From f1ff027bd846a8ec4fe4f0a93d5f7a8ebabdf108 Mon Sep 17 00:00:00 2001
From: Will Sackfield <sackfield@replicate.com>
Date: Tue, 10 Dec 2024 09:31:05 -0500
Subject: [PATCH] Ignore files within a .git directory (#2087)

* Ignore files within a .git directory

* Do not consider files within a .git directory
when we are using separate-weights

* Check for the .git directory explicitly
---
 pkg/weights/weights.go      | 14 ++++++++++++++
 pkg/weights/weights_test.go | 16 ++++++++++++++++
 2 files changed, 30 insertions(+)

diff --git a/pkg/weights/weights.go b/pkg/weights/weights.go
index a3b97f14a7..8ca54c252c 100644
--- a/pkg/weights/weights.go
+++ b/pkg/weights/weights.go
@@ -30,6 +30,9 @@ func FindWeights(fw FileWalker) ([]string, []string, error) {
 		if info.IsDir() {
 			return nil
 		}
+		if isGitFile(path) {
+			return nil
+		}
 		if isCodeFile(path) {
 			codeFiles = append(codeFiles, path)
 			return nil
@@ -97,6 +100,17 @@ func isCodeFile(path string) bool {
 	return ext == ".py" || ext == ".ipynb"
 }
 
+func isGitFile(path string) bool {
+	dir, _ := filepath.Split(path)
+	folders := strings.Split(filepath.Clean(dir), string(filepath.Separator))
+	for _, folder := range folders {
+		if folder == ".git" {
+			return true
+		}
+	}
+	return false
+}
+
 // filterDirsContainingCode filters out directories that contain code files.
 // If a dir is a prefix for any given codeFiles, it will be filtered out.
 func filterDirsContainingCode(dirs []string, codeFiles []string) []string {
diff --git a/pkg/weights/weights_test.go b/pkg/weights/weights_test.go
index 515af6ffe3..e00e1b3ceb 100644
--- a/pkg/weights/weights_test.go
+++ b/pkg/weights/weights_test.go
@@ -160,3 +160,19 @@ func TestSubDirMerge(t *testing.T) {
 	require.Empty(t, rootFiles)
 	require.Equal(t, []string{"models"}, dirs)
 }
+
+// Test case for ignoring files within a .git directory
+func TestIgnoreGitFiles(t *testing.T) {
+	mockFileWalker := func(root string, walkFn filepath.WalkFunc) error {
+		sizes := []int64{sizeThreshold, sizeThreshold, 1024}
+		for i, path := range []string{".git/root-large", "root-large", "predict.py"} {
+			walkFn(path, mockFileInfo{size: sizes[i]}, nil)
+		}
+		return nil
+	}
+
+	dirs, rootFiles, err := FindWeights(mockFileWalker)
+	require.NoError(t, err)
+	require.Equal(t, []string{"root-large"}, rootFiles)
+	require.Empty(t, dirs)
+}