Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 1 | # Tests universal newline support for both reading and parsing files. |
Benjamin Peterson | 4fa88fa | 2009-03-04 00:14:51 +0000 | [diff] [blame] | 2 | import io |
| 3 | import _pyio as pyio |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 4 | import unittest |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 5 | import os |
| 6 | import sys |
Benjamin Peterson | ee8712c | 2008-05-20 21:35:26 +0000 | [diff] [blame] | 7 | from test import support |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 8 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 9 | if not hasattr(sys.stdin, 'newlines'): |
Benjamin Peterson | e549ead | 2009-03-28 21:42:05 +0000 | [diff] [blame] | 10 | raise unittest.SkipTest( |
| 11 | "This Python does not have universal newline support") |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 12 | |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 13 | FATX = 'x' * (2**14) |
| 14 | |
| 15 | DATA_TEMPLATE = [ |
| 16 | "line1=1", |
Guido van Rossum | 8358db2 | 2007-08-18 21:39:55 +0000 | [diff] [blame] | 17 | "line2='this is a very long line designed to go past any default " + |
| 18 | "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] | 19 | "the uncommon case, naturally.'", |
| 20 | "def line3():pass", |
| 21 | "line4 = '%s'" % FATX, |
| 22 | ] |
| 23 | |
| 24 | DATA_LF = "\n".join(DATA_TEMPLATE) + "\n" |
| 25 | DATA_CR = "\r".join(DATA_TEMPLATE) + "\r" |
| 26 | DATA_CRLF = "\r\n".join(DATA_TEMPLATE) + "\r\n" |
| 27 | |
| 28 | # Note that DATA_MIXED also tests the ability to recognize a lone \r |
| 29 | # before end-of-file. |
| 30 | DATA_MIXED = "\n".join(DATA_TEMPLATE) + "\r" |
| 31 | DATA_SPLIT = [x + "\n" for x in DATA_TEMPLATE] |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 32 | |
Ezio Melotti | 28b0d9d | 2013-06-09 00:07:06 +0300 | [diff] [blame] | 33 | class CTest: |
| 34 | open = io.open |
| 35 | |
| 36 | class PyTest: |
| 37 | open = staticmethod(pyio.open) |
| 38 | |
| 39 | class TestGenericUnivNewlines: |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 40 | # use a class variable DATA to define the data to write to the file |
| 41 | # and a class variable NEWLINE to set the expected newlines value |
Guido van Rossum | 8358db2 | 2007-08-18 21:39:55 +0000 | [diff] [blame] | 42 | READMODE = 'r' |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 43 | WRITEMODE = 'wb' |
| 44 | |
| 45 | def setUp(self): |
Guido van Rossum | 9c62772 | 2007-08-27 18:31:48 +0000 | [diff] [blame] | 46 | data = self.DATA |
| 47 | if "b" in self.WRITEMODE: |
| 48 | data = data.encode("ascii") |
Philip Jenvey | a27c5bd | 2009-05-28 06:09:08 +0000 | [diff] [blame] | 49 | with self.open(support.TESTFN, self.WRITEMODE) as fp: |
| 50 | fp.write(data) |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 51 | |
| 52 | def tearDown(self): |
| 53 | try: |
Benjamin Peterson | ee8712c | 2008-05-20 21:35:26 +0000 | [diff] [blame] | 54 | os.unlink(support.TESTFN) |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 55 | except: |
| 56 | pass |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 57 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 58 | def test_read(self): |
Philip Jenvey | a27c5bd | 2009-05-28 06:09:08 +0000 | [diff] [blame] | 59 | with self.open(support.TESTFN, self.READMODE) as fp: |
| 60 | data = fp.read() |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 61 | self.assertEqual(data, DATA_LF) |
Walter Dörwald | 70a6b49 | 2004-02-12 17:35:32 +0000 | [diff] [blame] | 62 | self.assertEqual(repr(fp.newlines), repr(self.NEWLINE)) |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 63 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 64 | def test_readlines(self): |
Philip Jenvey | a27c5bd | 2009-05-28 06:09:08 +0000 | [diff] [blame] | 65 | with self.open(support.TESTFN, self.READMODE) as fp: |
| 66 | data = fp.readlines() |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 67 | self.assertEqual(data, DATA_SPLIT) |
Walter Dörwald | 70a6b49 | 2004-02-12 17:35:32 +0000 | [diff] [blame] | 68 | self.assertEqual(repr(fp.newlines), repr(self.NEWLINE)) |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 69 | |
