Fixed utf8 flag not being set correctly

Author: Pamparampam

pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "zipFly64"
-version = "1.3.1"
+version = "1.3.2"
 description = "Stream zip64 archives on the fly."
 readme = "README.md"
 authors = [{ name = "Pamparampampam" }]

src/zipFly/BaseFile.py

@@ -4,6 +4,7 @@
 
 from . import consts
 from .Compressor import Compressor
+from .consts import DATA_DESCRIPTOR_FLAG
 
 
 class BaseFile(ABC):
@@ -13,7 +14,7 @@ def __init__(self, name: str, compression_method: int = consts.NO_COMPRESSION):
         self.__offset = 0  # Offset to local file header
         self.__crc = 0
         self.__compression_method = compression_method
-        self.__flags = 0b00001000  # flag about using data descriptor is always on
+        self.__flags = DATA_DESCRIPTOR_FLAG  # flag about using data descriptor is always on
         self.__byte_offset_mode = False
         if name == "":
             raise KeyError("File name cannot be blank.")
@@ -102,8 +103,9 @@ def file_path_bytes(self) -> bytes:
         try:
             return self.name.encode("ascii")
         except UnicodeError:
+            print("UNICOE ERROR")
             self.__flags |= consts.UTF8_FLAG
-            return self.name.encode("utf-8")
+            return self.name.encode()
 
     @abstractmethod
     def _generate_file_data(self) -> Generator[bytes, None, None]:
@@ -127,6 +129,7 @@ def modification_time(self) -> float:
 
     @property
     def flags(self) -> int:
+        _ = self.file_path_bytes  # trigger to set utf8 flag if needed
         return self.__flags
 
     @property

src/zipFly/consts.py

@@ -4,7 +4,8 @@
 # ZIP CONSTANTS
 ZIP64_VERSION = 45
 VERSION_MADE_BY = 0x0A45  # Windows and ZIP version 45
-UTF8_FLAG = 0x800  # utf-8 filename encoding flag
+UTF8_FLAG = 0x0800  # utf-8 filename encoding flag
+DATA_DESCRIPTOR_FLAG = 0x08  # flag that signalises that a data descriptor is used
 
 # ZIP COMPRESSION METHODS
 NO_COMPRESSION = 0

tests/test_zipfly.py

@@ -1,5 +1,6 @@
 """Run tests of ZipFly."""
 import re
+import struct
 import time
 import zipfile
 import zlib
@@ -654,3 +655,58 @@ def test_EmptyFolder_creates_empty_directory(tmp_path):
         with zfp.open(folder_name) as folder_file:
             content = folder_file.read()
             assert content == b""
+
+def _read_local_flags(data, offset):
+    return struct.unpack_from("<H", data, offset + 6)[0]
+
+
+def _read_central_flags(data, offset):
+    return struct.unpack_from("<H", data, offset + 8)[0]
+
+
+def test_non_ascii_file_names_utf8_flag(tmp_path):
+    file_path = tmp_path / "input.bin"
+    file_path.write_bytes(lorem_ipsum)
+
+    files = [
+        LocalFile(file_path=file_path, name="zażółć.txt"),
+        LocalFile(file_path=file_path, name="файл.txt"),
+        LocalFile(file_path=file_path, name="αρχείο.txt"),
+        LocalFile(file_path=file_path, name="ファイル.txt"),
+        LocalFile(file_path=file_path, name="文件.txt"),
+    ]
+
+    zip_path = tmp_path / "non_ascii_flags.zip"
+    zipfly = ZipFly(files)
+
+    with open(zip_path, "wb") as f:
+        for chunk in zipfly.stream():
+            f.write(chunk)
+
+    data = zip_path.read_bytes()
+
+    UTF8_FLAG = 0x0800
+
+    # --- scan local headers ---
+    offset = 0
+    while True:
+        sig = data.find(b"PK\x03\x04", offset)
+        if sig == -1:
+            break
+
+        flags = _read_local_flags(data, sig)
+        assert flags & UTF8_FLAG, f"UTF-8 flag missing in local header at {sig}"
+
+        offset = sig + 4
+
+    # --- scan central directory ---
+    offset = 0
+    while True:
+        sig = data.find(b"PK\x01\x02", offset)
+        if sig == -1:
+            break
+
+        flags = _read_central_flags(data, sig)
+        assert flags & UTF8_FLAG, f"UTF-8 flag missing in central dir at {sig}"
+
+        offset = sig + 4