captures keep the combining marks of a decomposed name; test sources escape invisible characters

1 files changed, 7 insertions, 4 deletions

diff --git a/internal/norm/norm_test.go b/internal/norm/norm_test.go
index a165b32..dfae915 100644
--- a/internal/norm/norm_test.go
+++ b/internal/norm/norm_test.go
@@ -75,8 +75,9 @@ func TestFoldASCIINoAlloc(t *testing.T) {
 }
 
 // TestFoldMappedSource: a part of the folded text leads back to the original
-// characters it came from, widened to whole characters - so a capture can be
-// written with its diacritics.
+// characters it came from, widened to whole characters and to the combining
+// marks that follow them (a decomposed name, plan 11 review L5) - so a
+// capture can be written with its diacritics.
 func TestFoldMappedSource(t *testing.T) {
 	cases := []struct {
 		in, folded string
@@ -86,8 +87,10 @@ func TestFoldMappedSource(t *testing.T) {
 		{"Łódź-faktura.pdf", "Lodz-faktura.pdf", 0, 4, "Łódź"},
 		{"Łódź-faktura.pdf", "Lodz-faktura.pdf", 5, 12, "faktura"},
 		{"straße", "strasse", 4, 5, "ß"},
-		{"áb", "ab", 0, 1, "a"},
-		{"áb", "ab", 1, 2, "b"},
+		{"a\u0301b", "ab", 0, 1, "a\u0301"},
+		{"e\u0301-x.pdf", "e-x.pdf", 0, 1, "e\u0301"},
+		{"\u0141o\u0301dz\u0301-f.pdf", "Lodz-f.pdf", 0, 4, "\u0141o\u0301dz\u0301"},
+		{"a\u0301b", "ab", 1, 2, "b"},
 		{"\xff\xfeé", "�e", 0, 3, "\xff\xfe"},
 		{"plain.txt", "plain.txt", 0, 5, "plain"},
 		{"Łódź", "Lodz", 2, 2, ""},