def test_colon_handling_windows(self):
assert filesystem.sanitize_filename("test:aftertest") == "test-aftertest"
assert filesystem.sanitize_filename(":") == "-"
assert filesystem.sanitize_filename("test:") == "test-"
assert filesystem.sanitize_filename("test: ") == "test-"
# They should act the same
assert filesystem.sanitize_filename("test:aftertest") == filesystem.sanitize_foldername("test:aftertest")
def test_win_devices_not_win(self):
# Linux and Darwin are the same for this
assert filesystem.sanitize_filename(None) is None
assert filesystem.sanitize_filename("aux.txt") == "aux.txt"
assert filesystem.sanitize_filename("txt.aux") == "txt.aux"
assert filesystem.sanitize_filename("$mft") == "$mft"
assert filesystem.sanitize_filename("a$mft") == "a$mft"
def test_legal_chars_linux(self):
# Illegal on Windows but not on Linux, unless sanitize_safe is active.
# Don't bother with '/' which is illegal in filenames on all platforms.
char_ill = filesystem.CH_ILLEGAL_WIN.replace("/", "")
assert filesystem.sanitize_filename("test" + char_ill + "aftertest") == ("test" + char_ill + "aftertest")
for char in char_ill:
# Try at start, middle, and end of a filename.
assert filesystem.sanitize_filename("test" + char * 2 + "aftertest") == ("test" + char * 2 + "aftertest")
assert filesystem.sanitize_filename("test" + char * 2) == ("test" + char * 2).strip()
assert filesystem.sanitize_filename(char * 2 + "test") == (char * 2 + "test").strip()
def test_filename_too_long(self):
# Note: some filesystem can handle up to 255 UTF chars (which is more than 255 bytes) in the filename,
# but we stay on the safe side: max DEF_FILE_MAX bytes
# PART 1: Base cases: Nothing should happen:
# normal filename
name = "a" * 200 + ".ext"
sanitizedname = filesystem.sanitize_filename(name)
assert sanitizedname == name
# Unicode / UTF8 is OK ... as total filename length is not too long
name = "BASE" + "你" * 50 + "blabla.ext"
sanitizedname = filesystem.sanitize_filename(name)
assert sanitizedname == name
# filename with very long extension, but total filename is no problem, so no change
name = "hello.ext" + "e" * 200
sanitizedname = filesystem.sanitize_filename(name)
assert sanitizedname == name # no change
# PART 2: base truncating
name = "BASE" + "a" * 300 + ".mylongext"
sanitizedname = filesystem.sanitize_filename(name)
assert len(sanitizedname) <= DEF_FILE_MAX
assert sanitizedname.startswith("BASEaaaaaaaaaaaaaaa")
assert sanitizedname.endswith(".mylongext")
# too long filename, so truncate keeping the start of name and ext should stay the same
name = "BASE" + "a" * 200 + ".EXT" + "e" * 200
sanitizedname = filesystem.sanitize_filename(name)
assert len(sanitizedname) <= DEF_FILE_MAX
newname, newext = os.path.splitext(sanitizedname)
assert newname.startswith("BASEaaaaa")
assert newext.startswith(".EXTeeeee")
# PART 3: more exotic cases
# insert NON-ASCII chars, which should stay in place because overall length is no problem
name = "aaaa" + 10 * chr(188) + 10 * chr(222) + "bbbb.ext"
sanitizedname = filesystem.sanitize_filename(name)
assert sanitizedname == name
# insert NON-ASCII chars, which should get removed because overall length is too long
name = "aaaa" + 200 * chr(188) + 200 * chr(222) + "bbbb.ext"
sanitizedname = filesystem.sanitize_filename(name)
assert sanitizedname == "aaaabbbb.ext"
# Unicode / UTF8 ... total filename length might be too long for certain filesystems
name = "BASE" + "你" * 200 + ".ext"
sanitizedname = filesystem.sanitize_filename(name)
assert sanitizedname.startswith("BASE")
assert sanitizedname.endswith(".ext")
# Linux / POSIX: a hidden file (no extension), with size 200, so do not truncate at all
name = "." + "a" * 200
sanitizedname = filesystem.sanitize_filename(name)
assert sanitizedname == name # no change
def test_file_illegal_chars_linux(self):
assert filesystem.sanitize_filename("test/aftertest") == "test+aftertest"
assert filesystem.sanitize_filename("/test") == "+test"
assert filesystem.sanitize_filename("test/") == "test+"
assert filesystem.sanitize_filename(r"/test\/aftertest/") == r"+test\+aftertest+"
assert filesystem.sanitize_filename("/") == "+"
assert filesystem.sanitize_filename("///") == "+++"
assert filesystem.sanitize_filename("../") == "..+"
assert filesystem.sanitize_filename("../test") == "..+test"
def test_sanitize_safe_linux(self):
