fix(pipeline): read diff path headers only before a file's first hunk

internal/pipeline/diff.go

@@ -34,6 +34,10 @@ func parseUnifiedDiff(raw string) (ParsedDiff, error) {
 	var current *FilePatch
 	var patch strings.Builder
 	patchLine := 0
+	// inHunks is set at a file's first hunk header. The ---/+++ path headers
+	// only come before it; after it, a removed "-- x" or added "++ x" line
+	// reads the same as a header and must stay content.
+	inHunks := false
 
 	flush := func() {
 		if current == nil {
@@ -63,6 +67,7 @@ func parseUnifiedDiff(raw string) (ParsedDiff, error) {
 			current = &FilePatch{OldPath: oldPath, Path: newPath}
 			patch.Reset()
 			patchLine = 0
+			inHunks = false
 		}
 		if current == nil {
 			continue
@@ -71,12 +76,12 @@ func parseUnifiedDiff(raw string) (ParsedDiff, error) {
 		patchLine++
 		trimmed := strings.TrimRight(line, "\r\n")
 		switch {
-		case strings.HasPrefix(trimmed, "--- "):
+		case !inHunks && strings.HasPrefix(trimmed, "--- "):
 			path := parseHeaderPath(strings.TrimPrefix(trimmed, "--- "))
 			if path != "" {
 				current.OldPath = path
 			}
-		case strings.HasPrefix(trimmed, "+++ "):
+		case !inHunks && strings.HasPrefix(trimmed, "+++ "):
 			path := parseHeaderPath(strings.TrimPrefix(trimmed, "+++ "))
 			if path != "" {
 				current.Path = path
@@ -99,6 +104,7 @@ func parseUnifiedDiff(raw string) (ParsedDiff, error) {
 				return ParsedDiff{}, err
 			}
 			current.Hunks = append(current.Hunks, hunk)
+			inHunks = true
 		}
 	}
 	flush()

internal/pipeline/diff_test.go

@@ -323,3 +323,56 @@ func TestParseUnifiedDiffMarksAddedFiles(t *testing.T) {
 		t.Fatalf("added file with content = %#v, want Added with one hunk", withContent)
 	}
 }
+
+func TestParseUnifiedDiffKeepsHeaderLikeHunkLinesAsContent(t *testing.T) {
+	raw := strings.Join([]string{
+		"diff --git a/db/one.sql b/db/one.sql",
+		"index 1111111..2222222 100644",
+		"--- a/db/one.sql",
+		"+++ b/db/one.sql",
+		"@@ -1,3 +1,3 @@",
+		" SELECT 1;",
+		"--- TODO",
+		"+++ counter",
+		" SELECT 2;",
+		"diff --git a/db/two.sql b/db/two.sql",
+		"index 3333333..4444444 100644",
+		"--- a/db/two.sql",
+		"+++ b/db/two.sql",
+		"@@ -1,2 +1,1 @@",
+		"--- TODO",
+		" SELECT 3;",
+		"@@ -9,1 +8,2 @@",
+		" SELECT 4;",
+		"+++ /dev/null",
+		"",
+	}, "\n")
+
+	got, err := parseUnifiedDiff(raw)
+	if err != nil {
+		t.Fatalf("parseUnifiedDiff: %v", err)
+	}
+	want := []struct {
+		path  string
+		hunks int
+	}{
+		{path: "db/one.sql", hunks: 1},
+		{path: "db/two.sql", hunks: 2},
+	}
+	if len(got.Patches) != len(want) {
+		t.Fatalf("patches = %d, want %d", len(got.Patches), len(want))
+	}
+	for i, w := range want {
+		patch := got.Patches[i]
+		if patch.OldPath != w.path || patch.Path != w.path || patch.Deleted || patch.Added {
+			t.Errorf("patch %d = old %q path %q deleted %v added %v, want old and path %q, neither deleted nor added",
+				i, patch.OldPath, patch.Path, patch.Deleted, patch.Added, w.path)
+		}
+		if len(patch.Hunks) != w.hunks {
+			t.Errorf("patch %d hunks = %d, want %d", i, len(patch.Hunks), w.hunks)
+		}
+		if !strings.Contains(patch.Patch, "--- TODO\n") {
+			t.Errorf("patch %d lost its removed comment line:\n%s", i, patch.Patch)
+		}
+	}
+}