.Markup {
-		return "", 0, nil
-	}
-
-	return lang, uint64(object.Object.ObjectSize()), nil
-}