# Set sanitize_safe to on, simulating Windows-style restrictions.
assert filesystem.sanitize_filename("test" + filesystem.CH_ILLEGAL_WIN + "aftertest") == (
"test" + filesystem.CH_LEGAL_WIN + "aftertest"
)
for index in range(0, len(filesystem.CH_ILLEGAL_WIN)):
char_leg = filesystem.CH_LEGAL_WIN[index]
char_ill = filesystem.CH_ILLEGAL_WIN[index]
assert filesystem.sanitize_filename("test" + char_ill * 2 + "aftertest") == (
"test" + char_leg * 2 + "aftertest"
)
# Illegal chars that also get caught by strip() never make it far
# enough to be replaced by their legal equivalents if they appear
# on either end of the filename.
if char_ill.strip():
assert filesystem.sanitize_filename("test" + char_ill * 2) == ("test" + char_leg * 2)
assert filesystem.sanitize_filename(char_ill * 2 + "test") == (char_leg * 2 + "test")
def test_filename_dot(self):
# All dots should survive in filenames
assert filesystem.sanitize_filename(".test") == ".test"
assert filesystem.sanitize_filename("..test") == "..test"
assert filesystem.sanitize_filename("test.") == "test."
assert filesystem.sanitize_filename("test..") == "test.."
assert filesystem.sanitize_filename("test.aftertest") == "test.aftertest"
assert filesystem.sanitize_filename("test..aftertest") == "test..aftertest"
assert filesystem.sanitize_filename("test.aftertest.") == "test.aftertest."
assert filesystem.sanitize_filename("test.aftertest..") == "test.aftertest.."
def test_colon_handling_darwin(self):
assert filesystem.sanitize_filename("test:aftertest") == "aftertest"
assert filesystem.sanitize_filename(":aftertest") == "aftertest"
assert filesystem.sanitize_filename("::aftertest") == "aftertest"
assert filesystem.sanitize_filename(":after:test") == "test"
# Empty after sanitising with darwin colon handling
assert filesystem.sanitize_filename(":") == "unknown"
assert filesystem.sanitize_filename("test:") == "unknown"
assert filesystem.sanitize_filename("test: ") == "unknown"
def test_filename_empty_result(self):
# Nothing remains after sanitizing the filename
assert filesystem.sanitize_filename("\n") == "unknown"
assert filesystem.sanitize_filename("\r\n") == "unknown"
assert filesystem.sanitize_filename("\n\r") == "unknown"
assert filesystem.sanitize_filename("\t\t\t") == "unknown"
assert filesystem.sanitize_filename(" ") == "unknown"
assert filesystem.sanitize_filename(" ") == "unknown"
def test_win_devices_on_win(self):
assert filesystem.sanitize_filename(None) is None
assert filesystem.sanitize_filename("aux.txt") == "_aux.txt"
assert filesystem.sanitize_filename("txt.aux") == "txt.aux"
assert filesystem.sanitize_filename("$mft") == "Smft"
assert filesystem.sanitize_filename("a$mft") == "a$mft"
def test_colon_handling_other(self):
assert filesystem.sanitize_filename("test:aftertest") == "test:aftertest"
assert filesystem.sanitize_filename(":") == ":"
assert filesystem.sanitize_filename("test:") == "test:"
assert filesystem.sanitize_filename("test: ") == "test:"
def test_empty(self):
assert filesystem.sanitize_filename(None) is None
assert filesystem.sanitize_foldername(None) is None
