Index: Lib/test/test_pep277.py =================================================================== --- Lib/test/test_pep277.py (revision 79204) +++ Lib/test/test_pep277.py (working copy) @@ -1,6 +1,7 @@ # Test the Unicode versions of normal file functions # open, os.open, os.stat. os.listdir, os.rename, os.remove, os.mkdir, os.chdir, os.rmdir import sys, os, unittest +from unicodedata import normalize from test import test_support filenames = [ @@ -13,8 +14,22 @@ u'\u05d4\u05e9\u05e7\u05e6\u05e5\u05e1', u'\u66e8\u66e9\u66eb', u'\u66e8\u05e9\u3093\u0434\u0393\xdf', + # Specific code points: fn, NFC(fn) and NFKC(fn) all differents + u'\u2000\u2000\u2000', + u'\u2001\u2001\u2001', + u'\u2002\u2002\u2002', # == NFC(u'\u2000\u2000\u2000') + u'\u2003\u2003\u2003', # == NFC(u'\u2001\u2001\u2001') + u' ', # u' ' == NFKC(u'\u2000') == NFKC(u'\u2001') + # == NFKC(u'\u2002') == NFKC(u'\u2003') + u'\u1fee\u1ffd', + # Specific code points: NFC(fn), NFD(fn), NFKC(fn) and NFKD(fn) all differents + u'\u0385\u03d3\u03d4', + u'\u00a8\u0301\u03d2\u0301\u03d2\u0308', # == NFD(u'\u0385\u03d3\u03d4') + u'\u0020\u0308\u0301\u038e\u03ab', # == NFKC(u'\u0385\u03d3\u03d4') + u'\u1e9b\u1fc1\u1fcd\u1fce\u1fcf\u1fdd\u1fde\u1fdf\u1fed', ] + # Destroy directory dirname and all files under it, to one level. def deltree(dirname): # Don't hide legitimate errors: if one of these suckers exists, it's @@ -25,43 +40,51 @@ os.unlink(os.path.join(dirname, fname)) os.rmdir(dirname) + class UnicodeFileTests(unittest.TestCase): - files = [os.path.join(test_support.TESTFN, f) for f in filenames] + files = set(filenames) + normal_form = None def setUp(self): try: os.mkdir(test_support.TESTFN) except OSError: pass + files = set() for name in self.files: + name = os.path.join(test_support.TESTFN, self.norm(name)) try: f = open(name, 'w') except UnicodeEncodeError: if not os.path.supports_unicode_filenames: - raise unittest.SkipTest("only NT+ and systems with Unicode" - "-friendly filesystem encoding") + self.skipTest("only NT+ and systems with Unicode-friendly" + "filesystem encoding") f.write((name+'\n').encode("utf-8")) f.close() os.stat(name) + files.add(name) + self.files = files def tearDown(self): deltree(test_support.TESTFN) + def norm(self, s): + if self.normal_form and isinstance(s, unicode): + return normalize(self.normal_form, s) + return s + def _apply_failure(self, fn, filename, expected_exception, check_fn_in_exception = True): - try: + with self.assertRaises(expected_exception) as c: fn(filename) - raise test_support.TestFailed("Expected to fail calling '%s(%r)'" - % (fn.__name__, filename)) - except expected_exception, details: - # the "filename" exception attribute may be encoded - if isinstance(details.filename, str): - filename = filename.encode(sys.getfilesystemencoding()) - if check_fn_in_exception and details.filename != filename: - raise test_support.TestFailed("Function '%s(%r) failed with " - "bad filename in the exception: %r" - % (fn.__name__, filename, - details.filename)) + exc_filename = c.exception.filename + # the "filename" exception attribute may be encoded + if isinstance(exc_filename, str): + filename = filename.encode(sys.getfilesystemencoding()) + if check_fn_in_exception: + self.assertEqual(exc_filename, filename, "Function '%s(%r) failed " + "with bad filename in the exception: %r" % + (fn.