1# Test the Unicode versions of normal file functions 2# open, os.open, os.stat. os.listdir, os.rename, os.remove, os.mkdir, os.chdir, os.rmdir 3import os 4import sys 5import unittest 6import warnings 7from unicodedata import normalize 8from test.support import os_helper 9from test import support 10 11 12filenames = [ 13 '1_abc', 14 '2_ascii', 15 '3_Gr\xfc\xdf-Gott', 16 '4_\u0393\u03b5\u03b9\u03ac-\u03c3\u03b1\u03c2', 17 '5_\u0417\u0434\u0440\u0430\u0432\u0441\u0442\u0432\u0443\u0439\u0442\u0435', 18 '6_\u306b\u307d\u3093', 19 '7_\u05d4\u05e9\u05e7\u05e6\u05e5\u05e1', 20 '8_\u66e8\u66e9\u66eb', 21 '9_\u66e8\u05e9\u3093\u0434\u0393\xdf', 22 # Specific code points: fn, NFC(fn) and NFKC(fn) all different 23 '10_\u1fee\u1ffd', 24 ] 25 26# Mac OS X decomposes Unicode names, using Normal Form D. 27# http://developer.apple.com/mac/library/qa/qa2001/qa1173.html 28# "However, most volume formats do not follow the exact specification for 29# these normal forms. For example, HFS Plus uses a variant of Normal Form D 30# in which U+2000 through U+2FFF, U+F900 through U+FAFF, and U+2F800 through 31# U+2FAFF are not decomposed." 32if sys.platform != 'darwin': 33 filenames.extend([ 34 # Specific code points: NFC(fn), NFD(fn), NFKC(fn) and NFKD(fn) all different 35 '11_\u0385\u03d3\u03d4', 36 '12_\u00a8\u0301\u03d2\u0301\u03d2\u0308', # == NFD('\u0385\u03d3\u03d4') 37 '13_\u0020\u0308\u0301\u038e\u03ab', # == NFKC('\u0385\u03d3\u03d4') 38 '14_\u1e9b\u1fc1\u1fcd\u1fce\u1fcf\u1fdd\u1fde\u1fdf\u1fed', 39 40 # Specific code points: fn, NFC(fn) and NFKC(fn) all different 41 '15_\u1fee\u1ffd\ufad1', 42 '16_\u2000\u2000\u2000A', 43 '17_\u2001\u2001\u2001A', 44 '18_\u2003\u2003\u2003A', # == NFC('\u2001\u2001\u2001A') 45 '19_\u0020\u0020\u0020A', # '\u0020' == ' ' == NFKC('\u2000') == 46 # NFKC('\u2001') == NFKC('\u2003') 47 ]) 48 49 50# Is it Unicode-friendly? 51if not os.path.supports_unicode_filenames: 52 fsencoding = sys.getfilesystemencoding() 53 try: 54 for name in filenames: 55 name.encode(fsencoding) 56 except UnicodeEncodeError: 57 raise unittest.SkipTest("only NT+ and systems with " 58 "Unicode-friendly filesystem encoding") 59 60 61class UnicodeFileTests(unittest.TestCase): 62 files = set(filenames) 63 normal_form = None 64 65 def setUp(self): 66 try: 67 os.mkdir(os_helper.TESTFN) 68 except FileExistsError: 69 pass 70 self.addCleanup(os_helper.rmtree, os_helper.TESTFN) 71 72 files = set() 73 for name in self.files: 74 name = os.path.join(os_helper.TESTFN, self.norm(name)) 75 with open(name, 'wb') as f: 76 f.write((name+'\n').encode("utf-8")) 77 os.stat(name) 78 files.add(name) 79 self.files = files 80 81 def norm(self, s): 82 if self.normal_form: 83 return normalize(self.normal_form, s) 84 return s 85 86 def _apply_failure(self, fn, filename, 87 expected_exception=FileNotFoundError, 88 check_filename=True): 89 with self.assertRaises(expected_exception) as c: 90 fn(filename) 91 exc_filename = c.exception.filename 92 if check_filename: 93 self.assertEqual(exc_filename, filename, "Function '%s(%a) failed " 94 "with bad filename in the exception: %a" % 95 (fn.__name__, filename, exc_filename)) 96 97 def test_failures(self): 98 # Pass non-existing Unicode filenames all over the place. 