Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 1 | # Tests universal newline support for both reading and parsing files. |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 2 | |
Antoine Pitrou | 47a5f48 | 2009-06-12 20:41:52 +0000 | [diff] [blame] | 3 | # NOTE: this file tests the new `io` library backported from Python 3.x. |
| 4 | # Similar tests for the builtin file object can be found in test_univnewlines2k.py. |
| 5 | |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 6 | from __future__ import print_function |
| 7 | from __future__ import unicode_literals |
| 8 | |
| 9 | import io |
| 10 | import _pyio as pyio |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 11 | import unittest |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 12 | import os |
| 13 | import sys |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 14 | from test import test_support as support |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 15 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 16 | if not hasattr(sys.stdin, 'newlines'): |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 17 | raise unittest.SkipTest( |
| 18 | "This Python does not have universal newline support") |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 19 | |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 20 | FATX = 'x' * (2**14) |
| 21 | |
| 22 | DATA_TEMPLATE = [ |
| 23 | "line1=1", |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 24 | "line2='this is a very long line designed to go past any default " + |
| 25 | "buffer limits that exist in io.py but we also want to test " + |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 26 | "the uncommon case, naturally.'", |
| 27 | "def line3():pass", |
| 28 | "line4 = '%s'" % FATX, |
| 29 | ] |
| 30 | |
| 31 | DATA_LF = "\n".join(DATA_TEMPLATE) + "\n" |
| 32 | DATA_CR = "\r".join(DATA_TEMPLATE) + "\r" |
| 33 | DATA_CRLF = "\r\n".join(DATA_TEMPLATE) + "\r\n" |
| 34 | |
| 35 | # Note that DATA_MIXED also tests the ability to recognize a lone \r |
| 36 | # before end-of-file. |
| 37 | DATA_MIXED = "\n".join(DATA_TEMPLATE) + "\r" |
| 38 | DATA_SPLIT = [x + "\n" for x in DATA_TEMPLATE] |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 39 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 40 | class TestGenericUnivNewlines(unittest.TestCase): |
| 41 | # use a class variable DATA to define the data to write to the file |
| 42 | # and a class variable NEWLINE to set the expected newlines value |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 43 | READMODE = 'r' |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 44 | WRITEMODE = 'wb' |
| 45 | |
| 46 | def setUp(self): |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 47 | data = self.DATA |
| 48 | if "b" in self.WRITEMODE: |
| 49 | data = data.encode("ascii") |
| 50 | with self.open(support.TESTFN, self.WRITEMODE) as fp: |
| 51 | fp.write(data) |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 52 | |
| 53 | def tearDown(self): |
| 54 | try: |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 55 | os.unlink(support.TESTFN) |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 56 | except: |
| 57 | pass |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 58 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 59 | def test_read(self): |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 60 | with self.open(support.TESTFN, self.READMODE) as fp: |
Philip Jenvey | 6a11102 | 2009-05-28 05:58:44 +0000 | [diff] [blame] | 61 | data = fp.read() |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 62 | self.assertEqual(data, DATA_LF) |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 63 | self.assertEqual(set(fp.newlines), set(self.NEWLINE)) |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 64 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 65 | def test_readlines(self): |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 66 | with self.open(support.TESTFN, self.READMODE) as fp: |
Philip Jenvey | 6a11102 | 2009-05-28 05:58:44 +0000 | [diff] [blame] | 67 | data = fp.readlines() |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 68 | self.assertEqual(data, DATA_SPLIT) |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 69 | self.assertEqual(set(fp.newlines), set(self.NEWLINE)) |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 70 | |
| 71 | def test_readline(self): |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 72 | with self.open(support.TESTFN, self.READMODE) as fp: |
Philip Jenvey | 6a11102 | 2009-05-28 05:58:44 +0000 | [diff] [blame] | 73 | data = [] |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 74 | d = fp.readline() |
Philip Jenvey | 6a11102 | 2009-05-28 05:58:44 +0000 | [diff] [blame] | 75 | while d: |
| 76 | data.append(d) |
| 77 | d = fp.readline() |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 78 | self.assertEqual(data, DATA_SPLIT) |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 79 | self.assertEqual(set(fp.newlines), set(self.NEWLINE)) |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 80 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 81 | def test_seek(self): |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 82 | with self.open(support.TESTFN, self.READMODE) as fp: |
Philip Jenvey | 6a11102 | 2009-05-28 05:58:44 +0000 | [diff] [blame] | 83 | fp.readline() |
| 84 | pos = fp.tell() |
| 85 | data = fp.readlines() |
| 86 | self.assertEqual(data, DATA_SPLIT[1:]) |
| 87 | fp.seek(pos) |
| 88 | data = fp.readlines() |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 89 | self.assertEqual(data, DATA_SPLIT[1:]) |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 90 | |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 91 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 92 | class TestCRNewlines(TestGenericUnivNewlines): |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 93 | NEWLINE = '\r' |
| 94 | DATA = DATA_CR |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 95 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 96 | class TestLFNewlines(TestGenericUnivNewlines): |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 97 | NEWLINE = '\n' |
| 98 | DATA = DATA_LF |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 99 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 100 | class TestCRLFNewlines(TestGenericUnivNewlines): |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 101 | NEWLINE = '\r\n' |
| 102 | DATA = DATA_CRLF |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 103 | |
Guido van Rossum | ad8fb0d | 2007-09-22 20:18:03 +0000 | [diff] [blame] | 104 | def test_tell(self): |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 105 | with self.open(support.TESTFN, self.READMODE) as fp: |
Philip Jenvey | 6a11102 | 2009-05-28 05:58:44 +0000 | [diff] [blame] | 106 | self.assertEqual(repr(fp.newlines), repr(None)) |
| 107 | data = fp.readline() |
| 108 | pos = fp.tell() |
Guido van Rossum | ad8fb0d | 2007-09-22 20:18:03 +0000 | [diff] [blame] | 109 | self.assertEqual(repr(fp.newlines), repr(self.NEWLINE)) |
| 110 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 111 | class TestMixedNewlines(TestGenericUnivNewlines): |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 112 | NEWLINE = ('\r', '\n') |
| 113 | DATA = DATA_MIXED |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 114 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 115 | |
| 116 | def test_main(): |
Antoine Pitrou | 1969059 | 2009-06-12 20:14:08 +0000 | [diff] [blame] | 117 | base_tests = (TestCRNewlines, |
| 118 | TestLFNewlines, |
| 119 | TestCRLFNewlines, |
| 120 | TestMixedNewlines) |
| 121 | tests = [] |
| 122 | # Test the C and Python implementations. |
| 123 | for test in base_tests: |
| 124 | class CTest(test): |
| 125 | open = io.open |
| 126 | CTest.__name__ = str("C" + test.__name__) |
| 127 | class PyTest(test): |
| 128 | open = staticmethod(pyio.open) |
| 129 | PyTest.__name__ = str("Py" + test.__name__) |
| 130 | tests.append(CTest) |
| 131 | tests.append(PyTest) |
| 132 | support.run_unittest(*tests) |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 133 | |
| 134 | if __name__ == '__main__': |
| 135 | test_main() |