| 70 | def test_readline(self): |
Philip Jenvey | a27c5bd | 2009-05-28 06:09:08 +0000 | [diff] [blame] | 71 | with self.open(support.TESTFN, self.READMODE) as fp: |
| 72 | data = [] |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 73 | d = fp.readline() |
Philip Jenvey | a27c5bd | 2009-05-28 06:09:08 +0000 | [diff] [blame] | 74 | while d: |
| 75 | data.append(d) |
| 76 | d = fp.readline() |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 77 | self.assertEqual(data, DATA_SPLIT) |
Walter Dörwald | 70a6b49 | 2004-02-12 17:35:32 +0000 | [diff] [blame] | 78 | self.assertEqual(repr(fp.newlines), repr(self.NEWLINE)) |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 79 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 80 | def test_seek(self): |
Philip Jenvey | a27c5bd | 2009-05-28 06:09:08 +0000 | [diff] [blame] | 81 | with self.open(support.TESTFN, self.READMODE) as fp: |
| 82 | fp.readline() |
| 83 | pos = fp.tell() |
| 84 | data = fp.readlines() |
| 85 | self.assertEqual(data, DATA_SPLIT[1:]) |
| 86 | fp.seek(pos) |
| 87 | data = fp.readlines() |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 88 | self.assertEqual(data, DATA_SPLIT[1:]) |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 89 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 90 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 91 | class TestCRNewlines(TestGenericUnivNewlines): |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 92 | NEWLINE = '\r' |
| 93 | DATA = DATA_CR |
Ezio Melotti | 28b0d9d | 2013-06-09 00:07:06 +0300 | [diff] [blame] | 94 | class CTestCRNewlines(CTest, TestCRNewlines, unittest.TestCase): pass |
| 95 | class PyTestCRNewlines(PyTest, TestCRNewlines, unittest.TestCase): pass |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 96 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 97 | class TestLFNewlines(TestGenericUnivNewlines): |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 98 | NEWLINE = '\n' |
| 99 | DATA = DATA_LF |
Ezio Melotti | 28b0d9d | 2013-06-09 00:07:06 +0300 | [diff] [blame] | 100 | class CTestLFNewlines(CTest, TestLFNewlines, unittest.TestCase): pass |
| 101 | class PyTestLFNewlines(PyTest, TestLFNewlines, unittest.TestCase): pass |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 102 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 103 | class TestCRLFNewlines(TestGenericUnivNewlines): |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 104 | NEWLINE = '\r\n' |
| 105 | DATA = DATA_CRLF |
Tim Peters | 863ac44 | 2002-04-16 01:38:40 +0000 | [diff] [blame] | 106 | |
Guido van Rossum | 8ce8a78 | 2007-11-01 19:42:39 +0000 | [diff] [blame] | 107 | def test_tell(self): |
Philip Jenvey | a27c5bd | 2009-05-28 06:09:08 +0000 | [diff] [blame] | 108 | with self.open(support.TESTFN, self.READMODE) as fp: |
| 109 | self.assertEqual(repr(fp.newlines), repr(None)) |
| 110 | data = fp.readline() |
| 111 | pos = fp.tell() |
Guido van Rossum | 8ce8a78 | 2007-11-01 19:42:39 +0000 | [diff] [blame] | 112 | self.assertEqual(repr(fp.newlines), repr(self.NEWLINE)) |
Ezio Melotti | 28b0d9d | 2013-06-09 00:07:06 +0300 | [diff] [blame] | 113 | class CTestCRLFNewlines(CTest, TestCRLFNewlines, unittest.TestCase): pass |
| 114 | class PyTestCRLFNewlines(PyTest, TestCRLFNewlines, unittest.TestCase): pass |
Guido van Rossum | 8ce8a78 | 2007-11-01 19:42:39 +0000 | [diff] [blame] | 115 | |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 116 | class TestMixedNewlines(TestGenericUnivNewlines): |
Tim Peters | ea572b2 | 2002-04-21 06:12:02 +0000 | [diff] [blame] | 117 | NEWLINE = ('\r', '\n') |
| 118 | DATA = DATA_MIXED |
Ezio Melotti | 28b0d9d | 2013-06-09 00:07:06 +0300 | [diff] [blame] | 119 | class CTestMixedNewlines(CTest, TestMixedNewlines, unittest.TestCase): pass |
| 120 | class PyTestMixedNewlines(PyTest, TestMixedNewlines, unittest.TestCase): pass |
Jack Jansen | 518d261 | 2002-04-14 20:17:18 +0000 | [diff] [blame] | 121 | |
| 122 | if __name__ == '__main__': |
Ezio Melotti | 28b0d9d | 2013-06-09 00:07:06 +0300 | [diff] [blame] | 123 | unittest.main() |