99 for name in self.files: 100 name = "not_" + name 101 self._apply_failure(open, name) 102 self._apply_failure(os.stat, name) 103 self._apply_failure(os.chdir, name) 104 self._apply_failure(os.rmdir, name) 105 self._apply_failure(os.remove, name) 106 self._apply_failure(os.listdir, name) 107 108 if sys.platform == 'win32': 109 # Windows is lunatic. Issue #13366. 110 _listdir_failure = NotADirectoryError, FileNotFoundError 111 else: 112 _listdir_failure = NotADirectoryError 113 114 def test_open(self): 115 for name in self.files: 116 f = open(name, 'wb') 117 f.write((name+'\n').encode("utf-8")) 118 f.close() 119 os.stat(name) 120 self._apply_failure(os.listdir, name, self._listdir_failure) 121 122 # Skip the test on darwin, because darwin does normalize the filename to 123 # NFD (a variant of Unicode NFD form). Normalize the filename to NFC, NFKC, 124 # NFKD in Python is useless, because darwin will normalize it later and so 125 # open(), os.stat(), etc. don't raise any exception. 126 @unittest.skipIf(sys.platform == 'darwin', 'irrelevant test on Mac OS X') 127 @unittest.skipIf( 128 support.is_emscripten or support.is_wasi, 129 "test fails on Emscripten/WASI when host platform is macOS." 130 ) 131 def test_normalize(self): 132 files = set(self.files) 133 others = set() 134 for nf in set(['NFC', 'NFD', 'NFKC', 'NFKD']): 135 others |= set(normalize(nf, file) for file in files) 136 others -= files 137 for name in others: 138 self._apply_failure(open, name) 139 self._apply_failure(os.stat, name) 140 self._apply_failure(os.chdir, name) 141 self._apply_failure(os.rmdir, name) 142 self._apply_failure(os.remove, name) 143 self._apply_failure(os.listdir, name) 144 145 # Skip the test on darwin, because darwin uses a normalization different 146 # than Python NFD normalization: filenames are different even if we use 147 # Python NFD normalization. 148 @unittest.skipIf(sys.platform == 'darwin', 'irrelevant test on Mac OS X') 149 def test_listdir(self): 150 sf0 = set(self.files) 151 with warnings.catch_warnings(): 152 warnings.simplefilter("ignore", DeprecationWarning) 153 f1 = os.listdir(os_helper.TESTFN.encode( 154 sys.getfilesystemencoding())) 155 f2 = os.listdir(os_helper.TESTFN) 156 sf2 = set(os.path.join(os_helper.TESTFN, f) for f in f2) 157 self.assertEqual(sf0, sf2, "%a != %a" % (sf0, sf2)) 158 self.assertEqual(len(f1), len(f2)) 159 160 def test_rename(self): 161 for name in self.files: 162 os.rename(name, "tmp") 163 os.rename("tmp", name) 164 165 def test_directory(self): 166 dirname = os.path.join(os_helper.TESTFN, 167 'Gr\xfc\xdf-\u66e8\u66e9\u66eb') 168 filename = '\xdf-\u66e8\u66e9\u66eb' 169 with os_helper.temp_cwd(dirname): 170 with open(filename, 'wb') as f: 171 f.write((filename + '\n').encode("utf-8")) 172 os.access(filename,os.R_OK) 173 os.remove(filename) 174 175 176class UnicodeNFCFileTests(UnicodeFileTests): 177 normal_form = 'NFC' 178 179 180class UnicodeNFDFileTests(UnicodeFileTests): 181 normal_form = 'NFD' 182 183 184class UnicodeNFKCFileTests(UnicodeFileTests): 185 normal_form = 'NFKC' 186 187 188class UnicodeNFKDFileTests(UnicodeFileTests): 189 normal_form = 'NFKD' 190 191 192if __name__ == "__main__": 193 unittest.main() 194