__name__, filename, exc_filename)) def test_failures(self): # Pass non-existing Unicode filenames all over the place. @@ -82,39 +105,92 @@ f.close() os.stat(name) + def test_normalize(self): + files = set(f for f in self.files if isinstance(f, unicode)) + others = set() + for nf in set(['NFC', 'NFD', 'NFKC', 'NFKD']): + others |= set(normalize(nf, file) for file in files) + others -= files + if sys.platform == 'darwin': + files = set(normalize('NFD', file) for file in files) + for name in others: + if sys.platform == 'darwin' and normalize('NFD', name) in files: + # Mac OS X decomposes Unicode names, using Normal Form D. + # http://developer.apple.com/mac/library/qa/qa2001/qa1173.html + os.stat(name) + continue + self._apply_failure(open, name, IOError) + self._apply_failure(os.stat, name, OSError) + self._apply_failure(os.chdir, name, OSError) + self._apply_failure(os.rmdir, name, OSError) + self._apply_failure(os.remove, name, OSError) + # listdir may append a wildcard to the filename, so dont check + self._apply_failure(os.listdir, name, OSError, False) + def test_listdir(self): + sf0 = set(self.files) f1 = os.listdir(test_support.TESTFN) f2 = os.listdir(unicode(test_support.TESTFN, sys.getfilesystemencoding())) - sf2 = set(os.path.join(unicode(test_support.TESTFN), f) - for f in f2) - self.assertEqual(len(f1), len(self.files)) - self.assertEqual(sf2, set(self.files)) + if sys.platform == 'darwin': + # Mac OS X returns canonically decomposed Unicode (Normal Form D) + # http://developer.apple.com/mac/library/qa/qa2001/qa1173.html + sf0 = set(normalize('NFD', unicode(f)) for f in self.files) + f2 = [normalize('NFD', unicode(f)) for f in f2] + sf2 = set(os.path.join(unicode(test_support.TESTFN), f) for f in f2) + self.assertEqual(sf0, sf2) + self.assertEqual(len(f1), len(f2)) def test_rename(self): for name in self.files: - os.rename(name,"tmp") - os.rename("tmp",name) + os.rename(name, "tmp") + os.rename("tmp", name) def test_directory(self): - dirname = os.path.join(test_support.TESTFN,u'Gr\xfc\xdf-\u66e8\u66e9\u66eb') + dirname = os.path.join(test_support.TESTFN, + u'Gr\xfc\xdf-\u66e8\u66e9\u66eb') filename = u'\xdf-\u66e8\u66e9\u66eb' oldwd = os.getcwd() os.mkdir(dirname) os.chdir(dirname) - f = open(filename, 'w') - f.write((filename + '\n').encode("utf-8")) - f.close() - os.access(filename,os.R_OK) - os.remove(filename) - os.chdir(oldwd) - os.rmdir(dirname) + try: + with open(filename, 'w') as f: + f.write((filename + '\n').encode("utf-8")) + os.access(filename,os.R_OK) + os.remove(filename) + finally: + os.chdir(oldwd) + os.rmdir(dirname) + + +class UnicodeNFCFileTests(UnicodeFileTests): + normal_form = 'NFC' + + +class UnicodeNFDFileTests(UnicodeFileTests): + normal_form = 'NFD' + + +class UnicodeNFKCFileTests(UnicodeFileTests): + normal_form = 'NFKC' + + +class UnicodeNFKDFileTests(UnicodeFileTests): + normal_form = 'NFKD' + def test_main(): try: - test_support.run_unittest(UnicodeFileTests) + test_support.run_unittest( + UnicodeFileTests, + UnicodeNFCFileTests, + UnicodeNFDFileTests, + UnicodeNFKCFileTests, + UnicodeNFKDFileTests, + ) finally: deltree(test_support.TESTFN) + if __name__ == "__main__": test_main()