From eb7aa1d01061b79257e823ac1e0ac714ae8d7ab0 Mon Sep 17 00:00:00 2001
From: Erica Marigold <hi@devcomp.xyz>
Date: Wed, 8 Jan 2025 14:40:51 +0000
Subject: [PATCH] chore(tests): add test for utf8 encoded filenames

---
 tests/data/utf8_filenames.zip | Bin 0 -> 595 bytes
 tests/edge_cases.luau         |  13 +++++++++++++
 tests/extract.luau            |   2 +-
 3 files changed, 14 insertions(+), 1 deletion(-)
 create mode 100644 tests/data/utf8_filenames.zip

diff --git a/tests/data/utf8_filenames.zip b/tests/data/utf8_filenames.zip
new file mode 100644
index 0000000000000000000000000000000000000000..18ba8e6458637cf7f7af0856068bc2b208424451
GIT binary patch
literal 595
zcmWIWW@h1HVBlb2XerT%irdkAOO%0u0fhM(WEj#ib5i4<>|OA5>0Z5(ijvR}P6p=i
z-nw)UF0J5ZU}Sm0%)kI9LNYRo6@VyFp(M4qL;<8zk1GJD=@Ky0A2&{Z+%)-d<Lbwa
z3m-SGr<FBtrgpq;o%(j_j5jSi-praprZtRAa?H3wO9B{X3=E7w{J#;z!U#K7NZ6r;
sTYxuiLnRoH4UK~tiZ#e6HT@{ebgUtT%XC&YkXM<3P!Q;I9gy!C0AeJjHvj+t

literal 0
HcmV?d00001

diff --git a/tests/edge_cases.luau b/tests/edge_cases.luau
index 66a4d1b..681d1bf 100644
--- a/tests/edge_cases.luau
+++ b/tests/edge_cases.luau
@@ -44,5 +44,18 @@ return function(test: typeof(frktest.test))
 			-- the entire binary data in case there's a mismatch
 			check.equal(serde.hash("blake3", bin), serde.hash("blake3", expectedBin.stdout))
 		end)
+
+		test.case("Extracts files with UTF8 names", function()
+			local data = fs.readFile("tests/data/utf8_filenames.zip")
+			local zip = ZipReader.load(buffer.fromstring(data))
+
+			local entries = {}
+			for _, entry in zip:listDirectory("/") do
+				table.insert(entries, entry.name)
+			end
+
+			check.equal(#entries, 3)
+			check.table.equal(entries, { "file_こんにちは.txt", "file_你好.txt", "file_안녕하세요.txt" })
+		end)
 	end)
 end
diff --git a/tests/extract.luau b/tests/extract.luau
index b6fa02a..49f4850 100644
--- a/tests/extract.luau
+++ b/tests/extract.luau
@@ -14,7 +14,7 @@ local FALLIBLES = {
 	-- "invalid_cde_number_of_files_allocation_smaller_offset.zip",
 	"invalid_offset.zip",
 	"invalid_offset2.zip",
-	"chinese.zip", -- TODO: Support UTF8 data in file names
+	"chinese.zip", -- Contains non UTF8 data which can't parse without OS APIs
 	"non_utf8.zip", -- FIXME: Lune breaks for non utf8 data in process stdout
 	"pandoc_soft_links.zip", -- Soft links are tested separately in edge_cases
 }