From ec1003a33966cd4b3932d7dc9526f3e6d34c4bdc Mon Sep 17 00:00:00 2001
From: elsapet <elizabeth@bearer.sh>
Date: Thu, 17 Oct 2024 16:47:51 +0200
Subject: [PATCH] fix: filenames with spaces in diff scan

---
 pkg/git/diff.go      | 32 ++++++++++++++++++++++++++++----
 pkg/git/diff_test.go | 18 ++++++++++++++++++
 2 files changed, 46 insertions(+), 4 deletions(-)

diff --git a/pkg/git/diff.go b/pkg/git/diff.go
index 3fa0f5262..d0fb24749 100644
--- a/pkg/git/diff.go
+++ b/pkg/git/diff.go
@@ -28,6 +28,9 @@ type FilePatch struct {
 	Chunks Chunks
 }
 
+const SRC_PREFIX = "caf67b1d-c9b7-44ba-9b85-bd94727547af"
+const DST_PREFIX = "7dcfe536-96a9-49c5-8b86-7caa3d9e5e9c"
+
 func Diff(rootDir, baseRef string) ([]FilePatch, error) {
 	var result []FilePatch
 
@@ -40,7 +43,8 @@ func Diff(rootDir, baseRef string) ([]FilePatch, error) {
 			"--first-parent",
 			"--find-renames",
 			"--break-rewrites",
-			"--no-prefix",
+			"--src-prefix=" + SRC_PREFIX,
+			"--dst-prefix=" + DST_PREFIX,
 			"--no-color",
 			baseRef,
 			"--",
@@ -113,13 +117,13 @@ func parseDiff(scanner *linescanner.Scanner) ([]FilePatch, error) {
 }
 
 func parseDiffHeader(value string) (string, string, error) {
-	parts := strings.Split(value, " ")
-	fromPath, err := unquoteFilename(parts[2])
+	rawFromPath, rawToPath := splitPaths(value)
+	fromPath, err := unquoteFilename(rawFromPath)
 	if err != nil {
 		return "", "", fmt.Errorf("error parsing header 'from' path: %w", err)
 	}
 
-	toPath, err := unquoteFilename(parts[3])
+	toPath, err := unquoteFilename(rawToPath)
 	if err != nil {
 		return "", "", fmt.Errorf("error parsing header 'to' path: %w", err)
 	}
@@ -127,6 +131,26 @@ func parseDiffHeader(value string) (string, string, error) {
 	return fromPath, toPath, nil
 }
 
+func splitPaths(value string) (fromPath string, toPath string) {
+	split1 := strings.Split(value, SRC_PREFIX)[1]
+	split2 := strings.Split(split1, DST_PREFIX)
+
+	fromPath = strings.TrimSpace(split2[0])
+	toPath = strings.TrimSpace(split2[1])
+
+	// handle trailing whitespaces and missing quotation marks
+	// e.g. `foo\\t.txt \"`
+	if strings.Contains(fromPath, "\"") {
+		fromPath = "\"" + strings.TrimSpace(strings.ReplaceAll(fromPath, "\"", "")) + "\""
+	}
+
+	if strings.Contains(toPath, "\"") {
+		toPath = "\"" + strings.ReplaceAll(toPath, "\"", "") + "\""
+	}
+
+	return fromPath, toPath
+}
+
 func parseChunkHeader(value string) (Chunk, error) {
 	parts := strings.Split(value, " ")
 
diff --git a/pkg/git/diff_test.go b/pkg/git/diff_test.go
index 563e73a9e..41ea9baf7 100644
--- a/pkg/git/diff_test.go
+++ b/pkg/git/diff_test.go
@@ -105,6 +105,24 @@ var _ = Describe("Diff", func() {
 				{FromPath: fromPath, ToPath: toPath},
 			}))
 		})
+
+		fromPath = "from bar.txt"
+		toPath = "to foo.txt"
+
+		It("decodes the paths correctly with whitespace in both from and to path", func() {
+			Expect(git.Diff(tempDir, baseSHA)).To(ConsistOf([]git.FilePatch{
+				{FromPath: fromPath, ToPath: toPath},
+			}))
+		})
+
+		fromPath = "from bar.txt"
+		toPath = "to.txt"
+
+		It("decodes the paths correctly with whitespace in from path", func() {
+			Expect(git.Diff(tempDir, baseSHA)).To(ConsistOf([]git.FilePatch{
+				{FromPath: fromPath, ToPath: toPath},
+			}))
+		})
 	})
 
 	When("a file contains changes", func() {